2008 International Conference on Intelligent Computation Technology and Automation
An Advanced Control Method for ABS Fuzzy Control System
ZHANG Jing-ming1, SONG Bao-yu2, SUN Gang3
1. Automobile Engineering Department, Harbin Institute of Technology, Weihai, China E-mail:
whjingming@[Link]; 2. School of Mechanical and Electrical Engineering, Harbin Institute of
Technology, Harbin, China; 3. Automobile Engineering Department, Harbin Institute of
Technology, Weihai, China
Abstract relation and magic formula relation now [6]. Problem 2
and problem 3 are problems which this text will solve
The purpose of an antilock braking system (ABS) is mainly.
to prevent wheel lockup under heavy braking
conditions on any type of road surface. In this study, 2. ABS operation principles
multifactor input, d / ds and (d / ds) / dt , are
proposed in the fuzzy controller. This new method not The basic principle of an antilock braking system is
only offers a better braking performance, but also shown in Figure 1. It is a hydraulic braking system
improves the pedal pushing feeling. At last, the design with a hydraulically assisted power booster/master
is simulated in Matlab/Simulink environment. And the cylinder. A sensor at each wheel indicates rotational
results indicate multifactor input offer a good braking speed. The signal is transported to antilock braking
result and improves the active security when the controller. The controller operates solenoid valves to
automobile brakes promptly. reduce, maintain or increase braking pressure [7].
1. Introduction
ABS (Antilock Braking System), on the basis of
traditional braking system, adopts electronic control
technology to modulate the oil pressure of the brake
automatically. During to ABS system, the brake wheel
is prevented from lockup [1]. In general, there are two
kinds of ABS systems, one is based on the wheel Figure 1. Basic principle of an antilock braking
deceleration, the other is based on the wheel slip ratio. system
The first can work without signals of vehicle velocity,
but the full force offered by road cant be used If a wheel speed sensor signals severe wheel
completely. The second aimed to maintain the oil deceleration to the electronic controller that indicate
pressure around the optimum point. In a way, the the wheel is likely to lock up, the braking pressure at
maximal braking force can be used [2], [3]. the wheel involved is initially kept constant as opposed
Furthermore, the wheel can also be prevented from to being further increased. If the wheel still continues
lockup. Following problem should be solved during to decelerate, the pressure in the wheel brake cylinder
designing the second ABS system: 1) Identifying the is reduced so that the wheel is broken less heavily. The
road condition, in other words, how to calculate wheel will accelerate again as a result of the reduced
adhesive coefficient on the condition of the slip ratio braking pressure. Upon reaching a specified limitation,
which is already known. 2) The method to look for the controller registers the factor that the wheel is not
ideal slip ratio, which is corresponding with maximum braked sufficiently now. The former reduced pressure
adhesive coefficient. 3) The arithmetic to control slip is then increased so that the wheel is decelerated again.
ratio [4], [5].
The first problem is solved through experiment
mainly, a commonly used method include straight line
978-0-7695-3357-5/08 $25.00 2008 IEEE 845
DOI 10.1109/ICICTA.2008.300
3. Vehicle modeling
In order to certificate the new control arithmetic, a
vehicle model is needed. The model is proposed on the
supposed condition as follows during braking:
ABS dynamics model is supposed as follows while
applying the brake: 1) Automobile drives in a line. 2)
Neglecting the wind resistance. 3) The load of every
wheel is equal to a fourth of automobile weight. The
dynamics model of the automobile is described as
follows:
ma = 4Fz (1)
Fz = N (2) Figure 3. The analysis of the braking wheel
1 (3) According to the second law of Newton, We have
N = mg
4 received the equation below.
In above equations, m is the automobile quality, a J = Fz r M (5)
is the acceleration of automobile, and Fz is braking J is the rotating inertia of the wheel, is the
force on a wheel from the road. N is vertical force angular acceleration of the wheel, r is the radius of
offered by road, is adhesive coefficient, and g is the wheel, M is braking torque. Among them, braking
acceleration of gravity. is related with slip ratios as torque is related with oil pressure of the braking
cylinder:
follows.
M = KP
( s) = 0 + A sin{ Barctan
(6)
(4) In above equation, P is pressure of the brake
[Cs D(Cs arctan(Cs))] } cylinder; K is the brake coefficient. Because the
A, B, C and D undetermined coefficient can confirm switch of the electromagnetic valve responds quickly,
their values through the method that is tested on it can be regarded approximately as increasing
different road surface. Figure 2 shows s curve of proportionally as follows:
several types of typical road conditions:
P = Ut (7)
U is increasing or reducing pressure speed.
Corresponding maintaining, reducing and increasing
pressure, U can be shown as below equation.
a increasing
(8)
0 maintaining
a reducing
a is a variable value between maximum and
minimum according to the situation of inputs, and it
symbols the value of increasing or reducing pressure
speed.
Figure 2. Several types of typical road conditions
Every curve parameter is as follows: 4. ABS controller design
Curve 1 (general dry road surface):
A =0. 85; B =2. 4; C =4. 5; D =0. 9 ABS receives speed signal of wheels and calculates
Curve 2 (loose gravel road surface): automobile speed, slip ratio and acceleration of wheel.
A =0. 45; B =2. 4; C =5. 0; D =0. 9 It analyses these signals, thus to modulate the pressure
Curve 3 (ice road surface): regulator through sending corresponding order. In
A =0.1; B =2.4; C =6.0; D =0.9 order to satisfy the braking requires, and to apply the
While braking, the analysis of the braking wheel is maximum adhesive coefficient, the slip ratio should be
described by Figure 3. controlled near to ideal value. According to the
extremism theory, when d / ds is zero, adhesive
846
coefficient is the maximum. So we should control the
ratio near to the point.
In general ABS controller, input is single, d / ds
e.g. If we set the d / ds and (d / ds) / dt as input,
and braking pressure P as output. The key aspect is
that pressure modulating value U is variable. This can
be actuated by a PWM modulation electromagnetic
valve. First, the input and output variables are altered
to the scale [-5, +5], and then input is divided into five
discussed scale, {NL, NS, ZE, PS, PL}, witch is shown
as Figure 4. The fuzzifition of the output is divided
into seven fuzzy set, {NL, NM, NS, ZE, PS, PM, PL}
as Figure 4. The Gaussian method is applied.
Figure 5. ABS control system block diagram
Figure 6. ABS controller block diagram
Simulation parameter: in the magic formula
Coefficient A=0.85, B=2.4, C=4.5, D=0.9; Automobile
load m =4000kg, the wheel rotating
Figure 4. The fuzzy set of the input and output inertia J =2.23kg/m2, the radius of the wheel
After fuzzifition, the fuzzy rules is designed, Table 1 r =0.375m, the initial velocity V0 =50m/s, a has a
describes the fuzzy rules of the system. scale from 0 to 4000. The simulation result is as Figure
Table 1. Controls the regular form fuzzily from 7 to 10 shown.
5. Simulation and result analysis
Figure 5 shows matlab/simulink model according to
aforesaid ABS model.
Figure 7. The braking pressure after and before
improvement
847
general ABS system, the new antilock braking system
improves the pushback feeling on the braking pedal.
Fig. 10 indicates that the braking distance of the new
system is shorter than the general one. The reason of
this is easily found in Figure 8 and Figure 9. That is
because the wheel speed can maintain nearer to the
ideal speed than that of general one, and the adhesive
coefficient also can maintain nearer to the ideal value.
6. Experiment and result analysis
In order to test the correctness of ABS model, the
car was experimented which equipped with ABS
system when emergency braking occurred in the
condition of different initial velocity. This experiment
is in accordance with GB 7258-2004 test methods and
standards. The test results and the results of the
Figure 8. Vehicle speed and wheel speed after and simulation can be shown as in Table 2. MFDD is
before improvement described by the vehicle average braking deceleration
with full ability.
Table 2. Performances comparing between test
result and simulation result (1)
Vehicle Braking distance (m)
speed Experimental Simulation Relative
(km/h) data data error (%)
50 16.29 16.45 1.0
80 37.76 38.02 0.7
100 52.75 52.91 0.3
Performance comparing between test result and
simulation result (2)
Vehicle MFDD (m/s2)
speed Experimental Simulation Relative
(km/h) data data error (%)
Figure 9. The adhesive coefficient after and before 50 8.87 8.95 0.9
improvement 80 9.10 9.63 0.5
100 9.44 9.88 0.1
From Table 2 we can find the simulation results are
very close to the experimental results. The maximum
relative error of Braking distance is 1 percent. The
maximum relative error of MFDD value is 0.9 percent.
It can be proved that the model is fit for accuracy
.Besides, the simulation results of ordinary cars
equipped with the ABS braking performance and the
simulation results of the vehicles equipped with the
new ABS were compared. It can be stated as table 3.
Table 3. Performances comparing between general
ABS simulation result and new ABS simulation (1)
Vehicle Braking distance (m)
speed General New Increment Relative
Figure 10. Braking distance after and before (km/h) ABS ABS increment
improvement 50 16.45 15.49 -0.96 -5.83
Figure 7 indicates that the braking pressure reach the 80 38.02 36.76 -1.26 -3.31
ideal value and maintain still. Compared with the 100 52.91 51.94 -0.97 -1.83
848
Performances comparing between general ABS
simulation result and new ABS simulation (2) 8. References
Vehicle MFDD (m/s2)
speed Relative [1] MA Ming-xing, MAO Wu-ben, and ZHU Wei-xing,
General New Increment
(km/h) increment Vehicle ABS controls theory analysis of the algorithm and
ABS ABS
experimental study, Mechanical Journal for Agriculture.
50 8.95 9.05 0.10 1.12
April 2002, pp. 29-37
80 9.63 9.72 0.09 0.93
[2] MA Ming-xing, CHOU Qi-ya, The braking simulation
100 9.88 9.96 0.08 0.81 study of vehicle based MATLAB, Group Technology &
We can get a conclusion from Table 3 that braking Production Modernization, March 2004, pp. 11-14
distance is shorten maximum relatively 5.83 percent. [3] Long Xiao-lin, Du Xiao-fang, and Xu Da, The
automobile ABS analysis and simulation based on slip and
MFDD maximum relative value increment is 1.12 variable structure modulation control, ZHUAN YONG QI
percent. Simulation results fully demonstrated that this CHE, March 2004, pp. 25-26
new system can shorten the ABS braking distance. It [4] SUN Jun, The simulation of antilock braking system
can raise the braking average speed and improve the based on fuzzy control, Computer Simulation, October
braking performance of vehicles. 2004, pp. 160-162
[5] Li Jun, YU Fan, and ZHANG Jian-wu, The research of
7. Conclusion ABS fuzzy control system on the basis of identifications
automatically, Mechanical Journal for Agriculture. May
2002, pp. 26-29
This study aims to get the maximum adhesive [6] Qi Zhang, Guo-fu Liu, Yue-ke Wang, and Ting-ting
coefficient and control it to maintain near to the ideal Zhou. A Study of Calculation Method of Wheel Angular
value. Fuzzy control method is adopted which has two Acceleration in ABS System, Proceedings of 2004
input variable, d / ds and (d / ds) / dt . During the International Conference on Information Acquisition.
reducing, maintaining and increasing pressure, the September 2004, pp. 147-150.
[7] Dengfeng Yang, Zhihua Feng, Desheng Ma, and Gang
velocity is defined variable. This can carry out through
Sun, Development of Anti-lock Brake System in Virtual
a PWM modulating electromagnetic. This is the next Environment, IEEE Human-Computer Interfaces and
work. The ABS control system is simulated in Measurement Systems, June 2004, pp. 131-135
Matlab/Simulink environment. The result indicates
that, compared with the single input, the multi-input
can improve the pushback on the pedal.
849
