Documente Academic
Documente Profesional
Documente Cultură
3, MAY 2007
Abstract—This paper presents a nonlinear sliding-mode-type Tbi Brake torque at the ith wheel.
controller for slip regulation in a braking event for an electro- V̇x Vehicle longitudinal acceleration.
magnetic brake-by-wire (BBW)-system-equipped automobile. The Vx Vehicle longitudinal speed in road coordinate system.
electromagnetic BBW system under consideration consists of a
set of eddy current brakes (ECBs) attached to the automobile Vy Vehicle lateral velocity.
wheels. The antilock-braking controller modifies the brake torque Żs Sprung-mass velocity in the z-direction.
command generated by a supervisory controller which, in turn, is ωi Angular speed of the ith wheel.
based on the driver’s command sensed via brake pedal sensors. ω̇i Angular acceleration of the ith wheel.
The modified brake torque command is then generated by a
closed-loop actuator control algorithm to control the ECB system.
It is shown in this paper that the proposed antilock-braking I. I NTRODUCTION
control system is stable in the sense of Lyapunov. Experimental
results are presented for a test vehicle equipped with an eddy-
current-based BBW system. Experimental results show that the
proposed antilock brake control algorithm provides very good slip
A NTILOCK braking systems (ABSs) generally offer supe-
rior vehicle safety by limiting the longitudinal wheel slip
in a braking event with wheel lock-up condition. Drivers have
regulation in a braking event on low friction-coefficient surfaces
(wet jennite) when compared with that of a braking event without better directional control of the vehicle equipped with ABS.
the proposed antilock-braking control. Experimental results also Most of the ABS systems in the published literature utilize
indicate that the proposed antilock-braking control system pro- the wheel-slip estimate to control the wheel cylinder pressure
vides a smooth stop for the vehicle. through a set of valves and a pump in order to regulate the
Index Terms—Antilock braking, brake-by-wire (BBW), eddy wheel braking torque. These control systems are built around
current brake (ECB), sliding mode control. the actuators’ dynamic characteristics. With a new generation of
actuators on the horizon (e.g., electromagnetic, electromechan-
N OMENCLATURE ical, etc.), it may be necessary to design a supervisory brake
control system that would issue a torque command as control
Fxsum Sum of road forces in the x-direction at the input based on the vehicle dynamics. The subsystem controller
tire–ground interfaces. would then close the loop on the actuators based on actuator
Ftx Terrain forces at the center of gravity arising out of dynamics.
road slopes and grades. Watanabe and Noguchi [13] presented an ABS algorithm
Fax Aerodynamic-drag forces on the vehicle. that used a vehicle-deceleration threshold to activate the brake
Fxi Longitudinal friction force at the ith tire contact patch. pressure reduction algorithm, and the duration of the pressure
Frri Rolling resistance at the ith tire contact patch. pulse was determined by the road friction coefficient. Tan
Iwi ith wheel rotational inertia. and Chin [12] discussed an antilock-braking algorithm based
M Total vehicle mass. on sliding-mode-control theory. Based on a longitudinal one-
ms Sprung mass of the vehicle. wheel vehicle model, sufficient conditions for applying sliding
q Pitch velocity of the sprung mass. mode control to vehicle traction were derived via Lyapunov
R Effective wheel rolling radius. stability theory. Athan and Papalambros [3] presented a mul-
r Vehicle yaw velocity. ticriteria quasi-Monte Carlo method to optimize and compare
Tdi Drive torque at the ith wheel. the three ABS nonlinear control algorithms. An adaptive fuzzy
logic controller for an ABS was discussed by Kokes and
Singh [9]. Their controller initially employed a priori training
Manuscript received December 7, 2005; revised March 11, 2006 and June data to control the braking system, but it continued to train
21, 2006. This work was supported by the Chassis Advanced Technology online while continuously updating the confidence parameters
Department of Visteon Corporation. The review of this paper was coordinated
by Dr. M. Shahgir Ahmed.
and the placement of fuzzy sets by employing optimization
S. Anwar is with the Department of Mechanical Engineering, Purdue School algorithms. Suh et al. [11] presented a real-time simulator
of Engineering and Technology, Indiana University–Purdue University at for an ABS based on the methodology of hardware-in-the-
Indianapolis, Indianapolis, IN 46202 USA (e-mail: soanwar@iupui.edu).
B. Zheng was with Visteon Corporation, Van Buren Twp, MI 48111 USA. loop simulation using a personal computer. They also pro-
He is now with GE Aviation Systems, Cincinnati, OH 45246 USA (e-mail: vided an analysis and validation of the control logic that used
bing.zheng@ae.ge.com). commercial hardware. A genetic algorithm-based fuzzy logic
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. controller for ABS systems was the subject of a paper written
Digital Object Identifier 10.1109/TVT.2007.895604 by Chen and Liao [4]. The controller was based on a nonlinear
0018-9545/$25.00 © 2007 IEEE
ANWAR AND ZHENG: ANTILOCK-BRAKING ALGORITHM FOR AN EDDY-CURRENT-BASED BRAKE-BY-WIRE SYSTEM 1101
Tbi − Fxi R + Frri R − Tdi = − Iwi ω̇i . (2) IV. S LIDING -M ODE -T YPE ABS C ONTROLLER
Based on the above equations, the plant model for designing
The pitch dynamics of the vehicle in the first equation have a controller is written as follows:
negligible effect on the wheel braking forces and, hence, not
considered here. For simplicity, the effect of terrain forces 1
V̇x = − µi (κi )Fzi
arising out of road slopes and grades is neglected. The drive M
torque in a braking situation is generally small, and hence, it is 1
ω̇i = (−Tbi + µi (κi )Fzi R) . (4)
assumed to be insignificant in the second equation. In this in- Iwi
vestigation, only straight-line braking is considered to illustrate
Now
the concept. Therefore, the steering wheel angle assumed to be
zero resulting in zero lateral motion. Vx − Rωi Rωi
κi (t) = =1− .
Now, the following relationships are defined for rolling resis- Vx Vx
tance and braking forces:
Differentiating the above equation yields the following:
Fxi = µi (κ)Fzi Frri = ηFzi
Rω̇i Rωi V̇x
κ̇i = − +
where µi (κ) = friction coefficient, and η = rolling resistance Vx Vx2
coefficient. R R 1
Since a simple model is desired for the proposed controller =− µ(κi )Fzi − Tbi
Vx Iwi Iwi
development, the effect of aerodynamic drag on the above equa-
Rωi 1
tion is neglected (since the aerodynamic drag is small compared − 2 µi (κi )Fzi
to the brake forces for the normal driving speeds). Besides, Vx M
since this controller works in a closed-loop system, these ef- R 1 R2 1
= Tbi − µ(κi )Fzi
fects can be compensated through the feedback information. Iwi Vx Iwi Vx
The rolling-resistance force Frri is insignificant compared to Rωi 1
the braking force Fxi in a braking event and, hence, it is − 2 µi (κi )Fzi . (5)
Vx M
neglected. Thus
Fig. 3 shows the friction-coefficient curves for a number of
Fxsumr = − µi (κi )Fzi . tire–ground interfaces. It is evident that the peak of the friction-
coefficient curve varies significantly depending on the road
The simplified equations of motion are then given by conditions. The slip-ratio value at the peak friction coefficient
also varies between 0.1 and 0.2. It is clear that the friction-
− µi (κi )Fzi = M V̇x coefficient relationship with slip adds nonlinearity to (5). Since
all of the curves in Fig. 3 exhibit approximate linear relationship
Tbi − µi (κi )Fzi R = − Iwi ω̇i . (3) with slip ratio below the peak of the curve, this relationship
between the coefficient of friction and the slip can be approxi-
The above set of equations represents the plant model for a mated with a piecewise linear function. This concept is shown
straight-line braking event. The following section describes the in Fig. 4 [77]. The friction curves are approximated by a straight
controller development process. line with a slope of αsi and a slip threshold of κth . While the
ANWAR AND ZHENG: ANTILOCK-BRAKING ALGORITHM FOR AN EDDY-CURRENT-BASED BRAKE-BY-WIRE SYSTEM 1103
Vx Iwi
Tbi = κ̇th + Rαsi κi Fzi
R
Iwi ωi
+ αsi κi Fzi
Vx M
Iwi κth − κi
+η Vx ∗ SAT . (12)
R φ
Now, let us define the sliding surface as follows: V̇ = 2(κth − κi ) ∗ (κ̇th − κ̇i ). (15)
S = (κth − κi ). (8)
Assuming that κth is a constant reference value (i.e., κ̇th =
It is assumed here that the desired slip ratio is the same as 0) and substituting (7) in (15), the following expression is
the slip-ratio threshold. With the above definition of the sliding obtained:
surface, the sliding-mode-type control law is given by [55]
V̇ = −2 ∗ (κth − κi )
S 2 Rωi 1
Ṡ = −ηSAT (9) ∗ IR 1
wi Vx
Tbi − IRwi V1x αsi κi Fzi − Vx2 M αsi κi Fzi . (16)
φ
Rωi 1
κth − κi ∗ 2 αsi κi Fzi
= −ηSAT . (11) Vx M
(17)
φ
1104 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 56, NO. 3, MAY 2007
Fig. 5. Measured wheel speed for the front wheels in the non-ABS maneuver
on wet-jennite surface.
Fig. 8. Wheel slip ratio for the front wheels in the ABS maneuver on wet-
jennite surface.
Fig. 6. Measured wheel speed for the front wheels in the ABS maneuver on Fig. 9. Measured vehicle speed in the non-ABS maneuver on wet-jennite
wet-jennite surface. surface.
Fig. 7. Wheel slip ratio for the front wheels in the non-ABS maneuver on
wet-jennite surface. Fig. 10. Measured vehicle speed in the ABS maneuver on wet-jennite surface.
reach locked-up condition, the slip ratio for the front wheels activated. As shown, the front-wheel slip ratio is maintained at
reached a value of 0.75, indicating 75% skidding (or near lock- near the desired slip-ratio value of 0.2 over the same period
up condition). Fig. 8 shows the slip ratio for the straight-line when near lockup occurred without the ABS activated. Since
braking on wet jennite but with the proposed ABS algorithm the vehicle tests with and without the ABS activated were
1106 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 56, NO. 3, MAY 2007
Fig. 11. Wheel brake torque command in the non-ABS maneuver on wet-
jennite surface.
Fig. 14. Measured ECB current in the ABS maneuver on wet-jennite surface.
Fig. 12. Wheel brake torque command in the ABS maneuver on wet-jennite
surface. Fig. 15. Measured wheel torque for the front wheels in the non-ABS maneu-
ver on wet-jennite surface.
By limiting the slip value to 0.2, substantial lateral traction is [4] F. W. Chen and T. L. Liao, “Nonlinear linearization controller and genetic
made available to the wheel contact patch offering the driver algorithm-based fuzzy logic controller for ABS systems and their com-
parison,” Int. J. Veh. Des., vol. 24, no. 4, pp. 334–349, 2000.
full directional control of the vehicle. Basically, the tire contact [5] J. K. Hedrick, “Analysis and control of nonlinear systems,” J. Dyn. Syst.
patch dynamics can be explained via Kamm circle [88] which Meas. Control, vol. 115, no. 2B, pp. 351–361, Jun. 1993.
states that at higher longitudinal slip rates, the lateral coefficient [6] H.-P. Huang and C.-K. Wang, “Intelligent control of wheeled vehicles
with anti-braking systems,” Int. J. Veh. Des., vol. 26, no. 2/3, pp. 218–
of friction value decreases drastically diminishing the driver’s 238, 2001.
ability to steer. [7] S. Kimbrough, “Brake control algorithms,” Mech. Eng. Dept., Univ. Utah,
Figs. 9 and 10 show the measured vehicle speed for the Salt Lake City, 1997. Final Rep.
[8] U. Kiencke and L. Nielsen, Automotive Control System for Engine,
non-ABS and ABS maneuvers, respectively. It is clear from Driveline, and Vehicle, 2nd ed. New York: Springer-Verlag, 2005.
these plots that ABS provides a much smoother stopping of [9] G. Kokes and T. Singh, “Adaptive fuzzy logic control of an anti-
the vehicle than the non-ABS scenario. Figs. 11 and 12 show lock braking system,” in Proc. IEEE Conf. Control Appl., 1999, vol. 1,
pp. 646–651.
the commanded wheel braking torque from the supervisory [10] K. Lee and K. Park, “Optimal robust control of a contactless brake system
BBW controller for non-ABS and ABS activated controls, using an eddy current,” Mechatronics, vol. 9, no. 6, pp. 615–631, 1999.
respectively. Clearly, the commanded torques reached the max- [11] M. W. Suh, J. H. Chung, C. S. Seok, and Y. J. Kim, “Hardware-in-the-
loop simulation for ABS based on PC,” Int. J. Veh. Des., vol. 24, no. 2,
imum values quickly after brake is applied in a non-ABS case. pp. 157–170, 2000.
However, for the ABS activated condition, the commanded [12] H.-S. Tan and Y.-K. Chin, “Vehicle traction control. Variable-structure
torque is modified by the ABS control law and is modulated control approach,” Trans. ASME, J. Dyn. Syst. Meas. Control, vol. 113,
no. 2, pp. 223–230, 1991.
to keep the wheel slip ratio near the desired value. [13] M. Watanabe and N. Noguchi, “New algorithm for ABS to compensate
Figs. 13 and 14 show the measured current to the front for road disturbances,” SAE Trans., vol. 99, no. 6, pp. 271–279, 1990.
wheel ECBs for the non-ABS and ABS activated maneuvers,
respectively. It is clear that during the wheel lock-up event,
the current input to the ECBs modulates based on the torque
command in the ABS activated maneuver. Figs. 15 and 16 show
the measured wheel torque values for the front wheels in non-
ABS and ABS conditions, respectively. In the ABS mode, the
measured torque is reduced to a level to maintain the desired
slip ratio at each of the front wheels. Sohel Anwar received the Ph.D. degree in mechan-
ical engineering from the University of Arizona,
Tucson, in 1995.
VII. C ONCLUSION He then joined the Caterpillar, Inc., Peoria, IL,
as a Research Engineer, where he was responsi-
A nonlinear sliding-mode-type controller is presented for slip ble for developing the advanced control systems
for earthmoving equipments. From 1999 to 2004,
regulation in a braking event for an eddy-current-based BBW he worked as a Senior R&D Engineer with the
system in this paper. It is also shown that the closed-loop system Visteon Corporation, Dearborn, MI, where he led
is stable in the sense of Lyapunov. Experimental results show projects in the area of control-system development
for drive-by-wire systems. Since 2004, he has been
very good slip regulation in a braking event on the low friction- in the Department of Mechanical Engineering, Purdue School of Engineer-
coefficient surface when compared with the non-ABS braking ing and Technology, Indiana University–Purdue University at Indianapolis,
condition. The results show that the proposed ABS controller Indianapolis, as an Assistant Professor. He is the holder of 12 U.S. patents.
He is a registered Professional Engineer in the state of Michigan.
provided a smooth ABS stop as evident from the vehicle- Dr. Anwar is a member of American Society of Mechanical Engineers and a
speed plots. Faculty Advisor for Society of Automotive Engineers International.
ACKNOWLEDGMENT
The authors would like to thank the support of the technical
staff of Chassis Advanced Technology Department of Visteon
Corporation during the course of this paper. The authors would
also like to extend their gratitude to the Visteon management
team who encouraged the publication of this paper outside of Bing Zheng received the Ph.D. degree from the
Chinese Academy of Space Technology (CAST),
the corporation. Beijing, China, in 1987.
He then spent one year at the German Aerospace
R EFERENCES Center, Germany, as a Guest Scientist, conducting
research on large flexible space structure control.
[1] S. Anwar and B. Ashrafi, “Development of a hybrid electromag- He returned to CAST as a System Engineer, where
netic/electrohydraulic brake-by-wire system at Visteon,” Chassis Adv. he worked on the DFH-3 (Chinese) communica-
Technol., Visteon Corporation, Dearborn, MI, Tech. Rep. B600-027, tion satellite control system for the next five years.
May 2003. He then joined Yellow Spring Instrument, Yellow
[2] S. Anwar, “A parametric model of an eddy current electric machine for Springs, OH, as a Senior Electrical Engineer, where
automotive braking applications,” IEEE Trans. Control Syst. Technol., he worked for two years. He worked at Liebert Corporation, Columbus, OH, as
vol. 12, no. 3, pp. 422–427, May 2004. a Development Scientist for the next two years. From 1999 to 2006, he served
[3] T. W. Athan and P. Y. Papalambros, “Multicriteria optimization of anti- as a Technical Fellow with the Visteon Corporation, MI, leading the control
lock braking system control algorithms,” Eng. Optim., vol. 27, no. 3, and fault-tolerant communication systems development for x-by-wire systems.
pp. 199–227, 1996. Currently, he is at the GE Aviation Systems, Cincinnati, OH.