Adaptive Fuzzy PI of Double Wheeled Electric Vehicle Drive Controlled by Direct Torque Control

 

Brahim GASBAOUI1,*, Chaker ABDELKADER2, Adellah LAOUFI1,

and Boumedične ALLOUA1

1Bechar University, Faculty of Sciences and Technology, Department of Electrical Engineering, B.P 417 BECHAR (08000) Algeria.

2University of ENSET Oran, B.P 98 Oran (31000), Algeria.

E-mail: gasbaoui_2009@yahoo.com.

(* Corresponding author : E-mail: gasbaoui_2009@yahoo.com)

 

Received: 20 March 2010 / Accepted: 13 December 2010 / Published: 24 December 2010

 

 

Abstract

This paper presents a novel speed control design of electric vehicle (EV) to improve the comportment and stability under different road constraints condition. The control circuit using adaptive fuzzy PI controller is proposed. Parameters which guide the functioning of PI controller are dynamically adjusted with the assistance of fuzzy control. The EV is powered by two motors of 37 kilowatts each one, delivering a 476 Nm total torque. Its high torque (476Nm) is instantly available to ensure responsive acceleration performance in built-up areas. The electric drive canister of tow directing wheels and two rear propulsion wheels equipped with two induction motors thanks to their light weight simplicity and their height performance. Acceleration and steering are ensure by electronic differential, the latter control separately deriving wheels to turn at any curve. Electric vehicle are submitted different constraint of road using direct torque control. Electric vehicle are simulated in MATLAB SIMULINK. The simulation results have proved that the Adaptive fuzzy PI control method decreases the transient oscillations and assure efficiency comportment in all type of road constraints, straight, slope, descent and curved road compared to the conventional PI controller.

Keywords

Electric vehicle PI; Adaptive fuzzy PI; Fuzzy controller; Direct torque control

 

 

Introduction

 

Actually, electric vehicle (EV) including, full cell and hybrid vehicle have been developed very rapidly as a solution to energy and environmental problem.

Driven EV's are powered by electric motors through transmission and differential gears, while directly driven vehicles are propelled by in-wheel or, simply, wheel motors [1, 2]. The basic vehicle configurations of this research has two directly driven wheel motors installed and operated inside the driving wheels on a pure EV. These wheel motors can be controlled independently and have so quick and accurate response to the command that the vehicle chassis control or motion control becomes more stable and robust, compared to indirectly driven EV's. Like most research on the torque distribution control of wheel motor, wheel motors [3, 4] proposed a dynamic optimal tractive force distribution control for an EV driven by four wheel motors, thereby improving vehicle handling and stability [5, 6].

Direct torque control has become one of the most popular methods of control for induction motor drive systems [7-9]. DTC can decouple the interaction between flux and torque control, based on both torque and flux instantaneous errors, and provide good torque response in steady state and transient operation conditions. The main advantages of DTC are: absence of coordinate transformation and current regulator; absence of separate voltage modulation block; the actual flux-linkage vector position does not have to be determined, but only the sector where the flux-linkage vector is located, etc. In addition, DTC minimizes the use of machine parameters, so it is very little sensible to the parameters variation [3].

Research has shown that EV control methods such as, PI control are able to perform optimally over the full range of operation conditions and disturbances and it is very effective with constant vehicle torque, Moreover these non-linear vehicle torque are not fixed and change randomly. However EV with conventional PI control may not have satisfactory performance in such fast varying conditions, the system performance deteriorates. In addition to this, it is difficult to select suitable control parameters Kp and Ki in order to achieve satisfactory compensation results while maintaining the stability of EV traction, due to the highly complex, non-linear nature of controlled systems. These are two of the major drawbacks of the PI control. In order to overcome these difficulties, adaptive PI controller by fuzzy control has been applied both in stationary and under roads constraints, and is shown to improve the overall performance of EV.

 

 

Electric Traction System Elements Modeling

 

Figure 1 represent the general diagram of an electric traction system using an induction motor (IM) supplied by voltage inverter [5, 7].

Figure 1. Electrical traction chain

 

The Vehicle Load

The vehicle is considered as a load is characterized by many torqueses which are mostly considered as resistive torques [5, 6, 9, 10]. The different torques includes the vehicle inertia torque defined by the following relationship:

Tin = Jv(dwv/dt)

(1)

 

Aerodynamics Force

This part of the force is due to the friction of the vehicle body, moving through the air. It is q function of the frontal area shape protrusion such as side, mirrors, ducts and air passages spoilers, and any other factor the formula for this component is:

Faero = (ρSTxv2)/2

(2)

The aerodynamics torque is:

Taero = (ρSTxRrv2)/2

(3)

Rolling Force

The rolling resistance is primarily due to the traction of the tire on the rode. Friction in bearing and the gearing systems also play their part. The rolling resistance is appositely constant, depend on vehicle speed. It is proportional to vehicle weight. The equation is:

Ftire = Mgfr

(4)

The rolling torque is:

Ttire = MgfrRw

(5)

 

Hill Climbing Force

The force needed to drive the vehicle up a slope is the most straightforward to find. It is simply the component of the vehicle weight that acts along the slop. By simple resolution the force we see that [11]:

Fslope = Mgsin(β)

(6)

The slope torque is:

Tslope = Mgsin(β)Rw

(7)

       We obtain finally the total resistive torque (Figure 2):

Figure 2. The forces acting on a vehicle moving along a slope

 

 

Direct Torque Control (DTC)

 

The basic DTC strategy is developed in 1986 by Takahashi. It is based on the determination of instantaneous space vectors in each sampling period regarding desired flux and torque references. The block diagram of the original DTC strategy is shown in Figure 3. The reference speed is compared to the measured one. The obtained error is applied to the speed regulator PI whose output provides the reference torque.

The estimation value of flux and its phase angle is calculated in expression:

(8)

f = √f2 + f2

(9)

fs = artg(f / f)

(10)

and the torque is controlled by three-level Hysteresis. Its estimation value is calculated in Equation (11):

Tem = 3p(fi - fi)

(11)

Figure 3. Bloc Diagram of DTC for an EV Induction Motor

 

 

Conventional PI Controller

 

The reason behind the extensive use of proportional integral (PI) controller is its effectiveness in the control of steady-state error of a control system and also its easy implementation. However, one disadvantage of this conventional compensator is its inability to improve the transient response of the system. The conventional PI controller (Figure 4) has the form of Equation (12), where Tem is the control output, Kp and Ki are the proportional and integral gains respectively, these gains depend on the system parameters, ε is the error signal, which is the difference of the injected voltage to the reference voltage:

Tem(t) = Kpε(t) + Ki T ε(t)dt

(12)

Figure 4. Control of the injected speed using conventional PI controller

 

Equation (12) shows that the PI controller introduces a pole in the entire feedback system, consequently, making a change in its original root locus. Analytically the pole introduces a change in the control system’s response. The effect is the reduction of steady-state error. On the other hand, the constants Kp and Ki determine the stability and transient response of the system, in which, these constants rely on their universe of discourses:

 

where the values of the minimum and maximum proportional and integral constants (gains) are practically evaluated through experimentation and using some iterative techniques.

This makes the design of the conventional PI controller dependent on the knowledge of the expert. When the compensator constants exceed the allowable values, the control system may come into an unstable state. After the determination of the domain of the proportional and integral constants, the tuning of the instantaneous values of the constants takes place. Depending on the value of the error signal, ε, the values of the constants adjusts formulating an adaptive control system. The constants Kp and Kb changes to ensure that the steady-state error of the system is reduced to minimum if not zero.

 

 

Adaptive Fuzzy PI Controller

 

Fuzzy controllers have been widely applied to industrial process. Especially, fuzzy controllers are effective techniques when either the mathematical model of the system is nonlinear or no the mathematical model exists. In this paper, the fuzzy control system adjusts the parameter of the PI control by the fuzzy rule. Dependent on the state of the system, the adaptive PI realized is no more a linear regulator according to this principle. In most of these studies, the Fuzzy controller used to drive the PI is defined by the authors from a series of experiments [12-15]. The expression of the PI is given in the Equation (13):

(13)

where: y(t) - Output of the control; e(t) - Input of the control. The error of the reference current w*(t) and the injected speed w(t); Kp - Parameter of the scale; Ti - Parameter of the integrator.

The correspondent discrete equation is:

(14)

where: y(k) - Output on the time of kth sampling; e(k) - Error on the time of k sampling; T -Cycle of the sampling; Δe(k) = e(k) - e(k-1). Simple transformations applied to Equation 14 lead to:

,

 

 

 

On-line Tuning

 

The on-line tuning equation for kp and ki are show above:

kp = 20 + 0.8(Kp-2.5)

(15)

ki = 0.0125 + 0.003(Ki-2.5)

(16)

The frame of the fuzzy adaptive PI controller is illustrated in Figure 5.

Figure 5. PI gains online tuning by fuzzy logic controller

 

The linguistic variables are defines as {NL, NM, NS, Z, PS, PM, PB} meaning negative large, negative medium, negative small, zero, positive small, positive medium, positive big (tuning rules given in Table 1), and the Membership function is illustrated in the Figures 6-9. Using the settings given in Table 1 the fuzzy controllers were obtained and are given in Figures 10 and 11.

Figure 6. The Membership function of input e(k)

Figure 7. The Membership function of input Δe(k)

Figure 8. The Membership function of output kp

Figure 9. The Membership function of output ki

 

Table 1. Fuzzy tuning rules

kp and ki

 

           e(ω)

Δe(ω)

NL

NM

NS

ZE

PS

PM

PB

kp

N

L

M

S

M

S

M

L

Z

L

M

L

Z

L

M

L

P

L

M

L

Z

L

M

L

ki

N

Z

S

M

L

M

S

Z

Z

Z

S

M

L

M

S

Z

P

Z

M

L

L

L

M

Z

Figure 10. View plot surface of fuzzy controller for kp

 

Figure 11. View plot surface of fuzzy controller for ki

 

 

Implementation of Electronic Differential

 

The proposed control system principle could be summarized as follows:

¸        A speed control is used to control each motor torque;

¸        The speed of each rear wheel is controlled using speed difference feedback;

¸        Since the two rear wheels are directly driven by two separate motors, the speed of the outer wheel will need to be higher than the speed of the inner wheel during steering maneuvers (and vice-versa); this condition can be easily met if the speed estimator is used to sense the angular speed of the steering wheel;

¸        The common reference speed ωref is then set by the accelerator pedal command;

¸        The actual reference speed for the left drive ω*left and the right drive ω*right are then obtained by adjusting the common reference speed ω* using the output signal from the DTC speed estimator;

¸        If the vehicle is turning right, the left wheel speed is increased and the right wheel speed remains equal to the common reference speed ω*;

¸        If the vehicle is turning left, the right wheel speed is increased and the left wheel speed remains equal to the common reference speed ω* [16-18];

¸        Usually, a driving trajectory is adequate for an analysis of the vehicle system model. From the mode show in Figure 6, the following characteristic were calculated (where δ is the steering angle):

R = Lw/tg(δ)

(17)

¸        Therefore, the linear speed of each wheel drive is given by:

(18)

¸        The angular speed (where wv is the vehicle angular speed according to the center of turn):

(19)

(20)

¸        The difference between wheel drive angular speeds is:

(21)

¸        The steering angle indicates the trajectory direction:

(13)

The entire ensemble is given in the Figures 12-14.

Figure 12. Differential electronic

Figure 13. The driving wheels control system

 

Figure 14. Structure of vehicle in curve

 

 

Simulation Results

 

A. Case of Conventional PI Controller

In order to characterize the driving wheel system behavior, simulations were carried using the model of Figure 13. All simulation of electric vehicle is using the trajectory show in Figure 15. First application of DTC controller is with classical PI controller. The parameter of PI controllers are proportional and integral gains Kp=30 and Ki=200.

Figure 15. Different paths driving by electric vehicle

 

Electric vehicle has submitted a number of tests during the various routes:

¸     At 0 <t < 0.67s, the speed flows precisely the acceleration ramp (Figure 16.1); the electromagnetic torque decrease and settle around 222.80 Nm for both induction motors (Figure 16.4) and EV gives the same speed 70 km/h and driving wheels speeds stay always the same;

¸     At t = 2s, EV present in an acceleration and the speed increase from 70 km/h to 80 km/h (show in Figure 16.1), the electromagnetic torque and current growth sharply (Figures 16.2, 16.3 and 16.4);

¸     At 2 <t < 3s, right and left motor develop an electromagnetic torque around  262.40 Nm (Figure 16.4), and the variation of current for both motors are shown in Figures 16.2 and 16.3;

¸     At t = 3s, the vehicle is driving on a curved road on the right side; the road does not affect the control of the wheel and the DTC controller act immediately on the speed loop’s and rejects the disturbance and gives more and more efficiency to the electronic differential output references;

¸     At 3 < t < 4s, the EV are driving in straight road with constant speed, a good tracking of the speed step can be observed in Figure 16.1; EV reaches constant speed 80km/h; delay in reaching this speed follows from acceleration of the vehicle mass; in Figures 16.2 and 16.3, variation of phase currents for each motor is shown;

¸     At t = 4, the vehicle is driving on a curved road on the left side with 80km/h speed; the assumption is that the two motors are not disturbed; in this case the driving wheels follow different paths, and they turn in the same direction but with different speeds; the electronic differential acts on the two motor speeds by decreasing the speed of the driving wheel on the right side situated inside the curve, and on the other hand by increasing the wheel motor speed in the external side of the curve; the behavior of these speeds is given in Figure 16.4;

¸     At 4s < t < 5s, the EV are driving in straight road; call of current decreases and the EV torque jumps down to 127.73 Nm (Figure 16.4);

¸     At 5s < t < 6s, this test clarify the effect of the descent of vehicle moving on straight road; The presence of descent causes a great decrease in the phase current of each motor, as shown in Figures 16.2 and 16.3; the right and left motor are relieved and develops approximately 50.80 Nm; in the descent the slope torque become attractive torque; for that the consumption of energy decreases and the EV torque reach to 86.99 Nm;

¸     At 7s < t < 8s, this test explain the effect of the slope on the EV; speed of the driving wheels stay the same and the road drop does not influence the angular control of the wheels; the both motor develops more and more electromagnetic torque for vanquish the slop; the current increase speedily and still to its maximal value, and the speed maintain to 80 km/h; the variation of deriving force, are illustrate in Figure 16.5, and the motor absorbed more energy, the EV torque jumps back to 169 Nm;

¸     At 8s < t < 9s, EV are driving in straight road; referring to Figures 16.2, 16.3, 16.4 and 16.5 we show the variation of current, electromagnetic torques and deriving forces; the speed of the driving wheels stay the same  and speed vehicle settles at 80 km/h, and the EV torque increase to 127.73 Nm according to Figure 16.6.

Figure 16.1. Vehicle wheel speed

Figure 16.2. Phase current motor right

Figure 16.3. Phase current motor left.

Figure 16.4. Electromagnetic torques

Figure 16.5. Deriving forces

Figure 16.6. Aerodynamic torque, Rolling torque, slop torque, Vehicle torque

 

B. Case of Adaptive Fuzzy PI Controller

Secondly EV performance was examined with fuzzy adaptive PI controller. When the proposed method has been used, the vehicle speed, phase current, electromagnetic torques and driving forces waveform can be seen in Figures 16.7-16.11.

Figure 16.7. Vehicle wheel speed

Figure 16.8. Phase current motor right

Figure 16.9. Phase current motor left

Figure 16.10. Electromagnetic torques

Figure 16.11. Deriving  forces

 

 

Comparative Analysis of the Two Methods of Controlling

 

In simulations the two different methods to control the EV were used. Because of the sweeping of the kp on the interval [15..37] and the ki on the interval [228..243] as shown in the Figure 16.12 and 16.13. The DTC with Fuzzy Adaptive PI Control method improves EV performance. The adaptive fuzzy PI was proved in efficiency adaptation for variation of the speed in descent and slope result can be show in Figures 16.14 and 16.15.

Figure 16.12. Variation gain kp

of adaptive fuzzy PI controller

Figure 16.13. Variation gain kp

of adaptive fuzzy PI controller

Figure 16.14. Change of speed in descent and

slop using DTC  with adaptive fuzzy PI controller

Figure 16.15. Change of speed in descent and slop DTC

 using classical PI

Figure 16.16. Total harmonic distortion

using DTC with adaptive fuzzy PI controller

Figure 16.17. Total harmonic distortion

using DTC with classical PI

 

The results obtained by simulation show that this structure permits the realization of the robust control based on adaptive fuzzy inference system, with good dynamic and static performances for the multi-converters/multi-machines propelled system. Referring to figures 16.16 and 16.17 we show harmonic analyses of stator current. DTC with adaptive fuzzy PI controller present 29.96%, DTC with classical PI controller give 33.82%. The first controller offers a reduction of 13.84%. This remarkable change obtained enables us to say that the current inject by voltage source inverter in DTC classical PI controller is harmonics current polluting what to justify the great oscillations of the torque and the attraction force .as a consequence this ripple present negative effects on the autonomy of the battery and heating of the both motors and increase power losses.

 

 

Conclusions

 

The research outlined in this paper has demonstrated the feasibility of an improved vehicle stability which utilizes two independent back drive wheels for motion by using DTC controls. DTC with adaptive Fuzzy control is able to adapt itself the suitable control parameters which are the proportional and integral gains kp and ki to the variations of vehicle torque. This method was Improved EV steering and stability during different trajectory this. The advantage DTC controller is robustness and performance, there capacity to maintain ideal trajectories for two wheels control independently and ensure good disturbances rejections with no overshoot and stability of vehicle perfected ensured with the speed variation and less error speed. The DTC with adaptive fuzzy PI controller is more adaptive for propelled systems. The electric vehicle was proved best comportment and stability during different road path by maintaining the motorization error speed equal zeros and gives a good distribution for deriving forces. The electric vehicle was proved efficiency comportment in the different road constraints.

 

 

References

 

1.      Yang Y. P., Lo C. P., Current Distribution Control of Dual Directly Driven Wheel Motors for Electric Vehicles, Control Engineering Practice, 2008, 16(11), p. 1285-1292.

2.      He P., Hori Y., Kamachi M., Walters K., Yoshida H., Future Motion Control to be Realized by In-wheel Motored Electric Vehicle, In Proceedings of the 31st Annual Conference of the IEE Industrial Electronics Society, IEEE Press, Raliegh South Carolina, USA, 2005, p. 2632-2637.

3.      Kang K.-J., Sul S.-K., New Direct Torque Control of Induction Motor for Minimum Torque Ripple and Constant Switching Frequency, IEEE Trans Ind Applicat, 1999, 35(5), p. 1076-1082.

4.      Larminie J., Lowry J., Electric Vehicle Technology Explained, Edited by John Wiley and Sons, England, 2003.

5.      Chan C. C., Electric vehicles charge forward, IEEE Power Energy Mag, 2004, 2(6), p. 24-33.

6.      Zhu Z. Q., Howe D., Zhu Z., Electrical machines and drives for electric, hybrid, and fuel cell vehicles, Proc. IEEE, 2007, 95(4), p. 764-765.

7.      Chen Jr L., Chen K.-L. F., A Novel Direct Torque Control for Dual-Three-Phase Induction Motor, Conf. Rec. IEEE International Conference on Machine Learning and Cybernetics, 2003, 2, p. 876-881.

8.      Hartani K., Bourahla M., Miloud Y., Sekour M., Electronic Differential with Direct Torque Fuzzy Control for Vehicle Propulsion System, Turk J Elec Eng & Comp Sci, 2009, 17(1), p. 21-38.

9.      Vasudevan M., Arumugam R., New direct torque control scheme of induction motor for electric vehicles, 5th Asian Control Conference, 2004, 2, p. 1377-1383.

10.  Haddoun A., Benbouzid M.E.H., Diallo D., Abdessemed R., Ghouili J., Srairi K., Analysis, modeling and neural network traction control of an electric vehicle without differential gears, in Proc. IEEE IEMDC, Antalya, Turkey, May 2007, p. 854-859.

11.  Benbouzid M. E. H, Diallo D., Zeraoulia M., Advanced fault-tolerant control of induction motor drives for EV/HEV traction applications, IEEE Trans. Veh. Technol., 2007, 56(2), p. 519-528.

12.  Gao M., He S., Self-adapting Fuzzy-PID Control of Variable Universe in the Non-linear System, International Conference on Intelligent Computation Technology and Automation, ICICTA 2008, 1,p.473-478.

13.  Chen J.-Y., Tsai P.-S., Wong C.-C., Adaptive design of a fuzzy cerebellar model arithmetic controller neural network, IEE Proc.-Control Theory and Applications, 2005, 152(2), p. 133-137.

14.  Lin C.-M., Peng Y.-F., Adaptive CMAC-Based Supervisory Control for Uncertain Nonlinear Systems, IEEE Transactions on systems, man, and cybernetics-part b: cybernetics, 2004, 34(2), p. 1248-1260.

15.  Han H., Su C.-Y., Stepanenko Y., Adaptive control of a class of nonlinear systems with nonlinearly parameterized fuzzy approximators, IEEE transactions on fuzzy systems, 2001, 9(2), p. 3427-3431.

16.  Itoh K., Kubota H., Thrust ripple reduction of linear induction motor with direct torque control, Proceedings of the Eighth International Conference on Electrical Machines and Systems, ICEMS 2005, 1, p. 655-658.

17.  Emadi A., Handbook of automotive power electronics and motor drives, Taylor & Francis, 2005.

18.  Haddoun A., Benbouzid M., Diallo D., Abdesseme R., Ghouili J., Srairi K., Modeling Analysis and neural network control of an EV Electrical Differential, Transaction on Industrial Electronics, 2008, 55(6), p. 2286-2294.

 

 

Annex: Simulation Data

 

Table 2. Comparative between PI and adaptive fuzzy PI

 

Designation

PI controller

Adaptive fuzzy PI

THD [%]

33.82

29.96

Comportment  of speed in descent and slop

Less adaptive

More adaptive

Driving forces and electromagnetic torque

More oscillation

Less  oscillation

 

Table 3. Vehicle Parameters

Te

Motor traction torque

238 Nm

Je

Moment on inertia of the drive train

7.07 Kgm2

Rw

Wheel radius

0.32 m

a

Total gear ratio

10.0

η

Total transmission efficiency

0.93

M

Vehicle mass

1300 Kg

fe

Bearing friction coefficient

0.32

Kd

Aerodynamic coefficient

0.32

A

Vehicle frontal area

2.60 m2

fv

Vehicle friction coefficient

0.01

α

Grade angle of the road

Rad

Lw

Distance between two wheels and axes

2.5 m

dw

Distance between the back and the front wheel

1.5 m

 

 

Table 4. Induction Motors Parameters

Rr

Rotor winding resistance (per phase)

0.0503 Ω

Rs

Stator winding resistance (per phase)

0.08233 Ω

Ls

Stator leakage inductance (per pohase)

724 μH

Lm

Magnetizing inductance (per phase)

0.02711 H

Lr

Rotor leakage inductance (per phase)

724 μH

fc

Friction coefficient

0.02791

P

Number of poles

4

 

Based speed

1484 rpm

 

Rated power

50 hp

 

Table 5. Symbols, Designation and Units

Symbols

Nomenclature

units

P

Pole pairs

 

J

Rotor inertia

Kg.m2

Je

Moment of inertia of the drive train

Kg.m2

Jv

Vehicle inertia

Kg.m2

Tem

Electromagnetic torque

Nm

Tv

Vehicle torque

Nm

Tslope

Slope torque

Nm

Taero

Aerodynamique torque

Nm

Ttire

Tire torque

Nm

in

Inertia vehicle torque

Nm

Nred

Report of speed gear

%

G

Gear box

 

η

Transmission efficiency

%

L

Distance between two wheels

m

d

Distance between the back and the front wheel

m

R

Curve radius

m

ρ

Air density

 

S

Frontal vehicle surface

m2

Cx

Aerodynamic drag coefficient

 

M

Vehicle mass

Kg

g

Gravitational acceleration

N/m

α

Angle grade of road

rad

fr

Wheels Rolling resistance coefficient

 

V dc

Battery voltage

Volt

∆ωV

Angular speed variation given by electronic differential

Rad/sec

ωright

Right wheel angular speed

Rad/sec

ωleft

Left wheel angular speed

Rad/sec

ωref-right

Right wheel angular speed of reference

Rad/sec

ωref-left

Left wheel angular speed of reference

rad

δ

Reel angle wheel curve’s

rad

β

Vehicle slip angle

rad