Analysis and Optimization of Clutch Actuator on Automated ... · Combined with the clutch actuator module and clutch module, shifting process simulations with different TPS control,

05P-60

Analysis and Optimization of Clutch Actuator on Automated Manual Transmission System

Ching-Huan Tseng, Ming-Feng Hsieh Department of Mechanical Engineering, National Chiao Tung University, TAIWAN (R.O.C.)

Copyright © 2004 SAE International

ABSTRACT

This paper presents dynamic analysis, control simulation, and optimization for a clutch actuator on Automated Manual Transmission (AMT). The main object is to optimize a clutch actuator, both in mechanical component and control component, to be able to drive clutch faster and more stable. Using Matlab Simulink, both dynamic model and control model are integrated into a unit. According to the integrated model and using optimization technique, the clutch actuator is modified to capable of disengage/engage clutch faster and more stable in a real case. On the other hand, in order to obtain more information about the affection of clutch control while shifting, the dynamic model of powertrain from engine to vehicle loading is created in the Matlab Simulink program to perform a complete simulation from part control to vehicle dynamic.

INTRODUCTION

Automated Manual Transmission (AMT) is prevailing in recent years. Vehicles from high quality sports cars like BMW M3, Ferrari 355, etc., to general sedans like Opel Corsa, Renault Twingo, etc. have introduced such system. It uses auto clutch and shifting actuator to simulate Manual Transmission (MT) as an auto transmission system (Figure 1). Due to the low cost and high efficiency of manual transmission, AMT can either have a lower cost and higher efficiency comparing to traditional Automated Transmission (AT), which uses torque converter and planetary gear box for shifting.

Figure 1 Automated Manual Transmission (AMT)

However, the main disadvantage of AMT, which leads to a worse competitiveness comparing to AT, is comfort. AMT, like MT, needs a complete power interruption to provide a chance for synchronizer to actuate to shift gear ratio. Such interruption, not likely occur on AT, always causes uncomfortable accelerations and jerks. The uncomfortableness is mainly dominated by the control of clutch. It can be apparently experienced by a novice driver of MT car. To have a more comfortable shifting process, a faster clutch motion which can lead to a shorter shifting time and less power interruption and a more stable clutch control which can lead to a more comfortable re-acceleration while clutch is engaging are required. There are many strategies to deal with such problem, such as Double Clutch Transmission (DCT), [1], [2], etc., and torque tracking strategy [3]. Identically, both these methods require a fast and reliable auto clutch system. An approach to optimize a clutch actuator and its control according to the design requirements is proposed in this paper.

The main object of this paper is to optimize a clutch actuator and its controller to drive clutch faster and more stable. Besides, in order to completely understand the dynamic character while shifting, a modular simulation program which integrated engine, clutch, gear box, vehicle loading, and control of each component is created. Researchers can obtain information of vehicle dynamic response with different Throttle Position (TPS) control, clutch control, and gear ratio control, which can help to develop shifting strategy, clutch control strategies, and TPS control strategy.

Using Matlab Simulink, dynamic problems, control problems, and optimizations are integrated into a unit with modules to provide efficient computation, simultaneous analysis, and easy of modification. For dynamic module of the clutch system, a detailed dynamic model of the clutch actuator is created using free body analyses, and a black box model of the clutch is created according to experiments data and curve fitting methods. For control module, PID control strategy is used to control the clutch actuator. Ziegler-Nichols (Z-N) and Internal Model Control (IMC) control turning algorithms [4] are used at initial to design the controller parameters. With the combined model of clutch module, clutch actuator module, and

controller module, dimensions of some mechanical parts of the clutch actuator are optimized to perform a faster actuating speed, and control parameters are optimized to give more stable clutch travel control. The optimization result is practiced on a real model and is verified to have a faster shifting speed.

For the complete powertrain simulation program, engine module, gear box module, and vehicle loading module are also created by experiments data and free body analyses. Combined with the clutch actuator module and clutch module, shifting process simulations with different TPS control, clutch control, gear ration control, and driving condition are presented in the final part of this paper.

DYNAMIC MODEL

Vehicle dynamic models in this paper are mainly separated into two parts: clutch system and powertrain. Cultch system is the main objective of this paper which is to be optimized for faster actuating speed and more stable control. Dynamic model of the clutch system is to be created as detailed as possible by free body analysis. For other components of the powertrain, the models are mainly created by experiments data and simplified free body analysis, which also perform a well input-output relation. With the modular programming, such models can easily be modified to be more detailed if any of these components are to be studied more completely in the future.

For the assumption of the model creation, since rigid body assumption is well approximated in most cases on vehicle [5], all components, besides springs, are considered as rigid bodies for simplification.

CLUTCH SYSTEM

Clutch actuator

Structure of the clutch actuator is shown in Figure 2.

Figure 2 Structure of Clutch Actuator

The clutch actuator is mainly driven by a DC motor, through a couple of worm shaft and worm gear to magnify torque and serve as self-lock structure, to drive a linkage to move the clutch lever and to control the clutch position. A pre-deformed assist spring is assembled to assist clutch disengaging.

According to “law of motor” and KVL analysis [6], model of the DC motor is created. And by free body analysis of each part of the clutch actuator, the dynamic model of the clutch actuator is created. Combining the DC motor module and mechanical parts modules, clutch actuator module is created (Figure 3).

motor torque

worm fri coe

gear pin fri coe

actuator load

joint fri coe

actuator motion

actuator velocity

actuator acceleration

shaft velocity

clutch actuator

Motor Voltage

Motor w

Motor Torque

Motor Current

DC motor

Figure 3 Module of Clutch Actuator

Clutch

Structure of the clutch is shown in Figure 4, which is comprised by a clutch cover and a clutch disk.

Figure 4 Structure of Clutch [7]

Clutch cover is assembled on a fly wheel which is connected to engine shaft, and clutch disc is connected to input shaft of gear box. Torque from engine is transmitted from engine, through coupling of the clutch, to gear box. Actuating force from the clutch actuator, through release bearing, is exerted on the diaphragm spring of the clutch. Through the fulcrum ring, the diaphragm spring drives the pressure plat to couple/discouple with the clutch disc (Figure 5).

Figure 5 Assembly of Clutch and Clutch Actuator

The difficulty of modeling the clutch is diaphragm spring. The force-deformation relation of diaphragm spring is very nonlinear and not easy to be simulated or expressed theoretically. To be accurate, dynamic model of the clutch are created by experiment data.

Experiment result of diaphragm spring is shown in Figure 6 (real line).

Figure 6 Experiment Data of Diaphragm Spring

Using curve fitting method, relation between release bearing travel and release bearing load is obtained (Figure 6 dash line). The relation is the same between clutch actuator travel and clutch actuator loading.

608.302.35.1053726.194578.74)( 2

23

+−+−

=outout

outoutoutoutc DD

DDDDW

The cushion spring plays and important rule on clutch engage force, which is directly related to the clutch torque transmission ability. As shown in Figure 7 [7], region of cushion deflection dominates all region of engagement modulation zone. Thus, clutch engage force can be seen the same as cushion spring force and cushion spring deflection is the same as pressure plate travel. With experiment data of cushion spring deformation and cushion spring loading force, relation between pressure plate travel and clutch engage force is obtained. Combining with experiment data of pressure plate travel and release bearing travel, the relation between release bearing travel and clutch engage force is obtained.

Figure 7 Clutch Travel Description

For the relation between clutch engage force and release bearing travel, experiment results are shown in Figure 8 and Figure 9 (real line).

Figure 8 Experiment Result between Release Bearing Travel and Pressure Plate Travel

Fitted curve equation (dash line):

outoutoutoutout

outp

DDDDD

DD

004408.002388.002873.0005043.00002615.0

)(2345 +−+−

=

Figure 9 Experiment Result Cushion Spring

Fitted curve equation (dash line):

pppclu DDDN 5455819)( 2 −=

Combing these two equations to eliminate the parameter pD , relation between clutch engage force cluN and release bearing travel outD is obtained, where the release bearing travel is the same with clutch actuator travel.

Considering with relation between clutch engage force and friction force of the clutch, relation between clutch actuator travel and clutch torque transmission ability is obtained.

For the relation between input torque of clutch inT (torque from engine and applies on the clutch cover) and output

torque of clutch outT (torque from clutch disc and transmits to gear box), three cases is considered.

1. If clutchin TT > , then outin

outinclutchout TT

θθ

θθ&&

&&

−

−= .

2. If clutchin TT ≤ and outin θθ && ≠ , then outin

outinclutchout TT

θθ

θθ&&

&&

−

−= .

3. If clutchin TT ≤ and outin θθ && = , then outin TT = .

Where inθ& is rotation speed of clutch cover, and outθ& is rotation speed of clutch disk.

The clutch module is created as below:

Release Bearing Motion

Torque From Engine

Release Bearing Load

Torque to be Transfered

Torque avle to be Transfered

Clutch

Figure 10 Module of Clutch

POWERTRAIN

Dynamic model of powertrain includes engine, clutch, clutch actuator, gear box, transmission shaft, and vehicle loading, as shown in Figure 1.

Engine

The modeled engine is 1198 c.c. with 4 cylinders. Engine model is created according to experiment data and using “look-up table” [8] in Matlab Simulink. The engine model considers with Throttle Position (TPS), engine speed, and torque generated by engine. The engine map of the model is shown in Figure 11.

Figure 11 Engine Map

Gear box

The considered model is a 5 speed gear box used on Mitsubishi Veryca. The dynamic model is created by free body analysis considering with affections of mass and damping. And function of synchronizer is considered as a torque to synchronize the shifted gear. The magnitude of the torque is from experiments.

Vehicle loading

The vehicle loading module is created by vehicle loading components equation [9]. The components equation concerns affections of vehicle load, rolling resistance, aerodynamic force, and driving conditions of a car.

The components equation is in the form of:

loadcarhxAxxcarx WRDRaMF θsin++++=

Where carM is mass of the car, xa is longitudinal acceleration of the car, xR is rolling resistance forces on the four wheels, AD is aerodynamic drag force, hxR is hitch forces, loadθ is gradient angle of the road, xF is the force from drive wheels transmitted from engine, and carW is the force according to the gravity that acts on a car, which is known as cargM .

CONTROLLER MODEL

Controller model of the clutch actuator is designed using PID control strategy in order to have a stable control and good set point tracking performance.

PID CONTROL STRATEGY

The clutch actuator is controlled by two PID control plants. The first control plant controls the position of the clutch according to motor current. And the second control plant controls the motor current according to input voltage, the practiced input of the clutch actuator.

Figure 12 Controller Loops

The PID control equation is from general form, and modify with interacting modification, derivative limitation, and setpoint weighting [4].

))(1

1)1()(

1

1)1(()( tY

NsTsT

sTctY

NsTscT

sTbKsU

d

d

isp

d

d

i +

++−

+

++=

Where )(sU is the control plant output, )(tYsp is set point, )(tY is feedback of plant gain output, b and c are

setpoint weightings, N is the term for derivative limitation, and K , iT , and dT are the three PID control constants.

CONTROL PARAMETERS TURNING

Control parameters of the two control plants are determined by some control strategies at first to serve as initial values for control optimization. The initial values are important, because it can prevent many unfeasible local minimums which are prevailing in control optimization. For the parameter turning, it is more important to focus on set point tracking than on disturbance adjusting according to design requirements and system characters. Considering with the plant gains of each control plant, Z-N and IMC [10] are selected as the turning strategies.

The first control plant is adjusted by IMC method, since IMC is especially useful for setpoint tracking with simple plant gain [11]. And the second control plant is adjusted by Z-N method since the control gain is much more nonlinear and complicated.

OPTIMIZATION

Optimization is implemented in order to modify the clutch system to actuate more rapidly and steadily.

The design requirements for the clutch actuator on AMT are as quick as possible to shorten the shifting time and as smooth as possible to increase the comfort in order to lessen the feeling of shifting. The two objectives are conflicting, because fast drive of clutch always causes uncomfortable acceleration and jerk, and smooth drive of clutch always leads to longer shifting time. One way to compromise such contradiction is to disengage as quickly as possible and engage as smoothly as possible, which is the experience strategy for many MT drivers.

Thus, the main objects of the optimization are to shorten the disengage time and to increase the smoothness of engage. The smoothness is comprised of two parts: control strategy and control stability. This paper only concerns with control stability, which is the base of control strategy.

The two objectives are separated to two sub-problems. The disengage time is optimized by modification of mechanical parts and the stability is optimized by modification of control parameters, because the disengage time is mainly dominated by mechanism and stability is mainly dominate by controller according to exercise experiments.

OPTIMIZATION OF MECHANICAL PARTS

Cost function

Minimum disengage time of the clutch actuator.

Design variables

Considering with practical feasibility and optimization sensitivity, the following parts are chosen as design variables.

1. λ : lead angle of the worm gear. 2. spK : spring coefficient of the assist spring. 3. intL : pre-deformation of the assist spring. 4. R : clutch lever ratio. Constraints

The following are the optimization constraints.

1. ondTengage sec3.0≤ 2. °< 81.14λ

3. 0)22(max >−−× dwWdwWTlTlTlTl WtfWt

4. mmL 100int ≤ 5. 33.0 ≤≤ R

engageT is the engage time. The constraint ensures the clutch actuator being able to engage within the desired time while the disengage time is reduced. The constraint of lead angle restricts the worm gear and worm shaft to be able to self-lock. Such property reduces extra control at steady state. The third constraint constraints the directions of motor acceleration to be the same with directions of clutch command when maximum torque from electrical motor is applied. This constraint guarantees the clutch actuator for being drivable at any position with any speed. And the constraints of intL and R are used with consideration of space.

Implement

Using “fmincon” in Matlab optimization toolbox, the clutch actuator model is optimized as shown:

0 0.05 0.1 0.15 0.2 0.250

1

2

3

4

5

6

7

Time (sec.)

Clu

tch

Tra

vel (

mm

)

T=

0.08

9 se

c.

T=

0.23

4 se

c.

After OptimizationInitial Design

Fully Disengaged

Fully Engaged

Figure 13 Optimization Result of Clutch Actuator

The disengage time is reduced from 0.234 to 0.0896 second. The optimized design variables are shown below.

Initial Condition Optimization Result

Lead Angle λ 5˚ 11.14˚

Spring Coefficient spK 0.6 kg/mm 0.854 kg/mm

Pre-Deformation intL 40 mm 72.93 mm

Design Variables

Lever Ratio R 2 1.207

Cost Function 0.234 second 0.089 second

Table 1 Optimization Result Clutch Actuator

OPTIMIZATION OF CONTROL PARAMETERS

Cost function

The cost function of the control optimization is to increase control stability. The judgment of stability, also known as ability of setpoint tracking, is defined as Integrated Absolute Error (IAE) between setpoint and system output. A smaller IAE means A better control stability.

Cost function: min ∫ ∫=−=t t

sp dttedttytyIAE0 0

)()()(

where the setpoint )(ty is the general shifting command weighted steady state time as shown in Figure 14.

0 0.5 1 1.5 2 2.5−1

0

1

2

3

4

5

6

7

8

Time (sec.)

Setp

oint

of

Clu

tch

Posi

tion

(mm

)

Disengaging Time

0.8 sec.

0.8 sec.(weighted)

0.8 sec.

0.8 sec.(weighted)

Fully Disengaged

Fully Engaged

Figure 14 Setpoint Signal for Control Optimization

Design variables

Figure 15 shows simulation results of the second control plant, where sinusoid setpoint and impulse setpoint are implemented. It is obvious that setpoint tracking errors are small in both transient state and steady state. It also shows a good disturbance adjusting ability. In the scale view, the clutch actuator travels to the upper limit and suffered a collision, the controller can eliminate the interference quickly and steadily.

Figure 15 Control Plant 2 Simulation Results

Thus, only parameters of the first control plant are chosen as design variables, because the second control plant is already well set by IMC. Design variables, according to the PID control function, are: K , iT , dT , N , b , and c .

Constraints

Constrains of control optimization are set according to the commercial PID controller.

Design Variable K Ti Td N b c

Constraint 0<K<∞ 0<Ti<∞ 0<Td<∞ Integral 8<N<20

0 or 1

0 or 1

Table 2 Constraints of Control Optimization

Implement

Using “fmincon” and combining the controller model and dynamic model, the control parameters is optimized. The combination provides a more factual dynamic response than simplified quadratic or cubic model which is generally used in control optimization.

However, for such complicated plant gain with very nonlinearity, local minimum should be especially taken care. In order to prevent unsuitable local minimum solution, optimization is implemented with several different minimum step sizes of finite difference derivatives “DiffMinChange” [8].

The implementations are shown in Table 3 and Figure 16.

DiffMin Change K Ti Td N b c IAE

(1e-3)

1e-4 4.388 0.0082 0.001 20 1 1 23 5.6 1e-3 5.613 0.0185 0.115 10 1 1 137.2 1e-2 4.613 0.0185 0.117 18 1 1 136.8

1e-1 4.369 0.0447 0.086 20 1 1 115.5 Table 3 Optimization Implement Results

0 0.5 1 1.5 2 2.5−1

0

1

2

3

4

5

6

7

8

9

Time (sec.)

Clu

tch

Tra

vel (

mm

)

Diff.=1e−1Diff.=1e−2Diff.=1e−3Diff.=1e−4InitialSetpoint

Scale View

0.2 0.25 0.3 0.35 0.4 0.45

6

6.5

7

7.5

8

Time (sec.)

Clu

tch

Tra

vel (

mm

)

Diff.=1e−1Diff.=1e−2Diff.=1e−3Diff.=1e−4InitialSetpoint

Scale View

Figure 16 Optimization Implement Results

It is obvious that implementations with DiffMinChange bigger than 1e-3 has feasible setpoint tracking ability. The best result with DiffMinChange=0.1 is chosen as the optimization result.

The final optimization result is shown below:

0 0.5 1 1.5 2 2.5−1

0

1

2

3

4

5

6

7

8

9

Time (sec.)

Clu

tch

Tra

vel (

mm

)

SetpointInitialOptimization Result

Figure 17 Optimization Result

Design Variable K Ti Td N b c

IAE (1e-3)

Initial 4.740 0.0039 0.001 14 1 1 242.7Optimize

Result 4.369 0.0447 0.086 20 1 1 115.5

Table 4 Optimization Result

SHIFTING PROCESS SIMULATIONS

Modules of engine, clutch, gear box, vehicle loading, and control of TPS, clutch, and gear shifting are finally integrated to perform a complete shifting process simulation. The integrated model is shown in Figure 18.

Figure 18 AMT Simulation Program

The followings present some cases of simulations results. To verify the validity of the simulation program and observe the affections of each transmission control unit, the following cases are discussed:

1. General case 2. Clutch control 3. Gear ratio control 4. Vehicle loading 5. TPS control 1. GENERAL CASE

For a general shifting case of 1st-2nd-3rd, vehicle speed should increase gradually and vehicle acceleration should decrease with the increase of gear ratio. The shifting control command and simulation results are shown below, which indicate correct tendency according to general experiments. Note that cultch is fully disengaged at 7mm and TPS is fully opened at 10.

0 2 4 6 8 10

1.5

2

2.5

3

3.5

4

Gea

r R

atio

0 2 4 6 8 10

1.5

2

2.5

3

3.5

4

Gea

r R

atio

3rd Ratio

2nd Ratio

1st Ratio

0 2 4 6 8 10

0

1

2

3

4

5

6

7

8

9

10

Thr

ottle

Pos

ition

(T

PS)

Minimum TPS

Maximum TPS

0 2 4 6 8 10−1

0

1

2

3

4

5

6

7

8

Time (sec.)

Clu

tch

Posi

tion

(mm

)

Fully Disengaged

Fully Engaged

Figure 19 Shifting Control Command

0 2 4 6 8 100

10

20

30

40

50

60

70

Time (sec.)

Veh

icle

Spe

ed (

km/h

r)

Eng

agin

g

Dis

enga

ging

Dis

enga

ging

Eng

agin

g

Eng

agin

g

1st Ratio 3rd Ratio2nd Ratio

Figure 20 Simulation Result of Vehicle Speed

0 2 4 6 8 10−5

0

5

10

15

20

25

30

35

40

Time (sec.)

Veh

icle

Acc

eler

atio

n (k

m/h

r2 )

Fully

Eng

aged

Fully

Eng

aged

Fully

Eng

aged

Sync

hron

ize

Sync

hron

ize

1st Ratio 3rd Ratio2nd Ratio

Figure 21 Simulation Result of Vehicle Acceleration

2. CLUTCH CONTROL

For a clutch control with more gradual engagement (Figure 22), passengers always have a more comfortable feeling. The simulation results are shown in Figure 23 and Figure 24.

The simulation results verify with the experiences that a gradual clutch control can lead to a more gentle variation of speed and acceleration.

0 2 4 6 8 10−1

0

1

2

3

4

5

6

7

8

Time (sec)

Clu

tch

Posi

tion

Setpoint of Gratual Clutch ControlSetpoint of Swift Clutch Contrl

Figure 22 Clutch Travel Command

0 2 4 6 8 100

5

10

15

20

25

30

35

Time (sec.)

Veh

icle

Spe

ed (

km/h

r)

Gradual Clutch ControlSwift Clutch Control

Figure 23 Simulation Result of Vehicle Speed

0 2 4 6 8 100

5

10

15

20

25

30

Time (sec.)

Veh

icle

Acc

eler

atio

n (k

m/h

r2 )

Gradual Clutch ControlSwift Clutch Control

Figure 24 Simulation Result of Vehicle Acceleration

3. GEAR RATIO CONTROL

For this part two cases is discussed.

The first simulation case starts driving with second gear ration. The result is shown in Figure 25, Figure 26, and Figure 27. The results indicate the problem of starting with too large gear ratio, where the accelerating ability at initial is smaller and engine speed is lowered down to a speed which is possible to cause miss firing.

0 2 4 6 8 100

5

10

15

20

25

30

35

40

45

50

Time (sec.)

Veh

icle

Spe

ed (

km/h

r)

Start with 2nd RatioStart with 1st Ratio

Figure 25 Simulation Result of Vehicle Speed

0 2 4 6 8 100

5

10

15

20

25

30

Time (sec)

Veh

icle

Acc

eler

atio

n (k

m/h

r2 )

Start with 2nd RatioStart with 1st Ratio

Figure 26 Simulation Result of Vehicle Acceleration

0 2 4 6 8 100

1000

2000

3000

4000

5000

6000

7000

Time (sec.)

Eng

ine

Spee

d (r

.p.m

.)

Start with 2nd RatioStart with 1st Ratio

Fully Engaged

Fully Engaged

Figure 27 Simulation Result of Engine Speed

The second case is a down-shifting process. The shifting sequence is 1st-2nd-1st. The down-shifting movement always leads to engine brake, which decreases vehicle speed and increases engine speed. The shifting commands are the same with the general case besides the gear ratio command as shown in Figure 28. The simulation results are shown in Figure 29, Figure 30, and Figure 31.

0 2 4 6 8 102

2.2

2.4

2.6

2.8

3

3.2

3.4

3.6

3.8

4

Time (sec.)

Gea

r R

atio

1st Ratio 1st Ratio

2nd Ratio

Figure 28 Gear Ratio Command

0 2 4 6 8 100

10

20

30

40

50

60

70

Time (sec.)

Veh

icle

Spe

ed (

km/h

r)

Eng

agin

g

Dis

enga

ging

Dis

enga

ging

Eng

agin

g

Eng

agin

g

1st−2nd−3rd1st−2nd−1st

Figure 29 Simulation Result of Vehicle Speed

0 2 4 6 8 10−40

−30

−20

−10

0

10

20

30

40

Time (sec.)

Veh

icle

Acc

eler

atio

n (k

m/h

r2 )

Eng

agin

g

Eng

agin

g

1st−2nd−1st1sr−2nd−3rd

Figure 30 Simulation Result of Vehicle Acceleration

0 2 4 6 8 100

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

Time (sec.)

Eng

ine

Spee

d (r

.p.m

.)

Fully

Eng

aged

Eng

agin

g

Eng

agin

g

Fully

Eng

aged

Fully

Eng

aged

1sr Ratio 3rd Ratio2nd Ratio

Figure 31 Simulation Result of Engine Speed

4. VEHICLE LOADING

The case for starting with a mountain slope is important for AMT clutch control, since it is hard even for a MT vehicle driver.

Figure 32 and Figure 33 show the simulation results of starting with a mountain slope of 5° with the same clutch and TPS command as the general case. The car slides down and engine speed is dropped to a lower speed at initial. Such phenomena are valid with reality and show the difficulties of the control of clutch and TPS.

0 2 4 6 8 10−5

0

5

10

15

20

25

30

35

Time (sec.)

Veh

icle

Spe

ed

Mountain Slope=5Mountain Slope=0

Slide Down

Figure 32 Simulation Result of Vehicle Speed

0 2 4 6 8 100

1000

2000

3000

4000

5000

6000

7000

Time (sec.)

Eng

ine

Spee

d (r

.p.m

.)

Mountain Slope=0Mountain Slope=5

Fully Engaged

Fully Engaged

Figure 33 Simulation Result of Engine Speed

5. TPS CONTROL

Continuing with the above case of °= 5loadθ , this case shows the affection of different TPS control.

In this case the increase of TPS value is slowed down as shown in Figure 34. Besides, the clutch control is the same as the general case. Such control is a worse control for start driving with a mountain slope. The lower TPS value can lead to miss firing, because there is no enough power to resist the gravity.

The simulation result of engine speed indicates that the engine speed is dropped to negative speed, obviously causes miss firing.

0 2 4 6 8 100

1

2

3

4

5

6

7

8

9

10

11

Time (sec.)

TPS

Figure 34 TPS Control Command

0 2 4 6 8 10−1

0

1

2

3

4

5

6

7

8

Time (sec.)

Clu

tch

Posi

tion

Figure 35 Clutch Position Control

0 0.5 1 1.5 2 2.5 3−200

0

200

400

600

800

1000

1200

1400

Time (sec.)

Eng

ine

Spee

d (r

.p.m

.)

Miss Fire

Fully Engaged

Mountain Slope=5Slower TPS Control

Figure 36 Simulation Result of Engine Speed

CONCLUSION

An approach to create dynamic model of vehicle transmission, especially clutch and clutch actuator, and combining with control model to implement optimization is presented in this paper. The presented case of clutch actuator optimization is practiced on a real model and is verified to have a similar result of faster disengage time and more stable clutch control.

On the other hand, the created model of AMT vehicle transmission provides an essential base to understand the relation between dynamic response and control of clutch, TPS, and shifting strategy. Engineers can use the program to design and optimize the strategy of shifting control algorithm. Besides, with the modular creation of the program, it can easily be modified to simulate other systems such as double clutch transmission.

ACKNOWLEDGEMENTS

The support of this research by the National Science Council, Taiwan, R. O. C., under grant NSC 91-2212-E-

009-022 and Industrial Technology Research Institute, Taiwan, R. O. C., under grand C324K21CD0, is gratefully acknowledged.

REFERENCES

1. Richard, G., Reed, Jr., Carriere, D. L., “Electro-Mechanical Clutch Actuator,” United States Patent, Patent Number: 5996754, 1999.

2. Richard, G., Reed, Jr., Cherry, J. R., “Gear Preselect System for an Electro-Mechanical Automatic Transmission Having Dual Input Shafts”, United States Patent, Patent Number: 6286371, 2001.

3. Fischer, R., Berger, R., “Automation of Manual Transmission,” 6th LuK Symposium, pp. 95-121, 19-20 Mar. 1998.

4. Astron, K., Hagglund, T., “PID Controllers: Theory, Design, and Turning,” Instrument Society of America, 1995.

5. Genta, G., “Motor Vehicle Dynamics,” Word Scientific, 1997.

6. Ryff, P. F., “Electric Machinery,” Prentice Hall International, 1997.

7. Shaver, R., Chrysler Corporation, “Manual Transmission Clutch Systems,” Society of Automotive Engineers, 1997.

8. Matlab Version 6.5 R13, User Manual, MathWork, Inc. 9. Gillespie, T. D., “Fundamental of Vehicle Dynamics,”

Society of Automotive Engineers, 1992. 10. Pedret, C., Vilanova, R., Moreno, I., Serra, I., “A

Refinement Procedure for PID Controller Turning,” Computer and Chemical Engineering, Vol. 26, pp. 903-908, 2002.

11. Liang C.C., “Application of Taguchi Method for Parameter Turning of PID Controller” Master thesis, Department of Material Engineering, Chang Gung University, Tao-Yuan, Taiwan, 2002 (In Chinese).

CONTACT

1. Ching-Huan Tseng: Professor of department of Mechanical Engineering National Chiao Tung University (Taiwan) Hsichu, Taiwan 30050, R.O.C. Tel: 886-3-5717243 E-mail: [email protected]

2. Ming-Feng Hsieh: Graduate Student of Department of Mechanical Engineering National Chiao Tung University (Taiwan) Hsichu, Taiwan 30050, R.O.C. Tel: 886-3-5717243 E-mail: [email protected]