Anda di halaman 1dari 6

ANew MTTE Methodology for Electric

Vehicle Traction Control


Jia-Sheng Hu*, Dejun Yin**, Yoichi Hori**, Feng-Rung Hu***
* Department of Mechanical Engineering, National Cheng Kung University, Tainan, Taiwan
** Institute of Industrial Science, University of Tokyo, Tokyo, Japan
*** Department of Mathematical Education, National Taichung University, Taichung, Taiwan

Abstract--A new traction control which utilizes the vehicle stability and handling performance in critical
maximum transmissible torque estimation (MTTE) scheme dynamic situations. For example, the antilock braking
to execute the anti-slip control of electric vehicles is system (ABS) [5]-[6], electronic differential (ED) [7]-[8],
proposed in this study. Since the function of embedded direct yaw-posture control (DYC) [9]-[10], traction
knowledge mechanism, the chassis velocity and information
control (TC), and so on. Traction control is often of
about tire-road conditions is unnecessary. A closed-loop
observer with disturbance estimation performance is
interested in the performance of anti-slip control. When
employed to enhance the steering stability of MTTE the vehicle is driven or brakes on a slippery road,
approach. The proposed scheme which contains the closed- especially for light vehicles, traction control must not
loop friction torque estimator is not required to fulfill any only guarantee the effectiveness of the torque output to
differentiator. Moreover, the inversion of the controlled maintain vehicle stability, but also provide some
plant is also unnecessary. Evaluation examples are given to information about tire-road conditions to other vehicle
illustrate the performance and feasibility of the presented control systems. Moreover, a well-managed traction
anti-slip strategy. control system can cover the functions of ABS, because
motors can generate deceleration torque as easily as
Index Terms-- Traction control, anti-slip, electric vehicles. acceleration torque [11]. However, for practice, the actual
vehicle systems face challenges of restricting the
1. INTRODUCTION
development on traction control. For example, the real
From the beginning of the 21st century, industrial chassis velocity is not available, and the friction force
engineering has entered a period of rapid expansion and which drives the vehicle is immeasurable. In general,
development. The earth's limited petrochemical resources traction control systems need the chassis velocity due to
have been overused and consequently the price of oil has physical and economic reasons, the non-driven wheels
risen in recent years. Apparently, the worst consequences are utilized to provide an approximate vehicle velocity.
of overusing petrochemical energy are the air pollution However, this method is not applicable when the vehicle
problems and the greenhouse effect, and they both are is accelerated by 4WD systems or decelerated by brakes
becoming increasingly serious. In order to reduce our equipped in these wheels. For this reason, the
dependence on petrochemical resources and decrease the accelerometer measurement is also used to calculate the
deterioration of the earth's environment, the United velocity value, but it cannot avoid offset and error
Nations (UN) passed the "Kyoto Protocol" [1] in problems. Other sensors, e.g., optical sensors [12],
December of 1997. Basically, the Kyoto Protocol sensors of magnetic markers [13], etc., can also obtain the
progressively restricts the emissions of carbon dioxide chassis velocity. However, they are too sensitive and
(C02) in every industrialized country. Consequently, this reliant on the driving environment or too expensive to be
protocol has resulted in a massive impact on vehicle applied in actual vehicles. Some anti-slip control systems
engineering. The fully electric vehicles (EV) powered by [5]-[6], [14] try to realize optimal slip ratio controls
batteries can achieve quieter and pollution-free operation. according to the Magic Formula [15]. These systems not
Unlike the internal combustion engine vehicles, the only need extra sensors for the acquisition of chassis
electric vehicles using independently equipped motors to velocity or acceleration, but also more difficult to realize
drive each wheel. The individual output torque of the than expected. It is because of the tuned algorithms and
motor can be obtained from the motor current. This merit parameters for specific tire-road conditions cannot be
makes it easy to estimate the driving or braking forces adapted quickly to the significant variation in the
between tire and road surface in real time, which instantaneous, immeasurable relationship between slip
contributes a great deal to application of new control ratio and friction coefficient in different driving
strategies based on road condition estimation. The conditions. In order to overcome these problems, the
independently equipped motors provide higher Model Following Control (MFC) approaches
power/weight density, higher redundancy for safety and [12][13][16], do not need information on chassis velocity
better dynamic performance [2]-[4]. or even acceleration sensors is proposed. In these systems,
For advanced vehicles, many technologies embedded the controllers only make use of torque and wheel
in the micro controller unit (MCU), which enhances the rotation as input variables for calculation. Fewer sensors
contribute not only to lower cost, but also higher
This work was supported by Yen Tjing Ling Industrial Research reliability and greater independence from driving
Institute of Taiwan, R.O.C., under grant project 97PD03.
conditions, which are the most outstanding merits of this
F,lr- -+<:)--.J
class of control systems. Nevertheless, these control
designs based on compensation have to consider the
worst stability case to decide the compensation gain,
which impairs the performance of anti-slip control.
Furthermore, gain tuning for some specific tire-road
conditions also limits the practicability of this method.
Recently , the Maximum Transmissible Torque
Estimation (MTTE) approach [17] that requires neither
chassis velocity nor information about tire-road
conditions further upgrades the anti-slip performance of T- -+<:>--+I
electric vehicles. In this system, the use is only made for
the torque reference and the wheel rotation to estimate the Fig. 2. One wheel vehicle model with magic formula .
maximum transmissible torque to the road surface , then TABLE!
the estimated torque is applied for anti-slip control PARAM ETER LI ST

implementation. This approach also shows its benefits to Symbol Definition


a vehicle mass perturbed operation condition. Since a s; Wheel Inert ia
human being is involved in the operation of a vehicle, the Vw Wheel Velocity (Circumferential Velocity)
total mass potentially varies with different driver and
w Wheel Rotation
passengers. However, the MTTE approach is sensitive to
the wheel inertia. If the tires are getting flat or wheel T Driving Torque
inertia measurement is inaccurate, the anti-slip r Wheel Radius
performance ofMTTE will deteriorate gradually. Fd Friction Force (Driving Force)
This paper aims to make use of the advantages of M Vehicle Mass
electric vehicles to develop a new traction control
N Vehicle Weight
approach embedded the MTTE control scheme with
robust performance and is structured as follows . Section V Chassis Velocity (Vehicle Velocity)
II describes the MTTE approach for anti-slip control. r; Driving Resistance
Section III describes the closed-loop disturbance A Slip Ratio
estimation in general case. Details of the proposed robust
f.l Friction Coe fficient
MTTE approach are presented in Section IV. Section V
gives illustrated examples for evaluating the presented The MTTE approach for vehicle anti-slip control IS first
anti-slip strategy. Finally, Section VI offers some proposed in our pervious study [17]. Based on our
concluding remarks. previous MTTE approach, the following considerations
are concerned.
II. MTTE ApPROACH I) Whatever kind of tire-road condition the vehicle is
driven on, the kinematic relationship between the
Consider a longitudinal motion of a four-wheeled wheel and the chassis is always fixed and known.
vehicle, as depicted in Fig. 1, the dynamic differential 2) During the acceleration phase, considering stability
equations for the longitudinal motion of the vehicle can and tire abrasion, well-managed control of the
be described as velocity difference between wheel and chassis is
J ww = T-rFd (1) more important than the mere pursuit of absolute
maximum acceleration.
MV = Fd -F; (2) 3) If the wheel and the chassis accelerations are well
controlled, the difference between the wheel and the
Vw = rOJ (3) chassis velocities, i.e. the slip is also well controlled.
Fd(J.) = JiN From (1) and (3), the driving force, i.e. the friction
(4)
force between the tire and the road surface, can be
where the parameters' definition is listed in Table 1. calculated as (5).
Generally, the nonlinear interrelationships between the
slip ratio J. and friction coefficient Ji formed by tire's
F
d
.t. J wVw 2
(5)
r r
dynamics can be modelled by the widely adopted Magic
Under common road conditions, Fd is less than the
Formula [15] as shown in Fig. 2.
maximum friction force from the road and increases as T
goes up. However, when slip occurs , Fd cannot increase
by T. Hence when slip occurs , the difference between the
velocities of the wheel and the chassis becomes larger
and larger, i.e. the acceleration of the wheel is larger than
that of the chassis . Additionally, considering the Ji- J.
Fig. !. Dynami c longitudinal model of vehicl e.
relation described in the Magic Formula, an appropriate
difference between chassis velocity and wheel velocity is r
necessary to support the desired friction force. In this J s
paper, a defined as (6)
V· . (Fd - Fdr ) / M
a = V; ' r.e, a = (T -rFd)r/Jw
max
(6)

is a relaxation factor for smoothing the control system. In


order to satisfy the condition that slip does not occur or
become larger, a should be close to I . With a designed
a , when the vehicle enters a slippery road, Tmax must be
reduced adaptively according to the decrease of F d• If the
friction force Fd is estimable, the maximum
transmissible torque, Tmax can be formulated as
Fig. 4. Simplified MTTE control scheme.
i: =(~+I)rFd
aMr
(7) For practical convenience, two low pass filter (LPF) with
a time constant of Tj and T2 are introduced to smooth the
This formula indicates that a given estimated friction noises of digital signals and differentiator in which
force Fd allows a certain maximum torque output from follows. Note that the bandwidth of LPF is often designed
the wheel so as not to increase the slip. Hence , the to be double or higher than the system's bandwidth.
proposed control scheme utilizes Tmax to construct and Hence in motion control analysis, the LPFs can be
constrain the torque reference. Note that from (2) , it is ignored. Fig. 4 shows a simplified linear model of MTTE
clear that the driving resistance Fdr can be regarded as scheme where M n denotes the nominal value of vehicle
one of the perturbation source of the vehicle mass M . mass M and ~As) stands for the perturbation caused
Although the vehicle mass, M can also be estimated
by passenger and driving resistance Fdr • Here from Fig. 4,
online [18]-[21], in this paper, it is assumed to be a
constant. we have
Fig. 3 shows the main control scheme of MTTE. As
shown in Fig. 3, a limiter with a variable saturation value (8)
is expected to realize the control of torque output
according to the dynamic situation. Note that , if
The estimated disturbance force Fd is driven from the aMnr» J ; (9)
model inversion of controlled plant and command torque It is convinced that the condition of (9) is satisfied in
T . Consequently, a differentiator is needed. Under most of the commercial vehicles. Then
normal conditions, the torque reference is expected to
pass through the controller without any effect. T
~~-r (I 0)
Conversely, when on a slippery road, the controller can Fd
constrain the torque output to be close to Tmax • Based on Now consider the mass perturbation of ~ . From (8), it
(7), an open-loop friction force estimator is employed yields
based on the linear nominal model of wheeled motor to
produce the maximum transmissible torque.
i:F =-
(a(Mn+~M)r
J w
+r
)
(II)
i: d
Wheeled motor with tire
Obviously, from (II), the anti-slip performance of MTTE
will be enhanced when ~M is a positive value and
reduced when ~M is a negative value. Additionally, in
common vehicles, the MTTE approach is insensitive to
the varying of M n • Since passenger and driving
resistance are the primary perturbations of M; , the
MTTE approach reveals its merits for general driving
environments. The fact shows that the MTTE control
scheme is robust to the varying of the vehicle mass M .

III. DISTURBANCE E STIMATION


Fig. 5 shows the structure of open-loop disturbance
estimation. As can be seen in this figure, we can obtain
Fig. 3. Control system ofMTTE.
t;

T' y

i,
Fig. 5. Disturbance estimation based on open-loop observer.

Fig. 7 Proposed MTTE control system .


T' y
ob ser ver deteriorates the steering stability and causes some
abnormal faults in deriving. Here, let ~ s(s) denote the
1- - - - ; - - - :(~: - -:- -- - --,---------- -- -- slip perturbation on the wheeled motor. The uncertain
dynamics of ~s(s) represents the role of different
slippery driving situations. When ~s(s) '" 0, it means the
l . driving condition is fine. In contrast, for a slippery
Fig. 6 Disturbance estimation based on closed-loop observer. surface of road, the ~s(s)"* o. As discussed in Section
yG(s r J -T' =Td +(T' +Td)~(S)G(srl. (12) III, the open-loop disturbance observer has the following
drawbacks.
If ~(s)"* 0 , then t"*
Td • Without the adjustment I) Open-loop disturbance observer does not have a
mechanism, the estimation accuracy increases based on feedback mechanism to compensate the modeling
the deterioration of modeling error. errors. Therefore its robustness is often not good.
Fig. 6 shows the structure of closed-loop disturbance 2) Open-loop disturbance observer utilizes the
estimation. As can be seen in this figure, we can obtain inversion of controlled plant to acquire the
t; = C(s)~(s) T' + C(s) G(s) + C(s)~(s) T
d
• (13) estimation information of disturbance. However,
sometimes the inversion is not easy to be taken on.
1+C(s)G(s) 1+C(s)G(s)
Due to the closed-loop dynamics of PI compensator, the
If ~(s) = 0 , (13) becomes a low pass dynamics as
estimated disturbance torque i, in proposed system can
i; = C(s)G(s) Td • Obviously, the compensator in the enhance the performance of MTTE against the skid. It
1+C(s)G(s) also offers better robustness to the parameter varying.
closed-loop structure minimizes the modeling error Unlike the conventional MTTE approach, the proposed
caused by ~(s) in a short time. Consequently, the approach does not need to use the differentiator. The
compensator enhances the robust estimation performance proposed MTTE approach employs a closed-loop
observer to counteract the effects of disturbance. Hence it
against modeling error. Since the modeling error is
is sensitive to the phase of the estimated disturbance.
unpredictable, the disturbance estimation based on
Note that the delay element e- l.s is setup for
closed-loop observer is preferred.
compensating the digital delay of fully digital power
electronics driver.
IV. PROPOSED ANTI-SUP CONTROL STRATEGY
Fig. 7 shows the proposed MTTE control system V. E XAMPLES AND DISCUSSIONS
which embedded a closed-loop compensator into Fig. 7
with a friction detection unit. The disturbance torque Td In order to implement and evaluate the proposed
control system, a commercial electric vehicle, COMS3,
comes from the operation friction. When the vehicle is
which is assembled by TOYOTA Auto Body Co. Ltd.,
operated on a slippery road, it will cause the Td to shown in Fig. 8 was modified to carry out the
become very small due to the insufficient friction of the experiments ' requirements. As illustrated in Fig. 9, a
tires. Skid often happens in braking and racing of an control computer is embedded to take the place of the
operated vehicle when the tire 's adhesion cannot firmly previous Electronic Control Unit (ECU) to operate the
grip the surface of road. This phenomenon is often motion control. The corresponding calculated torque
referred as the magic formula. However, the f.1- A. reference of the left and the right rear wheel are
relation is immeasurable in real time. Therefore, in the independently sent to the inverter by two analog signal
proposed approach, the nonlinear behaviour between tire lines. Table 2 lists the main specifications.
and road is regarded as an uncertainty source which
compensator as the kernel of disturbance estimation. This
is also referred as the PI-observer [22]. Set the PI gains of
the PI compensator as K p = 350 and K; = 180 . As
shown in Fig. 10, the slippery road was set by an acrylic
sheet with a length of 1.2 m and lubricated with water.
As mentioned in Section IV, the command delay in the
proposed approach is exploited to adjust the phase of the
estimated disturbance. Fig. 11 shows the measured
results. , and obviously, the estimated disturbance
oscillates due to the digital delay of motor driver.
Fig. 8. Experimental electric vehicle: COMS3.
TABL E II
According to Fig. 11, with proper command delay of
SPECIFICATIONOF COMS3 29ms, the proposed approach can achieve a feasible
Total Weight 360kg
performance. Proceeding forward, the proposed approach
can compensate the reference torque to a limited value
Maximum Power/per wheel 2000W
when a slippery road has encountered. Based on the
Maximum Torque/per wheel 100Nm experiment al result of Fig. 12, the reference torqueses of
Wheel Inertia/per wheel O.Skgm' MTTE-based approaches are constrained without
Wheel Radius 0.22m divergence. Fig. 13 shows the performance evaluation of
Sampling Time O.Ols
the proposed approach against different wheel inertia
Pentium M1.8G , 1GB RAM
values. It is evident that the proposed MTTE approach
Controller has consistent performance to the varying of wheel inertia
using Linux
NO and O/A 12 bits J w • As shown in these figures, the proposed anti-slip
Shaft Encoder 36 pulses /round system offers an effective performance to maintain the
r--
"">'"
g
- driving stability of more common situations such that the
steering safety of the electric vehicles can be further
i a' tIl
~

1;'
"- ~
Q
~ 8" "r enhanced.
~
.fr
8: .;
'-r + --r VI. CONCL USIONS
! ! ! This paper has presented a new anti-slip control
approach which embedded the strategies of MTTE with
- feasible performance in electric vehicles. A closed-loop
'" () <
() 0 " compensator with disturbance estimation ability has been
c [ ~
- employed in the proposed scheme . Note that the proposed
- control system has benefits such as preventing possible
failures on a slippery driving situation. The experimental
results have substantiated that the proposed MTTE
strategy has the ability, in practice, to improve the EV's
safety.
-::; z
- 5 ;;
~
.3

=I
r-
-
Fig. 10. Setting of slippery road for exper iment.
W hee l veloc ity and chassis velocity
1 0 ,---~----r---r========]l
---- - No delay
-- --- 1Oms delay
..........20ms delay
30 ms delay
- v
..... J
4 4 .5

Fig. 9. Schematic of electrical system of COMS3. 100 '


----- No delay
In the following experiment, the relation factor is set as E
z ----- 1 Oms dela y
a = 0.9. The time constants of LPFs in the comparison
Qi" .......... 20 ms dela y
::> 50
e-
MTTE experiment are set as ,) = = 0.05 . As '2 o
f-
0
-
30ms dela y
Reference torque

mentioned in Section II, the passenger's weight variable 2.5 3 3.5 4 4 .5


is regarded as the model uncertainty in the problem Tim e(sec)

formulation. In addition, this paper adopts the PI Fig. II. Experimental results to different delay time L.
W heel velocity and chassis velocity mobile robot," Robotics and Autonomous Systems, vol. 17,
pp. 65-80, 1996.
[9] F. Tahami, S. Farhangi, and R. Kazemi, "A fuzzy logic
direct yaw-moment control system for all-wheel-drive
electric vehicles," Vehicle System Dynamics, vol. 41,
pp.203-221,2004.
3.5
I

4 4.5
[10] T. Mizushima, P. Raksincharoensak, and M. Nagai,
Time(sec) "Direct Yaw-moment Control Adapted to Driver Behavior
Reference torque and output torque Recognition," in Proceedings ofSICE-ICASE International
100
\o;~\ '!~ l.\ \ r/ ~ : Joint Conference 2006, Busan, Korea, 2006, pp. 534-539.
.,
I
E \,"\". ) .1' ~ \ Jj I ~ NO contro l(Reference) [II] N. Mutoh, Y. Hayano, 1-1. Yahagi, and K. Takita, "Electric
--
I
z 50 - - - - \ ;,:/;:1 1- \~,.t.,r -1- _·_·-·- MTI E
:J braking control methods for electric vehicles with
e-0 O
_ _ _ _ _ _ _ t., _ _ _ _ : ----- Proposed approach
independently driven front and rear wheels," IEEE
f- I I
I I Transaction on Industrial Electronics, vol. 54, pp.1168-
-5~ ';o
.5 -----!;----~----"--_---l
3 3.5 4 4 .5 1176,2007.
Time(sec)
[12] T. Saito, H. Fujimoto, and T. Noguchi, "Yaw-moment
Fig. 12. Practical comparisons between the MTTE and proposed stabilization control of small electric vehicle," in
approach. Proceedings of the Technical Meeting on Industrial
Wheel velocity and chassis veloc ity Instrumentation and Control, lEE Japan, Tokyo, Japan,
2002, pp.83-88.
[13] H. Fujimoto, T. Saito, and T. Noguchi, "Motion
stabilization control of electric vehicle under snowy
conditions based on yaw-moment observer," in
Proceedings of IEEE International Workshop on Advanced
Motion Control, Kawasaki, Japan, 2004, pp.35-40.
[14] K. Fuj ii and 1-1. Fuj imoto, "Traction control based on slip
1
ratio estimation without detecting vehicle speed for electric
100
vehicle," in Proceedings of 4th Power Conversion
.,E
z
:J
80
60
-- . -- Jw=O.2
----- J w=0.3
----- Jw=OA
Conference, Nagoya, Japan, 2007, pp.688-693.
[15] H. B. Pacejka and E. Bakker, "The magic formula tire
e-o J w=0.5
f- 40
~ R ef e re nce torque
model," Vehicle System Dynam ics, vol. 21, pp.I-18, 1992.
[16] S. Sakai and Y. Hori, "Advantage of electric motor for anti
2~ 5 3 3.5 4 4.5
Tim e(sec)
skid control of electric vehicle," European Power
Electronics Journal, vol. II, pp. 26-32, 2001.
Fig. 13. Experimental results of proposed approach to differcnt .Z; [17] D. Yin and Y. Hori, "A novel traction control for ev based
on maximum transmissible torque estimation," IEEE
RE FERENCES Transaction on Industrial Electronics, vol. 56, pp. 2086-
f1] Kyoto Protocol, http://unfccc.int/kyotoyrotocollitems/283 2094,2009.
O.php. [18] M. Ikeda, 1'. Ono, and N. Aoki, "Dynamic mass
[2] Y. Hori, "Future vehicle driven by electricity and Control- measurement of moving vehicles," Transactions of the
research on four-wheel-motored "UOT electric march II " " Society of Instrument and Control Engineers, vol. 28, No.
IEEE Transaction on Industrial Electronics, vol. 51, I, pp. 50-58, 1992.
pp.954-962,2004. [19] A. Vahidi, A. Stefanopoulou, and H. Peng, "Recursive
[3] P. He, Y. Hori, "Optimum traction force distribution for least squares with forgetting for online estimation of
stability improvement of 4WD EV in critical driving vehicle mass and road grade: theory and experiments,"
condition," in Proceedings of IEEE International Vehicle System Dynamics, vol. 43, pp. 31-55, 2005.
Workshop on Advanced Motion Control, Istanbul, Turkey, [20] V. Winstead and I. V. Kolmanovsky, "Estimation of road
2006, pp.596-60 I. grade and vehicle mass via model predictive control," in
[4] N. Mutoh, Y. Takahashi, and Y. Tomita, "Failsafe drive Proceedings of 2005 IEEE Conference on Control
performance of FRIO electric vehicles with the structure Applications, Toronto, Canada, pp. 1588-1593, 2005.
driven by the front and rear wheels independently," IEEE [21] P. Ratiroch-Anant, H. Hiroshi, A. Masatoshi, and O.
Transaction on Industrial Electronics, vol. 55, pp. 2306- Shigeto, "Adaptive controller design for anti-slip system of
2315,2008. EV," in Proceedings of 2006 IEEE Conf erence on
[5] M. Schinkel and K. Hunt, "Anti-lock braking control using Robotics, Automation and Mechatronics, Bangkok,
a sliding mode like approach," in Proceedings of the Thailand, 2006, pp. 1-6.
American Control Confe rence, Anchorage, USA, 2002, [22] J.-S. Hu, F.-R. Hu, and M.-Y. Cheng, "A partial pole
pp.2386-2391. placement approach to proportional-integral observer
[6] C. B. Patil, R. G. Longoria, and J. Limroth, "Control design," Proceedings of the Institution of Mechanical
prototyping for an anti-lock braking control system on a Engineers, Part I, Journal of Systems and Control
scaled vehicle," in Proceedings ofthe IEEE Conference on Engineering, vol. 223, pp. 253-262, 2009.
Decision and Control, Hawaii, USA, 2003, pp. 4962-4967.
[7] M.-C. Tsai and J.-S. Hu, "Pilot control of an auto-
balancing two-wheeled cart," Advanced Robotics, vol. 21,
pp. 817-827,2007.
[8] Y.-S. Ha and S. Yuta, "Trajectory tracking control for
navigation of the inverse pendulum type self-contained