Whole

DIRECT TORQUE CONTROLLED INDUCTION MACHINES FOR INTEGRATED

STARTER/ALTERNATOR SYSTEM

Jun Zhang

A thesis submitted for the degree of Doctor of Philosophy

School of Electrical Engineering and Telecommunications

The University of New South Wales

August 2006

ii

CERTIFICATE OF ORIGINALITY

I hereby declare that this submission is my own work and to the best of my knowledge

it contains no materials previously published or written by another person, or substantial

proportions of material which have been accepted for the award of any other degree or

diploma at UNSW or any other educational institution, except where due

acknowledgement is made in the thesis. Any contribution made to the research by

others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in

the thesis. I also declare that the intellectual content of this thesis is the product of my

own work, except to the extent that assistance from others in the project's design and

conception or in style, presentation and linguistic expression is acknowledged.

Signed …………………………….

Jun Zhang

iii

Dedicated to the memory of my grandmother

iv

ACKNOWLEDGMENTS

I would like to express my sincere acknowledgments to my supervisor, Professor M.

Fazlur Rahman, for his guidance and support during my PhD study. I would also like to

sincerely thank Professor Yuwen Hu for his kind help and encouragement during my

study.

I thank all my colleagues of the Energy Systems Research Group in the School of

Electrical Engineering and Telecommunications at University of New South Wales.

Special thanks are given to Dr. Lixin Tang and Dr. Zhuang Xu for their valuable

suggestions and help for my research.

I would like to express my deepest appreciation to my wife, my parents, my parents in

law and my younger brother for their love, patience and support.

v

ABSTRACT

An integrated starter/alternator (ISA) has been proposed for the future 42 V PowerNet,

which combines both starter and alternator functions into a single electrical machine

with bidirectional power flow ability. This thesis presents analysis, design, modeling

and experimental results of the direct torque controlled ISA system based on a low

voltage induction machine.

The classical direct torque controlled ISA based on switching-table is systematically for

an ISA evaluated in this thesis. The simulation and experimental results show that the

direct torque control (DTC) concept can be successfully extended to the ISA

application.

An improved DTC of the ISA based on direct stator flux vector is presented to reduce

the drawbacks of high torque and flux ripples of the classical DTC. Robust design of the

controller ensures the system is not sensitive to the variation of rotor resistance. By

controlling the electromagnetic torque of the induction machine quickly, the required dc

bus voltage can be well regulated within the 42 V PowerNet specifications. Another

improved DTC of the ISA with direct torque and flux control is also studied. Compared

to the direct flux vector control scheme, the calculation of the commanded voltage

vector in this scheme only requires the derivative of the stator flux magnitude, which is

a dc quantity. In addition, both torque and flux are regulated directly with two

independent closed-loops. This scheme is relatively insensitive to the noise.

The thesis proposed compensation methods to reduce the effects of switch voltage drops

and dead-time on the estimation of the stator flux. Experimental results confirm that the

estimation error is reduced with compensation for both motoring and generating modes

of the ISA.

A closed-loop type of sliding mode flux observer is proposed to reduce the estimation

error of the stator flux. Both Simulation and experimental results confirm that the

proposed sliding mode observer is insensitive to the stator resistance variation and

sensor offsets.

vi

A loss minimized scheme with power factor control for the ISA is proposed in this

thesis. It provides a simple solution for the efficiency improvement of the induction

machine without requiring any speed or load information.

The effectiveness of the direct torque controlled induction machine for an integrated

starter/alternator system has thus been confirmed and well supported by the studies

presented in this thesis.

vii

CONTENTS

CERTIFICATE OF ORIGINALITY...................................................................................................... II

ACKNOWLEDGMENTS .......................................................................................................................IV

ABSTRACT ........................................................................................................................................ V

CONTENTS ..................................................................................................................................... VII

LIST OF FIGURES .................................................................................................................................XI

LIST OF TABLES ................................................................................................................................XVI

LIST OF SYMBOLS........................................................................................................................... XVII

CHAPTER 1 INTRODUCTION.......................................................................................................... 1

1.1 42-VOLT POWERNET........................................................................................................ 1 1.2 INTEGRATED STARTER ALTERNATOR - ISA...................................................................... 5

1.2.1 Electrical specification .................................................................................................... 7 1.2.2 Machine technologies ...................................................................................................... 8 1.2.3 Electrical System configuration and Power converter topology.................................... 13 1.2.4 Machine controller- control of generator ...................................................................... 17

1.3 SCOPE OF THE THESIS...................................................................................................... 20 1.4 OUTLINE OF THE THESIS.................................................................................................. 21

CHAPTER 2 AN INDUCTION MACHINE BASED INTEGRATED

STARTER/ALTERNATOR USING ROTOR FIELD ORIENTED CONTROL

WITH SPACE VECTOR MODULATION............................................................... 22

2.1 INTRODUCTION ............................................................................................................... 22 2.2 INDUCTION MACHINE MODEL.......................................................................................... 22 2.3 ROTOR FLUX ORIENTED CONTROLLED ISA ..................................................................... 24 2.4 EXPERIMENTAL SETUP .................................................................................................... 27 2.5 EXPERIMENTAL RESULTS ................................................................................................ 28

2.5.1 Starting mode ................................................................................................................. 28 2.5.2 Generating mode - steady state...................................................................................... 31 2.5.3 Generating mode - dynamic response. ........................................................................... 32 2.5.4 High speed operation ..................................................................................................... 39

2.6 CONCLUSION .................................................................................................................. 39

viii

CHAPTER 3 CLASSICAL DIRECT TORQUE CONTROLLED INTEGRATED

STARTER/ALTERNATOR........................................................................................ 41

3.1 INTRODUCTION ............................................................................................................... 41 3.2 CLASSICAL DIRECT TORQUE CONTROL PRINCIPLE ........................................................... 42 3.3 ISA WITH CLASSICAL DTC ............................................................................................. 45 3.4 SIMULATION RESULTS..................................................................................................... 46

3.4.1 Starting mode ................................................................................................................. 47 3.4.2 Generating mode- steady state....................................................................................... 48 3.4.3 Generating mode - dynamic response............................................................................ 50

3.5 EXPERIMENTAL RESULTS ................................................................................................ 53 3.5.1 DTC-ST with constant switching frequency ................................................................... 53 3.5.2 Generating mode- steady state....................................................................................... 54

3.6 CONCLUSION .................................................................................................................. 56

CHAPTER 4 DIRECT FLUX VECTOR CONTROLLED INTEGRATED

STARTER/ALTERNATOR WITH SPACE VECTOR MODULATION .............. 57

4.1 INTRODUCTION ............................................................................................................... 57 4.2 DIRECT FLUX VECTOR CONTROL ..................................................................................... 58

4.2.1 Direct flux vector control scheme .................................................................................. 62 4.2.2 Design of the PI controller for torque regulation .......................................................... 64 4.2.3 Design of the PI controller with control delay............................................................... 66 4.2.4 Modeling results............................................................................................................. 71 4.2.5 Experimental results ...................................................................................................... 80

4.3 DIRECT FLUX VECTOR CONTROLLED INDUCTION GENERATOR FOR AN ISA..................... 85 4.3.1 Induction generator with DFC....................................................................................... 85 4.3.2 Experimental results ...................................................................................................... 88

4.4 CONCLUSION .................................................................................................................. 94

CHAPTER 5 DIRECT TORQUE AND FLUX CONTROLLED INTEGRATED

STARTER/ALTERNATOR WITH SPACE VECTOR MODULATION .............. 96

5.1 INTRODUCTION ............................................................................................................... 96 5.2 DIRECT TORQUE AND FLUX CONTROL PRINCIPLE ............................................................ 97 5.3 DIRECT TORQUE AND FLUX CONTROLLED INDUCTION GENERATOR FOR AN ISA........... 101 5.4 EXPERIMENTAL RESULTS .............................................................................................. 102

5.4.1 Starting mode ............................................................................................................... 102 5.4.2 Generating mode - steady state.................................................................................... 104 5.4.3 Generating mode - dynamic response. ......................................................................... 105 5.4.4 Performance High speed operation ............................................................................. 108

5.5 CONCLUSION ................................................................................................................ 109

ix

CHAPTER 6 NON-LINEAR BEHAVIOUR OF THE DC-AC CONVERTER AND ITS

COMPENSATION..................................................................................................... 110

6.1 INTRODUCTION ............................................................................................................. 110 6.2 EFFECT OF DEAD-TIME ................................................................................................. 111 6.3 EFFECT OF VOLTAGE DROP ON THE POWER DEVICE....................................................... 115 6.4 COMPENSATION ALGORITHM ........................................................................................ 117

6.4.1 Backward compensation .............................................................................................. 117 6.4.2 Forward compensation ................................................................................................ 118

6.5 EXPERIMENTAL RESULTS .............................................................................................. 119 6.5.1 Motoring mode............................................................................................................. 119 6.5.2 Generating mode.......................................................................................................... 130

6.6 CONCLUSION ................................................................................................................ 134

CHAPTER 7 AN IMPROVED STATOR FLUX ESTIMATION OF DIRECT TORQUE

CONTROLLED INTEGRATED STARTER/ALTERNATOR WITH SLIDING

MODE OBSERVER .................................................................................................. 136

7.1 INTRODUCTION ............................................................................................................. 136 7.2 DYNAMIC MODEL OF INDUCTION MACHINES ............................................................... 137 7.3 SLIDING MODE STATOR FLUX OBSERVER....................................................................... 139 7.4 SIMULATION RESULTS .................................................................................................. 142 7.5 EXPERIMENTAL RESULTS ............................................................................................. 146

7.5.1 Stator flux and torque estimation in motoring mode.................................................... 146 7.5.2 Stator flux and torque estimation in generating mode ................................................. 153

7.6 CONCLUSION ................................................................................................................ 158

CHAPTER 8 EFFICIENCY IMPROVEMENT FOR INTEGRATED

STARTER/ALTERNATOR WITH POWER FACTOR CONTROL................... 159

8.1 INTRODUCTION ....................................................................................................... 159 8.2 INDUCTION MACHINE LOSS MODEL ............................................................................. 160 8.3 PRINCIPLE OF POWER FACTOR CONTROL...................................................................... 161 8.4 MODELING RESULTS..................................................................................................... 163 8.5 EXPERIMENTAL RESULTS .............................................................................................. 166

8.5.1 Motoring mode............................................................................................................. 167 8.5.2 Generating mode.......................................................................................................... 168

8.6 CONCLUSION ................................................................................................................ 169

CHAPTER 9 CONCLUSIONS ........................................................................................................ 170

9.1 SUGGESTIONS FOR FUTURE WORK................................................................................. 175 9.1.1 Machine ....................................................................................................................... 175 9.1.2 Power converter........................................................................................................... 176

x

9.1.3 Direct torque controlled ISA based on permanent magnet synchronous machine....... 176

REFERENCES ..................................................................................................................................... 177

APPENDIX A LIST OF PUBLICATIONS....................................................................................... 187

APPENDIX B MODELLING OF THE DIRECT FLUX VECTOR CONTROL ......................... 189

APPENDIX C MODELLING OF THE DIRECT TORQUE AND FLUX CONTROL ............... 197

xi

LIST OF FIGURES

FIG. 1.1 ELECTRICAL AND ELECTRICS COMPONENTS IN AUTOMOBILES [2, 3] ............................................... 1 FIG. 1.2 MORE EXTENSIVE ELECTRONICS IN MODERN VEHICLES [4] .............................................................. 2 FIG. 1.3 GENERATOR PEAK POWER DEMAND OF AVERAGE PASSENGER VEHICLE [9] ..................................... 3 FIG. 1.4 VOLTAGE REGULATION OF 42 V ELECTRICAL SYSTEM [13] ............................................................. 3 FIG. 1.5 CONVENTIONAL 14V DC DISTRIBUTION SYSTEM ARCHITECTURE [1] .............................................. 4 FIG. 1.6 ADVANCED MULTIPLEXED AUTOMOTIVE POWER SYSTEM ARCHITECTURES OF THE FUTURE WITH

POWER AND COMMUNICATION BUSES [1] ....................................................................................... 4 FIG. 1.7 CRANKSHAFT MOUNTED STARTER ALTERNATOR [34]...................................................................... 5 FIG. 1.8 STARTING WITH ISA AND DC MOTOR [8] ........................................................................................ 6 FIG. 1.9 STARTER/ALTERNATOR STARTING AND APPROXIMATE GENERATING TORQUE REQUIREMENT (*)

AND THE TORQUE/SPEED CHARACTERISTIC (LINE) [15]. ................................................................. 7 FIG. 1.10 DC BUS VOLTAGE DYNAMIC REQUIREMENT [6] ............................................................................. 8 FIG. 1.11 ROTOR STRUCTURE OF IPM MOTORS ........................................................................................... 10 FIG. 1.12 COST COMPARISON OF THREE MACHINE SYSTEMS FOR A 6KW DIRECT-DRIVE

STARTER/ALTERNATOR APPLICATION [15] ................................................................................... 12 FIG. 1.13 HIGH VOLTAGE BUS CONFIGURATION .......................................................................................... 14 FIG. 1.14 HIGH VOLTAGE BUS CONFIGURATION WITH ULTRACAPACITOR.................................................... 14 FIG. 1.15 BLOCK DIAGRAM OF THE OVERALL SUPERVISORY CONTROL SCHEME [60] .................................. 15 FIG. 1.16 DUAL VOLTAGE (14V AND 42V) AUTOMOTIVE ELECTRICAL SYSTEM [59]................................... 16 FIG. 1.17 PROPOSED ISA ELECTRICAL SYSTEM CONFIGURATION ................................................................ 16 FIG. 1.18 DTC OF INDUCTION MOTOR ......................................................................................................... 18 FIG. 1.19 DTC OF INDUCTION GENERATOR ................................................................................................. 19

FIG. 2.1 DYNAMIC e ed q− EQUIVALENT CIRCUITS OF MACHINE (A)

eq AXIS CIRCUIT, (B) ed AXIS

CIRCUIT........................................................................................................................................ 24 FIG. 2.2 ROTOR FLUX ORIENTED CONTROLLED ISA WITH SVM.................................................................. 25 FIG. 2.3 FLUX MODEL IN THE ROTOR-FLUX-ORIENTED REFERENCE FRAME [75]........................................... 26 FIG. 2.4 VECTOR DIAGRAM OF THE INDUCTION MACHINE ........................................................................... 26 FIG. 2.5 EXPERIMENTAL SETUP ................................................................................................................... 28

FIG. 2.6 STARTING OF ISA WITH RFOC: (A) TORQUE, SPEED AND STATOR FLUX (B) TORQUE e

di AND e

qi 30

FIG. 2.7 ISA GENERATING WITH FULL LOAD ............................................................................................... 31 FIG. 2.8 SPECTRUM ANALYSIS OF THE STATOR CURRENT OF ISA WHILE OPERATING AS GENERATOR IN THE

STEADY-STATE............................................................................................................................. 32 FIG. 2.9 LOAD DUMP OF ISA WITHOUT BATTERY CONNECTED: (A) BUS VOLTAGE, TORQUE, STATOR FLUX

AND STATOR CURRENT (B) TORQUE, edi AND

eqi ........................................................................ 34

xii

FIG. 2.10 LOAD DUMP OF ISA WITH BATTERY CONNECTED: (A) BUS VOLTAGE, TORQUE, STATOR FLUX AND

STATOR CURRENT (B) TORQUE, edi AND

eqi ................................................................................ 35

FIG. 2.11 ISA PERFORMANCE AT ACCELERATION: (A) BUS VOLTAGE, SPEED, STATOR FLUX AND STATOR

CURRENT (B) SPEED, edi AND

eqi ................................................................................................ 37

FIG. 2.12 ISA PERFORMANCE AT DECELERATION: (A) BUS VOLTAGE, SPEED, STATOR FLUX AND STATOR

CURRENT (B) SPEED, edi AND

eqi ................................................................................................ 38

FIG. 2.13 ISA WITH FIELD WEAKENING AT HIGH SPEED............................................................................... 39 FIG. 3.1 EIGHT SWITCHING STATES AND THE VOLTAGE SPACE VECTORS ..................................................... 43 FIG. 3.2 MOVEMENT OF STATOR FLUX VECTOR BY SELECTION DIFFERENT VOLTAGE SPACE VECTORS ........ 43 FIG. 3.3 STRUCTURE OF CLASSICAL DIRECT TORQUE CONTROL................................................................... 44 FIG. 3.4 STATOR AND ROTOR FLUX VECTOR AT MOTORING AND GENERATION STATES ............................... 45 FIG. 3.5 CLASSIC DTC SCHEME FOR ISA .................................................................................................... 46 FIG. 3.6 STARTING PROCESS OF ISA (A) TS =150 sμ (B) TS =50 sμ .......................................................... 48

FIG. 3.7 ISA GENERATING WITH FULL LOAD (A) TS =150 sμ (B) TS =50 sμ .............................................. 49

FIG. 3.8 SPECTRUM ANALYSIS OF THE STATOR CURRENT WITH FFT (A) TS =150 sμ (B) TS =50 sμ .......... 50

FIG. 3.9 LOAD DUMPING PERFORMANCE OF ISA (A) TS =150 sμ (B) TS =50 sμ ........................................ 51

FIG. 3.10 ISA PERFORMANCE AT SPEED RAMP (TS =50 sμ ) (A) ACCELERATING (B) DECELERATION .......... 53

FIG. 3.11 ANALOG (A) AND DISCRETE (B) HYSTERESIS COMPARATOR [64].................................................. 54 FIG. 3.12 ISA GENERATING WITH DTC-ST ................................................................................................. 55 FIG. 3.13 STATOR FLUX VECTOR DIAGRAM ................................................................................................. 55 FIG. 4.1 EQUIVALENT SYSTEM MODEL OF THE TORQUE LOOP...................................................................... 62 FIG.4.2 PI CONTROL OF THE EQUIVALENT SYSTEM...................................................................................... 62 FIG.4.3 DIRECT FLUX VECTOR CONTROL SCHEME FOR INDUCTION MACHINE .............................................. 64 FIG.4.4 PI CONTROL OF EQUIVALENT SYSTEM WITH PRE-FILTER................................................................. 66 FIG.4.5 PI CONTROL OF EQUIVALENT TORQUE LOOP ................................................................................... 67 FIG.4.6 PI CONTROL OF EQUIVALENT SYSTEM WITH PRE-FILTER................................................................. 71 FIG.4.7 TORQUE DYNAMIC PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH ROTOR RESISTANCE

VARIATION OF 50% AND 100% .................................................................................................... 72 FIG.4.8 PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH SPEED LOOP ........................................... 73 FIG.4.9 TORQUE DYNAMIC PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH AND WITHOUT PRE-

FILTER.......................................................................................................................................... 75 FIG.4.10 TORQUE DYNAMIC PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH ROTOR RESISTANCE

VARIATION OF 50% AND 100% (PRE-FILTER ADDED) ................................................................... 75 FIG.4.11 PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH SPEED LOOP – NO PRE-FILTER ADDED.. 76 FIG.4.12 PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH SPEED LOOP –PRE-FILTER ADDED......... 77 FIG.4.13 TORQUE DYNAMIC RESPONSE OF RFOC ....................................................................................... 78 FIG.4.14 ROTOR FLUX ORIENTED CONTROL SCHEME WITH SVM ................................................................ 78

xiii

FIG.4.15 TORQUE DYNAMIC PERFORMANCE OF ROTOR FLUX ORIENTED CONTROL WITH VARIED ROTOR

RESISTANCE ................................................................................................................................. 79 FIG. 4.16 THE EXPERIMENT SETUP OF THE SYSTEM ..................................................................................... 80 FIG.4.17 TORQUE DYNAMIC PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH DIRECT SYNTHESIS OF

PI CONTROLLER ........................................................................................................................... 80 FIG.4.18 PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH SPEED LOOP ......................................... 81 FIG.4.19 TORQUE DYNAMIC PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH PRE-FILTER ............ 81 FIG.4.20 PERFORMANCE OF DIRECT FLUX VECTOR CONTROL WITH SPEED LOOP ......................................... 82 FIG.4.21 STEADY STATE PERFORMANCE WITH SPEED-LOOP......................................................................... 83 FIG.4.22 SPECTRUM ANALYSIS OF THE STATOR CURRENT ........................................................................... 83 FIG. 4.23 DFC SCHEME FOR ISA................................................................................................................. 84 FIG. 4.24 REFERENCE SPACE VOLTAGE VECTOR.......................................................................................... 86 FIG. 4.25 STARTING PROCESS OF ISA.......................................................................................................... 88 FIG. 4.26 ISA GENERATING WITH FULL LOAD ............................................................................................. 89 FIG. 4.27 SPECTRUM ANALYSIS OF THE STATOR CURRENT OF ISA .............................................................. 89 FIG. 4.28 LOAD DUMP OF ISA WITHOUT BATTERY CONNECTED .................................................................. 90 FIG. 4.29 LOAD DUMP OF ISA WITH BATTERY CONNECTED......................................................................... 91 FIG. 4.30 ISA PERFORMANCE AT ACCELERATION........................................................................................ 92 FIG. 4.31 ISA PERFORMANCE AT DECELERATION........................................................................................ 92 FIG. 4.32 ISA WITH FIELD WEAKENING AT HIGH SPEED............................................................................... 93 FIG. 5.1 VECTOR DIAGRAM OF THE INDUCTION MACHINE ........................................................................... 96 FIG. 5.2 DIRECT TORQUE AND FLUX CONTROLLED INDUCTION GENERATOR FOR ISA ............................... 100 FIG. 5.3 STARTING PROCESS OF ISA.......................................................................................................... 102 FIG. 5.4 ISA GENERATING WITH FULL LOAD ............................................................................................. 103 FIG. 5.5 SPECTRUM ANALYSIS OF THE STATOR CURRENT OF ISA .............................................................. 104 FIG. 5.6 LOAD DUMP OF ISA WITHOUT BATTERY CONNECTED .................................................................. 105 FIG. 5.7 LOAD DUMP OF ISA WITH BATTERY CONNECTED......................................................................... 105 FIG. 5.8 ISA PERFORMANCE AT ACCELERATION........................................................................................ 106 FIG. 5.9 ISA PERFORMANCE AT DECELERATION........................................................................................ 107 FIG. 5.10 ISA WITH FIELD WEAKENING AT HIGH SPEED............................................................................. 108 FIG. 6.1 ONE LEG OF THE CONVERTER ....................................................................................................... 110

FIG. 6.2（A） IDEAL GATE SIGNAL （B）PRACTICAL GATE SIGNAL WITH DEAD-TIME （C） aNV WITH

DEAD-TIME EFFECT ONLY（D）CONSIDERING ont AND offt OF THE POWER DEVICE.................. 111

FIG. 6.3 SWITCHING STATE OF VSI (A) AND SPACE VOLTAGE VECTORS (B)............................................... 111 FIG. 6.4 GATE SIGNAL WITHOUT DEAD-TIME............................................................................................. 112 FIG. 6.5 GATE SIGNAL WITH DEAD-TIME ................................................................................................... 113 FIG. 6.6 ANALYSIS OF THE VOLTAGE DROP ON THE POWER DEVICE ........................................................... 114 FIG. 6.7 GATE SIGNAL WITH VOLTAGE DROP............................................................................................. 115 FIG. 6.8 BACKWARD COMPENSATION STRUCTURE .................................................................................... 116

xiv

FIG. 6.9 FORWARD COMPENSATION STRUCTURE ....................................................................................... 117 FIG. 6.10 THE CONTROL SYSTEM WITH VOLTAGE DROP AND DEAD-TIME COMPENSATION......................... 118 FIG. 6.11 CURRENT MODE STATOR FLUX AND TORQUE ESTIMATOR .......................................................... 119 FIG. 6.12 VOLTAGE MODE STATOR FLUX AND TORQUE ESTIMATOR .......................................................... 120 FIG. 6.13 ROTOR SPEED, STATOR CURRENT, AND ESTIMATED TORQUE AND FLUX AT NO-LOAD -WITHOUT

COMPENSATION.......................................................................................................................... 121 FIG. 6.14 ESTIMATION ERRORS OF THE STATOR FLUX- WITHOUT COMPENSATION..................................... 122 FIG. 6.15 ROTOR SPEED, STATOR CURRENT, AND ESTIMATED TORQUE AT NO-LOAD - WITH BACKWARD

COMPENSATION.......................................................................................................................... 123 FIG. 6.16 ESTIMATION ERRORS OF THE STATOR FLUX- WITH BACKWARD COMPENSATION ........................ 123 FIG. 6.17 REFERENCE VOLTAGES AND ERROR VOLTAGES - WITH BACKWARD COMPENSATION ................. 124 FIG. 6.18 ROTOR SPEED, STATOR CURRENT, AND ESTIMATED TORQUE AT NO-LOAD - WITH FORWARD

COMPENSATION.......................................................................................................................... 125 FIG. 6.19 ESTIMATION ERRORS OF THE STATOR FLUX- WITH FORWARD COMPENSATION........................... 125 FIG. 6.20 REFERENCE VOLTAGES AND ERROR VOLTAGES - WITH FORWARD COMPENSATION .................... 126 FIG. 6.21 FLUX ESTIMATION ERRORS COMPARISON FOR WITH AND WITHOUT COMPENSATION.................. 127 FIG. 6.22 DYNAMICS OF THE TORQUE AND FLUX FOR THE DTC-SVM WITH AND WITHOUT COMPENSATION

................................................................................................................................................... 129 FIG. 6.23 PERFORMANCE COMPARISON WITH AND WITHOUT COMPENSATION WHILE ISA IS GENERATING AT

1500 RPM WITH NO-LOAD........................................................................................................... 131 FIG. 6.24 PERFORMANCE COMPARISON WITH AND WITHOUT COMPENSATION DURING LOAD DUMP AT 1500

RPM............................................................................................................................................ 133 FIG. 7.1 THE OVERALL STRUCTURE OF THE DIRECT TORQUE CONTROLLED INDUCTION MACHINE WITH

SLIDING MODE OBSERVER .......................................................................................................... 140

FIG. 7.2 OPEN-LOOP STATOR FLUX ESTIMATION WITH 50% ERROR IN sR ................................................. 141

FIG. 7.3 SLIDING MODE FLUX OBSERVER WITH 50% ERROR IN sR ............................................................ 142

FIG. 7.4 OPEN-LOOP STATOR FLUX ESTIMATION WITH 3A DC CURRENT OFFSET........................................ 143 FIG. 7.5 SLIDING MODE FLUX OBSERVER WITH 3A DC CURRENT OFFSET ................................................... 143 FIG. 7.6 DIRECT TORQUE CONTROLLED INDUCTION MACHINE WITH OPEN-LOOP STATOR FLUX ESTIMATOR

................................................................................................................................................... 144 FIG. 7.7 DIRECT TORQUE CONTROLLED INDUCTION MACHINE WITH SLIDING MODE FLUX OBSERVER........ 144 FIG. 7.8 ROTOR SPEED, STATOR CURRENT, AND ESTIMATED TORQUE AND FLUX AT NO-LOAD WITH OPEN-

LOOP STATOR FLUX ESTIMATION................................................................................................ 145 FIG. 7.9 ROTOR SPEED, STATOR CURRENT, AND ESTIMATED TORQUE AND FLUX AT NO-LOAD WITH SLIDING

MODE FLUX OBSERVER............................................................................................................... 146

FIG. 7.10 OPEN-LOOP STATOR FLUX ESTIMATION WITH 50% sR ERROR................................................... 147

FIG. 7.11 SLIDING MODE FLUX OBSERVER WITH 50% sR ERROR.............................................................. 147

FIG. 7.12 OPEN-LOOP STATOR FLUX ESTIMATION WITH 3A DC CURRENT OFFSET...................................... 148

xv

FIG. 7.13 SLIDING MODE FLUX OBSERVER WITH 3A DC CURRENT OFFSET ................................................. 149 FIG. 7.14 DYNAMIC PERFORMANCE WITH OPEN-LOOP STATOR FLUX ESTIMATION .................................... 150 FIG. 7.15 ESTIMATION ERRORS WITH OPEN-LOOP STATOR FLUX ESTIMATION ........................................... 150 FIG. 7.16 DYNAMIC PERFORMANCE WITH SLIDING MODE FLUX OBSERVER ............................................... 151 FIG. 7.17 ESTIMATION ERRORS WITH SLIDING MODE FLUX OBSERVER ...................................................... 151 FIG. 7.18 CURRENT ESTIMATION WITH SLIDING MODE FLUX OBSERVER.................................................... 152 FIG. 7.19 PERFORMANCE COMPARISON WITHOUT AND WITH COMPENSATION, AND SMO WHILE ISA IS

GENERATING AT 1500 RPM......................................................................................................... 155 FIG. 7.20 PERFORMANCE COMPARISON WITH/WITHOUT COMPENSATION AND WITH SMO DURING LOAD

DUMP AT 1500 RPM .................................................................................................................... 156 FIG. 8.1 THE OVERALL STRUCTURE OF THE DIRECT TORQUE CONTROLLED INTEGRATED

STARTER/ALTERNATOR .............................................................................................................. 161 FIG. 8.2 POWER FACTOR CONTROLLER..................................................................................................... 162 FIG. 8.3 POWER FACTOR OF THE INDUCTION UNDER DIFFERENT LOADS .................................................... 163 FIG. 8.4 STATOR VOLTAGE, STATOR AND ROTOR CURRENTS WITH 30% RATED LOAD ............................... 163 FIG. 8.5 CORE LOSS PERCENTAGE WITH AND WITHOUT POWER FACTOR CONTROL .................................... 164 FIG. 8.6 COPPER LOSS PERCENTAGE WITH AND WITHOUT POWER FACTOR CONTROL................................. 165 FIG. 8.7 EFFICIENCY COMPARISON OF THE INDUCTION MACHINE WITH AND WITHOUT POWER FACTOR

CONTROL IN MOTORING MODE AT 1200 RPM AND 1500 RPM...................................................... 166 FIG. 8.8 TRANSIENTS OF THE REGULATION OF THE POWER FACTOR CONTROLLER..................................... 167 FIG. 8.9 EFFICIENCY COMPARISON OF THE INDUCTION MACHINE WITH AND WITHOUT POWER FACTOR

CONTROL IN GENERATING MODE AT 1500 RPM AND 2100 RPM .................................................. 168 FIG. B.1 EQUIVALENT SYSTEM MODEL OF THE TORQUE LOOP ................................................................... 194 FIG.B.2 PI CONTROL OF THE EQUIVALENT SYSTEM ................................................................................... 194 FIG. C.1 VECTOR DIAGRAM OF THE INDUCTION MACHINE......................................................................... 196

xvi

LIST OF TABLES

TABLE 2.1 PARAMETERS OF THE INDUCTION MACHINE .............................................................................. 27 TABLE 3.1 SWITCHING TABLE OF INVERTER VECTORS ................................................................................ 44 TABLE 4.1 PARAMETERS OF THE INDUCTION MACHINE .............................................................................. 71 TABLE 4.2 PARAMETERS OF THE CONTROL SCHEME.................................................................................... 77

TABLE 6.1 DEAD-TIME EFFECT ANALYSIS ( 0ai > ; 0bi > ; 0ci < ) ....................................................... 113

TABLE 6.2 ERROR VOLTAGE VECTORS UNDER DIFFERENT CURRENT POLARITIES ...................................... 114 TABLE 6.3 ERROR VOLTAGE VECTORS UNDER DIFFERENT CURRENT POLARITIES ...................................... 116 TABLE 9.1 COMPARISON OF DIFFERENT CONTROL SCHEMES FOR THE ISA................................................ 172

xvii

LIST OF SYMBOLS

α −β stationary reference frame

d q− stator flux reference frame

e ed q− rotor reference frame

ia, ib, ic stator phase currents, A

Ic collector current of a power device, A

id, iq d and q axis stator currents, A

Is amplitude of the stator current, A

is stator current vector, A

isα, isβ α and β axis stator currents, A

p derivative

P number of pole pairs

sR stator resistance of the induction machine

sL stator inductance

rL rotor winding self-inductance

mL mutual inductance

lsL stator leakage inductance

lrL rotor leakage inductance

sΨ Stator flux vector

rΨ rotor flux vector

eT electromagnetic torque, Nm

TL load torque

xviii

Test, T estimated electromagnetic torque, Nm

td dead-time in the inverter, μs

toff turn off delay of the power device, μs

ton turn on delay of the power device, μs

Tref reference torque, Nm

Ts, Δt sampling interval, μs

Vce collector-emitter voltage, V

γ angle between the rotor and stator flux linkage vector, rad or degree

Superscripts

* reference value

^ estimated value

Subscripts

est estimated value

act actual value

ref reference value

k, k-1 kth and k-1 sampling interval

Abbreviation

ac, AC alternating current

dc, DC direct current

DSP digital signal processor

DTC direct torque control

DTFC direct torque and flux control

DFC direct flux vector control

EKF extended kalman filter

emf electromagnetic force

xix

FOC field oriented control

RFOC rotor field oriented control

FVD forward voltage drop

FW field weakening

IGBT insulated gate bipolar transistor

IPM interior permanent magnet

IPMSM interior permanent magnet synchronous motor

PI proportional and integral

PID proportional, integral and derivative

PM permanent magnet

PMSM permanent magnet synchronous motor

PWM pulse width modulation

SVM space vector modulation

SM sliding mode

THD total harmonic distortion

VC vector control

ISA integrated starter alternator

ISG integrated starter generator

VSI voltage source inverter

SPM surface permanent magnet machine

VRM variable reluctance machine

ICE internal combustion engine

rms root mean square.

Chapter 1 Introduction 1

CHAPTER 1

INTRODUCTION

1.1 42-Volt PowerNet

The electrical power demand in automobiles keeps increasing in recent years with

proliferation electrical systems installed in More Electric Cars (MEC) [1]. The electrical

systems in a MEC perform more duties other than conventional purposes of lighting,

cranking, and battery charging. The electric machines play an important role in current

and future automotive electrical system for propulsion, power steering, pumps, fans, air

conditioners, electrically active suspension, electric brakes electromechanical engine

valve, and so on [2]. Fig. 1.1 summarized partially current electrical and electric

applications and future products under development in automobiles.

Fig. 1.1 Electrical and Electrics components in automobiles [2, 3]


In addition, the automotive electronic systems are also kept growing. As shown in Fig.

1.2, many electric networks will be equipped in modern vehicles such as CAN

(controller area network), GPS (global positioning system), GSM (global system for

mobile communications), LIN (local interconnect network) and MOST (media-oriented

systems transport).

Fig. 1.2 more extensive electronics in modern vehicles [4]

As a result, the electrical power load on the alternator is expected to increase to 4 - 6

kW [5] and even to about 20 kW in the next decades [2]. The trend of power demand in

vehicles is shown in Fig. 1.3. This dramatic increase requires substantial changes in

automotive electrical generation and distribution systems. The present 14 V system

cannot meet the enhanced power requirements. Therefore, the electrical bus voltage of

automobiles is proposed to be increased from 14 V to 42 V, which is known as the 42 V

PowerNet [6-8].


Fig. 1.3 Generator peak power demand of average passenger vehicle [9]

Higher voltage system offers a lot of benefits, which includes:

• Saving in weight and improving in fuel efficiency. The current of 42 V systems

will be reduced by three times with same power output. Thus, the overall

efficiency of the system is improved with less copper loss. Furthermore, the

wiring resistance can be increased while retaining the same power loss over a

given length of wire. A lighter wiring harness can be achieved with a reduction

in the bundle diameter. Therefore, the duct arrangement becomes easier in the

limited space of automobiles.

• Reduction in the cost of semiconductor devices. The standard of the 42 V

electrical system is proposed [10], which stipulate a much tighter voltage

regulation than the current 14 V standard as shown in Fig. 1.4. The maximum

voltage is 58 V including transient voltages, whereas some auto manufacturers

allows 80 V [11] or even 100 V [12]. Therefore, the semiconductor devices can

be rated as lower voltage rating. Besides lower current rating, the lower voltage

rating results in significant reduction of the cost of semiconductor devices.

Fig. 1.4 Voltage regulation of 42 V electrical system [13]


• Flexibility in distribution of load and electrical system. The conventional 14 V

electrical system use point-to-point distribution architecture shown in Fig. 1.5.

The wiring and harness is heavy and complex. The 14 V system cannot handle

future higher power in MECs due to expensive cost and low efficiency [1]. The

electrical system can change from point-to-point architecture to multiplexed

architecture in 42 V system. As shown in Fig. 1.6, the loads are controlled by

intelligent remote modules. Power Management System can be realized by

interconnection between remote modules. The 42 V or a similar high voltage bus

for distributed application is inevitable in automobiles.

Fig. 1.5 Conventional 14V dc distribution system architecture [1]

Fig. 1.6 Advanced multiplexed automotive power system architectures of the future with power

and communication buses [1]


1.2 Integrated Starter Alternator - ISA

The existing Lundell alternator is not able to meet the requirements of high power,

efficiency and voltage transients. The maximum output power of Lundell alternator is

only 2 kW under force cooling [14]. New type of the alternator has to be used for high

power generation. With the introduction of 42 V PowerNet, an integrated starter

alternator (ISA) system has been proposed [2, 15-30], which is also named as integrated

starter generator [31-33] (ISG). In conventional system of automobiles, the dc starter

motor for cranking and the alternator for generation are separate as two units. The ISA

combines both starter and alternator functions into a single electrical machine with bi-

directional power flow. The ISA has attracted more and more research interest around

the world as an alternative to the current unsatisfactory generating system in

automobiles. The ISA provides a number of advantages listed as follows.

• The ISA can save space and reduce the cost and weight of the electrical system

with multifunctional integration, including starting, generating and reduction of

engine torque pulsations.

• The ISA can be mounted directly on the crankshaft of the engine and replaces

the flywheel. Fig. 1.7 shows crankshaft mounted starter alternator, which is so

called “Flywheel-starter-alternator”.

Fig. 1.7 Crankshaft mounted starter alternator [34]


• The ISA offers geater generating capacity and a start-stop facility that improves

fuel economy and reduces harmful gaseous emissions. Improved fuel efficiency

is obtained through implementation of start/stop cycles. With a conventional dc

starter, fuel is supplied shortly after engine cranking begins, but the engine does

not fire until about 500 ms later as shown in Fig. 1.8 because its torque

decreases as motor speed increases. In contrast, the ISA can start the engine

within about 250 ms [8] by producing torque without regard to speed. Therefore,

only the fuel necessary to maintain idle is supplied. Fuel saving and reduction of

hydrocarbon emissions are both achieved.

Integrated starter/alternator

DC starter Motor

Eng

ine

spee

d (r

pm)

Starting time (ms)

200 400 600 800 10000

0

200

400

600

800

1000

Fig. 1.8 Starting with ISA and DC motor [8]

• The ISA provides additional braking ability by converting kinetic energy to

electrical energy.

• The ISA also offers possibility of a soft hybrid configuration for acceleration

boost at low engine speeds. This feature allows the use of smaller internal

combustion engines.

Several important issues should be considered for the design of ISA system. They are:

electrical specification, selection of the machine, the power converter topology and the

control scheme.


1.2.1 Electrical specification

ISA operates on both starting mode and generation mode. For starting mode, The

MIT/industry consortium on automobiles recommended that the engine starting torque

requirement to be set at 150 Nm from standstill up to 100 rpm engine speed [15]. For

generation mode, the power level of the ISA is about 4 kW at 600 rpm engine (idle

condition) and it rises to 6 kW at 6000 rpm. The specification of the power level reveals

its high requirement over wide speed range. Due to the future fuel efficiency

consideration, the alternator system efficiency requirement was set to 75% for rated

base load for the combined electrical machine and converter. Fig. 1.9 shows the

resulting starting and generating torque requirements as a function of speed for the ISA.

Fig. 1.9 Starter/alternator starting and approximate generating torque requirement (*) and the

torque/speed characteristic (line) [15].

Although no decided specification of the dynamic performance of ISA is presented,

some general ideas are described in literatures. For example, the starting time of ISA is

accepted to 0.2 to 0.5 seconds. Faster starting characteristics ensures low emission and

fuel saving. On the other hand, the smooth transition from starting mode to generation

mode is also important for an ISA. In generation mode, the voltage regulation ability for

rated load and load dump plays an essential role to evaluate an ISA system. Finally, cost

and reliability determines whether certain ISA system can be applied in automobiles.


As shown in Fig. 1.4, the standard of the 42 V electrical system is proposed [10]. The

voltage transient is also defined for the worst case when full load is suddenly

disconnected from dc bus [6, 10]. This is known as the load dump condition [35-37].

The peak voltage of the 42 V is required to limits below 1.4 times (58 V) of rated

voltage during load dump. As shown in Fig. 1.10, the transient voltage of the 42 V

electrical system is required to be limited in 400 ms duration and magnitude lower than

58 V. in addition, the voltage has to be regulated back to 46.2 V within 430 ms since the

load dump happens.

Fig. 1.10 dc bus voltage dynamic requirement [6]

The above voltage transient specifications require the generation system of automobiles

has good dynamic regulation ability. This ability is determined by the controller of the

generator. The scalar control (V/f) scheme is not able to satisfy these requirements. The

advanced control scheme with field oriented control or vector control is thus adopted for

the existing ISA system [16, 20, 24-26, 38]. Another advanced control scheme with

direct torque control is explored in this thesis for the control of a generator for the ISA

application.

1.2.2 Machine technologies

For the application of ISA system, the selection of machine which determines the

performance of ISA system is very important. Although the DC machine has inherent

flexibility in control and capability of operation in both motor and generator, the

commutator brush makes it is impossible to be used in an ISA application due to the


limitation of speed and reliability. The conventional synchronous machine used in

today’s alternators has severe limitation due to its size and efficiency scaling

characteristics [22, 28, 38]. Therefore, the existing literatures on machine topologies for

ISA discuss and compare four alternative brushless machines. They are induction

machine (IM), surface permanent magnet machine (SPM), interior permanent magnet

machine (IPM), and variable (switched) reluctance machines (VRM or SRM).

1.2.2.1 Induction Machine (IM)

The induction machine is one of the most serious candidates for the starter/alternator

application because of its attractive characteristics such as robust rotor structure and

mature manufacturing technology. A primary advantage of the induction machine is the

simplicity and reliability. Since the power is transformed from the stator to the rotor

through transformer action, no commutators, brushes, or slip rings are required.

Therefore, the machine requires less maintenance, which makes it attractable in

automotive application. In addition, induction machines have very good efficiency,

smooth torque and wide speed range which match the specification of ISA application.

Control technologies for induction machines have been well studied over several

decades. These technologies are now quite mature.

In paper [16], the authors present the comparison results of using induction and variable

reluctance machines as the starter-alternator in a hybrid electric vehicle. Permanent

magnet machines are not considered due to the rotor heating under the dense packaging.

When both machines are compared against specified engine cranking and continuous

alternator output power requirements, they found that the induction machine has higher

average thermal duty cycle and benefits from the ability to use a simpler incremental

encoder for control. The variable reluctance machine has significant benefits for in-

vehicle packaging and low rotor inertia but suffered more in thermal performance and

its need for a high resolution encoder.

Researchers at Delphi describe their design of belt-driven starter-generator with

induction machine [39]. As mentioned in the paper, they considered PM machine as the

most expensive solution in this case because the inverter rating must handle the large

voltage range produced by the magnets (10:1 from idle to 6000 rpm). As for switched

reluctance machine, noise and vibrations would be the problem. And it is still an

emerging technology, which may complicate practical developments. By contrast, the


induction machine is an established technology with good efficiency and smooth torque.

Therefore, induction machine was selected for their project.

A integrated starter-alternator system was introduced by Visteon Automotive Systems

[18]. They also selected induction machine as the best machine for ISA application.

They consider that the induction machines have wide speed range, have a better failure

mode and are more reliable in the case of a winding short circuit; have high

performance at lowest possible cost.

1.2.2.2 Surface Permanent Magnet Machine (SPM)

The surface permanent magnet synchronous machine employs surface-mounted rotor

magnets to achieve high torque and power densities. Such characteristics make the SPM

well suited for delivering the high starting torque required in the starter/alternator

application. Unfortunately, the SPM has difficulty achieving wide constant-power speed

ranges because its back-EMF rises linearly with speed, and its phase inductances are

typically too low for effective flux weakening [15]. To overcome this obstacle,

additional DC-DC converter is required [15, 28] to regulate the voltage of the bus which

has a negative impact on the system cost.

1.2.2.3 Interior Permanent Magnet Machine (IPM)

S

N

N

S

SN S N

S

N

N

S

SN S N

(a) IPMSM-I (b) IPMSM-II

Fig. 1.11 Rotor structure of IPM motors

As shown in Fig. 1.11, the magnets are buried inside the rotor of IPM. As a result, the

IPM is inherently a ‘hybrid’ machine with torque contributions from both the magnets


and the iron saliency produced by the magnet cavities. Interior permanent magnet (IPM)

synchronous motors offer many advantages over induction motors, such as higher

overall efficiency, effective use of reluctance torque, smaller losses and compact motor

size. Moreover, the use of flux weakening control based on pole saliency supports a

wider range of speeds. In particular, proper balancing of the magnet strength and the

rotor saliency makes it possible to achieve very wide speed ranges of constant-power

operation. This is a major advantage over the SPM discussed above, further accentuated

by the IPM’s need for significantly less magnet material to deliver the same torque [15].

The benefits provided by IPM come with some disadvantages. Since the motor

magnetic field cannot be shut off, even with the stator winding disconnected from the

drive, the rotor will always create an induced voltage in the winding. In the event of a

winding failure or fault, the rotor will continue to pump fault current into the failed

region, even after the inverter trips the motor off line. This has the potential to do

considerable damage to motor components other than just the winding making repair

more costly. Moreover, the PM motor may have limited overload or peak torque

capability and can be demagnetized when the overload limit is exceeded. Overloads

limits for a PM motor may be as low as 120% of rated load. PM motors may have to be

oversized for applications that require overloads. In contrast, induction motors designed

for variable-speed application typically have a minimum of 250% overload capability

and have been applied for overloads as high as 600% [40]. In addition, IPM has the

thermal problem which is very harmful for practical application. Moreover, the relative

complexity of the IPM’s rotor structure represents an important technical risk in

comparison to the mature induction machine structure. The presence of the embedded

magnets contributes cost and manufacturing complications associated with the

installation and magnetization process. Relatively higher cost of the IPM high

efficiency magnetic material holds back its application.

1.2.2.4 Variable Reluctance Machine (VRM)

The variable (or switched) reluctance machine offers some attractive characteristics for

the integrated starter/alternator application including its robust rotor construction and a

torque-speed characteristic.

However, variable reluctance machines are excited with non-sinusoidal waveforms that

make it difficult to simultaneously minimize torque ripple while maximizing


torque/power density. In addition, it is still an emerging technology, which may

complicate the practical development.

From the literatures, it is found that the researches on the starter-alternator system with

switched reluctance machine are mainly for the aircraft application [41-46]. Only a few

papers discuss the SRM for automotive application [47-49].

Paper [48] concluded that thermal management of any permanent magnet machine

seems to be problematic due to the close distance from the engine. They announce that

the performance comparison of induction machine and SRM will be different with that

described in the paper [16] due to current intensive scenario (42 V,7.2 kW). Therefore, a

switched reluctance machine based ISA system was proposed by them. However, they

also concluded that there exists limitation on the performance of SRM in ISA which

depends on the development of ultra-fast DSP based processors and semiconductor.

Fig. 1.12 Cost comparison of three machine systems for a 6kW direct-drive starter/alternator

application [15]

Paper [15] estimated the cost of ISA system. The authors used cost estimation

algorithms into each of the analysis models in order to permit the cost of each individual

machine design to be estimated together with the cost of its accompanying converter.

The result of the cost optimization process for each of the four candidate machine types

is presented as a cost bar chart in Fig. 1.12. It can be observed from the bar diagram that

the induction and IPM are more attractive for the ISA application on the basis of

projected system cost compared to the surface PM and variable reluctance machines.


Their completed results of the trade-off study indicated that the induction machine and

IPM are both serious candidates for the direct-drive starter/alternator application.

1.2.2.5 Summary

As discussed in last sections, both induction machine (IM) and interior permanent

magnet machine (IPM) are suitable for of ISA application. Permanent-magnet machine

and induction machine based ISA systems have been proposed for automotive

applications by many researchers [16, 20, 24-26, 38, 50].

In comparison, IM has lower efficiency than that of IPM due to the loss in the rotor.

However, IM has higher reliability than IPM because of easing of thermal problem.

Compared with the permanent-magnet machine, the induction machine has robust

structure, low cost, mature technology and low maintenance requirement. Moreover, the

induction machine dose not retain magnetization, unlike a permanent-magnet machine,

when the system is turned off under fault condition. Therefore, the induction machine is

a viable option for ISA system design. The induction machine based integrated

starter/alternator systems have been reported in [16, 18-22, 24-26, 29, 34, 51-55]. Based

on the mature technology of previous research work, higher reliability can be achieved.

Therefore, induction machine is selected in this study.

1.2.3 Electrical System configuration and Power converter topology

Among several publications on induction generator for ISA [16, 20, 24-26, 38] and

stand-alone application [56, 57] to date, the Pulse Width Modulation (PWM) voltage

source converter is the most attractive hardware structure due to its excellent dynamic

performance.

Both high voltage [16, 20, 26] and low voltage [50, 58, 59] bus system of the ISA were

developed. Basically, high voltage electrical system has two stages of the power

converter as DC-DC-AC and low voltage electrical system has one stage of the power

converter as DC-AC.


1.2.3.1 High bus voltage with battery

Fig. 1.13 High voltage bus configuration

In paper [16], the authors presented a parallel hybrid structure with starter-alternator.

High voltage batteries (Pb-Acid) are used to provide about 300V bus voltage directly to

the inverter. In this kind of configuration, the machine acting as starter-alternator should

be high voltage machine and no boost converter is required between bus and machine.

The inverter is actually a bi-direction converter which transfers power flow between dc

bus and starter-alternator. However, DC-DC converters are needed to step-down the

high voltage for the supplying the low voltage electrical load of automobiles.

With this configuration, the bi-direction function is easy to fulfill by a simple full-

bridge without voltage-boasting part. But the cost of high voltage batteries may be the

problem for commercial application.

1.2.3.2 High bus voltage with ultracapacitor

Fig. 1.14 High voltage bus configuration with ultracapacitor

Visteon developed an integrated starter-alternator (V-ISA) system [20]. The V-ISA

system includes induction motor, inverter, DC-DC converters (boost and buck pattern),

ultracapacitors on the high voltage side and 42 Volt Battery. Therefore, this

configuration has high bus voltage. The ultracapacitor has a number of very attractive

features, offering high power density and extremely high cycling capability. The boost

converter powered by the 42 volt battery can charge capacitors up to 300 Volts in a few


seconds even during cold start. Fully charged capacitors can start the engine

consecutively before next recharge. Depending on battery’s state of charge, the

regenerative action charges the 42 volt battery, 12 volt battery and capacitors,

respectively. The main energy source during the start is from high voltage

ultracapacitors. Adding ultracapacitors to the system significantly reduces weight and

space by elimination high voltage batteries and also it makes braking regeneration

possible since capacitors and ideal for absorbing high inrush current.

A similar system configuration was proposed in paper [26], which is shown in Fig. 1.15.

Two DC-DC converters were applied in their scheme to realize the bi-direction power

flowing. The low-power DC-DC converter is enabled in start-up mode to provide

energy for engine cranking. And the high-power DC-DC converter transfers generation

power from inverter to the battery and loads. This function is achieved by a single bi-

direction DC-DC converter in paper [20], which required better tradeoff design for the

bi-directional converter.

Fig. 1.15 Block diagram of the overall supervisory control scheme [26]

1.2.3.3 Low bus voltage

Besides high bus voltage, a low bus with 42V is also adopted by some researchers [50,

58]. With low bus voltage, the DC-DC converter between the battery and inverter can

be removed when low voltage machine is used. Figure 2.7 shows a dual voltage (14V


and 42V) automotive electrical system with low voltage bus. In this system, the bi-

directional converter has no boost pattern with low voltage starter/alternator [59].

Fig. 1.16 Dual voltage (14V and 42V) automotive electrical system [59]

1.2.3.4 Summary

Single-stage bidirectional three phase DC-AC converter is selected in this study. In

comparison, the two-stage converter topology has a negative impact on the system cost

and raises special packaging issues in order to adequately protect humans from exposure

to the high voltages.

In the single-stage scheme, bi-directional DC-AC converter connects battery and the

machine. Single-stage scheme has higher power efficiency than the two-stage scheme

and is easier to control without considering the independent control of the two

converters. Moreover, isolation is not required in low voltage system.

Fig. 1.17 Proposed ISA electrical system configuration


1.2.4 Machine controller- control of generator

1.2.4.1 Field Oriented Control

The ISA requires sophisticated control that must monitor power demand and power

flow in and out of the motor/generator and batteries in all operating modes of

automobiles, whether the vehicle is cruising, braking, or accelerating. In the

publications related to ISA development, Field Oriented Control (FOC) of AC machines

[16, 20, 24-26, 38, 60] appears to have drawn much interest. Field oriented control has

been used in induction motor control for a long time and it was natural to extend it to

induction generator application [57, 61, 62]. Although field oriented control is an

advanced scheme, it has several disadvantages such as high computational requirement

for the co-ordinate transformation and high parameters dependency [63, 64]. In

addition, the rotor speed signal is essential for co-ordinate transformation in field

oriented control. Therefore, encoder based speed sensing or speed observer is needed

for both generation and motoring with field oriented control [20, 24-26, 38, 57, 61, 62,

65]. To avoid these drawbacks, efforts have gone into sensorless field oriented

controllers in the past two decades.

1.2.4.2 Direct torque control

Direct torque control (DTC) was introduced in 1980’s [66, 67]. Compared with field

oriented control, direct torque control is a very simple control scheme with low

computational requirement. Current regulator and co-ordinate transformation are not

required with DTC [63, 64]. The DTC and some of its variations have the merits like

inherent sensorless operation and reduced parameter sensitivity. DTC is increasingly

gaining wide acceptance in motor drives application from both academia and industry,

but has not yet been considered for ISA application. In generation application, the speed

of the machine is already determined by the prime mover or engine. No speed control

loop is thus needed for the controller. The speed sensorless controller is thus a natural

choice for ISA application. For starting mode, ISA system only requires large starting

torque and short starting time without concerning about the speed characteristics.

Therefore, the poor performance of DTC in low speed range is not a significant problem

in this application. For the application of ISA, the AC machine mostly operates in


generation state after the engine is started and runs above the idling speed. The DTC

scheme is thus more suitable than FOC scheme for ISA application.

This thesis is primarily concerned with the application of direct torque controlled

induction machines for the control of the ISA in both motoring (i.e. starting) and

generating modes.

By neglecting the loss of AC machine and the converter, the electromagnetic power of

generator should be balanced with the absorbing power of the load. In other words, the

following equation should be satisfied at any time.

e dc dcT V Iω = (1-1)

where ω is the speed of AC machine, which is determined by the engine. eT is the

electromagnetism torque of AC machine; dc dcV ,I are the output voltage and current on

the dc side.

Fig. 1.18 shows the structure of basic DTC scheme for induction motor. The

electromagnetic torque can be regulated as follows:

a

dcb s e

c

uV

u V TSwitching Signal

u

⎧⎧ ⎪→ → →Ψ →⎨ ⎨⎩ ⎪

⎩

(1-2)

By producing different voltage vector through the voltage source inverter (VSI), DTC

scheme restricts the flux and torque errors within respective flux and torque hysteresis

bands.

dcV

Fig. 1.18 DTC of induction motor


The concept of DTC for motor drive can be mirrored to generator mode of operation

directly. In motoring state, the desired voltage vector for torque control is produced by

VSI with certain switching signal. In other words, the dc bus voltage and voltage vector

are uniquely determined through the switching signals of the VSI inverter. Their

relationship depends on the switching signals that have been selected. In generating

state, this corresponding relationship still exists and it determines which switching

signal should be applied. On the other hand, the switching signal is also restricted by the

flux-linkage which is determined by desired electromagnetic torque. Based on above

analysis, DTC scheme of induction generator can be built as the structure shown in Fig.

1.19. The torque reference is given by voltage regulator. The dc bus voltage can be

regulated as follows:

a

bs e dc

c

uu

V T VuSwitching Signal

⎧⎪⎪→ Ψ → → →⎨⎪⎪⎩

(1-3)

dcU

Fig. 1.19 DTC of induction generator

A few papers have studied the classic switching table based DTC control with schemes

for the generator [68-71]. Switching table based DTC for integrated starter/alternator is

also reported in [72]. However, the switching table based classic DTC has some

drawbacks such as large torque and flux ripples, and variable switching frequency.

Faster sampling frequency has to be used to minimize the torque and flux ripples for

digital implementation of hysteresis controllers [63, 73].


The problems associated with the classic DTC can be solved by Proportional-Integral

(PI) controller plus Space Vector Modulation (SVM). This improved DTC scheme can

achieve better performance with reduced torque ripple and constant switching

frequency. This thesis proposed two improved DTC based control with space vector

modulation schemes for the integrated starter/alternator [29, 55]. Several papers arising

from this thesis have been published in proceeding and journal, which can be found in

the Appendix A. A DTC control scheme of a permanent magnet-assisted reluctance

synchronous machine (PM-RSM) with SVM for ISA application were reported [30].

This paper indicates further the potential of DTC for ISA application and also shows the

acceptance this idea in academia.

1.3 Scope of the thesis

The purpose of the thesis is to extend the application of direct torque control and its

variations in ISA system for the future 42 V PowerNet. This thesis presents the

modeling, design as well as experimental results of the direct torque controlled ISA

system, which includes

• Evaluation of the classical direct torque controlled integrated starter/alternator

• Study of improved direct torque control schemes for integrated starter/alternator

with space vector modulation

• Compensation of the non-linearity of the DC-AC converter due to dead-time and

voltage-drop of the power devices

• Design of an sliding mode observer for improvements on stator flux estimation

• Efficiency improvement of the integrated starter/alternator with power factor

control

The objective of this project is to develop a direct torque controlled induction machine

driven ISA meeting the strengthen requirements of the 42 V PowerNet. The solution

carried out in thesis in the above areas have proved the suitability of direct torque

controlled induction machine driven ISA, as is reported in Chapters 2-8.


1.4 Outline of the thesis

Chapter 1 gives a brief introduction of the Integrated Starter/Alternator (ISA). This

chapter also reviews the state-of-the-art for integrated starter/alternator and discusses

the machine selection, power converter topology and advance control schemes.

Chapter 2 presents a rotor field oriented controlled integrated starter/alternator with

space vector modulation in order to compare with the proposed DTC schemes.

Chapter 3 presents the analysis and implementation of the classical direct torque

controlled integrated starter/alternator.

Chapter 4 and Chapter 5 present two different direct torque control schemes for

integrated starter/alternator with space vector modulation. They have one-PI and two-PI

structures. Their controllers are analyzed and the design procedures are developed.

Chapter 6 analyzes the non-linear characteristics of the inverter and develops their

compensations. The compensation methods will be used in the control schemes in the

following chapters.

Chapter 7 presents a sliding observer to estimate the stator flux linkage based on the

motor current model. Compared to the open-loop estimator, the observer has exhibited

better dynamic behaviour, disturbance resistance and high accuracy estimation ability.

The experimental results show that the proposed observer is able to deliver more

accurate estimation than open-loop integrator estimator both in the steady state and

during transients.

Chapter 8 investigates an efficiency improvement method of the ISA with power factor

controller. The modeling and experimental results shows the efficiency of the induction

machine is improved with proposed method.

Chapter 9 gives the conclusions and suggestions for future research.

Chapter 2 An induction machine based ISA using RFOC with SVM 22

CHAPTER 2

AN INDUCTION MACHINE BASED

INTEGRATED STARTER/ALTERNATOR USING

ROTOR FIELD ORIENTED CONTROL WITH

SPACE VECTOR MODULATION

2.1 Introduction

As discussed in 1.2.4.1, rotor flux oriented control scheme has been used recently in a

few ISA designs [16, 20, 24-26, 38]. In order to compare the direct torque control with

rotor flux oriented control, the chapter presents a study of rotor flux oriented controlled

ISA. The structure of the rotor flux oriented controlled ISA is presented first, followed

by experimental results of an implemented ISA. Subsequent chapters present direct

torque controlled ISA solutions.

This chapter is organized as follows. Section 2.2 presents the dynamic model of the

induction machine. Section 2.3 proposes the rotor flux oriented controller for ISA.

Experimental results are shown in 2.4. At last, conclusion is drawn in Section 2.6.

2.2 Induction machine model

In the synchronously rotating reference frame ( e ed q− ), the dynamic of the induction

machine can be expressed as


( )

( )

ssd s sd e sq

sqsq s sq e sd

rdrd r rd e r rq

rqrq r rq e r rd

dv R idt

dv R i

dtdv R i

dtd

v R idt

αψ⎧ = + − ω ψ⎪⎪

ψ⎪ = + + ω ψ⎪⎨ ψ⎪ = + − ω −ω ψ⎪⎪ ψ⎪ = + + ω −ω ψ⎩

(2-1)

where

sdv and sqv are the stator voltages;

eω and rω are synchronous and rotor rotating frequency;

sdi , sqi , rdi and rqi are stator and rotor currents in d- and q-axis;

sdψ , sqψ , rdψ and rqψ are the stator and rotor flux linkages in d- and q-axis;

sR and rR are the stator and rotor resistances.

The dynamic equivalent circuits of the induction machine are shown in Fig. 2.1.

According to Fig. 2.1, the flux linkage can be expressed in term of the currents as

follows:

sd s sd m rd

sq s sq m rq

rd m sd r rd

rq m sq r rq

L i L iL i L i

L i L iL i L i

ψ = +⎧⎪ψ = +⎪⎨ψ = +⎪⎪ψ = +⎩

(2-2)

where sL , rL and mL are the stator self, rotor self and mutual inductances, respectively.


sqi rqi

sR

sqψmL

rR

ls s mL L L= −( )e r rdω ω ψ−

lr r mL L L= −

rqψ

e sdωψ

sqV rqV

(a)

sdi rdi

sR

sdψmL

rR

ls s mL L L= −( )e r rqω ω ψ−

lr r mL L L= −

rdψ

e sqωψ

sdV rdV

(b)

Fig. 2.1 Dynamic e ed q− equivalent circuits of machine (a) eq axis circuit, (b) ed axis circuit

2.3 Rotor flux oriented controlled ISA

The structure of ISA with rotor flux oriented control is shown in Fig. 2.2. There are two

control loops in this structure.

The outer-loop determines the torque and flux references for the inner-loop. In the

starting mode, the torque reference is the pre-determined starting torque startingT . In the

generating mode, torque reference is connected to the output of the dc bus voltage

controller by the Staring/Generating switch when the engine is started. A negative gain

is used for the dc bus voltage controller because the torque should be negative in

generation state. The flux reference is obtained from the output of the flux reference

block in Fig. 2.2 for both starting and generating modes. The flux reference is weakened


in proportional to 1 rω when the rotor speed is above the base speed of the induction

machine.

The two inner-loops implement the effective control of the torque and flux of the

induction machine for both starting and generating modes. The torque and flux are

independently regulated by the decoupled d and q axis current controllers. The q-axis

current reference is calculated by (2-3) and the d-axis current reference is the output of

the rotor flux PI controller to maintain the rotor flux level according to the flux

reference block. In this scheme, the angle and the amplitude of the rotor flux is

estimated by a conventional current mode estimator as shown in Fig. 2.3. The speed

signal and stator current are used as the inputs of the flux estimator. The outputs of the d

and q axis current controllers are the voltage references in the rotating frame ( e ed q− ),

which are transformed to the stationary frame (α −β ) by e ed q− to α−β block by (2-

4). Finally, the PWM signal is generated by the SVM unit according the voltage

reference in the stationary frame (α−β ).

r∠Ψ

−

Current Model Estimator

SVM IMPI

rω

PI

Encorder

sqV

sdV

e ed q−

α β−

,V Vα βPI

*sdi

*sqi

sqisdi

TK−

rΨ

dcV

*T

*dcV

startingT +

−

+

r∗Ψ

rω

bω

Fig. 2.2 Rotor flux oriented controlled ISA with SVM

2 23

r esq

m r est

L Tip L ψ

∗∗ = (2-3)


rω

sAisBisCi

si α

si β

11 rT s+

sdi

sqirje− θ

rθrT

mL

∫rθ

rψ

÷ +

Fig. 2.3 flux model in the rotor-flux-oriented reference frame [74]

de e

qe e

V Vcos sinV Vsin cosα

β

θ − θ⎡ ⎤ ⎡ ⎤⎡ ⎤=⎢ ⎥ ⎢ ⎥⎢ ⎥θ θ⎣ ⎦⎣ ⎦ ⎣ ⎦

(2-4)

where eθ is the angle between the rotor flux frame ( e ed q− ) and stationary frame

(α −β ), i.e. the angle of rotor flux linkage vector as shown in Fig. 2.4

rΨ

eθ α

β

eq edsI

sdisqi

eω

Fig. 2.4 Vector diagram of the induction machine

The ISA using a rotor flux oriented induction machine was extensively studied by

mathematical modeling and experiments. The simulation results were fully in agreement

with experimental results. In view of the fact that induction machine based ISA using

RFOC has been developed and tested by other researchers [16, 20, 24-26, 38, 60],

simulation results are not being included here. Only the experimental results of a fully

tuned RFOC for an ISA system are included in this chapter. These results are presented


in this chapter to serve as benchmarks for ISA using DTC which are described in

Chapters 3-8.

2.4 Experimental setup

The ISA using a rotor flux oriented induction machine, as shown in Fig. 2.2, was

implemented in the laboratory.

As mentioned in Chapter 1, a squirrel-cage induction machine was chosen in this study

to demonstrate the proposed control scheme of the ISA. The induction machine is

rewound with a 22 V 4-poles winding to work with the 42 V dc voltage bus. The

parameters of the induction machine are given in Table 2.1.

Table 2.1 Parameters of the Induction Machine

Rated output power(W) 1000

Rated Voltage (Volt) 22

Rated frequency (Hz) 50

Poles number p 4

Stator resistance sR ( mΩ ) 25.1

Rotor resistance rR ( mΩ ) 18.2

Mutual inductance (mH) 1.8

Stator leakage inductance lsL (mH) 0.07618

Rotor leakage inductance lrL (mH) 0.07618

Inertia (Kg.m²) 0.00824

The rated line to line voltage of the induction machine is chosen as 22 V (rms) by

considering (2-5) in a boost type three-phase PWM converter.

2 3 2 30.7855 42 25.7 V2 2l l dcV m V

π π−⎛ ⎞ ⎛ ⎞≤ ⋅ ⋅ = × × × =⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠

(2-5)

where m (=0.7855) is the modulation factor with sinusoidal PWM [75], dcV is the dc

bus voltage (with 42 V rating).

An ISA system based on the induction machine was implemented with the experimental

platform shown in Fig. 2.5. A DC machine is mechanically coupled with the induction


machine to simulate the engine during both starting and generation modes of operation.

The power converter for the experiment is a three-phase DC-AC voltage source

bidirectional converter, which is supplied with three 12 V batteries in series giving 36

V.

Fig. 2.5 Experimental setup

The control software is developed on a dSPACE DS1104 Controller Board with slave a

built-in Digital Signal Processor (TMS320F240). The rotor position and speed were

obtained from an incremental encoder with 5000 pulses per revolution. Voltage and

current sensors are used to detect the dc bus voltage and stator currents of the induction

machine, respectively. The feedback signals of dc bus voltage and two stator currents

are read by the DSP through A/D (Analog to Digital) converter. The control algorithm

is embedded in the DS1104 from dSPACE and the PWM (SVPWM) that gates the

power converter is generated through the slave DSP.

2.5 Experimental results

2.5.1 Starting mode

During the starting period, the induction machine produces full torque to drive the DC

machine which simulates the engine. In this experimental setup, the starting torque is set

as 6 Nm and the engine starting speed is 500 rpm to reflect the idle speed of the engine.


After DC machine simulated engine is started, both the DC machine and induction

machine are accelerated from 500 rpm (see A in (a) of Fig. 2.6) to 1200 rpm (see B in

(b) Fig. 2.6). For the study in this thesis, 1200 rpm was chosen as the simulated ISA

generating speed because of the limitation of the DC machine simulating the engine.

As shown in Fig. 2.6, the induction machine torque runs the whole set from 0 to 1200

rpm. After the speed reaches 1200 rpm, the DC machine simulated engine sets its speed

reference to 1500 rpm and regulated by its own controller. In this study, 1500 rpm is the

base speed of the induction machine. This speed is consequence of the 4-poles induction

machine chosen for the ISA. For a practical ISA, 10-12 poles machine should be more

appropriate. At the same time, the reference of the induction machine is switched from

torque to voltage to reflect the transition from motoring to generating. The induction

machine begins to act as a generator and provide power to the battery and the dc load.

The torque of the induction machine is thus changed from 6 Nm to -6 Nm as (i) in Fig.

2.6. The rotor flux [(iii) in Fig. 2.6] of the machine is kept constant by the rotor flux

control in Fig. 2.2. The decouple control of the torque and rotor flux is achieved by

rotor flux oriented control scheme. The part (b) of Fig. 2.6 shows the d and q axis

currents during starting and generating period. As discussed in Section 2.3, the d and q

axis currents control the flux and torque of the induction machine, respectively.

Therefore, the d axis current is fixed to maintain the rotor flux, and q axis current is

regulated according to the operation modes of the machine.


(a)

(b)

Fig. 2.6 Starting of ISA with RFOC: (a) torque, speed and stator flux (b) torque edi and e

qi


2.5.2 Generating mode - steady state

The steady state performance with full load of the induction machine is shown in Fig.

2.7. With voltage regulator the dc voltage of the ISA system is kept as 42 V when the

induction machine provides full torque to the load. And the stator flux of the induction

machine is still constant. The stator current waveform is captured by a digital

oscilloscope (LeCroy LT364L) and the data is analysed by FFT algorithm. It indicates

in Fig. 2.8 that the DC-AC converter of the ISA system runs at constant frequency 6.67

kHz, which is corresponding to the sampling time 150 sμ .

Fig. 2.7 ISA generating with full load


Fig. 2.8 Spectrum analysis of the stator current of ISA while operating as generator in the

steady-state

2.5.3 Generating mode - dynamic response.

The dynamic performance of the ISA also studied in this section. The performance of

the ISA system under load dumping and engine speed acceleration or deceleration is

presented as follows.

2.5.3.1 Performance during load dump

The dc load of the ISA is removed suddenly at the generating state when the speed is

1500 rpm. Two conditions are considered for the load dump of the ISA. They are load

dump without battery connected and load dump with battery connected as shown in Fig.

2.9 and Fig. 2.10, respectively

As shown in Fig. 2.9, the peak dc bus voltage of the ISA is well controlled below the

limitation of the 42 V PowerNet standard (58 V) [6] when the dc load is dumped. The

settling time of the dc bus voltage is about 150 ms. The induction machine’s torque is


changed from -5 Nm to about -1 Nm during load dumping. The edi is constant in Fig.

2.9 (b) to maintain the rotor flux. Consequently, the stator flux of in Fig. 2.9 (a) is kept

constant.

(a)


(b)

Fig. 2.9 Load dump of ISA without battery connected: (a) bus voltage, torque, stator flux and

stator current (b) torque, edi and e

qi

As shown in Fig. 2.10, the dc bus voltage of the ISA is also below the limitation of the

42 V PowerNet standard (58 V) [6] when the dc load is dumped. The torque of the

induction machine varies slower than last case because of the charging of the batteries.


(a)

(b)

Fig. 2.10 Load dump of ISA with battery connected: (a) bus voltage, torque, stator flux and

stator current (b) torque, edi and e

qi


2.5.3.2 Performance acceleration/deceleration

In normal operation, the engine speed may change quiet rapidly and frequently. The

ISA should cope with this and maintain the dc bus voltage to the 42 V PowerNet

specifications.

In this test, the DC machine’s speed reference is increased suddenly from 1500 rpm to

3000 rpm, while the induction machine is generating with full dc load. As shown in Fig.

2.11, the dc bus voltage of the ISA is well controlled as 42 V during speed acceleration.

Fig. 2.11 (a)-(iii) shows the flux of the machine is weakened when the speed is above

base speed (1500 rpm). Part (b) of Fig. 2.11 shows the d and q axis currents during

accelerating of the ISA.

(a)


(b)

Fig. 2.11 ISA performance at acceleration: (a) bus voltage, speed, stator flux and stator current

(b) speed, edi and e

qi

The deceleration of the ISA is also tested by dropping the speed suddenly from 3000

rpm to 1500 rpm, while the induction machine is generating with full dc load. As shown

in Fig. 2.12, the dc bus voltage of the ISA varies a little within the limitation of 42 V

specifications [6] during speed deceleration. Fig. 2.12 (a)-(iii) shows the flux of the

machine is increased when the speed returns to base speed (1500 rpm). Part (b) of Fig.

2.12 shows the d and q axis currents during decelerating of the ISA.


(a)

(b)

Fig. 2.12 ISA performance at deceleration: (a) bus voltage, speed, rotor flux and stator current

(b) speed, edi and e

qi


2.5.4 High speed operation

The operation of proposed ISA system in high speed range is also tested. When the

speed of the induction machine exceeds the base speed (1500 rpm), the stator flux

reference is weaken by the inverse proportional with the rotor speed. Fig. 2.13 shows

the ISA performance at 4000 rpm with full load. The induction machine’s torque is less

than 6 Nm due to the high speed operation. The flux of the induction machine is

reduced for field weakening.

In this thesis, it is possible to run the ISA only up to 4000 rpm due to limitation of the

experimental setup in the laboratory.

Fig. 2.13 ISA with field weakening at high speed

2.6 Conclusion

This chapter presents a rotor flux oriented control scheme of the integrated

starter/alternator. Extensive experimental results show its effectiveness in ISA

application. However, the current decoupling and co-ordinate transformation make the


control structure quite complex. Due to existence of current control loop, at least three

PI controllers have to be used for the torque and flux control of the induction machine.

In practice, these PI controllers’ gains are not easy to design and tune. Moreover, the

rotor flux estimation is sensitive to the variation of the induction machine’s parameters,

especially the rotor resistance. A mechanical speed sensor is also necessary for the

torque and flux control. This sensor requirement is a major disadvantage of the RROC

based ISA.

Because the above limitations of the flux oriented control, direct torque controlled ISA

is proposed in this thesis. Three different control structures based on direct torque

control concept are discussed in the following chapters under same conditions.

Subsequent chapters deal with a direct torque controlled ISA, starting with the simple

switching-table based DTC described in Chapter 3.

Chapter 3 Analysis and implementation of the classical direct torque controlled ISA 41

CHAPTER 3

CLASSICAL DIRECT TORQUE CONTROLLED

INTEGRATED STARTER/ALTERNATOR

3.1 Introduction

As stated in Chapter 1, classical direct torque control (DTC) for induction motors was

first introduced in 1980’s [66, 67]. Classical direct torque control is a very simple

control scheme with low computational requirement. A switching table is adopted to

select one of eight basic voltage space vectors determined by the torque and flux errors

and position of the stator flux vector. This classical direct torque control is a DTC with

a Switching-Table (DTC-ST). The torque and flux are estimated by a voltage mode

estimator in the stationary frame. Only stator resistance is involved in the calculation

and no axis transformation is required for DTC-ST. In addition, there is no rotor

velocity or position sensor required for the torque and flux control.

Since late 1980’s, DTC-ST has gained wide acceptance in motor drives application

from both academia [63, 76-80] and industry [73]. DTC-ST controlled generators has

attracted research interests in aircraft application [71], grid application [68, 69], wind

power generation [70] and ISA application [72] as well.

This chapter describes the principle of the classical direct torque control and the

classical direct torque controlled scheme for ISA. Both simulation and experimental

results are provided to confirm the feasibility of DTC-ST for ISA operation of the

induction machine. This chapter is organized as follows. Section 3.2 presents the

principle of classical direct torque control. The ISA control scheme with classic DTC is

developed in Section 3.3. Simulation results of ISA are given in Section 3.4. Section 3.5

provides the experimental results. Due to the limitation of hardware, only steady state of


the classic DTC controlled induction generator with 150 sμ (sampling time) is tested in

this chapter.

3.2 Classical direct torque control principle

In stationary frame, the dynamic behaviour of induction machine can be described as

following equations:

ss s s

dV R IdtΨ

= + (3-1)

0 rr r m r

dR I jdtΨ

= + − ω Ψ (3-2)

s s s m r

r m s r r

L I L I

L I L I

⎧Ψ = +⎪⎨Ψ = +⎪⎩

(3-3)

( )3 3

2 232

m me r s r s s r

s r s r

mr s

s r

L LT P P sinL L L LLP sinL L

= Ψ ×Ψ == Ψ Ψ θ − θσ σ

= Ψ Ψ γσ

(3-4)

where

2

1 m

s r

LL L

σ = − (3-5)

where sR and rR are the stator and rotor resistances, sL , rL and mL are the stator, rotor

and mutual inductances, respectively. And mω is rotor speed, P is the number of pole

pairs, sθ and rθ are the angles of stator and rotor flux vectors, respectively, and γ

(equal to s rθ − θ ) is the angle between the stator and rotor flux vectors.

The rotor flux vector changes slowly compared to the stator flux vector with a large

time constant. So it can be assumed to be constant. The stator flux vector can be

changed by applying proper stator voltage. Therefore, the torque can be rapidly changed

by varying γ in the required direction which is determined by the required torque

reference. This is the basic idea of the classic direct torque control scheme. With voltage

source inverter, the angle γ can be easily changed by producing appropriate stator


voltage space vectors according to (3-1). The voltage vector can be selected from eight

basic vectors including six non-zero active voltage ( )1 6V V→ vectors and two zero

voltage vectors ( )0 7,V V as shown in Fig. 3.1.

1AS =

0CS =

1CS =

0BS =0AS =

1BS =

2V3V

4V

5V 6V

0V7V

1Vα

β

Fig. 3.1 Eight switching states and the voltage space vectors

By applying different voltage vectors, the stator flux vector will move forward or

backward as indicated in Fig. 3.2.

1V

2V3V

4V

5V 6V

sψ

rψ

α

βsω

γ

Fig. 3.2 Movement of stator flux vector by selection different voltage space vectors

The structure of classical direct torque control for voltage-source inverter-fed induction

machine is shown in Fig. 3.3. Proper voltage vectors are selected from the switching

table by considering different output states of the torque and flux comparators and the

sectors where the stator flux vectors are located. The optimum switching table is shown


in Table 3.1, which is referred to [74, 75]. The torque and flux are independently

controlled by the torque and flux hysteresis comparators, respectively.

sθ∠

eT

eT ∗

sψ ∗

sψ

dcV

Fig. 3.3 Structure of classical direct torque control

Table 3.1 Switching table of inverter vectors

dψ edT Sector 1 Sector 2 Sector 3 Sector 4 Sector 5 Sector 6

1 V2 V3 V4 V5 V6 V1

0 V7 V0 V7 V0 V7 V0 1

1− V6 V1 V2 V3 V4 V5

1 V3 V4 V5 V6 V1 V2

0 V0 V7 V0 V7 V0 V7 0

1− V5 V6 V1 V2 V3 V4

The output of flux hysteresis comparator with two-level is dψ

1

0

s ref s

s ref s

d if

d if

⎧ ψ = ψ ≤ ψ − Δψ⎪⎨

ψ = ψ ≥ ψ + Δψ⎪⎩ (3-6)

where sΔψ is the error band of the flux comparator.

The output of torque hysteresis comparator with three-level is edT


1

0

1

0

e e ref e

e e ref

e e ref e

e e ref

dT if T T Tanticlockwise rotation :

dT if T T

dT if T T Tclockwise rotation :

dT if T T

⎧ ⎧ = ≤ − Δ⎪⎪ ⎨⎪ = ≥⎪⎩⎪⎪⎨⎪ ⎧ = ≥ − Δ⎪ ⎪

⎨⎪ = ≤⎪⎪ ⎩⎩

(3-7)

where eTΔ is the error band of the torque comparator.

3.3 ISA with classical DTC

γ

β

αMotoring State

sψ

rψ

γ

β

αGenerating State

sψ

rψ

Fig. 3.4 stator and rotor flux vector at motoring and generating states

The idea of the direct torque control can be extended from motoring mode to generation

mode with same control structure. The only difference in generation mode is that the

torque reference is negative and the stator flux vector lags to rotor flux vector as shown

in Fig. 3.4.

Based on above analysis, a complete scheme of classic direct torque controlled ISA is

developed and it is indicated in Fig. 3.5. It includes starting/generating state switch

which simulates the operation of ISA from stating mode to generating mode. During

starting mode, the induction acts as a motor to provide high torque for the starting of the

engine. During generating mode, the torque reference is switched to the output of

voltage controller to maintain the dc bus voltage with negative torque. As shown in Fig.

3.5, the dc load of ISA is connected at dc side of the DC-AC converter with the battery.

The converter of the induction machine supplies active power to the dc load during


generation state while it provides reactive power to the machine. In this scheme, one

voltage sensor for dc bus voltage and two current sensors for the stator current of the

induction machine are used for the controller.

dcV

*T

eT

sψ ∗

*dcV

sψ

startingT +

−

−

+

sθ∠

dcV

−

Fig. 3.5 Classic DTC scheme for ISA

The stator flux vector is estimated in the stationary frame avoiding co-ordination

transformation and involvement of more machine parameters. The estimation algorithm

is given in (3-8)

( )

32

s s s s

e s s

V R I dt

T P I

⎧Ψ = −⎪⎨

= Ψ ×⎪⎩

∫ (3-8)

3.4 Simulation results

The proposed scheme has been modeled with Matlab/Simulink in order to evaluate its

performance. The model of the induction machine in Simulink is modified to include

the engine speed as an input variable. The simulation results present in this chapter is

for 1.0 kW/22 V induction machine supplied by voltage source inverter with 42 V dc

bus. The parameters of the induction machine are shown in Table 2.1. The 42 V

batteries also modeled to provide dc voltage for starting. In the simulation, it is assumed

that the engine starts at 1200 rpm. After starting, the speed of induction machine is


determined by the engine. In order to illustrate the proposed scheme for the generator,

the simulation has been carried out under the following conditions.

3.4.1 Starting mode


machine, which simulates the engine. The starting torque is set as 6 Nm and the engine

starting speed is 1200 rpm. As shown in Fig. 3.6, the full induction machine’s torque

run the whole set from 0 to 1200 rpm. After the speed reaches 1200 rpm, the engine

speed in the model is set as 1500 rpm. At the same time, the reference of the induction

machine is switched from torque to the output of the voltage regulator. The induction

machine now begins to act as a generator and provide power to the battery and the dc

load. The stator flux of the induction machine is kept as constant with proposed direct

torque control method.

(a) Ts =150 sμ


(b) Ts =50 sμ

Fig. 3.6 Starting process of ISA (a) Ts =150 sμ (b) Ts =50 sμ

Two sampling times are used for the modeling. As shown in Fig. 3.6, the torque and

flux ripples are much less with sampling time as 50 sμ than that of the case with

sampling time as 150 sμ .

3.4.2 Generating mode- steady state



induction machine provides full torque to the load. Moreover, the stator flux of the

induction machine is kept constant within the error band. Similarly, small torque and

flux ripples are obtained with shorter sampling time (50 sμ ).


(a) Ts =150 sμ

(b) Ts =50 sμ

Fig. 3.7 ISA generating with full load (a) Ts =150 sμ (b) Ts =50 sμ


The spectrum analysis diagrams of the stator current Fig. 3.8 show the switching

frequency is variable with DTC-ST. in addition, the harmonic components is lower with

shorter sampling time (50 sμ ).

(a) Ts =150 sμ (b) Ts =50 sμ

Fig. 3.8 Spectrum analysis of the stator current with FFT (a) Ts =150 sμ (b) Ts =50 sμ

3.4.3 Generating mode - dynamic response

The dynamic performance of the ISA is also studied in this section. The performances

of the ISA system under load dump and engine speed acceleration or deceleration are

presented below.

3.4.3.1 Performance during Load dump


1500 rpm. As shown in Fig. 3.9, the dc bus voltage of the ISA is almost fixed at 42 V

when the dc load is dumping no matter the sampling time is 150 sμ or 50 sμ .

Certainly, high sampling frequency is preferred for lower torque and flux ripples. The

induction machine’s torque is changed from -6 Nm to about 0 Nm during load dumping.


(a) Ts =150 sμ

(b) Ts =50 sμ

Fig. 3.9 Load dumping performance of ISA (a) Ts =150 sμ (b) Ts =50 sμ


3.4.3.2 Dynamic performance during acceleration/deceleration

In this section, the DC machine’s speed reference is increased suddenly from 1500 rpm

to 2500 rpm and back to simulate the rapid change of the engine speed. During these

acceleration and deceleration, the induction machine is generating with full dc load. As

shown in Fig. 3.10, the dc bus voltage of the ISA is kept as 42 V whenever the

simulated engine is accelerated or decelerated. The stator flux of the induction machine

is reduced when the rotor speed is higher than the base speed (1500 rpm).

(a)


(b)

Fig. 3.10 ISA performance at speed ramp (Ts =50 sμ ) (a) accelerating (b) deceleration


Voltage mode stator flux estimator is used in the experiment. Due to the noise or

measurement error inherently present in the current sensor, the pure integrator in (3-8)

can to be saturated. Therefore, a low pass filter is used instead for the flux estimation.

( )11s s s s

c

V R Is T

Ψ = −+

(3-9)

Where ( )1 2c cT f= π and cf is the cut-off frequency of the filter.

3.5.1 DTC-ST with constant switching frequency

The hysteresis comparator based classic DTC has the disadvantages of variable

switching frequency. With a digital signal processor (DSP), the switching frequency can

be fixed with discrete hysteresis comparator as shown in Fig. 3.11. The discrete


hysteresis comparator is different with analog hysteresis comparator by using a fixed

sampling time sT for the output of the comparator. Moreover, the output voltage vector

selected from the Switching-Table is applied to the induction machine with equal

switching period. Therefore, the switching frequency is constant. The discrete hysteresis

comparator will operate like an analog hysteresis with low enough smaller sampling

time. However, it requires a fast DSP.

2e

eTT ∗ Δ

+

2e

eTT ∗ Δ

−1t 3t2t

eT ∗

2e

eTT ∗ Δ

+

2e

eTT ∗ Δ

−eT ∗

/S H

sT

sTsTsT

(a) (b)

Fig. 3.11 Analog (a) and discrete (b) hysteresis comparator [63]

In the experimental system, DSP slave processor TMS320F240 is used. It cannot

implement the control algorithm with low sampling time (50 sμ ) as mentioned in

simulation. Therefore, only the steady state experimental results with Classic DTC

based induction generator is given in this chapter with sampling time 150 sμ .

3.5.2 Generating mode- steady state

It shows in Fig. 3.12 that the torque of the induction machine is negative with large

ripples. The power generated by the induction machine transfers through the converter

to charge the batteries. The locus of the stator flux vector was shown in Fig. 3.13, which

indicates the vector is moving along a circle with an error band.


Fig. 3.12 ISA generating with DTC-ST

Fig. 3.13 Stator flux vector diagram


3.6 Conclusion

The classic direct torque controlled induction generator for integrated starter alternator

application has been analyzed and verified with simulation and experiments. The results

show that the direct torque control concept had been successfully extended to the

control of induction generator for an ISA. Although the torque and flux ripples are

rather large, their mean values are same as the RFOC based ISA for similar operating

conditions.

A discrete method was implemented to keep the switching frequency of the inverter

constant. High flux and torque ripples results from look-up table of the voltage vectors

and the hysteresis comparators of the torque and flux. Therefore, short sampling time

(as low as 25 sμ ) of the control system should be used [73].

The drawbacks of high torque and flux ripples of the classical DTC can be reduced by

using a voltage pulse width modulator instead of the switching table [81-91]. The DTC

strategies operating at constant switching frequency can be implemented by means of PI

controlled closed-loop schemes. The controllers calculate the required stator voltage

vector, averaged over a sampling period. The voltage vector is finally synthesized by a

PWM technique, which in most cases is the space-vector modulation (SVM).

The improved DTC schemes with SVM for ISA application have been proposed during

this study [29, 55]. One of improved DTC schemes with direct flux vector control is

presented in the next chapter.

Chapter 4 Direct flux vector controlled ISA with space vector modulation 57

CHAPTER 4

DIRECT FLUX VECTOR CONTROLLED

INTEGRATED STARTER/ALTERNATOR WITH


4.1 Introduction

Since direct torque control (DTC) was introduced in 1980’s, many schemes were

proposed to overcome the problems associated with the basic DTC [66, 67]: operation

with variable switching frequency and large torque ripple, due to the hysteresis control

and the switching table method. The variable switching frequency problem can be

addressed by Proportional-Integral (PI) controllers plus PWM instead of hysteresis

controller and the torque ripple can be reduced with space vector modulation (SVM)

technique [81-91]. However, few papers present the analytical design principle of the PI

controller parameters for DTC with SVM. The PI controllers seem to be determined

mainly by trial and error.

This chapter also presents the theory of direct flux vector control (DFC) scheme for an

induction machine, which is based on the basic DTC concept. The scheme proposed in

this chapter extends the works reported in [89], in which the relationships between

controlled variables and the torque were not fully developed. This DFC scheme controls

the electromagnetic torque of the induction machine by regulating the amplitude and the

rotating speed of the flux vector with only one Proportional-Integral (PI) controller and

the required voltage vector is applied to the induction machine by space vector

modulation. The speed sensor is eliminated and the torque and stator flux is estimated

with voltage mode estimator. This DFC scheme controls the torque of induction

machine with high dynamic performance. This thesis is concerned with the dc bus


voltage control in an ISA application to meet the specification of the 42 V PowerNet

using an induction machine under DFC.

This chapter analyzes the concept of proposed scheme in detail and presents the design

principle of the PI controller parameters. Two types of PI controller design schemes are

presented with direct synthesis and robust optimization methods based on the analysis

of the inner relationships between the control variables and the torque. Modeling results

show that the dynamic performance is not sensitive to the variation of the rotor

resistance. Fixed switching frequency and low torque ripple are obtained with SVM

technique. All the algorithms are based on stationary frame, and only stator resistance is

used for calculation of the stator flux vector. Modeling and experimental results for the

proposed direct flux vector control are presented for a 1.0 kW induction machine with a

PI controller.

4.2 Direct flux vector control



ss s s

dV R IdtΨ

= + (4-1)

0 rr r m r

dR I jdtΨ

= + − ω Ψ (4-2)

s s s m r

r m s r r

L I L I

L I L I

⎧Ψ = +⎪⎨Ψ = +⎪⎩

(4-3)

32

me r s

s r

LT PL L

= Ψ ×Ψσ

(4-4)

where

2

1 m

s r

LL L

σ = − (4-5)

where sR and rR are the stator and rotor resistances, sL , rL and mL are the stator self,

rotor self and mutual inductances, respectively. And mω is rotor speed, P is the number

of pole pairs.


The relationship between stator and rotor flux vectors sΨ and rΨ respectively is

derived from (4-2) and (4-3)

r r m rs m r

s r r

d R L R( j )dt L L LΨ

= Ψ + ω − Ψσ σ

(4-6)

By using Laplace transform of (4-6) and assuming the rotor speed mω is changing

slowly, the relationship between stator and rotor flux vectors sΨ and rΨ in the

frequency domain can be obtained

1

m

sr s

r rm

r r

LL( s ) ( s )

L Ls jR R

Ψ = Ψ⎛ ⎞

σ + − ω σ⎜ ⎟⎝ ⎠

(4-7)

(Please refer to Appendix B for further details of the derivation included in this chapter)

Assuming that s sj j t* *s s se eθ ωΨ = Ψ = Ψ and the amplitude of sΨ is kept constant, and

that sΨ rotates at an angular speed sω ,

1 *s s

s

( s )s j

Ψ = Ψ− ω

(4-8)

By substituting (4-8) into (4-7) and taking inverse Laplace transform

1 1

1

m

*sr s

r r sm

r r

L

L( t )

L L s js j

R R

−Ψ = Ψ− ω

σ + − ω σ

⎧ ⎫⎪ ⎪⎪ ⎪⎨ ⎬

⎛ ⎞⎪ ⎪⎜ ⎟⎪ ⎪⎩ ⎭⎝ ⎠

L (4-9)

Thus

( )( )

( )( )( )( )

21 2

21

1 1

mr

s

s

tt

e cos te s m( t )

s my* j tan tane s mx

LL

−τ−+ − ω − ωτ

Ψ =

+ τ ω − ω

− −− τ ω − ω× Ψ

⎛ ⎞⎜ ⎟⎜ ⎟⎝ ⎠

⎛ ⎞⎛ ⎞⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠

(4-10)


where

r

rt

s m

t

s m

LR

x cos( t ) cos( t )

y sin( t ) sin( t )

ee

−τ

−τ

⎧τ = σ⎪⎪⎪ = ω − ω⎨⎪⎪ = ω − ω⎪⎩

(4-11)

With small slip, (4-10) can be simplified as

( )( )1 11

tm s mr ss

* j( t ) eyL tan tane xL

− − −τΨ ≈ × Ψ

⎡ ⎤⎛ ⎞⎛ ⎞ − τ ω − ω⎜ ⎟− ⎢ ⎥⎜ ⎟ ⎝ ⎠⎣ ⎦⎝ ⎠ (4-12)

It shows that the rotor flux vector tracks stator flux vector in its amplitude and rotating

speed with a time constant, given by τ . Once the stator flux is built up and kept

constant, the rotor flux will also be kept constant. Therefore, the amplitude of the rotor

flux can be considered as fixed after establishing of the stator flux. Equation (4-12) can

be further simplified as

( )( )1 1m s mr ss

* j( t ) eyL tan tanxL

− −Ψ ≈ Ψ

⎡ ⎤⎛ ⎞ − τ ω − ω⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦ (4-13)

From (B-4), the torque can be expressed as

32

sj t*me r s

s r

LT ( t ) P ( t )L L e ω= Ψ × Ψ

σ (4-14)

By substituting (4-13) into (4-14), we obtain

( )( )

2

1 1

32

*m me s

s r s

s s m

L LT ( t ) PL L L

ysin t tan tanx

− −

⎧ ⎫= Ψ⎨ ⎬σ⎩ ⎭⎧ ⎫⎡ ⎤⎛ ⎞× ω − − τ ω − ω⎨ ⎬⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦⎩ ⎭

(4-15)

where


r

rt

s m

t

s m

LR

x cos( t ) cos( t )

y sin( t ) sin( t )

ee

−τ

−τ

⎧τ = σ⎪⎪⎪ = ω − ω⎨⎪⎪ = ω − ω⎪⎩

(4-16)

It clear that the dynamic response of torque is determined by the amplitude and rotating

speed of the stator flux vector with the non-linear relationship of (4-15). The torque of

the induction machine can be regulated by controlling rotating speed of the stator flux

vector sΨ as long as its amplitude is kept constant. As rotor flux vector tracks the stator

flux vector, its amplitude is also kept constant after establishing of constant stator flux

amplitude. In addition, the sin or tan computation results of a small angle is very close

to the angle by itself (in rad) as shown in (4-17). Therefore, the above torque expression

can be simplified to (4-18) in which the slip is small.

( ) ( ) ( ) ( )sin tan smallθ ≈ θ ≈ θ θ (4-17)

By considering (4-15) and (4-16) at same time, torque expression can be further

simplified as

( )2

2

2

31

2

t*m

e s s mr s

LT ( t ) P

R L e −τ≈ Ψ − ω − ω

⎧ ⎫⎛ ⎞⎨ ⎬ ⎜ ⎟

⎝ ⎠⎩ ⎭ (4-18)

Therefore

( )1t

e s mT ( t ) K e −τ= − ω − ω⎛ ⎞

⎜ ⎟⎝ ⎠

(4-19)

where

2 2

2

32

*ms

r s

r

r

LK PR L

LR

⎧= Ψ⎪⎪

⎨⎪τ = σ⎪⎩

(4-20)

By Laplace transform of (4-19), we have

{ }11e s mT ( s ) K

s⎛ ⎞= ω − ω⎜ ⎟τ +⎝ ⎠

L (4-21)


where { }s mω − ωL is the Laplace form of { }s mω − ω

The transfer function of the torque loop with input as { }s mω − ω can be written as

{ } 1

ep

s m

T ( s ) KG ( s )s

= =ω − ω τ +L

(4-22)

Equation (4-22) shows that the relationship between eT and sω is equivalent to a first

order system with a disturbance mω . The equivalent system block is shown as follows:

( )s sω ( )eT s

1Ksτ +

( )m sω−

Fig. 4.1 Equivalent system model of the torque loop

In order to achieve good performance of tracking a reference torque signal and

disturbance rejection, the PI controller of Fig.4.2 may be employed:

( )s sω ( )eT s1

Ksτ +

( )m sω−

( )cG s−

eT ∗

Fig.4.2 PI control of the equivalent system

where

p ic

K s KG ( s )

s+

= (4-23)

4.2.1 Direct flux vector control scheme

From above analysis, it is clear that a direct relationship exists between the torque and

the rotational speed of the stator flux vector when its amplitude is kept constant. This

means that it is possible to control the machine torque by directly controlling the


amplitude and rotating speed of stator flux vector. This is the basic idea of direct flux

vector control for induction machine. A complete scheme of direct flux vector control

that allows effective torque control has been developed and it is indicated in Fig.4.3.



is given in (3-8)

( )

32

s s s s

e s s

V R I dt

T P I

⎧Ψ = −⎪⎨

= Ψ ×⎪⎩

∫ (4-24)

The above scheme uses only one PI torque regulator to control the rotating speed of

stator flux vector. The desired amplitude and angle of the stator flux vector is given by

* *s s

s s

s s s

T∗

∗

⎧⎪⎪⎨⎪⎪⎩

ψ = Ψ

Δθ = ω Δθ = θ + Δθ

(4-25)

where TΔ is the sampling time, sΔθ is the increased angle of stator flux vector during

sampling period and sθ is the current angle of the stator flux vector. The reference

stator flux reference vector is compared with the estimated flux to obtain error flux

vector sΔΨ . With given sΔΨ , the exact stator voltage vector that changes the rotating

speed of stator flux vector to generate required torque while keeping its amplitude

constant is given by

sref s sV R I

TΔΨ

= +Δ

(4-26)

The space vector modulation method is used to apply the required stator voltage vector

with fixed switching frequency. In transient state, the reference voltage will be larger

than the available inverter voltage when the torque error is too large. In that case, the

speed s∗ω has to be limited to ensure that the reference voltage is lower or equal to the

maximum inverter voltage:

ref maxV V≤ (4-27)


where maxV is the maximum available inverter voltage. For under-modulation of SVM,

maxV equals to 1

3dcV , where dcV is the dc bus voltage of the inverter.

Therefore the limitation of the torque PI controller should be:

1

3dc

maxs * *

s s

VV∗ω ≤ =ψ ψ

(4-28)

*rω

*sψ

sΔΨsω∗

sθ

1TΔ

sψ

refV*

eT

eT

sψ ∗

TΔsθΔ sθ

∗ s sψ θ∗ ∗∠−

− +

&Torque Stator Flux Estimator

SVM IMPI

rω

PI

Encorder

Vdc

Fig.4.3 Direct flux vector control scheme for induction machine

4.2.2 Design of the PI controller for torque regulation

In this section, two different methods are presented for the design of PI controller in the

closed torque loop.

4.2.2.1 Direct synthesis of PI controller

Because pG ( s ) is a stable first order system, the PI can be synthesized for the desired

closed-loop transfer function. Assuming the desired close loop transfer function of

torque is

11

1G ( s )

s=λ +

(4-29)

where λ is the desired time constant.

thus


11

1 1c p

c p

G ( s )G ( s )G ( s )

s G ( s )G ( s )= =λ + +

(4-30)

1

p

i

KK

KK

τ⎧ =⎪⎪ λ⇒ ⎨⎪ =⎪ λ⎩

(4-31)

4.2.2.2 Robust PI controller

Equations (4-30) and (4-31) show that direct synthesized PI controller is very sensitive

to the parameters of the system, which are included in (4-20) and (4-21). Practically, the

rotor resistance rR may vary up to 100% of the nominal value due to the rotor heating

and the mutual inductance mL may also changes in the case of magnetic saturation. If

any parameter changes, the desired closed-loop characteristic cannot be realized by

designed PI controller parameters in (4-31). Therefore, robust control design method

based on performance index is used in this section.

The closed-loop transfer function of the system in Fig.4.2 is

2

2

1

1

c p

c p

p i

p i

G ( s )G ( s )G ( s )

G ( s )G ( s )

KK s KK

( KK ) KKs s

= =+

+τ=

++ +

τ τ

(4-32)

As presented in [92], the optimum coefficients of the performance index ITAE are

2 21 4 n ns . s+ ω + ω (4-33)

where nω is the natural frequency of the closed-loop system.

nω is selected to meet the settling time requirement. And the settling time is

4s

n

t =ζω

(4-34)

where ζ is the damping ratio.

Thus


2 2 211 4p i

n n

( KK ) KKs s s . s+

+ + = + ω + ωτ τ

(4-35)

Then

2

1 4 1np

ni

.KK

KK

ω τ −⎧ =⎪⎪⎨

ω τ⎪ =⎪⎩

(4-36)

To remove the zero of closed-loop system, a pre-filter fG ( s ) is added to 2G ( s ) . Then

the closed-loop transfer function changes to

3 2fG ( s ) G ( s )G ( s )= (4-37)

Since the desired closed-loop transfer function is

2

3 2 22 1 1 4

i

n

p i n n

KK

G ( s ) ( KK ) KK s . ss s

ωτ= =+ + ω + ω+ +τ τ

(4-38)

The pre-filter fG ( s ) is designed as

( )1

1fp i

G ( s )K K s

=+

(4-39)

With pre-filter fG ( s ) , the system change to

( )s sω ( )eT s

1Ksτ +

( )m sω−

( )cG s−

eT ∗

( )fG s

Fig.4.4 PI control of equivalent system with pre-filter

4.2.3 Design of the PI controller with control delay

Due to digital control structure, the control signal would be delayed for 1 to 1.5 times of

the sampling time. In this section, two different methods are presented for the design of

PI controller with considering the delay effect.


( )s sω ( )eT s1

dsTK esτ

−

+

( )m sω−

( )cG s−

eT ∗

Fig.4.5 PI control of equivalent torque loop

where

p ic

K s KG ( s )

s+

= (4-40)

and dT is the delay time

1

dsTp

KG ( s ) es

−=τ +

(4-41)


Because pG ( s ) is a stable first order system, the PI can be synthesized for the desired

closed-loop transfer function. Assuming the desired close loop transfer function of

torque is (as the time delay cannot be removed from the process)

11

1dsTG ( s ) e

s−=

λ + (4-42)

where λ is the desired time constant

thus

11

1 1d c psT

c p

G ( s )G ( s )G ( s ) e

s G ( s )G ( s )−= =

λ + + (4-43)

So

1

1

1 1 11 1

1dc sT

p

G ( s )G ( s ) KG ( s ) G ( s ) s es

−= =− λ + −

τ +

(4-44)

Suppose that dsTe− is approximated by a 1st order Taylor series expansion, i.e.


1dsTde T s− ≈ − (4-45)

So (4-44) is simplified as

( ) ( )

1 1 1 1 11 1

1 1

cd d d

sG ( s ) K Ks T s s T s K T ss s

τ += = =

λ + − − λ + λ +τ + τ +

(4-46)

Compared (4-46) with (4-40)

( )

( )1

pd

id

KK T

KK T

τ⎧ =⎪ λ +⎪⇒ ⎨⎪ =⎪ λ +⎩

(4-47)


Equations (4-46) and (4-47) also show that direct synthesized PI controller is very

sensitive to the parameters of the system. If any parameter changes, the desired closed-

loop characteristic cannot be realized by designed PI controller parameters in (4-47).

Therefore, robust control design method based on performance index is used in this

section.

The closed-loop transfer function of the system in Fig.4.2 is

21

1 11

d

d

p i sT

c p

p i sTc p

K s K K eG ( s )G ( s ) s sG ( s ) K s K KG ( s )G ( s ) es s

−

−

+τ += =

++ +τ +

(4-48)


( ) ( ) ( )2 2

2 2

2 2

2

111 1

1

1

d d

d

d

p i

p isT sT

p i p i dd

p i sT

p p d i i d

p i sT

p d p i d i

p i

p d

p i d

p d

K s K KK Ks K Ks sG ( s ) e eK s K K s s K Ks K K T sT s

s sK Ks K K

es s K Ks K KT s K K K KT s

K Ks K Ke

s K KT s s K Ks K KT s K K

K Ks K KK KT

K K K KTs s

K KT

− −

−

−

++τ += =

+ τ + + + −+ −τ ++

=τ + + − + −

+=τ − + + − +

+τ −

=+ −

+ +τ −

dsT

i

p d

eK KK KT

−

τ −

(4-49)

As presented in [14], the optimum coefficients of the performance index ITAE are

2 21 4 n ns . s+ ω + ω (4-50)

where nω is the natural frequency of the closed-loop system.

nω is selected to meet the settling time requirement. And the settling time is

4s

n

t =ζω

(4-51)

where ζ is the damping ratio.

Thus

2 2 211 4p i d i

n np d p d

K K K KT K Ks s s . sK KT K KT

+ −+ + = + ω + ω

τ − τ − (4-52)

Then


( )

( )

2

2 2

2 2

2

11 4

1 1 4 1 4 1 4

1 4 1 4 1

1 4 1 4 1

p i dn

p d

in

p d

p i d n p d n n p d

i n p d n

p n p d i d n

p d n i n

n d p i d n

p d n i

K K K KT.

K KT

K KK KT

K K K KT . K KT . . K KT

K K K KT

K K . K KT K KT .

K KT K K

K . KT K K KT .

K KT K K

+ −⎧= ω⎪ τ −⎪

⎨⎪ = ω⎪τ −⎩⎧ + − = ω τ − = ω τ − ω⎪⎨

= ω τ − ω⎪⎩+ ω − = ω τ −⎧⎪

⎨ω + = ω τ⎪⎩

+ ω − = ω τ −

ω + = 2

2 2

1 4 1 4 1n

n d d p n

d n i n

K . KT KT K .KT K K

⎧⎪⎨

ω τ⎪⎩+ ω − ω τ −⎡ ⎤ ⎡ ⎤ ⎡ ⎤

=⎢ ⎥ ⎢ ⎥ ⎢ ⎥ω ω τ⎣ ⎦ ⎣ ⎦ ⎣ ⎦

(4-53)

1

2 2

1 4 1 4 1p n d d n

i d n n

K K . KT KT .K KT K

−+ ω − ω τ −⎡ ⎤ ⎡ ⎤ ⎡ ⎤=⎢ ⎥ ⎢ ⎥ ⎢ ⎥ω ω τ⎣ ⎦ ⎣ ⎦ ⎣ ⎦

(4-54)

To remove the zero of closed-loop, a pre-filter fG ( s ) is added to 2G ( s ) . Then the

closed-loop transfer function changes to

3 2fG ( s ) G ( s )G ( s )= (4-55)

Since the desired closed-loop transfer function is

32 1

d

i

p d sT

p i d i

p d p d

K KK KT

G ( s ) eK K K KT K Ks sK KT K KT

−τ −==

+ −+ +

τ − τ −

(4-56)

The pre-filter fG ( s ) is designed as

( )1

1fp i

G ( s )K K s

=+

(4-57)

With pre-filter fG ( s ) , the system change to


( )s sω ( )eT s

1Ksτ +

( )m sω−

( )cG s−

eT ∗

( )fG s

Fig.4.6 PI control of equivalent system with pre-filter

Based on above analysis, the modified PI controller parameters can be obtained by

considering the delay effect. The modified PI controller parameters have been used in

the experiments.

4.2.4 Modeling results

A 1.0 kW, 22 V, 4-pole induction machine is considered to illustrate proposed direct

flux vector control scheme. The parameters of the induction machine are shown in

Table 4.1, which are same as that in Table 2.1. The whole system is modelled by

Simulink/Matlab.

Table 4.1 Parameters of the Induction Machine

Rated output power(W) 1000

Rated Voltage (Volt) 22

Rated frequency (Hz) 50

Poles number P 4

Stator resistance sR ( mΩ ) 25.1

Rotor resistance rR ( mΩ ) 18.2

Mutual inductance (mH) 1.8

Stator inductance sL (mH) 0.07618

Rotor inductance rL (mH) 0.07618

Inertia (Kg.m²) 0.00824

Using these parameters, the equivalent torque open-loop transfer function can be

express as follows

-3

0.49291 8.2 10 1p

KG ( s )s s

= =τ + × +

(4-58)


where

2 2

2

-3

3 0.49292

8 2 10

0 057

*ms

r s

r

r

*s

LK PR L

L .R

.

⎧= Ψ =⎪

⎪⎪τ = σ = ×⎨⎪⎪ Ψ =⎪⎩

(4-59)


The PI controller is designed with desired close loop performance and the machine

parameters by (4-31). For example, the desired close loop transfer function is

1 -4

1 11 4 10 1

G ( s )s s

= =λ + × +

(4-60)

It is corresponding to a first order system whose setting time is chosen as 2 ms for a

Step-function input.

Thus, the PI controller parameters are

1

31

41.5940

1 5.0716 10

p

i

KK

KK

τ⎧ = =⎪⎪ λ⎨⎪ = = ×⎪ λ⎩

(4-61)

Fig.4.7 Torque dynamic performance of direct flux vector control with rotor resistance variation

of 50% and 100%


Closed torque loop performance of proposed scheme with above PI controller is

investigated. The torque reference is 6 Nm and the sampling time is 150 sμ . Practically,

rotor resistance varies due to heating. The sensitivity of the system to parameter

variation should be studied. Fig.4.7 shows the torque dynamic response for square-wave

torque reversal reference input. It takes into account the effect of rotor resistance rR

variation, which varies by 50% and 100% of the original value in Table 4.1. The desired

response time of torque is achieved (2 ms) and the torque tracks the reference well

without steady state error as long as the parameters of the induction machine are

accurate. However, there will be a tracking error when the rotor resistance rR is

changed. The inaccurate rR results in the desired closed-loop behavior is not being

achieved.

Fig.4.8 Performance of direct flux vector control with speed loop

Closed speed loop performance of the system is also tested. Fig.4.8 shows the speed of

induction machine rises from standstill to 600 rpm, and then accelerates to rated speed


1500 rpm. There is an overshoot in torque which is caused by establishing of stator flux

during starting period (0-600 rpm).

Fig.4.8 also shows the amplitude of stator flux vector is kept constant by the controller

and that of rotor flux vector tracks stator flux with a time delay. After establishing of

stator and rotor flux, their amplitudes can be kept constant by the controller during

accelerating period (600-1500rpm). This indicates that the previous assumption of

constant rotor flux amplitude is valid. The stator flux and current diagram shows that

less torque ripple and current harmonics are obtained with the proposed scheme using

the designed PI controller by comparing with the classic direct torque control scheme.


Assuming the desired settling time is 2 ms, nω can be obtained by selecting ζ with (4-

34). For example

0 8

4 2500ns

.

t

ζ =⎧⎪⎨ω = =⎪ ζ⎩

(4-62)

then

2

25

2

1 4 1 56.2029

1.0398 10

np

ni

.KK

KK

ω τ −⎧ = =⎪⎪⎨

ω τ⎪ = = ×⎪⎩

(4-63)

-4

1 15.4049 10 11

fp

i

G ( s ) K ssK

= =× ++

(4-64)

Under same condition, closed torque loop performance of proposed scheme with PI

controller parameters in (4-63) is investigated. The effect of pre-filter is also studied.

Fig.4.9 shows that large torque overshoot occurs resulting from the zero of the closed-

loop transfer function. Therefore, pre-filter is required to remove the overshoot. As

shown in Fig.4.10, the system is very robust to the variation of rotor resistance even it

changes to two times of original value. Unlike the rotor flux oriented control scheme,


direct flux vector control need few parameters of the machine for the controller, which

increases its robust ability to parameters variation.

Fig.4.9 Torque dynamic performance of direct flux vector control with and without Pre-filter

Fig.4.10 Torque dynamic performance of direct flux vector control with rotor resistance

variation of 50% and 100% (pre-filter added)


Fig.4.11 Performance of direct flux vector control with speed loop – No pre-filter added

Similar with above section, closed speed loop performance of the system is also

investigated. Fig.4.11 and Fig.4.12 show the speed of induction machine rises from

standstill to 600 rpm, and then accelerates to rated speed 1500 rpm. There is an

overshoot in torque which is caused by establishing of stator flux during starting period

(0-600 rpm). With pre-filter added to the torque controller, the torque overshoot is less

in Fig.4.12.


Fig.4.12 Performance of direct flux vector control with speed loop –pre-filter added

4.2.4.3 Comparison with rotor flux oriented control (RFOC)

Table 4.2 Parameters of the control scheme

Inverter dc bus voltage dcV (V) 42

Stator flux reference sΨ for DFC (Wb) 0.057

Rotor flux reference rΨ for RFOC (Wb) 0.0547

Sampling time ( sμ ) 150

The direct flux vector control (DFC) is compared with rotor flux oriented control

(RFOC) presented in Chapter 2. The control structure of RFOC with SVM is shown in

Fig.4.14. The torque dynamic response of RFOC under same conditions is given in

Fig.4.13. The parameters of induction machine are same with Table 4.1 and the control

parameters are shown in Table 4.2. The torque response of RFOC is better than DFC

without overshoot. In this chapter, the settling time of DFC is designed as 2 ms as an


example, which is a little slower than that of RFOC. Under same design principle, the

torque response of DFC can be faster than 2 ms with difference PI parameters.

Fig.4.13 Torque dynamic response of RFOC

*rω

r∠Ψ

rψ ∗

−

Current Model

SVM IMPI

rω

PI

Encorder

qV

dVα β−

,V Vα βPI

*di

*qi

qi

di

PI−

rΨ

e ed q−

Fig.4.14 Rotor flux oriented control scheme with SVM

The sensitivity of RFOC is also investigated for comparison. As show in Fig.4.15,

RFOC is very sensitive to the variation of rotor resistance. Its torque performance is

poor even when the rotor resistance changes only by 20% of the original value. The

torque is totally out of control when the rotor resistance increases by 50%. Comparing


Fig.4.15 and Fig.4.10, it is obvious that DFC is more robust than RFOC to the variation

of rotor resistance rR . The DFC is more robust than RFOC to the variation of the rotor

resistance because of the following two facts. Firstly, the rotor resistance is not involved

in the control system of DFC whereas it is a critical parameter for the decoupling

controller of the RFOC. Secondly, the DFC with robust PI controller is not sensitive to

the variation of machine parameters.

Fig.4.15 Torque dynamic performance of rotor flux oriented control with varied rotor resistance


4.2.5 Experimental results

Fig. 4.16 The experiment setup of the system

As shown in Fig. 4.16, the system was implemented on a dSPACE DS1104 Controller

Board with TMS320F240 slave processor. A three phase VSI inverter is connected to

supply 42 V dc bus voltage, which is supplied from a rectifier.

Voltage mode stator flux estimator based on (4-1) is used in the system. Due to the

noise or measurement error inherently present in the current sensor, the pure integrator

in can lead to saturation. To avoid that, a low pass filter is used in stead for the flux

estimation.

( )11s s s s

c

V R Is T

Ψ = −+

(4-65)

where ( )1 2c cT f= π and cf is the cut-off frequency of the filter.


Closed-loop torque performance of proposed scheme with above the PI controller is

investigated. The torque reference is square wave signal with ± 6 Nm magnitudes and

the sampling time is 150 sμ .


Fig.4.17 Torque dynamic performance of direct flux vector control with direct synthesis of PI

controller

Closed-loop speed performance of the system is also tested. Fig.4.18 shows the speed of

induction machine rises from 600 rpm to rated speed 1500 rpm. The amplitude of stator

flux vector is kept constant by the controller.




Fig.4.19 Torque dynamic performance of direct flux vector control with pre-filter

Under same condition, closed torque loop performance of proposed scheme with PI

controller parameters in (4-63) is investigated. The torque response is shown in

Fig.4.19.

Similar with the above section, closed-loop speed performance of the system is also

investigated. Fig.4.20 shows the speed of induction machine rises from 600 rpm to rated

speed 1500 rpm. The amplitude of stator flux vector is kept constant by the controller.



Fig.4.21 shows the speed, torque and stator flux at steady state. Less torque and flux

ripples are obtain with proposed control method.


Fig.4.21 steady state performance with speed-loop

Fig.4.22 Spectrum analysis of the stator current


The spectrum of the stator current is analyzed by FFT algorithm by using the data

captured by a digital oscilloscope (Lecroy 364TL). The 6.67 kHz part of the frequency

in Fig.4.22 is corresponding to the sampling frequency of the system, which indicates

the switching frequency of the inverter is fixed by space vector modulation.

4.3 Direct flux vector controlled induction generator for

an ISA

4.3.1 Induction generator with DFC

dcU

*T

eT

sψ ∗

*dcU

*sψ sψΔ

sψ

refV

startingT

sω∗

1TΔ

+

−

−−

+

Fig. 4.23 DFC scheme for ISA

From above analysis, it becomes clear that a direct relationship exists between the

torque and the rotation speed of stator flux vector when its amplitude is kept constant.

This means that it is possible to control the machine torque by directly controlling the

amplitude and rotating speed of stator flux vector. This is the basic idea of direct flux

vector control for induction machine. A complete scheme of direct flux vector

controlled ISA that allows effective dc bus voltage and torque control has been

developed and it is indicated in Fig. 4.23. It includes a starting/generating mode switch

which simulates the operation mode of ISA from starter to generator. During starting

mode, the induction machine acts as a motor to provide high torque for the starting of

the engine. As shown in Fig. 4.23, the dc load of ISA is connected at dc side of the DC-


AC converter with the battery. The VSI converter for the induction machine supplies

active power to the dc load during generation state while the same converter also

provides reactive power to the machine for the excitation of its field.



is given in (3-8)

( )

32

s s s s

e s s

V R I dt

T P I

⎧Ψ = −⎪⎨

= Ψ ×⎪⎩

∫ (4-66)

The above scheme uses only one PI torque regulator to control the rotating speed of

stator flux vector. The desired reference stator flux vector *sψ is generated by Flux-

Vector-Combination block in Fig. 3.5, whose amplitude and angle is given by

s

**s s

s s

s s s

T∗∗

⎧⎪⎪⎨⎪⎪⎩

ψ = Ψ

Δθ = ωθ = θ + Δθ

(4-67)

where sT is the sampling time, sΔθ is the angular movement of the stator flux vector

during sampling period and sθ is the present angle of the stator flux vector. The

reference stator flux reference vector is compared with the estimated flux to obtain error

flux vector sΔΨ . With given sΔΨ , the exact stator voltage vector that changes the

rotating speed of stator flux vector to generate required torque while keeping its

amplitude constant is given by

sref s s

s

V R ITΔΨ

= + (4-68)

The space vector modulation method is used to apply the required stator voltage vector

with fixed switching frequency. In transient state, the reference voltage will be larger

than the available inverter voltage when the torque error is too large. In that case, the

speed s∗ω has to be limited to ensure the reference voltage is lower or equal to the

maximum inverter voltage:



where maxV is the maximum available inverter voltage. For under-modulation of SVM,

maxV equals to 1

3dcV , where dcV is the dc bus voltage of the inverter.

With SVM technique, the demand space voltage vector can be composed by two active

and one zero voltage vectors, which is illustrated in right part of Fig. 4.24.

2V3V

4V

6V

0V

7V1V

refV

5V

refV TΔ

sΨ*

sΨ α

sθ

α

β

α

β

Fig. 4.24 Reference space voltage vector

For example, when refV locates between 1V and 2V , it can be expressed as

0 1 20 1 2ref

s s s

T T TV V V VT T T

= + + (4-70)

where 0T , 1T , and 2T are the effective time intervals of 0V , 1V and 2V , respectively

within the sampling period sT .

From Fig. 4.24, the following can be obtained

1 21 2

22

3

3

refs s

refs

T TV cos V V cosT TTV sin V sinT

π⎧ α = +⎪⎪⎨ π⎪ α =⎪⎩

(4-71)

Thus


1

1

2

2

3

3

3

ref

s

refs

V sin( )T T

V sin

V sinT T

V sin

π⎧ − α⎪=⎪ π

⎪⎨⎪ α

=⎪ π⎪⎩

(4-72)

Hence

0 1 2sT T T T= − − (4-73)

4.3.2 Experimental results

4.3.2.1 Starting mode


machine, which simulates the engine. In this experimental setup, the starting torque is

set as 6 Nm and the engine starting speed is 500 rpm to reflect the idle speed of the

engine. After DC machine simulated engine is started, both the DC machine and

induction machine are accelerated from 500 rpm to 1200 rpm. For the study in this

thesis, 1200 rpm was chosen as the simulated ISA generating speed because of the

limitation of the DC machine simulating the engine.

As shown in Fig. 4.25, the full induction machine’s torque runs the whole set from 0 to

1200 rpm. After the speed reaches 1200 rpm, the DC machine simulated engine sets its

speed reference as 1500 rpm and regulated by its own controller. In this study, 1500

rpm is the base speed of the induction machine. At the same time, the reference of the

induction machine is switched from torque to voltage to reflect the transition from

motoring to generating. The induction machine now begins to act as a generator and

provide power to the battery and the dc load. The torque of the induction machine thus

changes from positive torque to negative torque as in (i) of Fig. 4.25. The stator flux

[(iii) in Fig. 4.25] of the machine is kept constant in this proposed direct flux vector

control method.


Fig. 4.25 Starting process of ISA

4.3.2.2 Generating mode - steady state





oscilloscope (LeCroy LT364L) and the data is analysed by FFT algorithm. The current

spectrum analysis in Fig. 4.27 indicates that the DC-AC converter of the ISA system

runs at constant frequency 6.67 kHz, which is corresponding to the sampling time 150

sμ .



Fig. 4.27 Spectrum analysis of the stator current of ISA


4.3.2.3 Generating mode - dynamic response.

The dynamic performance of the ISA is studied in this section. The performance of the

ISA system under load dump and engine speed acceleration or deceleration is presented

as follows.

4.3.2.3.1 Performance during load dump

The dc load of the ISA is removed suddenly in the generating mode when the speed is



4.28 and Fig. 4.29, respectively.

As shown in Fig. 4.28 and Fig. 4.29, the peak dc bus voltage of the ISA is well

controlled below the limitation of the 42 V PowerNet standard (58 V) [6] when the dc

load is dumped. The settling time of the dc bus voltage is only 100 ms. The induction

machine’s torque is changed from -6 Nm to about -1 Nm during load dumping. The

torque of the induction machine varies slower in Fig. 4.29 because of the charging of

the batteries.

Fig. 4.28 Load dump of ISA without battery connected


Fig. 4.29 Load dump of ISA with battery connected

4.3.2.3.2 Dynamic performance during speed acceleration/deceleration


to 3000 rpm, while the induction machine is generating with full dc load. As shown in

Fig. 4.30, the dc bus voltage of the ISA is well controlled as 42 V during speed

acceleration. The stator flux of the machine is weakened when the speed is above the

base speed (1500 rpm).


Fig. 4.30 ISA performance at acceleration

The deceleration of the ISA also tested by dropping the speed suddenly from 3000 rpm


Fig. 4.31, the dc bus voltage of the ISA is dropped a little from 42 V during speed

deceleration, but it still in the allowed voltage range of 42 V PowerNet [6]. Fig. 4.31

(iii) shows the flux of the machine is increased when the speed returns to base speed

(1500 rpm).

Fig. 4.31 ISA performance at deceleration


4.3.2.4 High speed operation





than 6 Nm due to the high speed operation. The stator flux of the induction machine is


In this thesis, the ISA only runs up to 4000 rpm due to limitation of the induction

machine in the laboratory.


4.4 Conclusion

In this chapter, an improved torque controller of induction machine based on direct

control of stator flux linkage vector is presented. The fundamental relationship between

the rotating speed of the stator flux linkage and torque is analyzed and the design


principle of controller is presented. A simple structure with only one Proportional-

Integral (PI) controller is shown to implement the torque and flux control adequately.

Parameters of PI controller are easily found in the proposed design principle. Robust

design of the controller ensures the system is not sensitive to the variation of rotor

resistance. Fixed switching frequency and low torque ripple are obtained with PI control

and space vector modulation (SVM) method. Satisfactory modeling and experimental

results indicate the feasibility of the proposed direct flux vector control scheme for

induction machines. The control scheme employs encoderless torque control structure,

and eliminates the disturbance of speed to the torque controller successfully. The

controller gives good torque and flux control performance.

A direct flux vector controlled scheme of induction generator has been proposed and

verified in this chapter for future 42 V automobiles application. A simple structure with

only one Proportional-Integral (PI) controller is shown to implement the torque and flux

control adequately. By controlling the electromagnetic torque of the induction machine,

the required dc bus voltage can be well regulated within the 42 V PowerNet

specifications. Simulation and experimental results indicate that the proposed scheme

provides a practical solution for an integrated starter alternator, avoiding the drawback

of rotor flux oriented control scheme.

However, the calculation of the commanded voltage vector requires the derivative of the

stator flux vector, which is kept moving. Thus, it is a potential source of error. Actually,

the stator flux linkage will be a dc quantity when the reference frame is fixed to the

stator flux vector. It should thus be possible to avoid calculation of the derivative of the

flux vector. In the next chapter, a control scheme of ISA based this on idea will be

presented.

Chapter 5 Direct torque and flux controlled ISA with space vector modulation 96

CHAPTER 5

DIRECT TORQUE AND FLUX CONTROLLED



5.1 Introduction

The direct flux vector control presented in Chapter 4 controls the rotating speed of the

stator flux vector by a torque feedback loop. No direct control for the amplitude of the

stator flux vector is included. In this chapter, the feedbacks of the torque and the flux

are both used in two independent control loops. It is a direct torque and flux control

(DTFC) scheme based on the basic DTC concept. In effect, the two hysteresis

comparators are replaced by two PI controllers.

A similar scheme to DTFC for induction motor drives application has been presented in

[83]. But its application in generators or the ISA has not been reported. This chapter

proposes a direct torque and flux control scheme for an induction generator in ISA

application. The relationships between controlled variables and the torque are fully

developed. Constant switching frequency and lower torque ripple are achieved with

Proportional-Integral (PI) controller and space vector modulation (SVM). The speed

sensor is eliminated and the torque and stator flux are estimated with voltage mode

estimator. As the torque of induction machine is controlled with DTFC with high

dynamic performance, the dc bus voltage can be regulated to meet the specification of

the 42 V PowerNet.

This chapter is organized as follows. Section 5.2 presents the detailed analysis of the

principle for direct torque and flux control based ISA system. In Section 5.4,

experimental results are presented. Finally, the conclusion is drawn in Section 5.5.


5.2 Direct torque and flux control principle

sΨ

sθ α

β

q d

rΨ

sI

sdisqi

Fig. 5.1 Vector diagram of the induction machine

In stator flux reference frame ( )d q− shown in Fig. 5.1, the dynamic behavior of

induction machine can be described as following equations:

( )0

32

ss s s s s

rr r s m r

e sd sq

dV R I jdt

dR I jdt

T P i

⎧ Ψ= + + ω Ψ⎪

⎪Ψ⎪

= + + ω −ω Ψ⎨⎪⎪ = Ψ ⋅⎪⎩

(5-1)

and

s s s m r

r m s r r

L I L I

L I L I

⎧Ψ = +⎪⎨Ψ = +⎪⎩

(5-2)

Therefore,


( )1

2

1

ss s s s s

rs m r r r

s s ss m r m

m r m ss r mr r r

d V R I jdt

d j R IdtI L L L L

L L L LL L LI

−

⎧ Ψ⎪ = − − ω Ψ⎪⎪ Ψ⎪ = − ω −ω Ψ −⎨⎪⎪ ⎡ ⎤ ⎡ ⎤⎡ ⎤ Ψ Ψ−⎡ ⎤ ⎡ ⎤

= =⎪ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥−−Ψ Ψ⎪⎢ ⎥ ⎣ ⎦ ⎣ ⎦⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦⎩

(5-3)

Equation (5-3) can be simplified as

( )

10

s s ms ss s r s

sr m rr r

s ms r r

R R Ld jL L Ldt V

R L Rd jL L Ldt

⎡ ⎤⎡ ⎤Ψ − − ω⎢ ⎥⎢ ⎥ ⎡ ⎤σ σ Ψ ⎡ ⎤⎢ ⎥⎢ ⎥ = +⎢ ⎥ ⎢ ⎥⎢ ⎥Ψ Ψ⎢ ⎥ ⎣ ⎦⎢ ⎥⎣ ⎦− ω −ω −⎢ ⎥⎢ ⎥ σ σ⎣ ⎦ ⎣ ⎦

(5-4)

where

2

1 m

s r

LL L

σ = − (5-5)

(The further details of the derivation included in this chapter can be found in Appendix

C)

So, the relationship between stator and rotor flux vector can be obtained from (5-4)

( )( )( ) ( )

1

m

sr s

s m

LLs s

s jΨ = Ψ

τσ + ω − ω τ + (5-6)

where r

r

LR

τ = .

It is known in the stator flux reference frame that

0

s ds qs

qs

j⎧Ψ = Ψ + Ψ⎪⎨Ψ =⎪⎩

(5-7)

The rotor flux vector in the stator flux reference frame can be expressed as

r rd rqjΨ = Ψ + Ψ (5-8)

With (5-6), (5-7) and (5-8), the dq component of rotor flux vector can be obtained


( )( )

( )

( ) ( )( )

( )

2

2

r m

s rrd sd

s m r

r r

r

r m

s m s rrq sd

r s m r

r r r

r

R LL Ls s

Rs R LsL

R LL Ls sR Rs sL R Ls

L

⎧⎪ σ⎪Ψ = Ψ

ω − ω⎪+ +⎪ σ⎪ +

σ⎪⎨⎪

− ω − ω⎪ σΨ = Ψ⎪

ω − ω⎪ + + +σ⎪ σ+⎪ σ⎩

(5-9)

The expression of stator current with stator and rotor flux vector is already shown in (5-

3), which is restated as

[ ]2

1 ss r m

s r m r

I L LL L L

⎡ ⎤Ψ⎡ ⎤ = − ⎢ ⎥⎣ ⎦ − Ψ⎢ ⎥⎣ ⎦

(5-10)

By substituting (5-9) into (5-10), it is derived that

( )

( )

( )

2

2

22 2 2 2 2

2

2

2 1

2 1

ms m

s rsq sd

s m

ms m

s rsd

LL LI ( s ) ( s )

s s

LL L ( s )

s

ω −ω τ= Ψτ σ + τσ + τ σ ω −ω +

ω −ω τ≈ Ψ

τσ +

(5-11)

The simplification in (5-11) is based on small τ and σ .

By inverse Laplace transform, the expression of sqI is time domain is obtained as

{ }

( )

( ) { }

2

21 1

22

2

2 1

1

ms m

s r sdsq sq

tms m sd

s r

LL LI ( t ) I ( s )

s s

L eL L

∗− −

−∗ τσ

⎧ ⎫ω − ω τ⎪ ⎪Ψ⎪ ⎪= = ⎨ ⎬τσ +⎪ ⎪

⎪ ⎪⎩ ⎭

= ω − ω τ Ψ −

L L (5-12)

It is assumed that the magnitude of the stator flux vector is kept constant with flux

regulator in axis d . By considering (5-1) and (5-12), the torque is obtained as follows.


( ) ( ){ }222

2

3 3 12 2

tm

e sd sq sd s ms r

LT ( t ) P ( t ) i ( t ) P eL L

−∗ τστ= Ψ ⋅ = Ψ ω −ω − (5-13)

By (5-1), the voltage equation in dq frame is

sd sd

sd s sd

sq s sq s sd s sd

d dV R idt dt

V R i

Ψ Ψ⎧ = + ≈⎪⎨⎪ = + ω Ψ ≈ ω Ψ⎩

(5-14)

By substituting (5-14) into (5-13), the relationship between the q voltage component

and the torque is developed as

{ } ( )2

22

3 12

tm

e sd sq ms r

LT ( t ) P e V fL L

−∗ τστ= ⋅Ψ − − ω (5-15)

where

( ) ( ) { }222

2

3 12

tm

m sd ms r

Lf P eL L

−∗ τστω = Ψ − ω (5-16)

Therefore, it is clear shown in (5-15) that the torque of induction machine can be

directly regulated by the q voltage component considering ( )mf ω as a disturbance to

the system. Similarly, the amplitude of stator flux vector can be regulated by the d

component of stator voltage directly as shown in (5-14). Above analysis forms the

principle of the direct torque and flux control (DTFC) scheme for the induction

machine.

The voltage vector should be transferred from the stator flux reference frame to the

stationary frame by (5-17) before using SVM algorithm.

s sds s

s sqs s

V Vcos sinV Vsin cos

α

β

θ − θ⎡ ⎤ ⎡ ⎤⎡ ⎤=⎢ ⎥ ⎢ ⎥⎢ ⎥θ θ⎣ ⎦⎣ ⎦ ⎣ ⎦

(5-17)

where sθ is the angle between the stator flux frame ( dq ) and stationary frame (αβ ), i.e.

the angle of stator flux linkage vector as shown in Fig. 5.1.

Then the reference voltage vector is

ref s sV V jVα β= + (5-18)


The gating signals can be generated by SVM algorithm as discuss in Section 4.3.1 by

inputting reference voltage vector.

5.3 Direct torque and flux controlled induction generator

for an ISA

Based on above analysis, a complete scheme of direct torque and flux control for ISA

that allows effective torque control has been developed and it is indicated in Fig. 5.2.

The torque and flux are regulated by two PI controllers. The design of this two PI

controller is based on (5-13) and (5-14). With same approaches discussed in Section

4.2.2, the PI controller parameters of the torque can also be found. The ISA system

includes starting/generating mode switch which simulates the operation of ISA from

starter to generator. After the switch changes to generating mode, the voltage regulator

will take effect to keep the dc bus voltage as 42 V and the torque reference will be

negative.

dcV

*T

eT

sψ ∗

*dcV

sψ

refV

startingT +

−

−

+sqV

sdV

d q−

α β−

sθ

dcV

−

Fig. 5.2 Direct torque and flux controlled induction generator for ISA

As shown in Fig. 5.2, only one voltage sensor for dc bus voltage and two current

sensors for stator current are adopted in proposed scheme. The voltage and current

signals are used for stator flux estimation. The stator flux vector is estimated in the

stationary frame avoiding co-ordination transformation and involvement of more


machine parameters. The estimation algorithm is given in (5-19). And the voltage signal

is also used as voltage feedback to maintain the dc bus voltage as 42 V.

( )( )1

32

s s s s

s s s

e s s

V R I dt

tan

T P I

−β α

⎧Ψ = −⎪⎪⎪θ = Ψ Ψ⎨⎪⎪ = Ψ ×⎪⎩

∫ (5-19)

In (5-19), current vector sI is constructed by the two line current with Park

transformation. And voltage vector sV can be obtained by

1refsV ( k ) V ( k )= − (5-20)

The time delay between refV and sV results from the SVM generating time sT .

In transient state, the reference voltage will be larger than the available inverter voltage

when the torque error is too large. In that case, the reference voltage has to be limited to

ensure the reference voltage is lower or equal to the maximum inverter voltage:


where maxV is the maximum available inverter voltage.

For under-modulation of SVM,

1

3max dcV V= (5-22)

where dcV is the dc bus voltage of the inverter.


5.4.1 Starting mode


machine, which simulates the engine. In this experimental setup, the starting torque is

set as 6 Nm and the engine starting speed is 500 rpm to reflect the idle speed of the

engine. After DC machine simulated engine is started, both the DC machine and


induction machine are accelerated from 500 rpm to 1200 rpm. For the study in this

thesis, 1200 rpm was chosen as the simulated ISA generating speed because of the

limitation of the DC machine simulating the engine.

As shown in Fig. 5.3, (i) of the full induction machine’s torque runs the whole set from

0 to 1200 rpm. After the speed reaches 1200 rpm, the DC machine simulated engine sets

its speed reference as 1500 rpm and regulated by its own controller. In this study, 1500

rpm is the base speed of the induction machine. At the same time, the reference of the

induction machine is switched from torque to voltage to reflect the transition from

motoring to generating. The induction machine begins to act as a generator to provide

power to the battery and the dc load. The torque of the induction machine is thus

changed from positive to negative torque as in (i) of Fig. 5.3. As shown in Fig. 5.3 (iii),

there is an overshoot of the stator flux when the rotor speed rises from standstill state.

This overshoot results from the PI regulation of the flux controller. The proposed design

of the PI parameters of the flux controller may eliminate the overshoot. After starting,

the stator flux is controlled as constant.

Fig. 5.3 Starting process of ISA


5.4.2 Generating mode - steady state





oscilloscope (LeCroy LT364L) and the data is analysed by FFT algorithm. It indicates

in Fig. 5.5 that the DC-AC converter of the ISA system runs at constant frequency 6.67

kHz, which is corresponding to the sampling time 150 sμ .



Fig. 5.5 Spectrum analysis of the stator current of ISA

5.4.3 Generating mode - dynamic response.

The dynamic performance of the ISA also studied in this section. The performance of

the ISA system under load dumping and engine speed acceleration or deceleration is

presented as follows.

5.4.3.1.1 Performance during load dump




5.6 and Fig. 5.7, respectively.

As shown in Fig. 5.6 and Fig. 5.7, the peak dc bus voltage of the ISA is well controlled

below the limitation of the 42 V PowerNet standard (58 V) [6] when the dc load is

dumped. The settling time of the dc bus voltage is only 100 ms. The induction

machine’s torque is changed from -6 Nm to about -2 Nm during load dumping. The


torque of the induction machine varies slower in Fig. 5.7 because of the charging of the

batteries. The stator flux of the induction machine is dropped a little resulting from the

load dump.

Fig. 5.6 Load dump of ISA without battery connected

Fig. 5.7 Load dump of ISA with battery connected


As shown in Fig. 5.6 and Fig. 5.7 it is found that the torque ripples are at same

frequency of stator current (voltage). The torque ripple is caused by the estimation error

in the stator flux. The stator flux is estimation by integration of the stator voltage as

shown in (3-8). That is why the torque error is at the same frequency of the stator

current (voltage).

5.4.3.2 Performance during speed acceleration/deceleration



Fig. 5.8, the dc bus voltage of the ISA is well controlled as 42 V during speed

acceleration. The stator flux of the machine is weakened when the speed is above the

base speed (1500 rpm).

Fig. 5.8 ISA performance at acceleration

The deceleration of the ISA also tested by dropping the speed suddenly from 3000 rpm


Fig. 5.9, the dc bus voltage of the ISA is dropped a little from 42 V during speed


deceleration, but it still in the allowed voltage range of 42 V PowerNet [6]. Fig. 5.9 (iii)

shows the flux of the machine is increased when the speed returns to base speed (1500

rpm).

Fig. 5.9 ISA performance at deceleration

5.4.4 Performance High speed operation





than 6 Nm due to the high speed operation. The stator flux of the induction machine is


In this thesis, the ISA only runs up to 4000 rpm due to limitation of the induction in the

laboratory.



5.5 Conclusion

This chapter presents a direct torque and flux control of the integrated starter/alternator.

This control scheme has been analyzed and verified with simulation and experiments.

The simulation and experimental results show that the direct torque control concept had

also been successfully extended to generator application. Simplicity of the system

structure and lower ripples of current and torque are both achieved with proposed

scheme. The modeling and experimental results confirm the effectiveness of the

proposed scheme to be a strong candidate for ISA system.

Compared to the direct flux vector control scheme proposed in last chapter, this scheme

is a little bit complex due to transformation computation. However, the calculation of

the commanded voltage vector by (5-14) requires the derivative of the stator flux

magnitude, which is a dc quantity. Thus, this scheme is less noisy [63] than the flux

vector calculation based direct flux vector control scheme.

Chapter 6 Non-linear behaviour of the converter and its compensation 110

CHAPTER 6

NON-LINEAR BEHAVIOUR OF THE DC-AC

CONVERTER AND ITS COMPENSATION

6.1 Introduction

In the direct torque control scheme, the stator flux can be estimated by sensing the stator

voltage and current of the machine. Sensing of line-line voltage waveforms required

filtering in order to eliminate the harmonics and noise created by PWM modulation, and

offset as well. Alternatively, it is preferred to reconstruct the stator voltage vector from

the gating signals and the dc link voltage which is in turn regarded as the reference

voltage vector. However, the reference voltage vector does not exactly represent the

voltage vector at the machine terminals due to the non-linear behaviour of the converter,

which are caused by the dead-time effect [93-95] and voltage drops on the power

devices [96, 97].

Specially, this case becomes critical in the 42 V ISA application because the machine’s

voltage is very low with the one-stage structure as stated in 1.2.3.3. For example, the

voltage of the induction machine used in this study is only 22 V. therefore; even 1 V

error for a power device will cause about 10% percentages of the reconstructed voltage.

The total effect of the dead-time and voltage drop introduce a large error of the

reference voltage. The inaccurate reference voltage can cause wrong stator flux

estimation, and further degrade the control ability of the DTC. Therefore, the non-linear

behaviour of the DC-AC converter has to be compensated.

This chapter analyzed the total effects of dead-time and voltage drops and developed a

combined compensation methods to compensate these two effects together, which are

normally considered separately in the existing literature [94-97]. Generally, the dead-


time and voltage drop compensation are based on normal PWM or SPWM fed inverters

[93, 96-98]. At present, the space vector PWM technique is widely used in voltage

source inverters. For the 42 V ISA, it is necessary to study the compensation method

also with space vector PWM of DC-AC converter. A novel dead-time compensation

method has been studied for space vector PWM in [95] based on time error resulting

from the dead-time. In this chapter, the effects of dead-time and voltage drops on the

three phase DC-AC converter with space vector modulation are both analyzed. An error

voltage vector based compensation method is proposed to reduce those two effects

together. Two compensation structures are developed and compared with feed-forward

and feed-backward manners.

6.2 Effect of Dead-time

To avoid direct short circuit across the dc bus voltage source, a blanking time or dead-

time is inserted into the gating signal of the switch that is to be turned on. There is a

time in each switching cycle where both the high and low side switches in the same leg

are off and the current flow is through the diodes.

+

dcV

+A

−A

+D

−D

ai-

P

N

0>

Fig. 6.1 one leg of the converter

The voltage level of each phase during dead-time is determined by the current direction

of each phase. As shown in Fig. 6.1, the positive direction is defined as the phase

current is flowing from converter to the load. By assuming the sign of phase current

doesn't change during the sampling period, the effect of the dead-time for PWM is

presented in Fig. 6.2. The shadow stands for the losing area due to the dead-time and

turn-on or turn-off time of the power device.


dt

+A

−A

+A

−A

aNV

)a

)b

)c

dt

)d

sT

dt

dt

aNV ′ 0>ai

0<aiaNV ′

ond tt +

offt ond tt +

aNV 0>ai

0<ai

offt

Fig. 6.2（a） ideal gate signal （b）practical gate signal with dead-time （c） aNV with dead-

time effect only（d）considering ont and offt of the power device

With space vector PWM, the voltage vectors diagram is shown in Fig. 6.3. The

reference voltage vector is synthesized by the two adjacent basic voltage vectors. The

gate signal of the converter in one sampling period sT is given in Fig. 6.4.

IM

Ai Bi Ci

1AS =

0CS =

1CS =

0BS =0AS =

1BS =dcV

2V3V

4V

6V

0V

7V1V

refV

5V

α

β

α

(a) (b)

Fig. 6.3 Switching state of VSI (a) and space voltage vectors (b)


AS

CS

BS

sT

0 0 0 1 0 0 11 0 111 11 0 1 0 0 0 0 0

Fig. 6.4 Gate signal without dead-time

Similar with PWM case in Fig. 6.2, the duration of the gate signal is reduced and

increased with dead-time effect. Table 6.1 shows a example with 0ai > , 0bi > and

0ci < .

AS

CS

BS

sT

0 0 0 1 0 0 11 0 111 11 0 1 0 0 0 0 0

ond tt +

ond tt +

offtond tt +

offt

offt


Fig. 6.5 Gate signal with dead-time

Table 6.1 Dead-time effect analysis ( 0ai > ; 0bi > ; 0ci < )

Lost parts

(shadowed

parts)

duration Extra Parts

(Non-shadowed

parts)

duration

A (1 0 0) =V1 td+ton (1 0 0) =V1 toff

B (0 1 0) =V3 td+ton (0 1 0) =V3 toff

C (0 0 1) =V5 toff (0 0 1) =V5 td+ton

The changes of the duration will introduce an error vector. The error vector is

determined by sign of the phase current. For example, the reference voltage vector is in

sector 1 and the sign of current are ( )+ + − , i.e. 0ai > ; 0bi > ; 0ci < . Then the actual

output voltage vector is

( )( )

( ) ( )

( ) ( )( ) ( )( )

1 3 5

1 3 5

2 5

5 5

52

ref s d on off

off d on

ref s d on off d on off

ref s d on off d on off

ref s d on off

V T V V t t V t

V V t V t t

V T V t t t V t t t

V T V t t t V t t t

V T V t t t

− + + −

+ + + +

= − + − + + −

= + + − + + −

= + + −

(6-1)

So, the error voltage vector for this case is ( ) ( )2

5423

d on off d on offdc

s s

t t t t t tV V a

T T+ − + −

=

where 2 3α j /e π= .

Table 6.2 summarizes the error voltage vectors caused by dead-time effect different

current polarities.


Table 6.2 Error voltage vectors under different current polarities

( )Asgn i ( )Bsgn i ( )Csgn i Error

vector 1

Error

vector 2

Error vector

3

total

+(0) +(0) -(1) -100(V1) -010(V3) +001(V5) 2V5

+ - + -100 +010 -001 2V3

+ - - -100 +010 +001 2V4

- + + +100 -010 -001 2V1

- + - +100 -010 +001 2V6

- - + +100 +010 -001 2V2

6.3 Effect of voltage drop on the power device

Fig. 6.6 analysis of the voltage drop on the power device

The effect of voltage drop on the output voltage vector depends on the polarity of the

current and the switching state of the power device as shown in Fig. 6.6. The letter s


indicates the switching state of the top power device on the leg A. For example, 0ai > ;

0bi > ; 0ci <

AS

CS

BS

sT

0 0 0 1 0 0 11 0 111 11 0 1 0 0 0 0 0

dV−

dV−

ceV

dc ceV V−

dc dV V+

dc ceV V−

Fig. 6.7 Gate signal with voltage drop

By assuming Vth=(Vce+Vd)/2, the error voltage vector will be only determined by the

sign of the current in each phase and it has nothing to do with switching state.

Therefore, the actual output voltage vector is

2 22 41 α α α3 3ref th ref thV V [ )] V V− + − = + (6-2)

where 2 3α j /e π= .

so, the error voltage vector is 222 α3

thdc

dc

VVV

⎛ ⎞⎜ ⎟⎝ ⎠

Table 6.3 lists the error voltage vectors caused by the voltage drop under different

current polarities.


Table 6.3 Error voltage vectors under different current polarities

Asgn( i ) Bsgn( i ) Csgn( i ) 1 a a2 total total

+(0) +(0) -(1) -1 -1 1 2a2 2V5

+ - + -1 1 -1 2a 2V3

+ - - -1 1 1 -2 2V4

- + + 1 -1 -1 2 2V1

- + - 1 -1 1 -2a 2V6

- - + 1 1 -1 -2a2 2V2

6.4 Compensation algorithm

By comparing Table 6.2 and Table 6.3, it is found that the total error voltage vectors are

identical in terms of the sign of current for both dead-time and voltage drop effects.

Therefore, their compensation can be combined together as shown in (6-3)

( ) ( )d on off d on offth th

error error errors dc s dc

t t t t t tu uV V V VT V T V

⎡ ⎤+ − + −Δ = + = +⎢ ⎥

⎢ ⎥⎣ ⎦ (6-3)

Where errorV is the error vector obtained from Table 6.2 and Table 6.3.

Basically, the error voltage vector can be compensated in backward and forward

manners.

6.4.1 Backward compensation

refV

realVVΔ

Fig. 6.8 Backward compensation structure


For the backward compensation structure, the error voltage vector VΔ is fed after

command voltage vector refV being applied to the SVM block. The real voltage vector

applied to the machine through the DC-AC converter will be

real refV V V= + Δ (6-4)

After compensation, the real voltage vector realV can be used for the estimation of the

flux vector and the controllers.

6.4.2 Forward compensation

refV

realV

VΔ

Fig. 6.9 Forward compensation structure

With backward compensation structure, the compensation process depends on the

controller of the system. In fact, the pressure of the controller can be eased by using

forward compensation. For forward compensation, the error voltage vector VΔ is fed

before command voltage vector refV being applied to the SVM block. Therefore, the

predicted error voltage vector can eliminate the effect of the dead-time and voltage drop

in advance. The real voltage vector applied to the machine through the DC-AC

converter will be

real refV V= (6-5)

The new command voltage vector of the SVM block changes to

ref new refV V V− = − Δ (6-6)

Similarly, the real voltage vector realV can be used for the estimation and the controller.



The effectiveness of the compensation schemes for the voltage drop and dead-time is

tested experimentally. Compensation logics are integrated with the real-time controller

of the induction machine as shown in Fig. 6.10. Only two current sensors and one dc

bus voltage sensor for the SVM DTC drive are used. No extra hardware is needed for

these schemes.

ˆsΨ

eT

θ

*sΨ

*eT

r e fV

realV

Fig. 6.10 The control system with voltage drop and dead-time compensation.

In order to evaluate the performance of the compensation method for the ISA, both

motoring and generating modes are studied in this section.

6.5.1 Motoring mode

In order to compare the accuracy of the stator flux estimation in DTC-SVM with the

actual stator flux of the induction machine, two possible methods could be used. One

would involve installing sensor coil in the stator frame. This would still not be very

accurate because of the stator resistance and leakage fluxes. The other method would

involve estimating stator flux from the rotating rotor flux frame using a current model.

The last approach was used in this thesis. Fig. 6.11 shows a current mode stator flux and

torque estimator based on the rotor flux estimator in the conventional rotor-oriented

reference frame as discussed in Chapter 2 (see Fig. 2.3).


rω

sAisBisCi

si α

si β

rθ

rψ

Polar to Cartesian

+

+

lrLlsL

m rL L +

+m rL L

lrL lsL

rαψ

rβψ

sαψ

sβψRotor flux estimator

(a) Stator flux estimation

( )32e s s s sT P i iα β β α= ψ −ψ

sαψ

sβψ

si α

si β

eT

(b) Torque estimation

Fig. 6.11 Current mode stator flux and torque estimator

The current mode stator flux and torque estimator in Fig. 6.11 are based on the

following equations.

( )

( )

( )32

ms ls s r lr s

r

ms ls s r lr s

r

r r r

r r r

e s s s s

LL i L iLLL i L iL

cos

sin

T P i i

α α α α

β β β β

α

β

α β β α

⎧ψ = + ψ +⎪⎪⎪ψ = + ψ +⎪⎪⎪ψ = ψ θ⎨⎪ψ = ψ θ⎪⎪

= ψ −ψ⎪⎪⎪⎩

(6-7)


where lsL , lrL and mL are the stator leakage, rotor leakage and mutual inductances,

respectively; sαψ , sβψ , rαψ , rβψ , si α and si β are the stator flux linkages, rotor flux

linkages and stator currents in stationary frame ( )α β− , respectively.

In DTC-SVM schemes, a voltage mode stator flux and torque estimator is used for the

feedback signals of the controller as shown in Fig. 6.12.

( )32e s s s sT P i iα β β α= ψ − ψ

sαψ

sβψ

si α

si β

eT( )

( )s s s s

s s s s

v R i dt

v R i dt

α α α

β β β

⎧ψ = −⎪⎨ψ = −⎪⎩

∫∫

sv α

sv β

si α

si β

Fig. 6.12 Voltage mode stator flux and torque estimator

where sv α and sv β are the stator voltages in stationary frame, si α , si β , sαψ and sβψ are

the stator and rotor current in stationary frames, respectively.

In practical, a low pass in (6-8) for stator fluxes is used instead of pure integration in

Fig. 6.12 to avoid saturation effect.

( )

( )

1111

s s s sc

s s s sc

v R is T

v R is T

α α α

β β β

⎧ψ = −⎪ +⎪⎨⎪ψ = −⎪ +⎩

(6-8)

where ( )1 2c cT f= π and cf is the cut-off frequency of the filter.

Therefore, the voltage mode stator flux and torque estimator in Fig. 6.12 is used for the

control of DTC-SVM while the current mode stator flux and torque estimator in Fig.

6.11 is working in parallel to verify the estimation accuracy of the voltage mode

estimation.

6.5.1.1 Results without compensation

Fig. 6.13 illustrates the rotor speed, stator current, and estimated torque and stator flux

at no-load state without compensation when the induction machine runs in the motoring


mode at 600 rpm. The torque estimated by the voltage mode estimator has large error

compared to the torque estimated by current mode estimator. It is shown in Fig. 6.14,

large stator flux estimation errors exist when compensation is not used. There is a six-

step like distortion in the current waveform. Due to inaccurate flux estimation, the

torque has large ripples.

Fig. 6.13 Rotor speed, stator current, and estimated torque and flux at no-load -without

compensation


Fig. 6.14 Estimation errors of the stator flux- without compensation

6.5.1.2 Results with backward compensation

The rotor speed, stator current, and torque at no-load state are plotted in Fig. 6.15 when

the induction machine is running at 600 rpm. With backward compensation, the

estimation errors of the stator flux are less as shown in Fig. 6.16 at same condition as

above section. The measured current waveforms were corrected and appeared the most

sinusoidal and the torque ripple is lower. The estimated torques with current mode and

voltage mode estimators are overlapped.


Fig. 6.15 Rotor speed, stator current, and estimated torque at no-load - with backward

compensation

Fig. 6.16 Estimation errors of the stator flux- with backward compensation


Fig. 6.17 Reference voltages and error voltages - with backward compensation

The reference voltages and error voltages are plotted in Fig. 6.17 with backward

compensation. The error voltages VαΔ and VβΔ are used for the calculation of the flux.

6.5.1.3 Results with forward compensation

Compared with backward compensation, the forward compensation method is also

tested under the same conditions. With forward compensation, the estimation of the

stator flux is further improved. The torque ripple is lower with improved stator current

waveform. The estimated torques with current mode and voltage mode estimators are

overlapped. Fig. 6.18 shows the speed, torque, stator current and stator flux results at

no-load with feed forward compensation. Fig. 6.19 shows the estimation errors of stator

flux with forward compensation.


Fig. 6.18 Rotor speed, stator current, and estimated torque at no-load - with forward

compensation

Fig. 6.19 Estimation errors of the stator flux- with forward compensation


Fig. 6.20 Reference voltages and error voltages - with forward compensation

Similar with backward compensation, the reference voltages and error voltages are

plotted in Fig. 6.20 with forward compensation.

6.5.1.4 Comparison

The stator flux estimation errors are calculated in percentage using (6-9) for the above

three cases.

( )( ) ( ) 100

current voltages ,s s ,s

,ref

errors %α β α β

α β

ψ −ψ= ×

ψ (6-9)

Where refψ is the magnitude of the flux reference, ( )current

s ,sα βψ is the estimated flux

with current mode estimator, and ( )voltage

s ,sα βψ is the estimated flux with voltage mode

estimator.


As shown in Fig. 6.21, the flux estimation errors are limited within 10% with those two

compensation methods, whereas the error is nearly 30% without compensation.

Fig. 6.21 Flux estimation errors comparison for with and without compensation

The dynamic performance of the compensation is also studied by comparing both

simulation and experimental results with and without compensation when the induction

machine speed is changed rapidly from 600 rpm to 1200 rpm. As shown in Fig. 6.22,

large torque and flux estimation error exists without compensation (part a), which

makes the dynamic response slower than that of with backward (part b) or forward (part

c) compensation. In the experiments, it takes 0.65 seconds for the speed rising from 600

rpm to 1200 rpm without compensation (part a), whereas only 0.5 seconds with

backward (part b) or forward (part c) compensation. The torque response is important

for the ISA during starting period. Therefore, compensation should be integrated into

the controller of the ISA.


(i) simulation results (ii) experimental results

(a) Without compensation


(b) With backward compensation



(c) With forward compensation

Fig. 6.22 dynamics of the torque and flux for the DTC-SVM with and without compensation

6.5.2 Generating mode

The performance of the compensation methods is also studied for both steady and

dynamic states under generating operation of the ISA.

6.5.2.1 Steady State performance

Fig. 6.23 compares the dc bus voltage, estimated torque, stator flux and stator current at

no-load state for with and without compensation when the ISA runs with generating

mode at 1500 rpm (rated speed). Although the torque is smoother and its estimation

error is smaller with backward or forward compensations, there is no significant

improvement in the dc bus voltage in the steady state. This is expected because the dc

bus voltage is regulated by PI feedback control action. The stator voltage is very much

larger at 1500 rpm than that at low speed range. Therefore, the error caused by the

voltage drop and the dead-time is no longer comparable with the stator voltage and their

effects on the performance of the system can be ignored at high speed range.


(a) without compensation

(b) with backward compensation


(c) with forward compensation

Fig. 6.23 performance comparison with and without compensation while ISA is generating at

1500 rpm with no-load

6.5.2.2 Dynamic performance

The dynamic performance of the compensation is studied by comparing the

experimental results with and without compensation during load dump. As shown Fig.

6.24, the torque of the induction machine changes from -6 Nm (full load) to about -1

Nm when the load dump happens at 1500 rpm. Very larger torque (almost 4 Nm) and

stator flux estimation errors exist when the compensation methods is not used in part (a)

of Fig. 6.24. However, the dc bus voltage is well regulated by the closed-loop control of

the voltage even without compensation.





(c) with forward compensation

Fig. 6.24 performance comparison with and without compensation during load dump at 1500

rpm

6.6 Conclusion

In this chapter, the effects of switch voltage drops and dead-time on the space vector

modulated DC-AC converter are analyzed. This analysis is necessary because of the low

voltage rating of the induction machine for ISA application. The experimental results

show that the effects of voltage drops and dead-time cause errors in estimated flux and

torque, lead to current distortion and generate oscillation in torque and flux linkage.

The proposed compensation schemes can reduce the above mentioned effects. No extra

hardware is needed for these compensators. Both steady state and dynamic performance

have been analyzed. Experimental results confirm their effectiveness in low speed range

and the torque response has been improved when the ISA runs in motoring mode. This

compensation algorithm has been integrated in the controllers which were described in

the Chapters 4 and 5. In the generating mode of ISA, the improvement of the


compensation on the dc bus voltage regulation is not significant because of closed-loop

control of the voltage. However, the estimation errors of the torque and flux can be

reduced with compensation, which could reduce the torque and flux ripples and increase

the stability of the control system. Therefore, the compensation is necessary for both

motoring and generating modes of the ISA.

The stator flux estimation with compensation is an open-loop type estimator, which is

sensitive to the noise and parameter variations. In addition, the compensation cannot

self-adjust due to open-loop structure. Therefore, a close-loop type estimator is needed

to improve the flux estimation further with self-adaptive ability. Next chapter describes

a close-loop estimator with a sliding mode observer for the ISA system discussed in this

thesis.

Chapter 7 Direct torque controlled ISA with sliding mode observer 136

CHAPTER 7

AN IMPROVED STATOR FLUX ESTIMATION

OF DIRECT TORQUE CONTROLLED


SLIDING MODE OBSERVER

7.1 Introduction

According to the operation principle of the direct torque control, the stator flux linkage

is estimated by integrating the stator voltage. However, a pure integrator has dc offset

and initial value problems. Moreover, uneven voltage drops on the power devices also

introduce errors in stator flux estimation even the compensation method is used. To

solve the problems, digital and programmable-cascaded low-pass filter is developed

[99-101]. These approaches are still open loop flux estimation methods, which are

sensitive to the noise, sensors offset and variation of stator resistance. Therefore, close-

loop flux estimation method is preferred in high performance applications, which are

generally known as flux observers [102]. Many research efforts have been made with

different observers, such as Kalman filter [103], Luenberger observer [104], etc. These

observers have some disadvantages, such as the complex matrix computation algorithm

and sensitivity to noise. Among different close-loop flux estimation schemes, sliding

mode observers have gained much research interests due to their order reduction,

disturbance rejection, simple implementation, and less computational burden [105-111].

In this chapter, a new sliding mode observer is developed to estimate the stator flux for

direct torque controlled integrated starter/alternator based on a simplified induction

machine model in the stationary reference frame. The sliding mode observer without

requiring any speed information is analyzed in detail. The simulation and experimental


results show that the proposed observer is able to deliver more accurate estimation than

open-loop integrator estimator for the stator flux.

7.2 Dynamic Model of Induction Machines

In stationary frame ( α −β ), the dynamic behavior of induction machine can be

described as

ss s s

ss s s

dv R idt

dv R i

dt

αα α

ββ β

ψ⎧ = +⎪⎪⎨ ψ⎪ = +⎪⎩

(7-1)

0

0

rr r m r

rr r m r

dR idt

dR i

dt

αα β

ββ α

ψ⎧ = + −ω ψ⎪⎪⎨ ψ⎪ = + + ω ψ⎪⎩

(7-2)

s s s m r

s s s m r

r m s r r

r m s r r

L i L iL i L i

L i L iL i L i

α α α

β β β

α α α

β β β

ψ = +⎧⎪ψ = +⎪⎨ψ = +⎪⎪ψ = +⎩

(7-3)

( )32e s s s sT P i iα β β α= ψ −ψ (7-4)

where sv α and sv β are the stator voltages in stationary frame, si α , si β , ri α and ri β are the

stator and rotor current in stationary frames, respectively, sαψ , sβψ rαψ and rβψ are the

stator and rotor fluxes, respectively, sR and rR are the stator and rotor resistances, sL ,

rL and mL are the stator, rotor and mutual inductances, respectively. And mω is rotor

speed, P is the number of pole pairs.

Thus


0 0 00 0 0

1 0

10

1 00 1

r s r mm

r s s r ss s

s sr s m rm

s sr s s s r

s ss

s

s

s

s s

R R RL L L L Li i

i id R R Rdt L L L L L

RR

Lv

L v

α α

β β

α α

β β

α

β

⎡ ⎤⎛ ⎞ ω− + −ω⎢ ⎥⎜ ⎟σ σ σ σ⎡ ⎤ ⎡ ⎤⎝ ⎠⎢ ⎥

⎢ ⎥ ⎢ ⎥⎢ ⎥⎛ ⎞ ω⎢ ⎥ ⎢ ⎥⎢ ⎥= ω − + −⎜ ⎟ψ ψ⎢ ⎥ ⎢ ⎥σ σ σ σ⎢ ⎥⎝ ⎠⎢ ⎥ ⎢ ⎥⎢ ⎥ψ ψ−⎣ ⎦ ⎣ ⎦⎢ ⎥⎢ ⎥−⎣ ⎦

⎡ ⎤⎢ ⎥σ⎢ ⎥

⎡ ⎤⎢ ⎥+ ⎢ ⎥⎢ ⎥σ ⎣ ⎦⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦

(7-5)

where

2

1 m

s r

LL L

σ = − (7-6)

In direct torque control scheme, the magnitude of stator flux vector sΨ will be

controlled as constant, that is sΨ = constant, or sddt

Ψ is equal to zero. So,

( )

( )

s s s s s s

s s s s s s

d sin tdtd cos tdt

α β

β α

⎧ ψ ≈ Ψ − ω ω = −ω ψ⎪⎪⎨⎪ ψ ≈ Ψ ω ω = ω ψ⎪⎩

(7-7)

Comparing with (7-1), it is found that

ss s s s s

ss s s s s

d v R idt

dv R i

dt

αα α β

ββ β α

ψ⎧ = − ≈ −ω ψ⎪⎪⎨ ψ⎪ = − ≈ ω ψ⎪⎩

(7-8)

Equation (7-5) can be reorganized as


1 0

10

sr r mm

r s r ss s

s sr m rm

sr s s r

s s s

s s

s

iR RL L L Li id

idt R RL L L L

L

L

α

α β

β α

β

β

α

⎡ ⎤⎛ ⎞ ω ⎡ ⎤− −ω⎢ ⎥⎜ ⎟ ⎢ ⎥σ σ σ⎡ ⎤ ⎝ ⎠⎢ ⎥ ⎢ ⎥≈⎢ ⎥ ⎢ ⎥ ψ⎢ ⎥⎛ ⎞ ω⎣ ⎦ ⎢ ⎥ω − − ⎢ ⎥⎜ ⎟ ψσ σ σ⎢ ⎥ ⎣ ⎦⎝ ⎠⎣ ⎦⎡ ⎤⎢ ⎥σ −ω ψ⎡ ⎤⎢ ⎥+ ⎢ ⎥ω ψ⎢ ⎥ ⎣ ⎦⎢ ⎥σ⎣ ⎦

(7-9)

With small slip, sω is close to mω . So, (7-9) can be simplified as

0

0

sr rm

r s rs s

s sr rm

sr s r

iR RL L Li id

idt R RL L L

α

α β

β α

β

⎡ ⎤⎛ ⎞ ⎡ ⎤− −ω⎢ ⎥⎜ ⎟ ⎢ ⎥σ σ⎡ ⎤ ⎝ ⎠⎢ ⎥ ⎢ ⎥≈⎢ ⎥ ⎢ ⎥ ψ⎢ ⎥⎛ ⎞⎣ ⎦ ⎢ ⎥ω − ⎢ ⎥⎜ ⎟ ψσ σ⎢ ⎥ ⎣ ⎦⎝ ⎠⎣ ⎦

(7-10)

The error between sω and mω can be considered as the disturbance to the system,

which can be compensated by the robust ability of a sliding mode observer.

7.3 Sliding mode stator flux observer

Based on (7-10), the encoder-less sliding mode observer can be designed without speed

signals mω as

1

2

1

2

r rs s s

r s r

r rs s s

r s r

ss s s

ss s s

d R Rˆ ˆ ˆ î i ndt L L L

d R Rˆ ˆ ˆ î i ndt L L L

ˆd v R i c signSdtˆd

v R i c signSdt

α α α

β β β

αα α

ββ β

⎧ ⎛ ⎞= − + ψ +⎪ ⎜ ⎟σ σ⎝ ⎠⎪

⎪ ⎛ ⎞⎪ = − + ψ +⎜ ⎟⎪ σ σ⎝ ⎠⎨⎪ ψ

= − + ⋅⎪⎪

ψ⎪ = − + ⋅⎪⎩

(7-11)

1

2

1 1

2 2

s s

s s

ˆS i iˆS i i

n k sign Sn k sign S

α α

β β

⎧ = −⎪

= −⎪⎨

= ⋅⎪⎪ = ⋅⎩

(7-12)


where si α , si β , sˆ αψ and sˆ βψ are the estimated stator currents and fluxes. c is a positive

number to be chosen. 1S and 2S are the current errors between measured and estimated

stator currents. 1n and 2n are discontinuous functions of the current errors and 0k > .

From (7-10) and (7-11), the error dynamics for current are obtained

r rs s s m s s

r s r

r rs s s m s s

r s r

d R Ri i i k sign idt L L L

d R Ri i i k sign idt L L L

α α α β α

β β β α β

⎧ ⎛ ⎞= − + ψ −ω −⎪ ⎜ ⎟σ σ⎪ ⎝ ⎠

⎨⎛ ⎞⎪ = − + ψ + ω −⎜ ⎟⎪ σ σ⎝ ⎠⎩

(7-13)

where

s s s

s s s

î i iî i i

α α α

β β β

⎧ = −⎪⎨

= −⎪⎩ (7-14)

By choosing Lyapunov candidate function as

( )2 212 s sV i iα β+= (7-15)

The time derivative of Lyapunov function V is

2 2

V i i i is s s s

Rr i i i f i f k i is s s s s sLr

= ⋅ + ⋅α α β β

⎛ ⎞ ⎛ ⎞⎛ ⎞⎜ ⎟= − + + + − +⎜ ⎟ ⎜ ⎟α β α α β β α β⎜ ⎟σ ⎝ ⎠ ⎝ ⎠⎝ ⎠

(7-16)

where

rs m s

s r

rs m s

s r

Rf iL LRf iL L

α α β

β β α

⎧ = ψ −ω⎪ σ⎪⎨⎪ = ψ + ω⎪ σ⎩

(7-17)

if k large enough, i.e. { }k max f , fα β> , then 0V < , until si α and si β are equal to zero,

which means that the estimated currents will converge to their actual values. So, the

sliding mode will occur in the intersection of the surfaces, and si α and si β are equal to

zero.


After sliding mode motion occurs, the error dynamics for flux estimation is obtained

from (7-1) and (7-11)

1

2

s

s

d c signSdt

dc signS

dt

α

β

ψ⎧ = − ⋅⎪⎪⎨ ψ⎪ = − ⋅⎪⎩

(7-18)

The equation (7-18) ensures that the flux errors converge to zero when c is a positive

gain. The valued of c is chosen for the desired convergence rates of the flux error. It

should be noted that a low-pass filter is used instead of direct integration to calculate the

fluxes in (7-11). This approach is introduced to overcome the problems of an ideal

integration such as the initial value effect.

Based on above analysis, the sliding mode observer is developed. Fig. 7.1 shows the

overall structure of direct torque controlled induction machine with sliding mode

observer.

ˆsΨ

( )3 ˆ ˆ2e s s s sT P i iα β β αψ ψ= −

( )7 11Equation −

eT

θ

*sΨ

2 2

1

ˆ ˆ

ˆˆˆ

s s s

s

stg

α α β

β

α

ψ ψ ψ

ψθ ψ

−⎛ ⎞⎜ ⎟⎜ ⎟⎝ ⎠

= +

=

*eT

,s sv vα β

,s si iα β

3 2

ˆ ˆ,s sα βψ ψ

ˆ ˆ,s si iα β

Fig. 7.1 The overall structure of the direct torque controlled induction machine with sliding

mode observer


7.4 Simulation Results

The proposed sliding mode flux observer is compared with the open-loop flux

estimator, which obtains stator flux by direct integration in (7-1). The performance of

these two types of stator flux estimators is investigated under the following cases when

the induction machine runs at 1200 rpm.

1. Stator resistance sR variation

Fig. 7.2 shows that there is a fixed estimation error with the open-loop estimator when

stator resistance varies by 50%. In comparison, the flux estimation error is converged

with sliding mode observer as shown in Fig. 7.3. It indicates that the sliding mode

observer is not sensitive to the sR variation.

Fig. 7.2 Open-loop stator flux estimation with 50% error in sR


Fig. 7.3 Sliding mode flux observer with 50% error in sR

2. dc offset in current measurement

The effect of dc offset in current for flux estimation is also studied. A 3A dc current

offset is deliberately added to stator current iα for open-loop estimator and sliding mode

observer. Due to the effect of integration in open-loop estimator, the estimation error of

stator flux keeps increased with time. Fig. 7.5 shows that the estimation error can be

limited in a small range with sliding mode observer.


Fig. 7.4 Open-loop stator flux estimation with 3A dc current offset

Fig. 7.5 Sliding mode flux observer with 3A dc current offset


3. Dynamic performance

The dynamics of direct torque controlled induction machine with open-loop estimator

and sliding mode observer are compared in Fig. 7.6 and Fig. 7.7. The speed of machine

is accelerated from 600 rpm to 1200 rpm under constant stator flux. They exhibit similar

dynamic response of the torque when there is no sR variation or current offset.

Fig. 7.6 Direct torque controlled induction machine with open-loop stator flux estimator

Fig. 7.7 Direct torque controlled induction machine with sliding mode flux observer


7.5 Experimental Results

In order to compare the performance of the sliding mode observer and open-loop

estimator (i.e. voltage mode estimator in Fig. 6.12), the current mode torque and stator

flux estimators described in Chapter 6 (see Fig. 6.11) are used as reference of the flux

estimation. Therefore, the sliding mode flux observer is used for the control of DTC-

SVM while the current mode stator flux and torque estimator is working in parallel to

verify the estimation accuracy of the stator flux estimation.

In the following sections, the flux estimation error is the difference between current

mode estimator and sliding mode observer (SMO), or open-loop estimator with low

pass filter (voltage mode estimator).

7.5.1 Stator flux and torque estimation in motoring mode

7.5.1.1 Steady state performance of the sliding mode flux observer

The steady state performance of direct torque controlled induction machine with open-

loop estimator and sliding mode observer are compared at 600 rpm with no-load. These

results in Fig. 7.8 and Fig. 7.9 indicate that the flux estimation with sliding mode

observer is more accurate than that of open-loop estimator.

Fig. 7.8 Rotor speed, stator current, and estimated torque and flux at no-load with open-loop

stator flux estimation


Fig. 7.9 Rotor speed, stator current, and estimated torque and flux at no-load with sliding mode

flux observer

7.5.1.2 Estimation error with Stator resistance variation

Fig. 7.10 shows that there is a fixed estimation error with the open-loop estimator when

stator resistance varied by 50%. In comparison, the flux estimation error is smaller with

sliding mode observer as presented in Fig. 7.11. It indicates that the sliding mode

observer is not sensitive with the sR variation.


Fig. 7.10 Open-loop stator flux estimation with 50% sR error

Fig. 7.11 Sliding mode flux observer with 50% sR error


7.5.1.3 Estimation error with dc offset in current measurement

The effect of current dc offset for flux estimation is also studied. 3 A dc current offset is

deliberately added to stator current iα for open-loop estimator and sliding mode

observer. With open-loop estimator, there exist constant errors at steady state. Fig. 7.13

shows that the estimation error can be limited in a small range with sliding mode

observer.

Fig. 7.12 Open-loop stator flux estimation with 3A dc current offset


Fig. 7.13 Sliding mode flux observer with 3A dc current offset

7.5.1.4 Effect of estimation errors on the dynamic performance

The dynamics of direct torque controlled induction machine with open-loop estimator

and sliding mode observer are compared in Fig. 7.14 and Fig. 7.16. The speed of

machine is accelerated from 600 rpm to 1200 rpm under constant stator flux. During

torque transient, the actual torque oscillates and deviates from the reference due to

inaccurate flux estimation by open-loop estimator. In comparison, the torque dynamic

behavior is better and the estimation error is small.


Fig. 7.14 Dynamic performance with open-loop stator flux estimation

Fig. 7.15 Estimation errors with open-loop stator flux estimation


Fig. 7.16 Dynamic performance with sliding mode flux observer

Fig. 7.17 Estimation errors with sliding mode flux observer


Fig. 7.18 Current estimation with sliding mode flux observer

As shown in Fig. 7.15 and Fig. 7.17, the estimation error of the open-loop estimator is

larger than that of the sliding mode observer. Fig. 7.18 shows the stator current

estimation of the sliding mode observer, which proves its tracking ability. The

oscillation of the estimated current results from the sliding mode operation of the

observer.

7.5.2 Stator flux and torque estimation in generating mode

The performance of the Sliding Mode Observer (SMO) is also studied for both steady

and dynamics states under generating operation of the ISA.

7.5.2.1 Steady State performance of the sliding mode flux observer

Fig. 7.19 compares the dc bus voltage, estimated torque, stator flux and stator current at

no-load state for with and without compensation, and with SMO when the ISA runs

with generating mode at 1500 rpm (rated speed). Compared to the cases of without/with

backward, the torque and flux estimation errors are greatly reduced. However, there is

no significant improvement in the dc bus voltage in the steady state due to the feedback


regulation of the voltage. The stator voltage is larger at 1500 rpm than that at low speed

range. Therefore, the error caused by the voltage drop and the dead-time is no longer

comparable with the stator voltage and their effects on the performance of the system

are not significant at high speed range.





(c) with SMO

Fig. 7.19 performance comparison without and with compensation, and SMO while ISA is

generating at 1500 rpm

7.5.2.2 Effect of estimation errors on the dynamic performance

The dynamic performance of the SMO is also studied by comparing the experimental

results during load dump. As shown Fig. 7.20, the torque of the induction machine is

increased from -6 Nm (full load) to about -1 Nm when the load dump happens at 1500

rpm. Very large torque (almost 4 Nm) and stator flux estimation errors exist when the

compensation method is not used in part (a) of Fig. 7.20. However, the dc bus voltage is

well regulated by the closed-loop control of the voltage even without compensation.

Compared to open-loop estimator with/without compensation, the torque and stator flux

estimation errors are reduced, which is helpful to stabilize the control system.




(c) with SMO

Fig. 7.20 performance comparison with/without compensation and with SMO during load dump

at 1500 rpm


7.6 Conclusion

This chapter presents a sliding mode flux observer for a direct torque controlled

integrated starter/alternator. The stator flux estimation accuracy is guaranteed when the

error between the actual current and observed current converges to zero. The algorithm

of the sliding mode observer is based on simple computation in the stationary frame,

which cost less time. Both simulation and experimental results confirm that the

proposed sliding mode observer is robust to the stator resistance variation and sensor

offset.

Experimental results confirm the effectiveness of SMO in low speed range and the

torque response has been improved when the ISA runs in motoring mode. Fast starting

of an ISA can thus be achieved with SMO. In the generating mode of ISA, the

improvement of the compensation on the dc bus voltage regulation is not significant

because of closed-loop control of the voltage. However, the estimation errors of the

torque and flux can be reduced with SMO, which could reduce the torque and flux

ripples and increase the stability of the control system. In addition, SMO is a close-loop

type estimator with self-adaptive ability and it is not sensitive to the variation of

parameters. Therefore, SMO can further improve the performance of the ISA for both

motoring and generating modes.

Chapter 8 Efficiency improvement for ISA with power factor control 159

CHAPTER 8

EFFICIENCY IMPROVEMENT FOR


POWER FACTOR CONTROL

8.1 INTRODUCTION

For the application of ISA, the induction machine works on both motoring and

generating state. The efficiency is an important factor to evaluate the performance. The

efficiency of induction machine is low at light load with rated flux. Because the ISA

operates in a wide load range, the efficiency can be improved significantly by optimal

control. The loss of an induction machine includes copper (Winding) losses; Core losses

and friction & windage losses. The copper and core losses belong to electromagnetic

losses, which can be minimized by optimal control of the flux level in the machine

[112].

Extensive work has been done previously for the adaptation of the flux. Most of them

are based on the following three methods.

1) Search method, where the output power of the machine is kept constant while

the flux level is iteratively adapted to find a minimum input power [21, 113-115]. It is

not a good choice for industry application because the slow adaptation, continuous

disturbances in the torque and the need for precise load information.

2) Loss model based method [116, 117] is a nature solution for field oriented

controlled machine whose control is already based on the knowledge of the machine.

Model-based control provides fast adaptation of the flux, but it requires knowledge of

the machine parameters, and it requires more computation than the other methods.


3) Power factor control method is based on ( )cos ϕ control. Compared with the

above two methods, it is a simple method requiring any speed or load information, and

its regulation speed is faster. Power factor control is implemented in both scalar

controlled (V/f) [118] and vector controlled drives [119-121]. It shows the drive loss

with power factor control is very close to the minimized loss. However, the application

of power factor control in direct torque control has not been reported yet.

In this chapter, a novel efficiency-optimized scheme based on power factor tuning for

direct torque controlled integrated starter/alternator is proposed. The power factor of the

induction machine is controlled to track the pre-determined power factor reference. A

new structure of the power factor controller is proposed. The power loss is reduced with

proper power factor under difference conditions. It is a simple method without requiring

any speed or load information, and it is a fast adaptation method. So, it is a good choice

for industry application.

This chapter is organized as follows. Section 8.2 introduced the loss model of the

induction machine. The principle of power factor control for direct controlled ISA is

presented in Section 8.3. Modeling analysis and experimental results are given in

Section 8.4-8.5. The conclusion is drawn in Section 8.6.

8.2 Induction Machine Loss Model

In stationary frame ( α −β ), the dynamic behaviour of induction machine can be

described as

ss s s

ss s s

dv R idt

dv R i

dt

αα α

ββ β

ψ⎧ = +⎪⎪⎨ ψ⎪ = +⎪⎩

(8-1)

0

0

rr r m r

rr r m r

dR idt

dR i

dt

αα β

ββ α

ψ⎧ = + −ω ψ⎪⎪⎨ ψ⎪ = + + ω ψ⎪⎩

(8-2)


s s s m r

s s s m r

r m s r r

r m s r r

L i L iL i L i

L i L iL i L i

α α α

β β β

α α α

β β β

ψ = +⎧⎪ψ = +⎪⎨ψ = +⎪⎪ψ = +⎩

(8-3)

( )32e s s s sT P i iα β β α= ψ −ψ (8-4)

where sv α and sv β are the stator voltages in stationary frame, si α , si β , ri α and ri β are the

stator and rotor current in stationary frames, respectively, sαψ , sβψ , rαψ and rβψ are

the stator and rotor fluxes, respectively, sR and rR are the stator and rotor resistances,

sL , rL and mL are the stator, rotor and mutual inductances, respectively. And mω is

rotor speed, P is the number of pole pairs.

The total copper loss is

( ) ( )2 2 2 232copper s s s r r rP i i R i i Rα β α β⎡ ⎤= + + +⎣ ⎦ (8-5)

The core loss contains hysteresis and eddy current losses, whose density [122] can be

express as

2

2 2

h h m

e e m

P K fB W kg

P K f B W kg

⎧ =⎪⎨

=⎪⎩ (8-6)

where hK and eK are the hysteresis and eddy current loss coefficients, f is the

frequency, mB is the maximum flux density.

mB is determined by the flux level in the magnetic field. Therefore, the flux level has

significant effect on the core loss with higher speed at light load. That is the case when

the integrated starter/alternator is generating at high speed.

8.3 Principle of Power Factor Control

Fig. 8.1 shows the complete structure of the direct torque controlled integrated

starter/alternator.


dcV

*dcV

startingT +

−

dcV

ˆsΨ

( )3ˆ ˆ ˆ2e s s s sT P i iα β β αψ ψ= −

eT

θ

*sΨ

1

2 2ˆ ˆ ˆ

ˆˆˆ

s s s

s

stg

α β

β

α

ψ ψ

ψθ ψ

−⎛ ⎞⎜ ⎟⎜ ⎟⎝ ⎠

Ψ

=

= +

*eT

,s sv vα β

,s si iα βˆ ˆ,s sα βψ ψ

3 2( )8 1Equation −

*PF

PF∧

Fig. 8.1 The overall structure of the direct torque controlled integrated starter/alternator

The torque and flux are regulated by the controllers. The ISA system includes

starting/generating state switch which simulates the operation of ISA from starter to

generator. After the switch changes to generating mode, the voltage regulator will take

effect to keep the dc bus voltage as 42 V and the torque reference will be negative. As

shown in Fig. 8.1, only one voltage sensor for dc bus voltage and two current sensors

for stator current are adopted in proposed scheme. The voltage and current signals are

used for stator flux estimation. The stator flux vector is estimated in the stationary frame

avoiding co-ordination transformation and involvement of more machine parameters.

The estimation algorithm is given in (8-1). In practice, the pure integrator in (8-1) could

be saturated due to the noise or measurement error inherently present in the current

sensor. Therefore, a low pass filter should be used in stead for the flux estimation. In

Fig. 8.1, the voltage signal is also used as voltage feedback to maintain the dc bus

voltage as 42 V.


As shown in Fig. 8.1, the reference flux is obtained from the Power Factor (PF)

controller by maintaining the power factor at given PF reference.

In this scheme, the power factor controller is design as in Fig. 8.2. A negative gain is

used because power factor will be increased with lower flux level.

*PF

PF∧

PI1−*

sΨ

+−

+

Rated Flux

Minmum Flux

Rated Flux

Fig. 8.2 Power factor controller

The reference voltage vector is adopted for the estimation of power factor without using

line voltage sensors. The power factor can be calculated by [123, 124]

( ) ( )2 2 2 2 2 2

s s s s

s s s s

v i v iPPFP Q v v i i

α α β β

α β α β

+= =

+ + + (8-7)

where P and Q are the instantaneous active power and reactive power of the induction

machine, respectively.

8.4 Modeling Results

A 1kW/22V integrated starter/alternator is modelled by Simulink/Matlab to verify the

proposed power factor scheme. The parameter of the induction machine is given in

Appendix. The rated flux used in simulation is 0.0572. Constant PF reference is chosen

as 0.75.

Fig. 8.3 shows the variation of the power factor under different loads when the

induction machine is running at 1500 rpm with constant flux. The power factor is low

when the load is small. So, the efficiency of the induction machine is low under smaller

load.


Power factor@1500 rpm

00.10.20.30.40.50.60.70.80.9

0 0.2 0.4 0.6 0.8 1Load ( x100% rated Te)

Fig. 8.3 Power factor of the induction under different loads

Fig. 8.4 Stator voltage, stator and rotor currents with 30% rated load


Fig. 8.4 shows the stator voltage, stator current and rotor current when the power factor

controller is added to the system. With power factor control, the flux level is reduced

with decreased stator voltage. The rotor current is increased with low flux level.

In order to evaluate the performance of the power factor control under different loads,

the power loss is calculated in percentage by (8-8) with considering of (8-5) and (8-6).

( ) ( )( ) ( )

2

2 2

2 2

100

100

100

100

core

core _ rated

m

m _ rated

copper

copper _ rated

s s r r

s rated s r rated r

Pcore loss% %P

%

Pcopper loss% %

P

i R i R%

i R i R− −

⎧ = ×⎪⎪⎪ ⎛ ⎞ψ⎪= ×⎜ ⎟⎪ ⎜ ⎟ψ⎪ ⎝ ⎠⎨⎪ = ×⎪⎪⎪ +⎪= ×

+⎪⎩

(8-8)

where core _ ratedP and copper _ ratedP are the core loss and copper loss at rated load,

respectively.

It is shown in Fig. 8.5 that the core loss is greatly reduced by power factor control.

More power is saved by power factor control under lower load. Fig. 8.6 indicates that

the copper loss also deceased by power factor control within low load range (< 50%

rated load). Because more current is required to maintain the higher electromagnetic

torque, the copper loss is increased with power factor control when the load is larger

than 50% rated load.

Core loss% @ 1500rpm

0

20

40

60

80

100

120

0 0.2 0.4 0.6 0.8 1 1.2

Load (x100% rated Te)

core loss%-with PFcontrolcore loss%-withoutPF control

Fig. 8.5 Core loss percentage with and without power factor control


Copper loss% @1500rpm

0

20

40

60

80

100

120

0 0.2 0.4 0.6 0.8 1 1.2

Load (x100% rated Te)

copper loss%-with PFcontrolcopper loss%-withoutPF control

Fig. 8.6 Copper loss percentage with and without power factor control


In order to evaluate the power factor controller, the efficiency of the induction machine

is tested with the ISA experimental platform as shown in Fig. 2.5. The electrical power

of the induction machine is obtained with YOKOGAWA Power Analyzer (PZ4000).

The mechanical torque of the induction machine is calculated by the torque of the DC

drive machine and the torque to overcome friction loss.

Both motoring and generating modes are investigated for the efficiency improvement.

The efficiencies in different modes are calculated by (8-9)

( )

( )

Analyzer Analyzer

Analyzer Analyzer

100 100

100 100

DCM friction mreal mM

Greal m DCM friction m

T TT % %P P

P P% %

T T T

⎧ + ωωη = × = ×⎪⎪⎨⎪η = × = ×⎪ ω − ω⎩

(8-9)

where Mη and Gη are the efficiencies of the induction machine in motoring and

generating modes; AnalyzerP is the electrical power measured from the power analyzer;

mω is the rotor speed of the induction machine in rad/s; realT , DCMT and frictionT are the

real torque of the induction machine, the real torque of the dc drive machine, and the

torque caused by the friction loss, respectively.


8.5.1 Motoring mode

It is shown in Fig. 8.7 that the efficiency of the induction machine in motoring mode is

improved by power factor controller. Specially, the efficiency is increased almost 10 %

at small load range (0 - 30% rated load).

Fig. 8.7 Efficiency comparison of the induction machine with and without Power Factor (PF)

control in motoring mode at 1200 rpm and 1500 rpm

Similar with modeling, the transients of the regulation of the power factor controller is

also recorded in experiment at 1200 rpm. As shown in Fig. 8.8, the stator voltage is

reduced gradually while the power factor controller is taking effect. Therefore, the core

loss of the machine will be minimized with reduced flux level.


Fig. 8.8 Transients of the regulation of the power factor controller

8.5.2 Generating mode

In generating mode, the efficiencies of the induction machine are compared in Fig. 8.9

when it is running at 1500 rpm and 2100rpm. It is indicated that the efficiency of the

induction machine in generating mode is also improved by power factor controller. The

efficiency improvements are significant in the low load range (0-30% rated load) as

expected from the analysis.


Fig. 8.9 Efficiency comparison of the induction machine with and without power factor control

in generating mode at 1500 rpm and 2100 rpm

8.6 Conclusion

A loss minimization scheme for the direct torque controlled integrated starter/alternator

is proposed in this chapter. With proper power factor control, both core loss and copper

loss are minimized under different loads. It provides a simple solution for the efficiency

improvement of the induction machine without speed or load information. The results

confirm the effectiveness of the proposed control scheme.

Chapter 9 Conclusions 170

CHAPTER 9

CONCLUSIONS

In this thesis, an integrated starter/alternator (ISA) for automobiles based on direct

torque controlled induction machines has been modeled, analyzed, designed and

implemented. The simulation and experimental results show that effective control of the

ISA has been achieved for both starting and generating modes. This study provides a

high performance control solution for an ISA in the future 42-V PowerNet application,

other than the widely applied rotor flux oriented control scheme [16, 20, 24-26, 38]

which is sensitive to the variation of machine parameters and requires accurate speed

sensor signal for the flux orientation and decoupling. Considering the moist, hot and

severe environment in automobiles, direct torque control scheme is more reliable and

attractive without involving many machine parameters and requiring speed sensor

signal for the control of torque and flux.

In summary, the contributions made in this thesis are:

• Investigation on a classical direct torque controlled integrated starter/alternator

based on switching table

• Investigation and experimental verification of two improved direct torque

controlled integrated starter/alternator schemes based on space vector

modulation (DTC-SVM)

• Theoretical analysis of two improved DTC-SVM schemes and design of their

controllers

• Design, analysis and implementation of an encoder-less sliding mode observer

for the stator flux estimation

• Design, analysis and implementation of a power factor control structure to

improve the efficiency of the induction machine in a prototype ISA system


• Development of compensation methods of the non-linear characteristics of the

inverter used in an ISA

The classic direct torque controlled induction generator for integrated starter alternator

application has been analyzed and verified with simulation and experiments in Chapter

3. Discrete hysteresis comparator is used to keep the switching frequency of the inverter

constant. High flux and torque ripples results from the look-up table of the voltage

vectors and the hysteresis comparators of the torque and flux. Therefore, higher

sampling time of the control system has to be used (25 sμ or less) [73]. All the above

difficulties can be eliminated by using a voltage space vector modulator instead of the

switching table [81-91].

In this thesis, an improved torque controller of induction machine based on direct

control of stator flux linkage vector is presented in Chapter 4. The fundamental

relationship between the rotating speed of the stator flux linkage and torque is analyzed

and the design principle of controller is presented. Parameters of PI controller are easily

found using the proposed design principle. Robust design of the controller ensures the

system is not sensitive to the variation of rotor resistance. Fixed switching frequency

and low torque ripple are obtained with the combination of PI control and space vector

modulation (SVM) method. Satisfactory modeling and experimental results indicate the

feasibility of the proposed direct flux vector control scheme for induction machines.

The control scheme employs encoderless torque control structure, and eliminates the

disturbance of speed to the torque controller successfully. The controller gives good

torque and flux control performance. The direct flux vector controlled scheme of

induction generator has been proposed and verified for the future 42 V automobiles

application. A simple structure with only one Proportional-Integral (PI) controller is

shown to implement the torque and flux control adequately. By controlling the

electromagnetic torque of the induction machine, the required dc bus voltage can be

well regulated within the 42 V PowerNet specifications.

Another DTC concept based improved direct torque and flux control of the integrated

starter/alternator is also proposed in Chapter 5. This control scheme has been analyzed

and verified with simulation and experiments. Compared to the direct flux vector

control scheme proposed in Chapter 4, this scheme is a little more complex due to

transformation from stator flux frame ( d q− ) to stationary frame (α β− ). However, the


extra complexity is minor because no mechanical sensor signal is required. The direct

flux vector control presented in Chapter 4 controls the rotating speed of the stator flux

vector by a torque feedback loop. The amplitude of the stator flux vector is regulated

indirectly. In Chapter 5, the torque and the amplitude of the stator flux are regulated by

two independent control loops. In addition, only derivative of a dc quantity is involved

in the calculation of the commanded voltage vector, whereas derivative of an ac

quantity is involved in the direct flux vector control scheme. Thus, this scheme is not

sensitive to the noise which is generated when the flux vector is differentiated [63]. The

simulation and experimental results show that the scheme has achieved similar

performance to the direct flux vector control scheme. This scheme provides an

alternative solution for the ISA application with direct torque control concept.

The voltage rating of the induction machine used in this study is very low (22 V). The

effects of voltage drops on the power devices and dead-time of the converter are

significant when the stator flux is estimated by reconstruction of the stator voltage

vector from the gating signals and the dc link voltage. This non-linear behaviour

introduces large error in the stator flux estimation leading to slower dynamic response

and instability due to the oscillation of torque and flux. The effects of voltage drops and

dead-time on the space vector modulated DC-AC converter are analyzed in Chapter 6.

Compensation schemes have been proposed to reduce the abovementioned effects.

Moreover, the compensation of the non-linear behaviour of the converter has been

implemented through experimental works. No extra hardware is needed for these

compensators. Experimental results confirm that the compensation is necessary for both

motoring and generating modes of the ISA. These compensation algorithms have been

integrated in the controller in the direct torque controlled ISA system discussed in the

Chapters 4 and 5.

The stator flux estimation with compensation discussed in Chapter 3-6 is an open-loop

type estimator, which is sensitive to the offset in sensors and variation of stator

resistance. In Chapter 7, a closed-loop sliding mode stator flux observer for a direct

torque controlled integrated starter/alternator has been developed to improve the stator

flux estimation. The sliding mode stator flux observer is based on the error between the

actual current and observed current converging to zero. The algorithm of the sliding

mode observer is simple and all computation is in the stationary frame, which leads to


low computation burden of the DSP. Both Simulation and experimental results confirm

that the proposed sliding mode observer is insensitive to the stator resistance variation

and measurement offset in sensor outputs.

In this study, DTC schemes for the control of the integrated starter/alternator are

compared with a rotor flux oriented scheme (RFOC-ISA). Three direct torque controlled

induction machine for ISA system are presented. They are: Classic DTC-ST in Chapter

3 (DTC-ST-ISA), two DTC-SVM schemes in Chapter 4 (DFC-ISA) and Chapter 5

(DTFC-ISA). These schemes are compared with RFOC for ISA application.

Table 9.1 lists general comparison of the control schemes for the ISA discussed in this

thesis in terms of the control ability, structure, etc. The shadowed parts indicate the

drawbacks of the schemes.

Table 9.1 Comparison of different control schemes for the ISA

DTC-ST-ISA DFC-ISA DTFC-ISA RFOC-ISA

Torque control

Directly torque

control by

hysteresis

comparator

Directly torque

control by PI

action

Directly torque

control by PI

action

Indirectly torque

control by PI

control of q axis

current

Flux control

Directly flux

control by

hysteresis

comparator

Indirectly flux

control by PI

action

Directly flux

control by PI

action

Indirectly flux

control by PI

control of d axis

current

dc bus voltage

control

Satisfied ISA

specifications

Satisfied ISA

specifications

Satisfied ISA

specifications

Satisfied ISA

specifications

PWM

generation Not required SVM SVM SVM

Switching

frequency

Variable (could

be constant with

discrete

hysteresis

comparator)

Constant Constant Constant

Current &

Torque ripples

Highest (the

maximum peak-

peak torque

low(the maximum

peak-peak torque

ripple is 16.7 % of

low(the maximum

peak-peak torque

ripple is 16.7 % of

low(the maximum

peak-peak torque

ripple is 14.7 % of


ripple is 183.3 %

of rated torque

with 150sT sμ= )

rated torque with

150sT sμ= )

rated torque with

150sT sμ= )

rated torque with

150sT sμ= )

Current

controller Not required Not required Not required Required

Coordinate

transformation

using rotor

speed signal

Not required Not required Not required

Rotor flux vector

frame to stationary

frame ( e ed q to

αβ

Induction

machine’s

parameters

involved

sR sR sR sR , rR , sL ,

rL and mL

Flux orientation

and decoupling

algorithm

Not required Not required Not required required

Implementation

Complexity simplest Medium Medium complex

Flux estimation Voltage mode:

LPF; SMO

Voltage mode:

LPF; SMO

Voltage mode:

LPF; SMO

current mode;

could be voltage

mode, but it still

involves many

induction machine

parameters ( sR ,

rR , sL , rL and

mL )

High speed

performance Good Good Good

Good, but has

instability during

high-speed

generation [125]

It can be concluded that both DTC and RFOC schemes can effectively control the

induction machine for the ISA application. By considering the parameters dependency,

complexity of the structure and cost, it is clear that DTC is superior to FOC.


A tradeoff between performance and simplicity is needed for the comparison of DTC-

ST and DTC-SVM schemes (DFC-ISA, DTFC-ISA). Although lower flux & torque

ripples and constant switching frequency are achieved with DTC-SVM scheme, SVM

unit makes the control structure complex. On the other hand, DTC-ST scheme require

fast sampling frequency to minimize the flux & torque ripples within acceptable limits.

Therefore, DSP interfaced with hardware to determine the switching logic of the

inverter, such as ASIC (Application-Specific Integrated Circuit) [73], FPGA (Field-

Programmable Gate Array), and CPLD (Complex Programmable Logic Device) is

needed for DTC-ST scheme.

High efficiency of the automotive electrical system is required for the economy of fuel.

A loss minimized scheme for the direct torque controlled integrated starter/alternator is

thus proposed in Chapter 8. With proper power factor control, both core loss and copper

loss are minimized under different loads. It provides a simple solution for the efficiency

improvement of the induction machine without requiring speed or load information

when the load is small. The experimental results confirm the effectiveness of the

proposed control scheme.

The effectiveness of the direct torque controlled induction machine for an integrated

starter/alternator system has thus been confirmed and well supported by the studies

presented in this thesis.

9.1 Suggestions for future work

9.1.1 Machine

The induction machine used in an ISA runs in both motoring and generating modes.

Therefore, special design of the induction machine is needed to satisfy the requirement

of the ISA during starting (high torque) and generating (constant power over a wide

speed range). In addition, higher voltage rating than 22 V of the machine is worthy of

further investigation in an ISA application with different topologies. The power losses

on the connection and winding of an induction machine and semiconductor switches

could be reduced with higher voltage rating.


9.1.2 Power converter

High electrical power requirement (6–15 kW) of the future 42-V PowerNet imposes

great challenge on the bidirectional power converter in an ISA system. The bidirectional

power converter has to handle several hundred-amperes of the current with compact size

due to the limited space in automobiles. Thermal design of the converter is also an

important issue for the environment of a vehicle, which can be very hostile. Research

related to this area has been reported in papers [60].

With higher voltage rating of the machine, investigation on new bidirectional DC-DC-

AC converter topologies is required for the ISA application. Comparison study on this

topic has been presented in paper [33].

9.1.3 Direct torque controlled ISA based on permanent magnet synchronous

machine

High efficiency makes the permanent magnet synchronous machine (PMSM) also a

strong candidate for an ISA system. The direct torque control for PMSM drives has

been studied in the last decade [80, 85, 86, 88, 126-129], but not for an ISA application.

Direct torque controlled ISA based on PMSM is worthy of investigation. Recently,

direct torque and flux control of a permanent magnet-assisted reluctance synchronous

machine (PM–RSM) for the ISA system in hybrid electric vehicles has been reported

[30].

Many new innovations in machine design, converter and control may therefore be

possible.

References 177

REFERENCES

[1] J. M. Miller, A. Emadi, A. V. Rajarathnam, and M. Ehsani, "Current status and future trends in More Electric Car power systems," in Vehicular Technology Conference, 1999 IEEE 49th, 1999, pp. 1380-1384 vol.2.

[2] C. P. Cho and R. H. Johnston, "Electric motors in vehicle applications," in Vehicle Electronics Conference, 1999. (IVEC '99) Proceedings of the IEEE International, 1999, pp. 193-198 vol.1.

[3] "Automotive Electrical & Electronics," DuPont Automotive, 2006.

[4] G. Leen and D. Heffernan, "Expanding automotive electronic systems," Computer, vol. 35, pp. 88-93, 2002.

[5] J. M. M. a. N. T. J.G. Kassakian, "Automotive Electronics Power Up," IEEE Spectrum, 2000.

[6] D.-I. H.-D. Hartmann, "Standardization of the 42 V Powernet - Intention, Values," sci-worx GmbH, Hannover, 2000.

[7] J. G. Kassakian, H. C. Wolf, J. M. Miller, and C. J. Hurton, "Automotive electrical systems circa 2005," Spectrum, IEEE, vol. 33, pp. 22-27, 1996.

[8] J. G. Kassakian, J. M. Miller, and N. Traub, "Automotive electronics power up," Spectrum, IEEE, vol. 37, pp. 34-39, 2000.

[9] T. A. Keim, "42 Volts –The View from Today," Massachusetts Institute of Technology, 2005.

[10] "Road vehicles - Conditions for electrical and electronic equipment for a 42 V powernet – Part -2: Electrical loads," lSO/WD 21848.2 I S 0 Working Document, April 2000.

[11] P. Nicastri and H. Huang, "42V PowerNet: Providing the Vehicle Electrical Power for the 21st Century," SAE Technical Papers, 2000.

[12] R. Kunkel, R. Spriessler, and S. Young, "Revision of the vehicle transient test specification ISO 7637," in Electromagnetic Compatibility for Automotive Electronics (Ref. No. 1999/134), IEE Seminar on, 1999, pp. 3/1-3/10.

[13] C. S. Namuduri, B. V. Murty, and M. G. Reynolds, "Load dump transient control of a 42 V automotive generator," in Power Electronics Specialists Conference, 2004. PESC 04. 2004 IEEE 35th Annual, 2004, pp. 389-394 Vol.1.

[14] J. Neubert, "Powering up [42 V automotive electronics]," IEE Review, vol. 46, pp. 21-25, 2000.

[15] E. C. Lovelace, T.M. Jahns, and J. L. K. a. J. H. Lang, "An Interior PM Starter Alternator for Automotive Applications," in International Conference on Electric Machines and Drives Conference, Istanbul, Turkey, 1998.

References 178

[16] J. M. Miller, A. R. Gale, P. J. McCleer, F. Leonardi, and J. H. Lang, "Starter-alternator for hybrid electric vehicle: comparison of induction and variable reluctance machines and drives," in Industry Applications Conference, 1998. Thirty-Third IAS Annual Meeting. The 1998 IEEE, 1998, pp. 513-523 vol.1.

[17] C. P. Cho and D. R. Crecelius, "Vehicle alternator/generator trends toward next millennium," in Vehicle Electronics Conference, 1999. (IVEC '99) Proceedings of the IEEE International, 1999, pp. 433-438 vol.1.

[18] G. S. Kahlon, R. Mohan, N. Liu, and H. Rehman, "A case study of starting power requirement: for Visteon Integrated Starter-Alternator system," in Digital Avionics Systems Conference, 1999. Proceedings. 18th, 1999, pp. 8.B.4-1-8.B.4-7 vol.2.

[19] J. W. Kolar, R. S. Wieser, and H. Ertl, "Analysis of a wide speed range starter/alternator system based on a novel converter topology for series/parallel stator winding configuration," in Industry Applications Conference, 1999. Thirty-Fourth IAS Annual Meeting. Conference Record of the 1999 IEEE, 1999, pp. 2631-2641 vol.4.

[20] H. Rehman, X. Xu, N. Liu, G. S. Kahlon, and R. J. Mohan, "Induction motor drive system for the Visteon Integrated Starter-Alternator," in Industrial Electronics Society, 1999. IECON '99 Proceedings. The 25th Annual Conference of the IEEE, 1999, pp. 636-641 vol.2.

[21] R. B. Sepe, Jr., J. M. Miller, and A. R. Gale, "Intelligent efficiency mapping of a hybrid electric vehicle starter/alternator using fuzzy logic," in Digital Avionics Systems Conference, 1999. Proceedings. 18th, 1999, pp. 8.B.2-1-8.B.2-8 vol.2.

[22] R. F. Wall and H. L. Hess, "Induction machines as an alternative for automotive electrical generation and starting systems," in Electric Machines and Drives, 1999. International Conference IEMD '99, 1999, pp. 499-501.

[23] F. Leonardi and M. Degner, "Integrated starter generator based HEVs: a comparison between low and high voltage systems," in Electric Machines and Drives Conference, 2000. IEMDC 2001. IEEE International, 2001, pp. 622-628.

[24] C. Shaotang, B. Lequesne, R. R. Henry, X. Yanhong, and J. J. Ronning, "Design and testing of a belt-driven induction starter-generator," Industry Applications, IEEE Transactions on, vol. 38, pp. 1525-1533, 2002.

[25] C. P. Mudannayake and M. F. Rahman, "An integrated starter alternator for the 42 V PowerNet," in Power Electronics and Drive Systems, 2003. PEDS 2003. The Fifth International Conference on, 2003, pp. 648-653 Vol.1.

[26] D. Spillane, D. O' Sullivan, M. G. Egan, and J. G. Hayes, "Supervisory control of a HV integrated starter-alternator with ultracapacitor support within the 42 V automotive electrical system," in Applied Power Electronics Conference and Exposition, 2003. APEC '03. Eighteenth Annual IEEE, 2003, pp. 1111-1117 vol.2.

[27] I. Boldea, "Starter/Alternator systems for HEV and their control: a review," KIEE International Transactions on Electrical Machinery and Energy Conversion Systems, vol. Volume 4-B pp. pp.157-169, 2004.

References 179

[28] W. Cai, "Comparison and review of electric machines for integrated starter alternator applications," in Industry Applications Conference, 2004. 39th IAS Annual Meeting. Conference Record of the 2004 IEEE, 2004, p. 393.

[29] J. Zhang, M. F. Rahman, and L. Tang, "A direct flux controlled induction generator with space vector modulation for integrated starter alternator," in Industrial Electronics Society, 2004. IECON 2004. 30th Annual Conference of IEEE, 2004, pp. 330-334 Vol. 1.

[30] I. Boldea, C. I. Pitic, C. Lascu, G. D. Andreescu, L. Tutelea, F. Blaabjerg, and P. Sandholdt, "DTFC-SVM motion-sensorless control of a PM-assisted reluctance synchronous machine as starter-alternator for hybrid electric vehicles," Power Electronics, IEEE Transactions on, vol. 21, pp. 711-719, 2006.

[31] K. Tae-Suk, L. Dong-Hoon, and S. Seung-Ki, "Reduction of engine torque ripple at starting with belt driven integrated starter generator," in Applied Power Electronics Conference and Exposition, 2005. APEC 2005. Twentieth Annual IEEE, 2005, pp. 1035-1040 Vol. 2.

[32] A. Walker, P. Anpalahan, P. Coles, M. Lamperth, and D. Rodgert, "Automotive integrated starter generator," in Power Electronics, Machines and Drives, 2004. (PEMD 2004). Second International Conference on (Conf. Publ. No. 498), 2004, pp. 46-48 Vol.1.

[33] L. Xu and J. Liu, "Comparison study of DC- DC-AC combined converters for integrated starter generator applications," in Power Electronics and Motion Control Conference, 2004. IPEMC 2004. The 4th International, 2004, pp. 1130-1135 Vol.3.

[34] K. Bolenz, "Design modifications of the electrical system to use intermittent engine operation," in Machines for Automotive Applications (Digest No. 1996/166), IEE Colloquium on, 1996, pp. 2/1-2/7.

[35] J. D. Dimech, "Standardized automotive load dump testing," in Electromagnetic Compatibility, 1991. Symposium Record. IEEE 1991 International Symposium on, 1991, pp. 355-359.

[36] P. Le Bars and A. Regini, "42 V load dump transient and centralised active suppression," in Passenger Car Electrical Architecture (Ref. No. 2000/088) IEE Seminar, 2000, pp. 4/1-4/3.

[37] T. Efland, M. Manternach, A. Marshall, and J. Mings, "The load dump [automobiles]," in Electronic Applications in Transportation, 1990. IEEE Workshop on, 1990, pp. 73-78.

[38] M. Naidu and J. Walters, "A 4-kW 42-V induction-machine-based automotive power generation system with a diode bridge rectifier and a PWM inverter," Industry Applications, IEEE Transactions on, vol. 39, pp. 1287-1293, 2003.

[39] B. H. Shaotang Chen; Lequesne, R.R.; Yanhong Xue; Ronning, J.J., "Design and testing of a belt-driven induction starter-generator," Industry Applications, IEEE Transactions on, vol. 38, pp. 1525-1533, 2002.

[40] S. Evon and R. Schiferl, "Direct-drive induction motors: using an induction motor as an alternative to a motor with reducer," Industry Applications Magazine, IEEE, vol. 11, pp. 45-51, 2005.

References 180

[41] E. Richter and C. Ferreira, "Performance evaluation of a 250 kW switched reluctance starter generator," in Industry Applications Conference, 1995, pp. 434-440 vol.1.

[42] A. V. Radun, C. A. Ferreira, and E. Richter, "Two-channel switched reluctance starter/generator results," Industry Applications, IEEE Transactions on, vol. 34, pp. 1026-1034, 1998.

[43] S. R. MacMinn and W. D. Jones, "A very high speed switched-reluctance starter-generator for aircraft engine applications," in Aerospace and Electronics Conference, 1989. NAECON 1989., Proceedings of the IEEE 1989 National, 1989, pp. 1758-1764 vol.4.

[44] S. R. Jones and B. T. Drager, "Sensorless switched reluctance starter/generator performance," IEEE Industry Applications Magazine, vol. 3, pp. 33-38, 1997.

[45] I. Husain, A. Radun, and J. Nairus, "Fault analysis and excitation requirements for switched reluctance-generators," Energy Conversion, IEEE Transactions on, vol. 17, pp. 67-72, 2002.

[46] C. A. Ferreira, S. R. Jones, W. S. Heglund, and W. D. Jones, "Detailed design of a 30-kW switched reluctance starter/generator system for a gas turbine engine application," Industry Applications, IEEE Transactions on, vol. 31, pp. 553-561, 1995.

[47] A. de Vries, Y. Bonnassieux, M. Gabsi, F. d'Oliveira, and C. Plasse, "A switched reluctance machine for a car starter-alternator system," in Electric Machines and Drives Conference, 2000. IEMDC 2001. IEEE International, 2001, pp. 323-328.

[48] B. Fahimi, "On the suitability of switched reluctance drives for starter/generator application VO - 4," in Vehicular Technology Conference, 2002. VTC Spring 2002. IEEE 55th, 2002, pp. 2070-2075 vol.4.

[49] D. A. Torrey, "Switched reluctance generators and their control," Industrial Electronics, IEEE Transactions on, vol. 49, pp. 3-14, 2002.

[50] J. Wai and T. M. Jahns, "A new control technique for achieving wide constant power speed operation with an interior PM alternator machine," in Industry Applications Conference, 2001. Thirty-Sixth IAS Annual Meeting. Conference Record of the 2001 IEEE, 2001, pp. 807-814 vol.2.

[51] L. Ferraris, G. Griva, M. Lazzari, F. Profumo, and A. Tenconi, "Iron losses analysis with different supply techniques in high poles number induction motors for automotive applications," in Industrial Electronics, 1999. ISIE '99. Proceedings of the IEEE International Symposium on, 1999, pp. 651-655 vol.2.

[52] P. J. McCleer, J. M. Miller, A. R. Gale, M. W. Degner, and F. Leonardi, "Nonlinear model and momentary performance capability of a cage rotor induction machine used as an automotive combined starter-alternator," Industry Applications, IEEE Transactions on, vol. 37, pp. 840-846, 2001.

[53] K. Junha, J. Jinhwan, and N. Kwanghee, "Dual-inverter control strategy for high-speed operation of EV induction motors," Industrial Electronics, IEEE Transactions on, vol. 51, pp. 312-320, 2004.

References 181

[54] C. P. Mudannayake, M. F. Rahman, and B. Karanayil, "Sensorless induction machine based integrated starter alternator with loss minimization," in Industry Applications Conference, 2005. Fourtieth IAS Annual Meeting. Conference Record of the 2005, 2005, pp. 1058-1065 Vol. 2.

[55] J. Zhang and M. F. Rahman, "Direct torque and flux controlled induction generator for integrated starter alternator with minimized sensor numbers," in Vehicle Power and Propulsion, 2005 IEEE Conference, 2005, p. 5 pp.

[56] D. Seyoum, M. F. Rahman, and C. Grantham, "Terminal voltage control of a wind turbine driven isolated induction generator using stator oriented field control," in Applied Power Electronics Conference and Exposition, 2003. APEC '03. Eighteenth Annual IEEE, 2003, pp. 846-852 vol.2.

[57] R. O. C. Lyra, S. R. Silva, and P. C. Cortizo, "Direct and indirect flux control of an isolated induction generator," in Power Electronics and Drive Systems, 1995., Proceedings of 1995 International Conference on, 1995, pp. 140-145 vol.1.

[58] W. L. Soong, N. Ertugrul, E. C. Lovelace, and T. M. Jahns, "Investigation of interior permanent magnet offset-coupled automotive integrated starter/alternator," in Industry Applications Conference, 2001. Thirty-Sixth IAS Annual Meeting. Conference Record of the 2001 IEEE, 2001, pp. 429-436 vol.1.

[59] S. M. Lukic and A. Emadi, "Effects of Electrical Loads on 42V Automotive Power Systems," SAE transactions 2003.

[60] J. Liu, J. Hu, and L. Xu, "Design and control of a kilo-amp DC/AC inverter for integrated starter-generator (ISG) applications," in Industry Applications Conference, 2004. 39th IAS Annual Meeting. Conference Record of the 2004 IEEE, 2004, pp. 2754-2761 vol.4.

[61] Y. Tang and L. Xu, "Stator field oriented control of doubly-excited induction machine in wind power generating system," in Circuits and Systems, 1992., Proceedings of the 35th Midwest Symposium on, 1992, pp. 1446-1449 vol.2.

[62] T. Ahmed, K. Nishida, and N. Mutsuo, "Advanced voltage control of induction generator using rotor field-oriented control," in Industry Applications Conference, 2005. Fourtieth IAS Annual Meeting. Conference Record of the 2005, 2005, pp. 2835-2842 Vol. 4.

[63] G. S. Buja and M. P. Kazmierkowski, "Direct torque control of PWM inverter-fed AC motors - a survey," Industrial Electronics, IEEE Transactions on, vol. 51, pp. 744-757, 2004.

[64] D. Casadei, F. Profumo, G. Serra, and A. Tani, "FOC and DTC: two viable schemes for induction motors torque control," Power Electronics, IEEE Transactions on, vol. 17, pp. 779-787, 2002.

[65] J. Sastry, O. Ojo, and W. Zhiqiao, "High performance control of a boost AC-DC PWM rectifier-induction generator system," in Industry Applications Conference, 2005. Fourtieth IAS Annual Meeting. Conference Record of the 2005, 2005, pp. 1007-1014 Vol. 2.

[66] I. Takahashi, and Noguchi, T, "A New Quick-Response and High-Efficiency Control Strategy of an Induction Motor," Industry Application, IEEE Transactions on, vol. IA-22, pp. 820-827, 1986.

References 182

[67] M. Depenbrock, "Direct self-control (DSC) of inverter-fed induction machine," Power Electronics, IEEE Transactions on, vol. 3, pp. 420-429, 1988.

[68] R. Datta and V. T. Ranganathan, "Direct power control of grid-connected wound rotor induction machine without rotor position sensors," Power Electronics, IEEE Transactions on, vol. 16, pp. 390-399, 2001.

[69] S. A. Gomez and J. L. R. Amenedo, "Grid synchronisation of doubly fed induction generators using direct torque control," in IECON 02 [Industrial Electronics Society, IEEE 2002 28th Annual Conference of the, 2002, pp. 3338-3343 vol.4.

[70] G. Poddar, A. Joseph, and A. K. Unnikrishnan, "Sensorless variable-speed controller for existing fixed-speed wind power generator with unity-power-factor operation," Industrial Electronics, IEEE Transactions on, vol. 50, pp. 1007-1015, 2003.

[71] W. Huang and Y. Hu, "Research of the DTC control strategy for cage-type induction generator," Diangong Jishu Xuebao/Transactions of China Electrotechnical Society, vol. 17, p. 30, 2002.

[72] A. K. Jain, S. Mathapati, V. T. Ranganathan, and V. Narayanan, "Integrated starter generator for 42-V powernet using induction machine and direct torque control technique," Power Electronics, IEEE Transactions on, vol. 21, pp. 701-710, 2006.

[73] P. Tiitinen and M. Surandra, "The next generation motor control method, DTC direct torque control," in Power Electronics, Drives and Energy Systems for Industrial Growth, 1996., Proceedings of the 1996 International Conference on, 1996, pp. 37-43 vol.1.

[74] P. Vas, Sensorless vector and direct torque control. Oxford ; New York: Oxford University Press, 1998.

[75] B. K. Bose, Modern power electronics and AC drives. Upper Saddle River, NJ: Prentice Hall, 2002.

[76] I. Takahashi and Y. Ohmori, "High-performance direct torque control of an induction motor," Industry Applications, IEEE Transactions on, vol. 25, p. 257, 1989.

[77] H. Y. Zhong, H. P. Messinger, and M. H. Rashad, "A new microcomputer-based direct torque control system for three-phase induction motor," Industry Applications, IEEE Transactions on, vol. 27, pp. 294-298, 1991.

[78] G. Buja, D. Casadei, and G. Serra, "DTC-based strategies for induction motor drives," in Industrial Electronics, Control and Instrumentation, 1997. IECON 97. 23rd International Conference on, 1997, pp. 1506-1516 vol.4.

[79] I. Takahashi and T. Noguchi, "Take a look back upon the past decade of direct torque control [of induction motors]," in Industrial Electronics, Control and Instrumentation, 1997. IECON 97. 23rd International Conference on, 1997, pp. 546-551 vol.2.

[80] L. Zhong, M. F. Rahman, W. Y. Hu, and K. W. Lim, "Analysis of direct torque control in permanent magnet synchronous motor drives," Power Electronics, IEEE Transactions on, vol. 12, pp. 528-536, 1997.

References 183

[81] D. Casadei, G. Serra, and K. Tani, "Implementation of a direct control algorithm for induction motors based on discrete space vector modulation," Power Electronics, IEEE Transactions on, vol. 15, pp. 769-777, 2000.

[82] T. G. Habetler, F. Profumo, M. Pastorelli, and L. M. Tolbert, "Direct torque control of induction machines using space vector modulation," Industry Applications, IEEE Transactions on, vol. 28, pp. 1045-1053, 1992.

[83] Y.-S. Lai and J.-H. Chen, "A new approach to direct torque control of induction motor drives for constant inverter switching frequency and torque ripple reduction," Energy Conversion, IEEE Transactions on, vol. 16, pp. 220-227, 2001.

[84] C. B. Lascu, I.; Blaabjerg, F., "A modified direct torque control for induction motor sensorless drive," Industry Applications, IEEE Transactions on, vol. 36, pp. 122-130, 2000.

[85] T. Lixin, Z. Limin, M. F. Rahman, and Y. Hu, "A novel direct torque controlled interior permanent magnet synchronous machine drive with low ripple in flux and torque and fixed switching frequency," Power Electronics, IEEE Transactions on, vol. 19, pp. 346-354, 2004.

[86] T. Lixin, Z. Limin, M. F. Rahman, and H. Yuwen, "A novel direct torque control for interior permanent-magnet synchronous machine drive with low ripple in torque and flux-a speed-sensorless approach," Industry Applications, IEEE Transactions on, vol. 39, pp. 1748-1756, 2003.

[87] J. Rodriguez, J. Pontt, C. Silva, S. Kouro, and H. Miranda, "A novel direct torque control scheme for induction machines with space vector modulation," in Power Electronics Specialists Conference, 2004. PESC 04. 2004 IEEE 35th Annual, 2004, pp. 1392-1397 Vol.2.

[88] D. Swierczynski and M. P. Kazmierkowski, "Direct torque control of permanent magnet synchronous motor (PMSM) using space vector modulation (DTC-SVM)-simulation and experimental results," in IECON 02 [Industrial Electronics Society, IEEE 2002 28th Annual Conference of the], 2002, pp. 751-755 vol.1.

[89] L. Tang and M. F. Rahman, "A new direct torque control strategy for flux and torque ripple reduction for induction motors drive by using space vector modulation," in Power Electronics Specialists Conference, 2001. PESC. 2001 IEEE 32nd Annual, 2001, pp. 1440-1445 vol. 3.

[90] L. Tang, L. Zhong, A. F. Rahman, and Y. Hu, "An investigation of a modified direct torque control strategy for flux and torque ripple reduction for induction machine drive system with fixed switching frequency," in Industry Applications Conference, 2002. 37th IAS Annual Meeting. Conference Record of the, 2002, pp. 837-844 vol.2.

[91] A. Tripathi, A. M. Khambadkone, and S. K. Panda, "Space-vector based, constant frequency, direct torque control and dead beat stator flux control of AC machines," in Industrial Electronics Society, 2001. IECON '01. The 27th Annual Conference of the IEEE, 2001, pp. 1219-1224 vol.2.

References 184

[92] R. C. Dorf and R. H. Bishop, Modern control systems, 7th ed. Reading, Mass.: Addison-Wesley, 1995.

[93] Y. H. Liu and C. L. Chen, "Novel dead time compensation method for induction motor drives using space vector modulation," Electric Power Applications, IEE Proceedings-, vol. 145, pp. 387-392, 1998.

[94] C. Jong-Woo and S. Seung-Ki, "A new compensation strategy reducing voltage/current distortion in PWM VSI systems operating with low output voltages," Industry Applications, IEEE Transactions on, vol. 31, pp. 1001-1008, 1995.

[95] C. Jong-Woo and S. Seung-Ki, "Inverter output voltage synthesis using novel dead time compensation," Power Electronics, IEEE Transactions on, vol. 11, pp. 221-227, 1996.

[96] J. Holtz and Q. Juntao, "Drift- and parameter-compensated flux estimator for persistent zero-stator-frequency operation of sensorless-controlled induction motors," Industry Applications, IEEE Transactions on, vol. 39, pp. 1052-1060, 2003.

[97] J. Holtz and Q. Juntao, "Sensorless vector control of induction motors at very low speed using a nonlinear inverter model and parameter identification," Industry Applications, IEEE Transactions on, vol. 38, pp. 1087-1095, 2002.

[98] L. Jong-Lick, "A new approach of dead-time compensation for PWM voltage inverters," Circuits and Systems I: Fundamental Theory and Applications, IEEE Transactions on [see also Circuits and Systems I: Regular Papers, IEEE Transactions on], vol. 49, pp. 476-483, 2002.

[99] K. D. Hurst, T. G. Habetler, G. Griva, and F. Profumo, "Zero-speed tacholess IM torque control: simply a matter of stator voltage integration," Industry Applications, IEEE Transactions on, vol. 34, pp. 790-795, 1998.

[100] H. Jun and W. Bin, "New integration algorithms for estimating motor flux over a wide speed range," Power Electronics, IEEE Transactions on, vol. 13, pp. 969-977, 1998.

[101] B. K. Bose and N. R. Patel, "A programmable cascaded low-pass filter-based flux synthesis for a stator flux-oriented vector-controlled induction motor drive," Industrial Electronics, IEEE Transactions on, vol. 44, pp. 140-143, 1997.

[102] P. L. Jansen, R. D. Lorenz, and D. W. Novotny, "Observer-based direct field orientation: analysis and comparison of alternative methods," Industry Applications, IEEE Transactions on, vol. 30, pp. 945-953, 1994.

[103] D. Pai A, L. Umanand, and N. J. Rao, "Direct torque control of induction motor with extended Kalman filter," in Power Electronics and Motion Control Conference, 2000. Proceedings. PIEMC 2000. The Third International, 2000, pp. 132-137 vol.1.

[104] I. G. Bird and H. Zelaya de La Parra, "Practical evaluation of two stator flux estimation techniques for high performance direct torque control," in Power Electronics and Variable Speed Drives, 1996. Sixth International Conference on (Conf. Publ. No. 429), 1996, pp. 465-470.

References 185

[105] M. Tursini, R. Petrella, and F. Parasiliti, "Adaptive sliding-mode observer for speed-sensorless control of induction motors," Industry Applications, IEEE Transactions on, vol. 36, pp. 1380-1387, 2000.

[106] F. Chen and M. W. Dunnigan, "Comparative study of a sliding-mode observer and Kalman filters for full state estimation in an induction machine," Electric Power Applications, IEE Proceedings-, vol. 149, pp. 53-64, 2002.

[107] S. Xepapas, A. Kaletsanos, F. Xepapas, and S. Manias, "Sliding-mode observer for speed-sensorless induction motor drives," Control Theory and Applications, IEE Proceedings-, vol. 150, pp. 611-617, 2003.

[108] L. Jingchuan, X. Longya, and Z. Zheng, "An adaptive sliding-mode observer for induction motor sensorless speed control," Industry Applications, IEEE Transactions on, vol. 41, pp. 1039-1046, 2005.

[109] C. Lascu, I. Boldea, and F. Blaabjerg, "Direct torque control of sensorless induction motor drives: a sliding-mode approach," Industry Applications, IEEE Transactions on, vol. 40, pp. 582-590, 2004.

[110] H. Rehman, A. Derdiyok, M. K. Guven, and X. Longya, "A new current model flux observer for wide speed range sensorless control of an induction machine," Power Electronics, IEEE Transactions on, vol. 17, pp. 1041-1048, 2002.

[111] H. U. Rehman and X. Longya, "A voltage model flux observer design requiring no stator resistance or voltage signal information," in Applied Power Electronics Conference and Exposition, 2002. APEC 2002. Seventeenth Annual IEEE, 2002, pp. 299-303 vol.1.

[112] D. S. Kirschen, D. W. Novotny, and W. Suwanwisoot, "Minimizing induction motor losses by excitation control in variable frequency drives," IEEE Transactions on Industry Applications, vol. 1A-20, pp. 1244-50, 1984.

[113] D. S. Kirschen, D. W. Novotny, and T. A. Lipo, "Optimal efficiency control of an induction motor drive," IEEE Transactions on Energy Conversion, vol. EC-2, pp. 70-6, 1987.

[114] A. Kusko and D. Galler, "Control means for minimization of losses in ac and dc motor drives," IEEE Transactions on Industry Applications, vol. IA-19, pp. 561-570, 1983.

[115] R. B. Sepe, Jr.; Morrison, C.M.; Miller, J.M.; Gale, A.R., "High efficiency operation of a hybrid electric vehicle starter/generator over road profiles," in Industry Applications Conference, 2001. Thirty-Sixth IAS Annual Meeting. Conference Record of the 2001 IEEE, 2001, pp. 921-925 vol.2.

[116] S. Taniguchi, T. Yoshizumi, and K. Matsuse, "A method of speed sensorless control of direct-field-oriented induction motor operating at high efficiency with core loss consideration," in Applied Power Electronics Conference and Exposition, 1999. APEC '99. Fourteenth Annual, 1999, pp. 1244-1250 vol.2.

[117] K. S. Rasmussen and P. Thogersen, "Model based energy optimiser for vector controlled induction motor drives," in Proceedings of 7th European Conference on Power Electronics and Applications, Trondheim, Norway, 1997, pp. 711-16.

[118] H. R. Andersen and J. K. Pedersen, "Low cost energy optimized control strategy for a variable speed three-phase induction motor," in Power Electronics

References 186

Specialists Conference, 1996. PESC '96 Record., 27th Annual IEEE, 1996, pp. 920-924 vol.1.

[119] F. Abrahamsen, F. Blaabjerg, J. K. Pedersen, and P. B. Thoegersen, "Efficiency-optimized control of medium-size induction motor drives," Industry Applications, IEEE Transactions on, vol. 37, pp. 1761-1767, 2001.

[120] F. Abrahamsen, F. Blaabjerg, J. K. Pedersen, P. Z. Grabowski, and P. Thogersen, "On the energy optimized control of standard and high-efficiency induction motors in CT and HVAC applications," Industry Applications, IEEE Transactions on, vol. 34, pp. 822-831, 1998.

[121] Y. Sheng-Ming and L. Feng-Chieh, "Loss-minimization control of vector-controlled induction motor drives," in Power Electronics and Drive Systems, 2001. Proceedings., 2001 4th IEEE International Conference on, 2001, pp. 182-187 vol.1.

[122] I. Boldea and S. A. Nasar, The induction machine handbook. Boca Raton, Fla.: CRC Press, 2002.

[123] H. Akagi, Y. Kanazawa, and A. Nabae, "Instantaneous reactive power compensators comprising switching devices without energy storage components," IEEE Transactions on Industry Applications, vol. IA-20, pp. 625-30, 1984.

[124] P. Fang Zheng and L. Jih-Sheng, "Generalized instantaneous reactive power theory for three-phase power systems," Instrumentation and Measurement, IEEE Transactions on, vol. 45, p. 293, 1996.

[125] X. Yanhong and C. Shaotang, "Instability issues for control system in induction generator," in Industry Applications Conference, 2001. Thirty-Sixth IAS Annual Meeting. Conference Record of the 2001 IEEE, 2001, pp. 110-117 vol.1.

[126] L. Zhong, M. F. Rahman, W. Y. Hu, K. W. Lim, and M. A. Rahman, "A direct torque controller for permanent magnet synchronous motor drives," Energy Conversion, IEEE Transactions on, vol. 14, pp. 637-642, 1999.

[127] Y. Hu, C. Tian, Y. Gu, Z. You, L. X. Tang, and M. F. Rahman, "In-depth research on direct torque control of permanent magnet synchronous motor," in IECON 02 [Industrial Electronics Society, IEEE 2002 28th Annual Conference of the], 2002, pp. 1060-1065 vol.2.

[128] M. F. Rahman, L. Zhong, M. E. Haque, and M. A. Rahman, "A direct torque-controlled interior permanent-magnet synchronous motor drive without a speed sensor," Energy Conversion, IEEE Transactions on, vol. 18, pp. 17-22, 2003.

[129] M. F. Rahman, M. E. Haque, T. Lixin, and Z. Limin, "Problems associated with the direct torque control of an interior permanent-magnet synchronous motor drive and their remedies," Industrial Electronics, IEEE Transactions on, vol. 51, pp. 799-809, 2004.

Appendix A 187

APPENDIX A

LIST OF PUBLICATIONS

Journal publications:

1. Jun Zhang, M. F. Rahman, “A Direct Flux Vector Controlled Induction Generator

with Space Vector Modulation for Integrated Starter Alternator”, fully accepted by

the IEEE Transactions on Industrial Electronics.

2. Jun Zhang, M. F. Rahman, “Direct Torque and Flux Controlled Induction

Generator for Integrated Starter/alternator with Minimized Sensor Numbers”, under

review of the IEEE Transactions on Vehicular Technology.

Conference publications:

3. Jun Zhang, M.F. Rahman, "Non-Linear Behaviour Compensation of the Converter

for Direct Torque Controlled Induction Machines ", proceeding of Australasian

Universities Power Engineering Conference, Melbourne, Australia, December 10 -

13, 2006.

4. Jun Zhang, M. F. Rahman, “A Sliding Mode Flux Observer for Direct Torque

Controlled Integrated Starter/Alternator”, proceeding of 41st Annual Meeting of the

IEEE Industry Applications Society, October 8 - 12, 2006, Tampa Florida, USA

(IAS 2006).

5. Jun Zhang, M. F. Rahman, “Efficiency-Optimized Direct Torque Controlled

Integrated Starter/Alternator with Power Factor Control”, the 37th IEEE Power

Electronics Specialists Conferences, June 18 - 22, 2006, Jeju, Korea (PESC 2006).

6. Jun Zhang, M. F. Rahman, “A New Scheme to Direct Torque Control of Interior

Permanent Magnet Synchronous Machine Drives for Constant Inverter Switching

Frequency and Low Torque Ripple”, the 5th International the Power Electronics

and Motion Control Conference, 13-16 August, 2006 Shanghai, P. R. China

(IPEMC 2006).

7. Jun Zhang, Zhuang Xu, Lixin Tang and M. F. Rahman, “A Novel Direct Load

Angle Control for Interior Permanent Magnet Synchronous Machine Drives with

Appendix A 188

Space Vector Modulation”,The Sixth IEEE International Conference on Power

Electronics and Drive Systems, 28 Nov – 1 Dec 2005, Kuala Lumpur, Malaysia

(PEDS 2005).

8. Jun Zhang, M. F. Rahman, “Direct Flux Vector Control Scheme for Induction

Machine Drives with Space Vector Modulation”, IEEE Industry Applications

Society, 40th Annual General Meeting, October 2-6, 2005, Hong Kong (IAS 2005).

9. Jun Zhang, M. F. Rahman, “Sliding Mode Controlled Low Voltage Induction

Machine for 42V Automotive Systems”, Australasian Universities Power

Engineering Conference, The University Of Tasmania, Hobart, Australia, 25

September – 28 September 2005.

10. Jun Zhang, M. F. Rahman, “Direct Torque and Flux Controlled Induction

Generator for Integrated Starter Alternator with Minimized Sensor Numbers”,

2005 IEEE Vehicle Power and Propulsion Conference, September 7-9, 2005,

Illinois Institute of Technology, Chicago, Illinois, USA (VPP 2005).

11. Jun Zhang, M. F. Rahman , “Analysis and Design of a Novel Direct Flux Control

Scheme for Induction Machine”, Proceeding of IEEE International Electric

Machines and Drives Conference, San Antonio, USA, May 15 – 18, 2005, ISBN: 0-

7803-8988-3 (CD ROM) (IEMDC 2005).

12. Jun Zhang, M.F. Rahman and L. Tang, “A direct flux controlled induction

generator with space vector modulation for integrated starter alternator”, Industrial

Electronics Society, 2004. 30th Annual Conference of IEEE, Vol.1, Iss., 2-6 Nov.

2004, Pages: 330- 334 Vol. 1 (IECON 2004).

13. Jun Zhang, M.F. Rahman and L. Tang, "A Direct Torque Controlled Integrated

Starter Alternator with Space Vector Modulation", Proc. AUPEC 2004, Brisbane,

Australia, 29 Sept. - 2 Oct. 2004.

14. Jun Zhang, M.F. Rahman and L. Tang, “Modified direct torque controlled

induction generator with space vector modulation for integrated starter alternator”,

Power Electronics and Motion Control Conference, 2004. The 4th International,

Vol.1, Iss., 14-16 Aug. 2004, Pages: 405- 408 Vol.1, (IPEMC 2004).

Appendix B Modelling of the direct flux vector control 189

APPENDIX B

MODELLING OF THE DIRECT FLUX VECTOR

CONTROL



ss s s

dV R IdtΨ

= + (B-1)

0 rr r m r

dR I jdtΨ

= + − ω Ψ (B-2)

s s s m r

r m s r r

L I L I

L I L I

⎧Ψ = +⎪⎨Ψ = +⎪⎩

(B-3)

32

me r s

s r

LT PL L

= Ψ ×Ψσ

(B-4)

where

2

1 m

s r

LL L

σ = − (B-5)

where sR and rR are the stator and rotor resistances, sL , rL and mL are the stator, rotor

and mutual inductances, respectively. And mω is rotor speed, P is the number of pole

pairs.

The stator and rotor current vectors can be denoted by the stator and rotor flux vectors

from (B-3), respectively.

1

21s s ss m r m

m r m ss r mr r r

I L L L LL L L LL L LI

− ⎡ ⎤ ⎡ ⎤⎡ ⎤ Ψ Ψ−⎡ ⎤ ⎡ ⎤= =⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥−−Ψ Ψ⎢ ⎥ ⎣ ⎦ ⎢ ⎥ ⎣ ⎦ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦

(B-6)

By considering (B-1), (B-2) and (B-6), we get


1

21

rm r r r

s s ss m r m

m r m ss r mr r r

d j R IdtI L L L L

L L L LL L LI

−

⎧ Ψ= Ψ −⎪

⎪⎨ ⎡ ⎤ ⎡ ⎤⎡ ⎤ Ψ Ψ−⎡ ⎤ ⎡ ⎤⎪ = =⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎪ −−Ψ Ψ⎢ ⎥ ⎣ ⎦ ⎢ ⎥ ⎣ ⎦ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦⎩

ω

(B-7)

The relationship between stator and rotor flux vectors sΨ and rΨ is derived from (B-7)

r r m rs m r

s r r

d R L R( j )dt L L LΨ

= Ψ + ω − Ψσ σ

(B-8)

where

2

1 m

s r

LL L

σ = − (B-9)

By using Laplace transform of (B-8) and assuming the rotor speed mω is changing

slowly, the relationship between stator and rotor flux vectors sΨ and rΨ in the

frequency domain can be obtained

1

r m m

s r sr s s

r r rm m

r r r

R L LL L L( s ) ( s ) ( s )

R L Ls j s jL R R

σΨ = Ψ = Ψ

⎛ ⎞ ⎛ ⎞− ω − σ + − ω σ⎜ ⎟ ⎜ ⎟σ⎝ ⎠ ⎝ ⎠

(B-10)

Assuming that s sj j t* *s s se eθ ωΨ = Ψ = Ψ and the amplitude of sΨ is kept constant, and

that sΨ rotates at an angular speed sω , the Laplace form of the stator flux vector sΨ is

1 *s s

s

( s )s j

Ψ = Ψ− ω

(B-11)

By substituting (B-10) into (B-11) and taking inverse Laplace transform

1 1

1

m

*sr s

r r sm

r r

L

L( t )

L L s js j

R R

−Ψ = Ψ− ω

σ + − ω σ

⎧ ⎫⎪ ⎪⎪ ⎪⎨ ⎬

⎛ ⎞⎪ ⎪⎜ ⎟⎪ ⎪⎩ ⎭⎝ ⎠

L (B-12)

Thus


( )

( )

1 *

1 *

1

1

1

1 1

1 1

1 1

1 1

m

sr s

sr rm

r r

r

m rs

s sr r r rm s m

r r r r

r

m r

rs s rs m

r r

LLL

s jL Ls jR R

LL RLL s jL L L Lj j s j

R R R R

LL RL LL s j Lj sR R

−

−

−

⎧ ⎫⎪ ⎪⎪ ⎪Ψ = Ψ⎨ ⎬−⎛ ⎞⎪ ⎪+ −⎜ ⎟⎪ ⎪⎝ ⎠⎩ ⎭⎧ ⎫⎛ ⎞⎪ ⎪⎜ ⎟⎪ ⎪⎜ ⎟= − Ψ⎨ ⎬⎜ ⎟−⎛ ⎞ ⎛ ⎞⎪ ⎪− − − + −⎜ ⎟⎜ ⎟ ⎜ ⎟⎜ ⎟⎪ ⎪⎝ ⎠ ⎝ ⎠⎝ ⎠⎩ ⎭

= −−+ − +

ωσ ω σ

σ

ωω σ σ ω σ ω σ

σ

ωσ ω ω σ

( )

( )

( )

*

1

*

1a*

2

ar g

1

1

rm

rs

r

r

rr ms mrr s

r

r

r

sr

mr

m LjRt ts L

s Rrs m

r

LL jj rctg RR t tm Ls R

sr

s mr

Lj ctR

m

s

LjR

LL

LjR

LL L

R

LL

e e

e e e

−−

⎛ ⎞ −− −⎜ ⎟⎝ ⎠ −

−

⎧ ⎫⎛ ⎞⎪ ⎪⎜ ⎟⎪ ⎪⎜ ⎟ Ψ⎨ ⎬⎜ ⎟⎛ ⎞⎪ ⎪−⎜ ⎟⎜ ⎟⎜ ⎟⎪ ⎪⎝ ⎠⎝ ⎠⎩ ⎭

⎧ ⎫⎪ ⎪= Ψ −⎨ ⎬⎪ ⎪+ − ⎩ ⎭

⎧ ⎫⎪ ⎪= Ψ −⎨ ⎬⎪ ⎪⎛ ⎞ ⎩ ⎭+ −⎜ ⎟

⎝ ⎠

=

ω σω

σ

ω σσ ω ωω

σ

σ

ω σ

σ ω ω

σ ω ω

( )

( )( )

( )

( )( )

1*

2

ar g 1*

2

cos( ) sin( ) cos( ) sin( )

1

cos( ) sin( ) cos( ) sin( )

1

s mr

r r

rs m

r

r r

tL Rs s s m m

rs m

r

Lj ctR tm L Rs s s m m

sr

s mr

t j t t j tLR

L t j t t j tL L

R

e e

e e

⎛ ⎞−⎜ ⎟

⎝ ⎠ −

⎛ ⎞− −⎜ ⎟

⎝ ⎠ −

⎧ ⎫Ψ + − +⎨ ⎬⎩ ⎭⎛ ⎞

+ −⎜ ⎟⎝ ⎠

⎧ ⎫= Ψ + − +⎨ ⎬⎩ ⎭⎛ ⎞

+ −⎜ ⎟⎝ ⎠

ω ω

σ

σ ω ω

σ

ω ω ω ω

σ ω ω

ω ω ω ω

σ ω ω

(B-13)

Equation (B-13) can be further simplified as


( )

( )

( )( )

( )

( )

2 2ar g ar g*

2

2

ar g ar g*

2

( )

1

1 2 cos

1

rs m

r

r r

rr r s mr

y Lj ct j ctm x Rr ss

rs m

r

t tL L

s m y LR R j ct ctm x Rss

rs m

r

tx yL

L LR

tLL L

R

e e

e ee

⎛ ⎞⎛ ⎞ − −⎜ ⎟⎜ ⎟⎝ ⎠ ⎝ ⎠

− −

⎛ ⎞⎛ ⎞⎛ ⎞− −⎜ ⎟⎜ ⎟⎜ ⎟⎜ ⎟⎝ ⎠ ⎝ ⎠⎝ ⎠

Ψ+

= Ψ⎛ ⎞

+ −⎜ ⎟⎝ ⎠

⎛ ⎞+ − −⎜ ⎟⎝ ⎠= Ψ

⎛ ⎞+ −⎜ ⎟⎝ ⎠

σ ω ω

σ σσ ω ω

σ ω ω

ω ω

σ ω ω

(B-14)

where

cos( ) cos( )

sin( ) sin( )

r

r

r

r

tL

s mR

tL

s mR

x t t

y t t

e

e

−

−

⎧ = −⎪⎪⎨⎪ = −⎪⎩

σ

σ

ω ω

ω ω (B-15)

That is

( )( )

( )( )( )( )

21 2

21

1 1

mr

s

s

tt

e cos te s m( t )

s my* j tan tane s mx

LL

−τ−+ − ω − ωτ

Ψ =

+ τ ω − ω

− −− τ ω − ω× Ψ

⎛ ⎞⎜ ⎟⎜ ⎟⎝ ⎠

⎛ ⎞⎛ ⎞⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠

(B-16)

where

r

rt

s m

t

s m

LR

x cos( t ) cos( t )

y sin( t ) sin( t )

ee

−τ

−τ

⎧τ = σ⎪⎪⎪ = ω − ω⎨⎪⎪ = ω − ω⎪⎩

(B-17)

With small slip, (B-16) can be simplified as


( )( )

( )( )1 1

21 2

1 1

1

1

mr s

s

tm s mss

tt

eey* j tan tan( t ) e s mx

* je

LL

yL tan tane xL− − −τ

−τ−+ −τ − −− τ ω − ωΨ ≈ × Ψ

× Ψ

⎛ ⎞⎜ ⎟⎜ ⎟ ⎛ ⎞⎛ ⎞⎝ ⎠ ⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠

⎡ ⎤⎛ ⎞⎛ ⎞ − τ ω − ω⎜ ⎟= − ⎢ ⎥⎜ ⎟ ⎝ ⎠⎣ ⎦⎝ ⎠

(B-18)

It shows that the rotor flux vector tracks stator flux vector in its amplitude and rotating

speed with a time constant, given by τ . Once the stator flux is built up and kept

constant, the rotor flux will also be kept constant. Therefore, the amplitude of the rotor

flux can be considered as fixed after establishing of the stator flux. Equation (B-18) can

be further simplified as

( )( )1 1m s mr ss

* j( t ) eyL tan tanxL

− −Ψ ≈ Ψ

⎡ ⎤⎛ ⎞ − τ ω − ω⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦ (B-19)

From (B-4), the torque can be expressed as

32

sj t*me r s

s r

LT ( t ) P ( t )L L e ω= Ψ × Ψ

σ (B-20)

By substituting (B-19) into (B-20), we obtain

( )( )

( )( )

1 1

2 1 1

32

32

sj t*m m s me s ss r s

*m ms s s m

s r s

* jeyL L tan tanT ( t ) P xL L L

L L yP sin t tan tanL L L x

e− − ω

− −

Ψ⎧ ⎫⎡ ⎤⎛ ⎞⎪ ⎪− τ ω − ω⎜ ⎟= × Ψ⎢ ⎥⎨ ⎬⎝ ⎠⎣ ⎦σ ⎪ ⎪⎩ ⎭

⎧ ⎫⎧ ⎫ ⎡ ⎤⎛ ⎞≈ Ψ × ω − − τ ω − ω⎨ ⎬ ⎨ ⎬⎜ ⎟⎢ ⎥σ ⎝ ⎠⎣ ⎦⎩ ⎭ ⎩ ⎭

(B-21)

where

r

rt

s m

t

s m

LR

x cos( t ) cos( t )

y sin( t ) sin( t )

ee

−τ

−τ

⎧τ = σ⎪⎪⎪ = ω − ω⎨⎪⎪ = ω − ω⎪⎩

(B-22)

It clear that the dynamic response of torque is determined by the amplitude and rotating

speed of the stator flux vector with the non-linear relationship of (B-21). The torque of

the induction machine can be regulated by controlling rotating speed of the stator flux


vector sΨ as long as its amplitude is kept constant. As rotor flux vector tracks the stator

flux vector, its amplitude is also kept constant after establishing of constant stator flux

amplitude. In addition, the sin or tan computation results of a small angle is very close

to the angle by itself (in rad) as shown in (B-23). Therefore, the above torque expression

can be simplified to (B-25) in which the slip is small.

( ) ( ) ( ) ( )sin tan smallθ ≈ θ ≈ θ θ (B-23)

So, torque expression can be further simplified as

( )( )

( )( )

2 1

2

32

32

*m me s s s m

s r s

*m ms s s m

s r s

L L yT ( t ) P sin t tanL L L x

L L yP tL L L x

−⎧ ⎫⎧ ⎫ ⎡ ⎤⎛ ⎞= Ψ × ω − − τ ω − ω⎨ ⎬ ⎨ ⎬⎜ ⎟⎢ ⎥σ ⎝ ⎠⎣ ⎦⎩ ⎭ ⎩ ⎭⎧ ⎫ ⎧ ⎫⎛ ⎞= Ψ × ω − + τ ω − ω⎨ ⎬ ⎨ ⎬⎜ ⎟σ ⎝ ⎠⎩ ⎭⎩ ⎭

(B-24)

By considering (B-24) and (B-22) at same time, we obtain

( )( )

( )2

2

2

2

31

2

3 12e

t*m

s s mr s

t*m m

s s ms r s

T ( t )

LP

R L

L LPL L L

e

e−τ

−τ≈

= Ψ − ω − ω

⎧ ⎫ ⎧ ⎫⎛ ⎞Ψ − τ ω − ω⎨ ⎬ ⎨ ⎬⎜ ⎟σ ⎝ ⎠⎩ ⎭⎩ ⎭⎧ ⎫⎛ ⎞⎨ ⎬ ⎜ ⎟

⎝ ⎠⎩ ⎭

(B-25)

The simplification in (B-25) is based on the fact that

( )( ) ( )1t

s ms s mytx e −

τ− ω − ω⎧ ⎫⎛ ⎞ ⎛ ⎞ω − + τ ω − ω ≈⎨ ⎬ ⎜ ⎟⎜ ⎟ ⎝ ⎠⎝ ⎠⎩ ⎭

(B-26)

by considering the dynamic in (B-22).

Therefore

( )

( )

22

2

31

2

1

t*m

e s s mr s

t

s m

LT ( t ) P

R L

K

e

e

−τ

−τ

≈ Ψ − ω − ω

= − ω − ω

⎧ ⎫⎛ ⎞⎨ ⎬ ⎜ ⎟

⎝ ⎠⎩ ⎭⎛ ⎞⎜ ⎟⎝ ⎠

(B-27)

where


2 2

2

32

*ms

r s

r

r

LK PR L

LR

⎧= Ψ⎪⎪

⎨⎪τ = σ⎪⎩

(B-28)

By Laplace transform of (B-27), we have

{ }11e s mT ( s ) K

s⎛ ⎞= ω − ω⎜ ⎟τ +⎝ ⎠

L (B-29)

where { }s mω − ωL is the Laplace form of { }s mω − ω .

The transfer function of the torque loop with input as { }s mω − ω can be written as

{ } 1

ep

s m

T ( s ) KG ( s )s

= =ω − ω τ +L

(B-30)

Equation (B-30) shows that the relationship between eT and sω is equivalent to a first

order system with a disturbance mω . The equivalent system block is shown as follows:

( )s sω ( )eT s

1Ksτ +

( )m sω−

Fig. B.1 Equivalent system model of the torque loop

In order to achieve good performance of tracking a reference torque signal and

disturbance rejection, a PI controller of Fig.B.2 can be employed:

( )s sω ( )eT s1

Ksτ +

( )m sω−

( )cG s−

eT ∗

Fig.B.2 PI control of the equivalent system


where

p ic

K s KG ( s )

s+

= (B-31)

Fig.B.2 is the equivalent torque loop of the direct flux vector control scheme discussed

in Chapter 4.

197

APPENDIX C

MODELLING OF THE DIRECT TORQUE AND

FLUX CONTROL

sΨ

sθ α

β

q d

rΨ

sI

sdisqi

Fig. C.1 Vector diagram of the induction machine

In stator flux reference frame ( )d q− shown in Fig. C.1, the dynamic behavior of

induction machine can be described as following equations:

( )0

32

ss s s s s

rr r s m r

e sd sq

dV R I jdt

dR I jdt

T P i

⎧ Ψ= + + ω Ψ⎪

⎪Ψ⎪

= + + ω −ω Ψ⎨⎪⎪ = Ψ ⋅⎪⎩

(C-1)

and

s s s m r

r m s r r

L I L I

L I L I

⎧Ψ = +⎪⎨Ψ = +⎪⎩

(C-2)

198

Therefore,

( )1

2

1

ss s s s s

rs m r r r

s s ss m r m

m r m ss r mr r r

d V R I jdt

d j R IdtI L L L L

L L L LL L LI

−

⎧ Ψ⎪ = − − ω Ψ⎪⎪ Ψ⎪ = − ω −ω Ψ −⎨⎪⎪ ⎡ ⎤ ⎡ ⎤⎡ ⎤ Ψ Ψ−⎡ ⎤ ⎡ ⎤

= =⎪ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥−−Ψ Ψ⎪⎢ ⎥ ⎣ ⎦ ⎣ ⎦⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦⎩

(C-3)

Equation (C-3) can be simplified as

( )

( ) 2

0 010 00

0 01 10 00

10

s

s s s ss

s m r rr r

s s sr mss

s m m sr s r mr r

dj R Idt V

j R Iddt

j L LRV

j L LR L L L

⎡ ⎤Ψ⎢ ⎥ ⎡ ⎤− Ψ ⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤⎢ ⎥ = + −⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥ − −⎢ ⎥Ψ Ψ⎣ ⎦ ⎣ ⎦⎢ ⎥⎣ ⎦ ⎣ ⎦⎣ ⎦⎢ ⎥⎣ ⎦

⎡ ⎤ ⎡ ⎤− Ψ Ψ−⎡ ⎤ ⎡ ⎤⎡ ⎤⎡ ⎤= + −⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ − − −−Ψ Ψ⎣ ⎦ ⎣ ⎦⎢ ⎥ ⎣ ⎦ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦

=

ωω ω

ωω ω

( )

( )

0 0 10 0

10 01

0 00 1

s s sr mss

s m m sr s rr r

m

s s s rs sss

s m r mr r

s r r

j L LRV

j L LR L L

Lj L L LR

Vj R L

L L L

⎡ ⎤ ⎡ ⎤− Ψ Ψ−⎡ ⎤ ⎡ ⎤⎡ ⎤⎡ ⎤+ −⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ − − −Ψ Ψ⎣ ⎦ ⎣ ⎦⎢ ⎥ ⎣ ⎦ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦

⎡ ⎤−⎢ ⎥⎡ ⎤ ⎡− Ψ Ψ⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎢ ⎥= + −⎢ ⎥ ⎢⎢ ⎥ ⎢ ⎥⎢ ⎥ − − ⎢ ⎥Ψ Ψ⎣ ⎦ ⎣ ⎦⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣−⎢ ⎥⎣ ⎦

ωω ω σ

ω σ σω ω

σ σ

( )

( )

0100

10

s s m

s s s rs ss

s m r m rr r

s r r

s s ms

s s r ss

r m r rs m

s r r

R R Lj L L L

Vj R L R

L L L

R R LjL L L

VR L RjL L L

⎤⎥

⎢ ⎥⎦

⎡ ⎤−⎢ ⎥⎡ ⎤ ⎡ ⎤− Ψ Ψ⎡ ⎤⎡ ⎤ ⎢ ⎥= + +⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ − − ⎢ ⎥Ψ Ψ⎣ ⎦ ⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦ ⎣ ⎦−⎢ ⎥⎣ ⎦

⎡ ⎤− −⎢ ⎥ ⎡ ⎤Ψ ⎡ ⎤⎢ ⎥= +⎢ ⎥ ⎢ ⎥⎢ ⎥ Ψ ⎣ ⎦⎢ ⎥⎣ ⎦− − −⎢ ⎥⎣ ⎦

ω σ σω ω

σ σ

ωσ σ

ω ωσ σ

(C-4)

where

2

1 m

s r

LL L

σ = − (C-5)

So, the relationship between stator and rotor flux vector can be obtained by Laplace

transform from (C-4)

199

( )r m rr s s m r

s r r

R L Rs jL L L

⎛ ⎞Ψ = Ψ + − − − Ψ⎜ ⎟

⎝ ⎠ω ω

σ σ (C-6)

thus

( ) ( )

( )( ) ( )( )

1

1 1

r m m

s r sr s s

r r rs m s m

r r r

m m

s ss s

s m s m

R L LL L L

R L Ls j s jL R R

L LL L

s j s j

σΨ = Ψ = Ψ

⎛ ⎞ ⎛ ⎞− − ω − ω − σ − − ω − ω σ −⎜ ⎟ ⎜ ⎟σ⎝ ⎠ ⎝ ⎠

= Ψ = Ψτσ − − ω − ω τ − τσ + ω − ω τ +

(C-7)

where r

r

LR

τ = .

It is known in the stator flux reference frame that

0

s ds qs

qs

j⎧Ψ = Ψ + Ψ⎪⎨Ψ =⎪⎩

(C-8)

The rotor flux vector in the stator flux reference frame can be expressed as

r rd rqjΨ = Ψ + Ψ (C-9)

With (C-6), (C-8) and (C-9), the dq component of rotor flux vector can be obtained

200

( )

( ) ( ) ( )

( )

( )

( )

r m rr s s m r

s r r

r m rrd rq sd s m rd rq

s r r

r m rrd sd s m rq rd

s r r

rrq s m rd rq

r

r m rrd sd s m rq

s r

R L Rs jL L L

R L Rs j j jL L L

R L RsL L L

RsL

R L RsL L L

⎛ ⎞Ψ = Ψ + − ω − ω − Ψ⎜ ⎟σ σ⎝ ⎠

⎛ ⎞⇒ Ψ + Ψ = Ψ + − ω − ω − Ψ + Ψ⎜ ⎟σ σ⎝ ⎠

⎧ ⎛ ⎞Ψ = Ψ + ω − ω Ψ − Ψ⎪ ⎜ ⎟σ σ⎪ ⎝ ⎠⇒ ⎨

⎛ ⎞⎪ Ψ = − ω − ω Ψ − Ψ⎜ ⎟⎪ σ⎝ ⎠⎩

Ψ = Ψ + ω − ω Ψ −σ

⇒ ( )

( )

( )

( )

( )

2

2

rdr

s mrq rd

r

r

s mr m rrd sd rd rd

rs r r

r

s mrq rd

r

r

s m r mrrd sd

r r s r

r

s mrq rd

r

r

RsL

R L Rs RL L LsL

RsL

R LRs R L L LsL

RsL

⎧ ⎛ ⎞Ψ⎪ ⎜ ⎟σ⎝ ⎠⎪⎪

⎨ − ω − ωΨ = Ψ⎪⎪ +

σ⎪⎩⎧ ⎛ ⎞⎪ ⎜ ⎟− ω − ω⎪ ⎜ ⎟Ψ = Ψ + Ψ − Ψ

σ σ⎪ ⎜ ⎟+⎪ ⎜ ⎟σ⇒ ⎝ ⎠⎨⎪ − ω − ω⎪Ψ = Ψ⎪ +⎪ σ⎩⎧⎛ ⎞⎜ ⎟ω − ω⎜ ⎟+ + Ψ = Ψ

σ σ⎜ ⎟+⎜ ⎟σ⇒ ⎝ ⎠⎨− ω − ω

Ψ = Ψ+

σ

( )

( )( )

2

2

r m

s rrd sd

s m r

r r

r

r m

s m s rrq sd

r s m r

r r r

r

R LL L

Rs R LsL

R LL L

R Rs sL R LsL

⎪⎪⎪⎪

⎪⎪⎪⎪⎩⎧⎪ σ⎪Ψ = Ψ

ω − ω⎪+ +⎪ σ⎪ +

σ⎪⇒ ⎨⎪

− ω − ω⎪ σΨ = Ψ⎪

ω − ω⎪ + + +σ⎪ σ+⎪ σ⎩

(C-10)

201

The expression of stator current with stator and rotor flux vector is already shown in (C-

3), which is restated as

[ ]2

1 ss r m

s r m r

I L LL L L

⎡ ⎤Ψ⎡ ⎤ = − ⎢ ⎥⎣ ⎦ − Ψ⎢ ⎥⎣ ⎦

(C-11)

By substituting (C-10) into (C-11), it is derived that

( )( )

( )( )

( )

( )

2

2

2

2

1

1 11

1 1 1

m rqsq r sq m rq

s r m s r

r m

s mm s rsq sd

rs r s m r

r r r

r

m

s mm ssq sd

s r s m

s mm m

s r ss m

LI L L

L L L L LR L

L L LI RL L Rs sL R LsL

LL LI

L L s ss

L LL L L s s s

− Ψ⎡ ⎤⇒ = Ψ − Ψ =⎣ ⎦−

− −−⇒ = Ψ

−+ + ++

−⇒ = Ψ

−+ + ++

−=

⎛ ⎞ ⎛ ⎞+ + − + +⎜ ⎟ ⎜⎝ ⎠ ⎝ ⎠

σ

ω ω σσ ω ω

σ σσ

ω ω τσσ ω ω

τσ τστσ

ω ωσ τσ ω ω

τσ τσ τσ( )

( )

( )( )

2

2 2 222 2

2 22

22 2 2 2 2 2 2

1 1 1

2 1

sd

s mmsq sd

s rs m

s mmsq sd

s r s m

LIL L s s s

LIL L s s

Ψ

⎟

−⇒ = Ψ

+ + − + +

−⇒ = Ψ

+ + − +

ω ωτσ ω ω

τσ τσ τ σω ω τ σ

τσ τ σ τσ τ σ ω ω

(C-12)

Thus

( )

( )

( )

( )

( )

2

2

22 2 2 2 2

2

2

22 2

2

2

2 1

2 1

2 1

ms m

s rsq sd

s m

ms m

s rsq sd

s m

ms m

s rsq sd

LL LI

s s

LL LI

s

LL LI

s

ω − ω τ⇒ = Ψ

τ σ + τσ + τ σ ω − ω +

ω − ω τ⇒ ≈ Ψ

τσ + τ σ ω − ω +

ω − ω τ⇒ ≈ Ψ

τσ +

(C-13)

202

where

2

1 m

s r

r

r

LL L

LR

⎧σ = −⎪⎪⎨⎪τ =⎪⎩

(C-14)

The simplification in (C-13) is based on small τ and σ .

By inverse Laplace transform, the expression of sqI is time domain is obtained as

{ }( )

( )

( ) { }

2

21 1

2

21

22

2

2 1

2 1

1

ms m

s rsq sq sd

ms m

s r sd

tms m sd

s r

LL LI ( t ) I ( s ) ( s )

s

LL L

s s

L eL L

− −

∗−

−∗ τσ

⎧ ⎫ω − ω τ⎪ ⎪⎪ ⎪= = Ψ⎨ ⎬τσ +⎪ ⎪

⎪ ⎪⎩ ⎭⎧ ⎫ω − ω τ⎪ ⎪Ψ⎪ ⎪= ⎨ ⎬τσ +⎪ ⎪

⎪ ⎪⎩ ⎭

= ω − ω τ Ψ −

L L

L (C-15)

It is assumed that the magnitude of the stator flux vector is kept constant with flux

regulator in axis d . By considering (C-1) and (C-15), the torque is obtained as follows.

( ) { }( ) ( ){ }

22

2

222

2

32

3 12

3 12

e sd sq

tmsd s m sd

s r

tmsd s m

s r

T ( t ) P ( t ) i ( t )

LP eL L

LP eL L

−∗ ∗ τσ

−∗ τσ

= Ψ ⋅

= Ψ ω − ω τ Ψ −

τ= Ψ ω − ω −

(C-16)

By (C-1), the voltage equation in dq frame is

sd sd

sd s sd

sq s sq s sd s sd

d dV R idt dt

V R i

Ψ Ψ⎧ = + ≈⎪⎨⎪ = + ω Ψ ≈ ω Ψ⎩

(C-17)

By substituting (C-17) into (C-16), the relationship between the q voltage component

and the torque is developed as

203

{ } ( )2

22

3 12

tm

e sd sq ms r

LT ( t ) P e V fL L

−∗ τστ= ⋅Ψ − − ω (C-18)

where

( ) ( ) { }222

2

3 12

tm

m sd ms r

Lf P eL L

−∗ τστω = Ψ − ω (C-19)

Therefore, it is clear shown in (C-18) that the torque of induction machine can be

directly regulated by the q voltage component considering ( )mf ω as a disturbance to

the system. Similarly, the amplitude of stator flux vector can be regulated by the d

component of stator voltage directly as shown in (C-17). Above analysis forms the

principle of the direct torque and flux control (DTFC) scheme for the induction

machine.

Whole

Documents

space vector

average passenger

30th annual

allowed voltage

commanded

sliding mode

space voltage

voltage space