Fuzzy Logic

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

World Academy of Science, Engineering and Technology 29 2007

Intelligent ABS Fuzzy Controller for Diverse Road


Surfaces
Roozbeh Keshmiri, and Alireza Mohamad Shahri

the difference between the velocity of the vehicle and the


translational velocity of the wheel to the velocity of the
vehicle [1].

AbstractFuzzy controllers are potential candidates for the


control of nonlinear, time variant and also complicated systems. Anti
lock brake system (ABS) which is a nonlinear system, may not be
easily controlled by classical control methods. An intelligent Fuzzy
control method is very useful for this kind of nonlinear system. A
typical antilock brake system (ABS) by sensing the wheel lockup,
releases the brakes for a short period of time, and then reapplies again
the brakes when the wheel spins up. In this paper, an intelligent fuzzy
ABS controller is designed to adjust slipping performance for variety
of roads. There are tow major sections in the proposing control
system. First section consists of tow Fuzzy-Logic Controllers (FLC)
providing optimal brake torque for both front and rear wheels.
Second section which is also a FLC provides required amount of slip
and torque references properties for different kind of roads.
Simulation results of our proposed intelligent ABS for three different
kinds of road show more reliable and better performance in compare
with two other break systems.

Te

Fz

Fz

Fig. 1 Front wheel free-body diagram [9]

The goal of the Anti-lock brake system is to hold each tire


of the vehicle operating near the peak of the curve for
that tire, which implies performance of an ABS is strongly
related to the surface condition. Up to now different control
methods have been developed to keep wheels slip in desired
interval. Anti lock brake systems based on Sliding Mode
Control [2], Neural Network [3], and Fuzzy Logic Controller
[4-7] are a few examples of the ABS design. As shown in Fig.
2, it can not be expected that an anti-lock brake system which
is optimized for dry asphalt, performs as reliable as on a wet
or icy surfaces.
However, an intelligent method based on identifying the
type of surface, which is proposed in this paper, may adapt
itself for different condition of road surfaces to have optimized
wheel slip. The Main propos of this control method is to
identify road surface condition which leads to an optimized
brake efficiency for different surface conditions.

KeywordsFuzzy Logic Control, ABS, Anti lock Braking


System.

Rw

Tb

I. INTRODUCTION

HE main disadvantage of the ordinary brakes is that the


driver can not precisely control the brake torque applied to
the wheels. Moreover, as the driver does not have enough
information of the road conditions, he may cause locking up
the wheels by applying extra pressure on the brake pedal. The
wheel lock up not only ends to have maximum stopping
distance, but also causes lateral instability of the vehicle. All
modern vehicles are equipped with anti lock braking system
that prevents locking of wheels. Anti lock brake decreases the
stopping distance of vehicle and improves controllability of
vehicle in compare with other brake systems lacking ABS.
When a vehicle accelerates or brakes, the tractive forces Ftf

Road adhesion coefficient


()

1.2

and Ftr developed by the front and rear tire, respectively are
proportional to the normal forces of the road acting on the tire
( Fzf and Fzr ) as illustrated in Fig. 1. The coefficient of

Dry asphalt

0.8

Wet asphalt

0.6

proportionality, denoted by , is called the road coefficient of


adhesion and it varies depending on the road surface type, as
shown in Fig. 2. The wheel slip, denoted by , is the ratio of

0.4

Icy asphalt

0.2

Manuscript received June 30, 2007. Roozbeh Keshmiri is with the member
of Young Researchers Club, Tabriz Islamic Azad University, and member of
IAUT-CRL (computer Research Laboratory), Tabriz, Iran (phone: +98 914
411 1418; e-mail: roozbeh.keshmiri@gmail.com).
Ali M. Shahri is with the Electronic Research Center, Iran University of
Science and Technology, Narmak, Tehran, 16844, Iran (phone: +98 21
73913704; fax: +98 21 77893174; e-mail: shahri@iust.ac.ir ).

0.2

0.4

0.6

( ) Wheel slip

0.8

Fig. 2 - curves for different road conditions

292

1.2

World Academy of Science, Engineering and Technology 29 2007

Fuzzy control system presented in this paper is consisted of


three sections which are shown as FLC1, FLC2 and FLC3. All
Fuzzy controllers are designed and simulated by MATLABTM
software. The FLC1 and FLC2 which are designed for front
and rear wheels are exactly similar and have similar FuzzyLogic-Components. In these two FLCs the optimal brake
torque for front and rear sections are considered as FLC
output. The FLC3 determines the amount of optimal wheel
slip based on the vehicle acceleration.

The front and rear tractive forces are denoted Ftf and Ftr
respectively. The total tractive force, denoted Ftot , is;
Ftot = Ftf + Ftr
Ftf = ( f ) Fzf , Ftr = ( r ) Fzr
Where
Fzf and Fzr are normal forces acting on the front and rear

tires, respectively, and


( ) is the road coefficient of adhesion which is a function

II. VEHICLE BRAKE SYSTEM MODEL

of the wheel slip . For the front tire, = f while for the

The development of FLC for vehicle brake system which is


discussed in this part is generally based on the work carried
out by Will and Zak [9]. A straight-line braking with no
steering is assumed in this work. We have also neglected the
effects of pitch and roll for simplicity.

rear tire = r .The wheel slip is defined as;

V wR w
wR w
= 1
V
V
Where V is the vehicle velocity, w is the angular velocity of
the tire, and R w is the radius of the tire. It should be noted that
0 1 and it is common to present the values of in
percent.
Wheel lockup occurs when = 100% or ( = 1 ) which
equivalently, means the angular velocity of wheel is w = 0 .
However, as can be seen in Fig. 2, a wheel lockup does not
have the maximum coefficient of adhesion, and thus, does not
lead to the maximal braking force. The maximum braking
force is achieved when the wheels are slipping. For example,
in Fig. 2, for an icy road, the maximal braking force is
achieved when the wheel slip is about 1. It also should be
noted that tractive force is a functions of normal force and
there are two components in the normal force. One component
is due to the mass distribution of the vehicle, while the other
component comes from the mass transfer of the vehicle. Based
on the vehicle dynamics equations and performing some
manipulations, we obtain;

ms
hs
mr

mf

hr

hf
Ftr

a Ftf

Fig. 3 Vehicle free-body diagram

A vehicle-free body diagram for the straight-line braking


maneuver is shown in Fig. 3. The symbols, parameters and
values which are used in our simulation are described in Table
I.

TABLE I
VEHICLE BRAKE SYSTEM DATA
acceleration due to gravity

9.81m/s2

distance from center of gravity to front axle

1.186m

distance from center of gravity to rear axle

1.258m

hs

height of the sprung mass

0.6 m

hf
m

height of front unsprung mass

0.3 m

height of rear unsprung mass

0.3 m

m3 = (m f h f + m s h s + m r hr ) /( a + b)

mtot

total mass of the vehicle

1500 kg

The state variables could be defined as;

ms

sprung mass of the vehicle

1285 kg

mf

front unsprung mass

96 kg

mr

rear unsprung mass

119 kg

Jf

moment of inertia of the front wheel

1.7 gm2

Jr

moment of inertia of the rear wheel

1.7
kgm2

Rw

radius of tire

0.326 m

Te

engine torque

0 Nm

K br

brake displacement proportionality


constant

0.5

x = g

( f )m1 + ( r )m3
mtot ( f )m3 + ( r )m3

(1)

Where
m1 = bmtot /(a + b) ,
m 2 = amtot /(a + b) ,

x1 = x, x 2 = x , x 3 = w f , x 4 = wr
Therefore, the vehicle brake state space model and its
output could be presented as follow,
x1 = x2

x = f = g ( f )m1 + (r )m2
2
2
mtot + m3 ( (r ) ( f ))

1


x3 = f 3 = 2 J (Tbf + ( f )m1Rw g ( f ) m3 Rw x2 + Te )
f

1
(Tbr + (r )m2 Rw g + (r ) m3 Rw x 2 )
x 4 = f 4 =
2J f

293

(2)

World Academy of Science, Engineering and Technology 29 2007

y = x1 , x 2 , f , r

]T

is the optimal torque of each surface that results from the first
and third outputs of FLC3. This error for front and rear wheels
is;
eTbf = Tbfref Tbf (t 1)

(3)

Note that our brake system model is composed of only four,


nonlinear, differential equations, which makes it suitable for
our design purposes. Ref. [10] could be referred for more
details of the derivations of the vehicle brake system model.

eTbr = Tbrref Tbr (t 1)


This input is fuzzified by five linguistic values which are
defined below in the universe of discourse [5000 -5000].
Terror = [vfew, few, zero, big, vbig]
Fig. 6 shows membership function of second input for
FLC1 and FLC2.

III. DESIGNING FUZZY CONTROLLER


The Fuzzy controllers which are designed in this paper, are
shown in Fig. 4. As it is explained it has three sections FLC1,
FLC2 and FLC3. In FLC1 and FLC2, optimal brake torques
for front and rear wheels are determined based on given inputs
which are 1) error of slip ratios and 2) error of optimal torque
with torque of wheels. FLC3 also determines optimal slip and
torque according to the vehicle acceleration as its input. First
input for FLC1 and FLC2 is error of slip ratios, i.e. It is
difference between wheel slip and reference slip.

FLC1
Front
Wheel

Tbf (t 1)
+-

zero

-2500

pos

vpos

-5000

2500

5000

Out put of these two Fuzzy controllers is the variation of


torques Tb which adds up to the previous applied torques to
generate required braking torque for each wheels.
Tb (t ) = Tb (t 1) + Tb

Tbf

+
+

Tbf (t 1)

Slip
f
reference
+-

Vehicles
Accelerate

neg

Fig. 6 Membership function for second input of FLC1&FLC2

Front optimum
torque
+-

FLC3

vneg

FLC2
Rear
Wheel

+Rear optimum
torque T (t 1)

+
+

Where Tb refers to brake torques. Membership function of


Tb which are defined by five linguistics values as below are
shown in Fig. 7.
Tb = [vneg, neg, zero, pos, vpos]

Tbr

vneg

Tbr (t 1)

neg

zero

pos

vpos

br

Fig. 4 Schematic of the proposed fuzzy ABS controller

Reference slip is amount of optimal slips for all road


surfaces which is outcome of the second output of FLC3. This
error for front and rear wheels is;
e f (t ) = fref f (t )

-5000

-2500

2500

5000

Fig. 7 Membership functions for output of FLC1 & FLC2

There are 9 IF-THEN Fuzzy Rules for each of FLC1 and


FLC2. A 3-D input-output map for FLC1&FLC2 is illustrated
in Fig. 8.

e r (t ) = rref r (t )

This input has 5 linguistic values as shown Figure 5.


error = [vneg, neg, zero, pos, vpos]
vneg

neg

zero
pos

vpos

-1

-0.6

0.6

Fig. 8 3-D input-output map for FLC1 & FLC2

Fig. 5 Membership functions for first input of FLC1&FLC2

FLC3 which is shown in Fig. 4 is applied to determine slip


and torque references for different road surfaces. According to
Fig. 2 which shows slip and adhesions - curves for different
road conditions, it is obvious that there is no constant slip

Second input of FLC1 & FLC2 is error of brake tourqe or


Tb , that is difference between the last applied brake torque
and optimal torque. The amount of required torque reference

294

World Academy of Science, Engineering and Technology 29 2007

values for maximum road adhesion in different road surfaces


such as; dry, wet or icy asphalts. Therefore, it is possible to
define optimum slip reference for FLC1 & FLC2 by FLC3
which consequently results in maximum adhesion applied to
the brake system. Input of the FLC3 is vehicle acceleration
which is shown in Fig. 9 by 5 linguistic membership functions
as below;
Accelerate = [vbig, big, med., small, vsmall]
In this fuzzy controller, we may identify the type of road
surface by vehicle acceleration.
1

big

vbig

-10

medium

small

hydraulic systems and brake pad travel as in [6], [8].


Simulation results are compared based on braking or stopping
distance, vehicle speed profile and also wheel slip for three
different states of a vehicle a) without ABS, b) ABS with
constant slip and finally c) with proposed ABS. In the process
of simulation, three different surfaces are applied to vehicle
braking systems in series, i.e. a simulation surface starting
with 10 meters of dry asphalt, and then 20 meters of wet
asphalt and finally change to an icy condition. As it is shown
in Figs. 12 and 13 for a vehicle without ABS, stopping
distances is approximately 80 meters and stopping time is 10
seconds. As in this case, there is no ABS system wheel slip for
front and rear wheels will be one and wheels will be locked up
which it consequently results in larger stopping distance (Fig.
14).

vsmall

-5

Fig. 9 Membership function for input of FLC3


Position (m)

Without ABS

There are three optimum outputs for FLC3, front brake


torque, slip and rear brake torque. All three output
membership functions in FLC3 are defined with 5 similar
linguistics values as below;
Reference = [Icy, Icy-wet, wet, wet-dry, dry]
Fig. 10 illustrates output membership functions for optimal
front and rear wheel brake torque which could be adjusted
based on slip and road adhesion relationship.
Icy

IcyWet

Wet

Wetdry

Fixed slip ABS


Proposed ABS

Time (sec)

Fig. 12 Plots of the vehicle position on the surface changing from dry
asphalt to wet asphalt after 10m and icy asphalt after 20m for three
different of braking systems

Dry

1000

2000

3000

Velocity (m/s)

3500

Fig. 10 Membership function for first & third outputs of FLC3

Fig. 11 also illustrates output membership function for


optimal slip.

Icy

IcyWet

Wet

Wetdry

Proposed ABS

Time (sec)

Wheel slips ()

1.5

Fixed slip ABS

Fig. 13 Plots of the vehicle velocity on the surface changing from dry
asphalt to wet asphalt after 10m and icy asphalt after 20m for
different of braking maneuvers

Dry

Without ABS

0.2

Fig. 11 Membership function for second outputs of FLC3

IV. SIMULATION RESULTS

Time (sec)

In order to simulate the proposed intelligent fuzzy controller


and compare the simulation results with other methods, it is
assumed that the vehicle is moving at 20 m/s, (72 km/h). It
should be noted that all the simulation results are based on the
assumption of straight-line braking in the brake system model
and neglect of transportation delay caused by the brake

Fig. 14 Plots of wheel slip on brake without ABS

As the second braking method, ABS with constant slip (0.2


for all three different surfaces) is applied. As it is shown in
Figs. 12 and 13, stopping distance has been reduced to
approximately 43 meters and stopping time is about 6 seconds.

295

World Academy of Science, Engineering and Technology 29 2007

Wheel slips ()

Brake torque

Figs. 15 and 16 illustrate the wheel slips for front and rear
wheels.

Time (sec)

Fig. 19 Front wheel applied brake torque by the proposed ABS


Time (sec)

Wheel slips ()

Brake torque

Fig. 15Plot of front wheel slip on brake with fixed slip ABS

Time (sec)

Time (sec)

Fig. 16 Plot of rear wheel slip on brake with fixed slip ABS.

Fig. 20 Rear wheel applied brake torque by the proposed ABS

Finally, the proposed ABS is applied to the simulation


process and according to the Figs. 12 and 13 there is an
evident superior performance in simulation results in compare
to other two methods. Stopping distance has been reduced to
40m and stopping time has been reduced to 5 seconds. Figs.
17 and 18 also illustrate adaptive wheel slip based on the
changing surface conditions. This result verifies that the
proposed fuzzy controller (FLC3) is able to generate different
wheel slip according to the surface condition.
Wheel slips ()

Acceleration (m/s2)

Acceleration of Vehicle for different surfaces which is an


input of the FLC3 is also shown in Fig. 21. This acceleration
is actually the reason to determine adaptive slip wheel based
on the surface condition.

Time (t)

Fig. 21 Plot of vehicle acceleration in proposed ABS.

V. CONCLUSION

Time (sec)

In this paper an intelligent ABS fuzzy controller is


designed. The vehicle dynamic based on the half of vehicle
equations are presented in the form of state space model. This
model is applied for all simulation process using MATLAB
SIMULINK toolbox. Simulation results verify the superior
performance and quality of the proposed Intelligent ABS
fuzzy controller over other methods without identifying road
conditions. This controller has excellent ability to apply
continuous brake torque and accordingly determines optimal
slip to reduce stopping interval and distance. Two parallel
fuzzy controllers have been designed to determine braking
torque for front and rear wheels respectively. Furthermore,
third fuzzy controller is designed to determine optimal slip and
torque for different road surfaces.

Wheel slips ()

Fig. 17 Plot of front wheel slip in proposed ABS

Time (sec)

Fig. 18 Plot of rear wheel slip in proposed ABS.

Applied brake torque to the front and rear wheels by the


proposed ABS control system have been shown in Figs. 19,
20.

REFERENCES
[1]

296

J. Y. Wong, Theory of Ground Vehicles. New York: J. Wiley & Sons,


second ed., 1993.

World Academy of Science, Engineering and Technology 29 2007

[2]

S. Drakunov, . zgner, P. Dix, and B. Ashrafi, ABS control using


optimum search via sliding modes, IEEE Trans. Contr. Syst.
Technol.,vol. 3, pp. 7985, Mar. 1995.
[3] L. I. Davis, Jr., G. V. Puskorius, F. Yuan, and L. A. Feldkamp. Neural
Network Modeling and Control of an Anti-Lock Brake System
Research Laboratory, Ford Motor Company,IEEE ,1993.
[4] A. Mirzaei, M. Moallem, B. Mirzaeian, B.Fahimi, Design of an
Optimal Fuzzy Controller for Antilock Braking Systems IEEE, 2005.
[5] J. R. Layne, K. M. Passino, and S. Yurkovich, Fuzzy learning control
for antiskid braking systems, IEEE Trans. Contr. Syst.Technol., vol.
1,pp. 122129, June 1993.
[6] G. F. Mauer, A fuzzy logic controller for an ABS braking
system,IEEE Trans. Fuzzy Syst., vol. 3, pp. 381388, Nov. 1995.
[7] W. K. Lennon and K. M. Passino, Intelligent control for brake
systems, IEEE Trans. Contr. Syst. Technol., vol. 7, pp. 188202, March
1999.
[8] B. Friedland, Advanced Control System Design. Englewood Cliffs, NJ:
Prentice-Hall, 1996.
[9] A. B. Will and S. H. Zak, Antilock brake system modeling and fuzzy
control, Int. J. of Vehicle Design, vol. 24, no. 1, pp. 1-18, 2000.
[10] Y. Lee and S.H.Zak, Designing a genetic neural fuzzy anti-lock brake
system controller, IEEE Trans. on Evolutionary Computation,vol. 6,
no. 2, pp. April 2002 in press.

297
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy