DIRECT TORQUE COTROL- SVM

7/30/2019 DIRECT TORQUE COTROL- SVM

1/12

8 IEEE INDUSTRIAL ELECTRONICS MAGAZINE FALL 2007 1932-4529/07/$25.002007IEEE

DIGITALVISION

LECH M. GRZESIAK AND

MARIAN P. KAZMIERKOWSKI

Exploring the Problems

and Remedies

Digital Object Identifie r 10.1109/MIE.2007.901483


2/12

H

IGH-PERFORMANCE

control methods for

converter-fed ac motors

such as field-oriented

control (FOC), direct

torque control (DTC),adaptive, and nonlinear

or sliding mode control [1][5], [42]

require complete information about

state and output variables. However,

sensors used in feedback loops

increase costs and decrease reliability

and immunity of the drive system;

therefore, they should be avoided.

Since the late 1980s, many efforts have

been made to reconstruct such state

variables of ac motors as rotor or stator

flux vectors and rotor angular speed,

e.g. [6][12].Artificial neural networks (ANNs)

are well suited for ac drive control and

estimation, because of their known

advantages, such as the ability to

approximate any nonlinear functions to

a desired degree of accuracy, learning

and generalization, fast parallel compu-

tation, robustness to input harmonic

ripples, and fault tolerance [3], [4].

These aspects are important in the case

of nonlinear systems, like converter-fed

ac drives, where linear control theory

cannot be directly applied. Additionally,

high-efficiency power electronic con-

verters used for ac motors operate in

switch mode, which results in very

noisy signals. For these reasons, ANNs

are attractive for signal processing and

control of ac drives. The commonlyused feed-forward ANN (FF-ANN) [Fig-

ure 1(a)] as universal nonlinear func-

tion approximator is suitable only for

steady-state systems. For a system

dynamics approximation task, a few

modifications of the FF-ANN architec-

ture are commonly used (Figure 1):

placement of tapped delay lines

(TDLs) at inputs of the FF-ANN [see

Figure 1(b)]

recurrent ANNs (RANNs) [see Figure

1(c)(d)].

All presented dynamic models[Figure 1(b)(d)] are widely applied as

nonlinear models of plants, estimators,

and controllers. The motivation for

using such an architecture is simple

mathematical description and avail-

ability of a very efficient training

method (e.g., available in the MATLAB

Neural Networks Toolbox).

However, presented schemes are

marked by serious limitations inher-

ited from TDLs and recurrent archi-

tecture, which requires that the

initially selected sampling time is

applied to the data in learning and

working phases.

The tapped delay neural architec-

ture [Figure 1(b)] has several limita-

tions caused by sampling and accuracy

of measurements [10]. These factorsinfluence rank of a matrix called the

input teaching matrix which is essen-

tial for neural network parameter tun-

ing. It is quite obvious that the ANN

with TDLs shown in Figure 1(b) (also

used in an ac drive control system) has

to act between two domains marked by

very different frequency spectra, at

inputs and at output. For instance, a

neural speed estimator works in steady

state at constant speed, which must be

reconstructed from periodical sinu-

soidal input signals (stator voltage andstator current). Limitations of certain

neural schemes based on time

instances of periodic signals are dis-

cussed in detail in [13] and [14].

The sampling time selected for a

continuous signal must fulfil Shannons

condition, so it is bounded from the

top as T < (1/2fmax) , but it is also

bounded from below. If a very small

sampling time is selected, then two

neighboring columns (each column

representing data at given time) of the

input teaching matrix become very

FIGURE 1 Example of ANNs used as models of static or dynamic plant: (a) a simple static FF-ANN, (b) adynamic model comprised of TDLs and static FF-ANN, (c) a Jordan network, and (d) a recurrent network with

TDLs on input and output signals.

ANN

Output

Input

DUDU

DU

DU

ANN

DU

DU

DU

DU

DU

DU

Output

OutputOutput

Input

InputInput

ANN

ANN

(a) (b)

(c) (d)

FALL 2007 IEEE INDUSTRIAL ELECTRONICS MAGAZINE 9


3/12

close. So if their differences are within

the accuracy of measurement, then the

input teaching matrix may loose its fea-

ture of being full column rank, reducing

the dimension of the approximation

space. The lower bound of sampling

time, below which two sinusoidal sig-

nals shifted by the sampling time are

fully located within accuracy of themeasurement, has been determined.

Using the concept of indistinguishable

signals, it was demonstrated in [10],

[14], and [26] that the range of allowed

sampling is related to maximal and

minimal useful frequencies of the signal

and accuracy of the measurement in

the following manner:

fmax

fmin

100

p, (1)

wherep is the accuracy of measurement

signal in %. This equation expresses thatthe useful frequency band of every

input signal is limited. One has to

acknowledge that one value of the sam-

pling time must be selected for all

inputs and the output signal. So, if there

are signals with different spectrumi.e.,

different fmin and fmax parametersand

the spectrum of one signal is not includ-

ed in another, then according to (1) it is

impossible to select just one T parame-

ter (sampling time) as being appropriate

for all the inputs and the output. To

avoid such limitation, it is necessary to

select either different physical signals or

use a nonlinear (or dynamic) transfor-

mation (called here preprocessing) of

existing signals to achieve the inputs

and output spectra confinement. In such

cases, new structures for modelingdynamic systems using a neural archi-

tecture can be created.

This article presents selected exam-

ples of ANN-based flux vector and

mechanical speed estimators. By using

appropriate preprocessing of input sig-

nals, the performances of flux vector

and speed estimators are considerably

improved (compared to estimators

based on TDL FF-ANN and RANN) in

terms of accuracy and sensitivity to

parameter changes. The properties of

the discussed estimators are illustratedin an example of the induction motor

drive system with the direct torque con-

trol and space vector modulation (DTC-

SVM) scheme, as shown in Figure 2.

This system has been selected as a

very practical high-performance

scheme that can be used for induction

and/or permanent magnet synchronous

motor drives [3]. All presented experi-

mental oscillograms have been meas-

ured using the laboratory set-up (Figure

B) described in the Appendix.

Stator Flux Estimation

Based on ANN

Overview

Operation of high-performance ac motor

drives (DTC or FOC) depends stronglyon stator (or rotor) flux estimation accu-

racy. Several approaches are used for

flux vector estimation: model-based [15],

[16], using Kalman filters [17], Luenberg-

er observers [8], [18], [19], and many

other closed-loop estimators (e.g., [6],

[20]). Nevertheless, in many applica-

tions the stator flux is calculated from

the so-called voltage model

s(t) =

s(t0)+

tt0

(vs(t)

Rsis

(t))dt,(2)

where s

denotes stator flux space vec-

tor, vs, is denote stator voltage and cur-

rent space vectors, respectively, and R sdenotes the estimated stator resistance.

Accurate knowledge of stator resist-

ance is important for speed sensorless

drives operating at a wide speed con-

trol range including zero speed in

steady states and transients. In order to

implement (2), numerous improved

FIGURE 2 Block scheme of a DTC-SVM induction motor drive apply-ing flux and speed estimators based on ANN.

M

Control System (DTC SVM)

refs

Tes s

Load

ANN-a ANN-b

Vdc

LPF

m

refm

FIGURE 3 RANN architecture for stator flux estimation.

(k)us

(k)is

(k)is

(k)us

DU

ANN

DU

s(k+1)

s(k+1)

10 IEEE INDUSTRIAL ELECTRONICS MAGAZINE FALL 2007


4/12

integration algorithms have been pro-

posed [6], [7], [21], [22]. These deal with

drift and initial condition problems relat-

ed to pure integration. However, any sta-

tor flux estimator which includes back

electromotive force integration requires

an additional algorithm for Rs adapta-

tion. This is because even relatively

small modeling errors due to motor tem-perature rise can lead to large errors in

flux estimation accuracy [6].

Simple RANN Architecture

for Estimating Stator Flux VectorTo reconstruct flux directly from stator

currents and voltages (instead of back

EMF), a nonlinear transformation

between stator flux and stator voltage

and current will be reflected via neural

model. From a basic ac motor mathe-

matical model, stator voltage and flux

current equations can be written as[24], [25]

d

dt

s(t) = Rsis(t)+ vs(t) .

(3)

Now, if is and vs are assumed to be

measurable and the model parameters

Rs, Lm, and Lr are constant, then the

system becomes linear and no typical

neural network attributes are needed.

Having collected enough data points

in time and using a simple linear

autoregressive moving average model

(ARMA), one can find a linear approxi-

mation [is(k), vs(k)] s(k+ 1) as

a direct pseudo-inverse calculation.

However, if some of the parameters

change in time, they will influence is;therefore, a nonlinear approximation

is necessary.

The stator resistance variations

should be included in a training set;

thus, some level of robustness could

be achieved. Note that the relationship

between (vs(k), is(k)), and s(k) is

not a static one. This in turn implies

that there is no such function f1 that

satisfies s

(k) = f1(vs(k), is(k)). One

can expand the approximation space

to turn the problem into a static one

s

(k+ 1) = f2(v s(k), is(k),

s(k)) . (4)

To build a neural estimator using

(4), it is necessary to write down (3) in

stationary orthogonal coordinates.

Now the mathematical model of flux

components is given by

d

dts(t) = vs(t) Rsis(t)

d

dts (t) = vs (t) Rsis (t), (5)

where v(k)s , v

(k)s , i

(k)s , i

(k)s ,

(k )s , and

(k)s denote real and imaginary com-

ponents of the stator voltage, stator

current, and stator and rotor flux

space vectors, respectively.A neural architecture for stator

flux vector estimation is presented in

Figure 3. The ANN estimates stator

flux from the stator voltage and cur-

rent data. Mathematically, the prob-

lem can be classified as a nonlinear

dynamic system approximation. The

ANN can use only a single-hidden-

layer architecture. Training data can

be generated from a simulated mathe-

matical model operated in various

working conditions or from the data

collected from motor measurements.To train the network, many different

methods can be applied. Mostly, back-

propagation methods are used. For

such a training method, the architec-

ture of the network (number of hid-

den layers and neurons) must be set

up in advance. One can also use incre-

mental learning methods (originated

in [27] and [28]), presented for exam-

ple in [10] and [26], to design an FF-

ANN for function approximation. If an

incremental learning method is cho-

sen for ANN design and training (Fig-

ure 3), the number of hidden sigmoid

neurons is not fixed in advance. At

FIGURE 4 Illustration of incremental approximation concept shown in sequence situation when first, second, and nth neuron are added.

f1

f2

fm

x1

x2

g1 g1

g2 g2

xd

g1x1

x2

xd

x1

x2

xd

f1

f2

fm

f1

f2

fmgn

(a) (b) (c)

Artificial neural networks are well suited

for ac drive control and estimation.



5/12


each iteration, one hidden neuron isoptimized and added to the network,

but all other hidden neuron parame-

ters remain unchanged. Then all out-

put weights are recalculated. The red

lines in Figure 4 indicate connection

parameters being recalculated when a

neuron is added to the network. The

iterative process is terminated when

final conditions such as error level or

number of hidden neurons are met.This type of ANN can approximate

any continuous function with any

accuracy, provided that one may use

as many hidden neurons as needed

[10], [27], [28].

However, this ANN architecture

(Figure 3) is marked by serious limita-

tions inherited from recurrent ANN.

First, it requires that initially selected

sampling time is applied to the data in

learning and recall phases, as dis-

cussed earlier. Second, there is no

guarantee of an ANN estimators stable

operation when it is trained in a series-

parallel identification scheme (during

the learning process, inputs and out-

puts signals of plant have been used)

and then works in a parallel configura-tion (only real input signals are avail-

able) using an estimated output signal.

FF-ANN with Dynamical Preprocessing

as a Stator Flux Estimator

To improve stator flux estimation accu-

racy, it is possible to use, rather than

the FF-ANN with TDLs or RANN, a sim-

ple dynamical preprocessing method

FIGURE 5 An FF-ANN-based stator flux estimator with dynamic signal preprocessing: (a) with ortogonal (, ) components of voltage and currentspace vector on input and (b) with natural (a, b, c) terminal voltage and phase current on input.

vab0v1vs

vs

is

K1u T1u

FF-ANN FF-ANN

K2u T2u

K3u T3u

K1i T1i

K2i T2i

K3i T3i

is

K1u T1uK0u T0u

K2u T2u

K3u T3u

vab

vbc

ia

K1i T1i

K2i T2i

K3i T3i

ib

v1

i1

s

s

s

si1

(a) (b)

Considerable improvement of flux vector estimation

can be achieved using FF-ANN with dynamic signal

preprocessing.


6/12

[29]. It was shown in [29] and [30] that

dynamical filtering of ANNs input sig-

nals can turn a believed nonfunction

modeling problem into a function mod-

eling one. The natural evaluations of

FF-ANN with TDLs on the input signals

(vs, vs, is, is) conduct to the

solution presented in [29], which is

shown in Figure 5(a). Input signals

(vs, vs, is, is) prompt different

low-pass filters (first-order dynamic

units). Outputs of filters are connected

to FF-ANN. Vector flux components

have been achieved on two outputs of

the FF-ANN. Time constants can be

chosen by trial and error. The best

results are when two filters act roughly

as integrators, whereas the third one

FIGURE 6 Performance of DTC-SVM drive: (a) results for programmable LPF as flux estimator, (b) results for neural flux estimator, and (c) s com-ponents, speed, and torque signals.

1.6 1.7 1.8 1.9 2 2.1 2.2

0 2 4 6 8 10

0 2 4 6 8 10

0 2 4 6 8 10

1.7 1.8 1.9 2 2.1 2.21.6

1.6 1.7 1.8 1.9 2 2.1 2.2

0 2 4 6 8 10

0 2 4 6 8 10

0 2 4 6 8 10

1.7 1.8 1.9 2 2.1 2.21.6

1

0.5

0

0.5

0.5

0

0.5

50

0

50

20

0

20

20

0

20

1

Te

[Nm]

T

load[Nm]

m

[Nm]

1

0.8

0.6

0.4

0.2

0.2

0.4

0.5 0 0.5 1

0.6

0.8

11 0.5 0 0.5 11

0

s[Vs] s[Vs]

s

[Vs]

s

[Vs]

Error[Vs]

1

0.5

0

0.5

0.5

0

0.5

50

0

50

20

0

20

20

0

20

1

Te

[Nm]

T

load[Nm]

m

[rad/s]

1

0.8

0.6

0.4

0.2

0.2

0.4

0.6

0.8

1

0

s

[Vs]

s

[Vs]

Error[Vs]

s

s-sest

s -sest

sest

m

Tload

Te

s-sest

s -sest

sest

m

Tload

Te

s

t [s] t [s]

Time (s) Time (s)

(a)

(b)

(c)



7/12

has a very small time constant and

performs only the duty of noise attenu-

ation. The similar flux estimator based

on FF-ANN and dynamic signal prepro-

cessing, but uses direct stator electri-

cal signals (a, b, c phase current and

terminal voltage), is shown in Figure

5(b) [30]. Note that measured stator

voltages and currents are not trans-formed into components, whereas

estimated flux components are

expressed in . The Clark transform

is automatically formed within the FF-

ANN during the learning process. Each

stator electrical signal, (vab, vbc, ia, ib),

excites three different first-order

dynamical systems. As shown in [29],

those almost dynamically unprocessed

signals, combined with others, span an

approximation space that turns a mod-

eling problem into a static one. It was

found that there exists such a functionf3 that satisfies

s

(t) = f3(v1(t), . . . ,v6(t),

i1(t), . . . , i6(t)) . (6)

There are several advantages related

to that scheme:

It is a very simple learning algo-

rithm because a typical feed for-

ward architecture of neural network

is used.

The resultant estimator is stable

(stable dynamics and static nonlin-

earity connected in series).

Contrary to an uncompensated pure

integrator, there is no drift problem

due to LPFs presence.

There are different sampling times

available in learning and recall

mode (contrary to the tapped delay

or recurrent architecture).

A significant level of robustness to sta-

tor resistance variations is achieved.The Matlab/Simulink model of the

DTC-SVM drive is shown in Figure A in

the Appendix. The performance of the

DTC-SVM drive with ANN stator flux

estimator (Figure 5) is compared with

the programmable low-pass filter (PLPF)

algorithm presented in [22], [43], and

[44]. After off-line supervised training,

ANN estimators have been implemented

in the DTC drive. Figure 6 shows select-

ed signals in the DTC-SVM drive. A sta-

tor resistance rise in the amount of 30%

has been used. Figure 6(a) (with PLPFestimator) shows clearly that such iden-

tification error leads to significant

dynamics deterioration and even nonex-

clusion of unstable behavior.

A contrary ANN-based estimator with

dynamic signal preprocessing with the

same conditions assures stable opera-

tion and decent dynamics [Figure 6(b)].

Experimental Verification

The block diagram of the experimen-

tal system is shown in the Appendix

(Figure B). The ANN was learned with

the help of patterns taken from tested

drives equipped with a conventional

PLPF-based estimator.

Estimated resistancethe one pres-

ent in the algorithmic estimatorwas

updated simultaneously with imposed

stator resistance variations. If there is

an error in the stator resistance (Rs) of

about 30%, a response of the system to

step change of reference speed visiblydeteriorates [Figure 7(a)]. Under simi-

lar conditions, the drive that takes

advantage of feedback signals provided

by the ANN estimator with dynamic

signal preprocessing reverses smooth-

ly [Figure 7(b)]. So the idea of dynami-

cal preprocessing at the inputs of the

neural approximator was successfully

implemented and shows that common

problems related to recurrent and

tapped delayed ANNs can be avoided.

Selected Problemsof ANN-Based Speed Estimation

Overview

In the past few years, many speed-sen-

sorless techniques have been pro-

posed to cope with the speed sensing

problem [5]. Developed speed estima-

tion algorithms are more or less

parameter dependent and/or computa-

tionally time-consuming, so further

investigation is justified. Speed estima-

tors are designed on the basis of the

very common belief that information

about actual speed is contained in

FIGURE 7 Performance of the experimental drive: (a) with incorrect stator resistance (Rs) identification (PLPF estimator) and (b) with ANN-basedestimator (visible improvement compare to the PLPF estimator).

0

50

0

50

20

0

20

20

0

20

50

0

50

20

0

20

20

0

20

0.5 1 1.5 2 2.5 3 3.5 4 4.5

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

0

Test [Nm]

Tref

[Nm]

[rad/s]

Test [Nm]

Tref

[Nm]

[rad/s]

0.5 1 1.5 2 2.5 3 3.5 4 4.5

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

e e

m

Tref

Teste

m

Tref

Teste

Time (s) Time (s)

(a) (b)



8/12

easily accessible electrical signals.

Available solutions can be divided into

two main groups. The first group

includes methods that should be

regarded as algorithmic ones. These

methods exploit a mathematical model

of an induction motorextended Luen-

berger observer [18], sliding-mode

observers [33], extended Kalman filter(EKF) [17], model reference adaptive

systems (MRAS) [16], indirect flux

detection by online reactance meas-

urement (INFORM) [2], high-frequency

signal injection [34], low-frequency sig-

nal injection [12], slip calculation [16],

pseudoinversion [10], and their muta-

tions. Unfortunately, any mathematical

modelings introduce some simplifica-

tion, which in turn entails deteriora-

tion of estimation. On the other hand,

there is a set of solutions that does not

take advantage of any mathematicalmodel of discussed plant. This group

consists of estimators based on ANN.

In such a case, only inputs and outputs

are known, whereas nonlinear relation-

ships between them are not. ANN solu-

tions can incorporate online trained

nets [11], [31], [35] or off-line trained

ones [36], [37]. These estimators natu-

rally possess robustness to noise and

parameter disturbances.

However, most popular ANNs dedi-

cated to system dynamics approxima-

tion, consisting of a multilayer FF-ANN

with additional TDLs or recurrences

[Figure 1(b)(d)], cause problems and

limitations that have been briefly

discussed earlier. Therefore, the speed

reconstruction problem using as

inputs current and voltage signals has

exactly the same limitation like flux

vector estimation. Therefore, it is also

necessary to use a nonlinear transfor-

mation (called here preprocessing) of

existing signals to achieve the inputs

and output spectra confinement.

Speed Estimators

with Nonlinear Preprocessing

and Neural Function Approximator

Nonlinear preprocessing of the elec-

trical signals should be applied to

solve problems occurring with TDLs.

Its task is to provide input signals

with a spectrum similar to the speed

signal. At the same time, a dimension

of the approximation space is

enlarged and the number of delay

units can be reduced even to zero

[10], [13]. There is a large number ofnonlinear functions useful at this

stage of preprocessing [see (7)]. The

nonlinear preprocessing is performed

in order to create a set of signals

marked by suitable spectrum. These

transformations are designed with

the aid of cross and dot products of

stator voltage vs and current is space

vectors, giving for instance the fol-

lowing equations:

u1 = abs(v s) =v2s + v

2s

u2 = abs(is) =

i2s + i

2s

u3 = Re(v s i

s )

= isvs + isvs

u4 = Im(vs i

s )

= isvs isvs

u5 = Re

v s

is

=isvs + isvs

i2s + i

2s

u6 = Imvs

is

=isvs isvs

i2s + i

2s

...

un = f(is, vs , is , vs) . (7)

Given the first set of six signals

(7), one can enlarge this set by

including their product ratios and

powers [13]. The basic structure of

the neural speed estimator with non-

linear input signal preprocessing is

shown in Figure 8. It is a difficult task

to take the best of the signal results(including most of the important

FIGURE 8 Speed estimator for ac motor with nonlinear preprocessing of input signals and neuralfunction approximator.

FF-ANN

Nonlinear

Preprocessing

vs

vs

is

is

m

FIGURE 9 Improved speed estimator for acmotor with nonlinear preprocessing, neural(PCA) linear preprocessing and neural func-tion approximator.

NonlinearPreprocessing

isa isb vsa vsb

is is vs vs

Self-Organizing

PCA

FF-ANN

Transformationabc/

Online

Offline

m



9/12

information) from preprocessing. In

[13] and [32], few methods of final

selection from candidate signals have

been described and discussed.

Speed Estimator with Nonlinear

Preprocessing and NeuralImplementation of Principal

Component Analysis andFunction ApproximatorFurther improvement of the speed esti-

mator can be achieved by adding an

extra system that can automatically

select the best signals from the large

number of candidates [29].

The structure of such a speed esti-

mator (Figure 9) consists of three

stages connected in series: a nonlinear

preprocessing, a linear preprocessing

(self-organizing principal component

analysis, or PCA) and nonlinear func-tion approximator (FF-ANN) stage. The

second stage is designed to supply

optimal signals for the neural function

approximator. The goal of the second

stage is to decorrelate variables and to

maximize eigenvalues of the autocorre-

lation matrix at the same time. Such a

problem can be solved by PCA [32].

The PCA model generates a new set of

a few decorrelated variables called

principal components (PCs). The pur-

pose of the PCA is to derive a smallnumber of decorrelated linear combi-

nations of a set of zero-mean variables

while retaining as much of the

APPENDIX

FIGURE A Simulink model of a DTC-SVM drive with neural flux and neural speed estimators.

PITorque Controller

PISpeed Controller

k2k1

gamma

U_s_d

U_s_q

U_s_alfa_r

U_s_beta_r

Transformation

Visualization_Data.mat

Training_Data.mat

To File1

Load Torque

Signal Builder

Reference SpeedSignal Builder

In

U1

U2

U3

U4

U5

U6

NonlinearPreprocessing

y{1}p{1}

Neural Network 2Speed Estimator

y{1}p{1}

Neural Network 1Stator Flux Estimator

MS3

MS2

MS1

In

U1

U2

U3

U4

U5

U6

Linear DynamicPrprocessing

U_s_alfa_r

U_s_beta_r

U_s_alfa

U_s_beta

Inverter

[abs_Psi_s]

[gamma]

[Speed_est]

[Psi_s_est]

[Psi_s]

[Psi_r]

[Speed]

[Us_Is]

[Torque]

[Psi_s]

[Psi_r]

[Us_Is]

[gamma]

[Us_Is]

[abs_Psi_s]

[Speed_est]

[Torque]

[Speed_est]

[Psi_s_est]

[Torque]

[Speed]

[Psi_s_est]

[Psi_s]

[Us_Is]

[Us_Is]

[Psi_s]

[Speed]

[Speed]

PIFlux Controller

Reference_Flux

In

gamma

abs(Psi_s)

Calculation

Us_alfa

Us_beta

T_load

I_s_alfa

I_s_ beta

Torque

Speed

Psi_r_alfa

Psi_r_beta

Psi _s_alfa

Psi _s_beta

ac Motor

To File2

+

+

+

The Simulink simulation model of a DTC-SVM drive with neural flux

and speed estimators is shown in Figure A.

The block scheme of the laboratory set-up used for experimental

verification is shown in Figure B. It consists of a 1.5-kW, six-pole ac

motor which is fed by an IGBT voltage source inverter. For control

tasks, a dSpace DS1103 card programmed in the Matlab/Simulink

Real-Time Workshop and monitored via the ControlDesk interface

has been used. The control and estimation algorithms run at a com-

putation time of 0.5 ms. The angular speed of the drive is measured

with an encoder.



10/12

information from the original variables

as possible [38]. This, in turn, implies

reduction of a number of signals in

comparison to the number of nonlinear

transformations. The dimensionality

reduction is one of the basic features of

PCA. The PCA model involves second-

order statistics and guarantees data

decorrelation with simultaneousdenoising [38]. PCA is normally done

by analytically solving an eigenvalue

problem of the input correlation func-

tion. The eigenvectors must be put indecreasing order with respect to eigen-

values. However, in [38] it was demon-

strated that PCA can be accomplished

by a single-layer, linear self-organizingneural network trained with a modified

Hebbian learning rule [37], [39], [40].

Several different types of stopping

rules have been developed to deter-

mine the number of extracted compo-

nents from a given analysis.

One of the most popular is a per-

centage of variance criterion in which

successive eigenvectors are extracted

until some absolute percentage of the

total variance has been explained (com-

monly used with typical threshold set

to 95%). Such a type of PCA implemen-tation is shown in Figure 9 as the sec-

ond stage of signal preprocessing for a

neural speed estimator. Computer test

results of a DTC-SVM drive operated

with a feedback signal from a neural

speed estimator is shown in Figure 10.

One can see the very good performance

of drive systems.

Experimental Verification

The experimental has been performed

on drive system shown in Figure B

with data given in the Appendix. The

control and estimation algorithms

have been running with a sampling

time of 0.5 ms. The drive was tested at

random speed and load profiles.

The speed of the drive was meas-

ured additionally with an encoder.

After offline training, the nonlinear

ANN was implemented in the DSP

board. Figure 11 shows some experi-

mental results of a drive operated with

speed estimator of Figure 9(b).

Conclusions

This article presented problems and

selected improvements related to the

application and implementation of

ANN-based flux vector and mechani-

cal speed estimators for control of

high-performance pulse width modu-

lated (PWM) inverter-fed induction

motor drives. Key conclusions include

the following:

The tapped delay neural architecture

has several limitations caused by sampling and

accuracy of measurements.

FIGURE B Block diagram of experimental set-up DTC-SVM drive for testing of flux and speed estimators.

acGrid

M

Rectifier Inverter

ac Motor

dSPACEDS1103

Visualization

Power Circuit

Load

Encoder

PC Computer(ControlDesk)

Speed

Calculation

Flux EstimatorPLPF

NeuralFlux Estimator

NeuralSpeed Estimator

us,

is

S1

S2

is

SpeedController

PWM-SVM

DTCTorque and Flux

Controller

Input: Keyboard Output: Screen

CurrentandVoltage

Measurement

M ENC

Parameters Set UpConfiguration Set Up

est

est m

ref

ref

est

MOTOR DATA:

Induction cage motor Sf100L6K:

PN= 1.5 kW, UN= 220 V, IN= 6.8 A, fN= 50 Hz, cos N= 0.75, N= 0.76,

nN= 930 rpm, Rs = 1.54 , Rr= 1.29 , Xs = 31.56 , Xr= 30.43 ,

Xm = 28.74

INVERTER DATA:

IPM-IGBT module:Vdc = 330 V, switching frequency= 10 kHz, Iout = 50A



11/12

Recurrent ANNs can be used for flux

vector estimation only under the

assumption of very fast sampling

equal in learning and recall mode.

Considerable improvement of flux

vector estimation can be achieved

using FF-ANN with dynamic signal

preprocessing (Figure 5). This estima-

tor poses the following advantages:very simple learning algorithm, sta-

ble operation, no drift problems due

to LPFs presence, different sampling

times available in learning and recall

mode (contrary to the tapped delay

or recurrent architecture), and some

level of robustness to stator resist-

ance variations.

For reliable mechanical speed esti-

mation, an FF-ANN with nonlinear

signal preprocessing has been

described. Two-stage preprocessing

(nonlinear preprocesing and linearPCA) of four terminal signals is com-

bined with FF-ANN. The first one

guarantees enlarged approximation

space using a linearly independent

set of input signals based on cross

and dot products of motor voltage

and current vectors. The second one

(linear stage, PCA) takes advantage

of self-organizing principal compo-

nent analysis to optimize the

approximation space without loss of

information. Proposed estimators

are very convenient to implement.

The presented experimental oscillo-

grams measured in a 1.5-kW labora-

tory induction motor drive verify

estimators based on ANN with non-

linear preprocessing of input signals.

Achieved results can be easily

extended to DTC-SVM permanent

magnet synchronous motor drives.

It is expected that, thanks to contin-

uous developments in digital signal

processing technology, ANN-based

techniques will have a strong impacton drive control, estimation, and

monitoring in the coming decades.

BiographiesLech M. Grzesiak graduated from the

Electrical Engineering Faculty of

Warsaw University of Technology in

1976. He received the Ph.D. in 1985 and

the Dr.Sc. degree in 2002, respectively,

from the same university. Since 1977, he


FIGURE 10 Simulation results of DTC-SVM drive with speed feedback signal taken from a neuralspeed estimator (as shown in Figure 8).

100

0

0 2 4 6 8 10 12 14 16 18 20

0 2 4 6 8 10 12 14 16 18 20

0 2 4 6 8 10 12 14 16 18 20

0 2 4 6 8 10

Time (s)

12 14 16 18 20

100

5

0

5

20

0

20

10

0

Tload

[Nm]

Te

[Nm

]

Errorm

[%]

m

[rad/s]

10

FIGURE 11 Experimental waveforms of speed estimation based on seven principal components(estimator structure shown in Figure 9).

0

50

0

50

50

0

50

5

0

5

10 20 30 40 50 60 70 80

0 10 20 30 40 50 60 70 80

0 10 20 30 40

Time (s)

50 60 70 80

[

rad/s]

m

[rad/s]

es

[rad/s]

By using appropriate preprocessing of input signals,

the performances of flux vector and speed estimators

are considerably improved in terms of accuracy and

sensitivity to parameter changes.


12/12

has been employed at Warsaw Universi-

ty of Technology, currently as a profes-

sor. He was also codirector of the Centre

of Excellence on Power Electronics Intel-

ligent Control for Energy Conservation

(PELINCEC) from 20032005. He is an

associate editor ofIEEE Transactions on

Industrial Electronics since 2004. His

main research interests and publica-tions are in the theory and application

of control, generally dedicated for elec-

tric drives, power electronics, and ener-

gy generating systems. From 1994, his

research interest has focused on devel-

opment and applications of neural sys-

tems. He is an author and coauthor of a

variety of papers related to this subject.

He is a Senior Member of the IEEE.

Marian P. Kazmierkowski re-

ceived the M.S., Ph.D., and Dr.Sci.

degrees in electrical engineering from

the Institute of Control and IndustrialElectronics (ICIE), Warsaw University of

Technology, Warsaw, Poland, in 1968,

1972, and 1981, respectively. Since

1987, he has been a professor and

director of ICIE. He was also head of

PELINCEC from 20032005 (European

Framework Program V) at ICIE, Warsaw

University of Technology, Poland. He

coedited (with R. Krishnan and F.

Blaabjerg) and coauthored the com-

pendium Control in Power Electronics

(Academic Press, 2002). He received an

Honorary Doctorate degree from Aal-

borg University in 2004 and the Dr.

Eugene Mittelmann Achievement

Award from the IEEE Industrial Elec-

tronics Society in 2005. He was the edi-

tor-in-chief of IEEE Transactions on

Industrial Electronics (20042006). He is

past-chair of the IEEE Poland Section.

References[1] M.P. Kazmierkowski and T. Kowalska-Orlowska,

ANN based estimation and control in converter-fed induction motor drives, inSoft Computing in

Industrial Electronics, S.J. Osaka and L. Sztandera,

Eds. Heidelberg, Germany: Physica Verlag, 2001.

[2] P. Vas, Arti ficia l-Int ellig ence- Based Elect rica lMachines and Drives: Application of Fuzzy, Neural,Fuzzy-Neural, and Genetic-Algorithm-Based Tech-niques. London: Oxford Univ. Press, 1999.

[3] M.P. Kazmierkowski, R. Krishnan, and F.Blaabjerg, Control in Power ElectronicsSelected

Problems. New York: Academic, 2002.

[4] B.K. Bose,Modern Power Electronics and AC Dri-ves. Pearson Education, Englewood Cliffs, NJ:Prentice-Hall, 2002.

[5]K. Rajashekara, A. Kawamura, and K. Matsue,Sensor-less Control of AC Motor Drives, Speed and Position

Sensorless Operation. New York: IEEE Press, 1996.

[6] J. Holtz and J. Quan, Drift- and parameter-com-pensated flux estimator for persistent zero-stator-frequency operation of sensorless-controlledinduction motors,IEEE Trans. Ind. Applicat., vol.39, no. 4, pp. 10521060, July-Aug. 2003.

[7] J. Hu and B. Wu, New integration algorithms forestimating motor flux over a wide speed range,

IEEE Trans. Power Electron., vol. 13, no. 5, pp.964977, Sept. 1998.

[8] J. Maes and J.A. Melkebeek, Speed-sensorlessdirect torque control of induction motors using

an adaptive flux observer,IEEE Trans. Ind. Appli-cat., vol. 36, no. 3, pp. 778785, MayJune 2000.

[9] J. Zhao and B.K. Bose, Neural-network-basedwaveform processing and delayless filtering inpower electronics and AC drives,IEEE Tran. Ind.

Electron., vol. 51, no. 5, pp. 981991, Oct. 2004.

[10] B. Bel iczy nski and L. Grzesiak, Inductionmotor speed estimation: Neural versus phenom-enological model approach, Neurocomput. ,vol. 43, no. 1-4, pp. 1736, 2002.

[11] S.-H. Kim, P.T.-S. Ark, J.-Y. Yoo, and G.-T. Park,Speed-sensorless vector control of an inductionmotor using neural network speed estimator,IEEETrans. Ind. Electron., vol. 48, no. 3, pp. 609614, 2001.

[12] V.-M. Leppnen, Low-frequency injection-basedspeed sensorless control of induction motorsApplicability and implementation aspects, in

Proc. EPE Conf., 2003 [CD-ROM].

[13] L.M. Grzesiak, B. Beliczynski, and B. Ufnalski,Input preprocessing in tapped delay neuralarchitecture for induction motor speed estima-tion, inProc. EPE Conf., 2003 [CD-ROM].

[14] B. Beliczynski, On input discretisation process-es for tapped delay neural archirecture, inProc.

Int. Conf. Artificial Neural Networks Genetic Algo-rithms (ICANNGA), 2003, pp. 2832.

[15] P.L. Jansen and R.D. Lorenz, A physicallyinsightful approach to the design and accuracyassessment of flux observers for field orientedinduction machine drives,IEEE Trans. Ind. Appli-cat., vol. 30, no. 1, pp. 101110, Jan.-Feb. 1994.

[16] P. Vas,Sensorless Vector and Direct Torque Control(Monographs in Electrical and Electronic Engi-neering, vol. 42). London: Oxford Univ. Press, 1998.

[17] Sensorless control with Kalman filter on

TMS320 fixed-point DSP, Texas InstrumentsEurope, Literature no. BPRA057, 1997, Dallas, TX.

[18] H. Kubota and K. Matsuse, Speed sensorlessfield-oriented control of induction motor with rotorresistance adaptation,IEEE Trans. Ind. Applicat.,vol. 30, no. 5, pp. 12191224, Sept.Oct. 1994.

[19] H. Kubota, K. Matsuse, and T. Nakano, DSP-based speed adaptive flux observer of inductionmotor, IEEE Trans. Ind. Applicat., vol. 29, no. 2,pp. 344348, Mar.Apr. 1993.

[20] D. Casadei, G. Serra, and A. Tani, Steady-stateand transient performance evaluation of a DTCscheme in the low speed range,IEEE Trans. Power

Electron., vol. 16, no. 6, pp. 846851, Nov. 2001.

[21] G. Andreescu and A. Popa, Flux estimatorbased on integrator with DC-offset correctionloop for sensorless direct torque and flux con-trol, inProc. Int. Conf. on Electrical Machines

(ICEM), 2002, [CD-ROM].

[22] M. Hinkkanenand J. Luomi, Modified integratorfor voltage model flux estimation of inductionmotors, inProc. IEEE Ind. Electronics Society Conf.(IECON), vol. 2, Nov. 2001, pp. 13391343.

[23] L.M. Grzesiak and B. Beliczynski, Simple neuralcascade architecture for estimating of stator androtor flux vectors, inProc. EPE99 Conf., 1999,[CD-ROM].

[24] M.P. Kazmierkowski and H. Tunia,AutomaticControl of Converter-Fed Drives. Warsaw, Poland:Elsevier-Amsterdam and PWN, 1994.

[25] W. Leonhard, Control of Electric Drives. NewYork: Springer-Verlag, 1985.

[26] B. Beliczynski and L. Grzesiak, Dynamic modelsand learning: application of neural approach to esti-mation of stator and rotor fluxes in an inductionmotor, inProc. Conf. Engineering Application of

Neural Networks, Warsaw, Sept.1999, pp. 110116.

[27] A.R. Barron, Universal approximation bounds forsuperpositions of a sigmoidal function,IEEE Trans.

Inform. Theory, vol. 39, no. 3, pp. 930945, 1993.

[28] L.K. Jones, A simple lemma on greedy approxi-mation in Hilbert space and convergence rates forprojection pursuit regression and neural network

training, inAnnals of Statistics. Beachwood, OH:Institute of Mathematical Statistics, 1992.

[29] L.M. Grzesiak and B. Ufnalski, Neural stator fluxestimator with dynamical signal preprocessing,inProc. IEEE Conf. AFRICON, 2004, pp. 11371142.

[30] L.M. Grzesiak and B. Ufnalski, DTC drive withANN-based stator flux estimator, in Proc. 11th

EPE05, Sept. 1114, 2005, Dresden, Germany,[CD-ROM].

[31] D. L. Sobczuk, Application of ANN for control ofPWM inverter fed induction motor drives, Ph.D.thesis, Warsaw University of Technology, Facultyof Electrical Engineering, 1999.

[32] B. Ufnalski, Application of artificial neural net-works for estimation of rotor angular speed andstator flux in cage induction motor drive, Ph.D.thesis, Warsaw University of Technology, Facultyof Electrical Engineering, 2005 (in Polish).

[33] A. Derdiyok, A novel speed estimation algo-rithm for induction machines,Elect. Power Sys-tems Res., vol. 64, no. 1, pp. 7380, 2003.

[34] R.D. Lorenz, Advances in electric drive con-trol, inProc. IEEE Int. Electric Machines and DrivesConf. (IEMDC), 1999, pp. 916.

[35] L. Ben-Brahim, S. Tadakuma, and A. Akdag,Speed control of induction motor without rota-tional transducers, IEEE Trans. Ind. Applicat.,vol. 35, no. 4, pp. 844850, 1999.

[36] T. Orlowska-Kowalska and P. Migas, Analysis ofthe induction motor speed estimation qualityusing neural networks of different structure,

Archives Elect. Eng., vol. L, no. 4, pp. 411425, 2001.

[37] A. Cichocki and S. Amari,Adaptive Blind Signaland Image Processing: Learning Algorithms and

Applications. New York: Wiley, 2002.

[38] A. Hyvrinen and E. Oja, Independent compo-nent analysis: A tutorial, Helsinki University ofTechnology, Laboratory of Computer and Infor-mation Science, 1999 [Online]. Available:http://www.cis.hut.fi/projects/ica/

[39] P. Paplinski, Generalized Hebbian learning andits application in dimensionality reduction, Fac-ulty of Computing and Information Technology,Department of Digital Systems, MONASH Aus-tralias International University, Technical Rep.97-2, 1997.

[40] NeuroDimension,NeuroSolutions 4.2 Manual,2003 (free evaluation copy) [Online] Available:http://www.nd.com/

[41] K. Simoes and B.K. Bose, Neural network basedestimation of feedback signals for a vector con-trolled induction motor drive, IEEE Trans. Ind.

Applicat., vol. 34, no. 3, pp. 620629, 1995.

[42] G.S. Buja and M.P. Kazmierkowski, Directtorque control of PWM inverter-fed ac motorsAsurvey, IEEE Trans. Ind. Elect., vol. 51, no. 4, pp.744757, Aug. 2004.

[43] B.K. Bose and N.R. Patel, A sensorless stator fluxoriented vector controlled induction motor drivewith neuro-fuzzy based performance improvement,inIEEE IAS Conf. Rec., 1997, pp. 393400.

[44] L. Borges, B.K. Bose, and J. Pinto, Recurrentneural network based implementation of pro-grammable cascaded low pass filter used in sta-tor flux synthesis of vector controlled inductionmotor drive,IEEE Trans. Ind. Elect., vol. 46, pp.662665, June 1999.


DIRECT TORQUE COTROL- SVM

Documents