Genetic fuzzy self-tuning PID controllers for antilock braking systems

Abdel Badie Sharkawy

Mechanical Engineering Department, Faculty of Engineering, Assiut University, Assiut 715 16, Egypt
a r t i c l e i n f o
Article history:
Received 16 June 2008
Received in revised form
18 June 2010
Accepted 23 June 2010
Available online 14 July 2010
Keywords:
Antilock braking system (ABS)
Braking basics
Slip ratio
Decoupled rule bases
TakagiSugeno (TS) fuzzy systems
Self-tuning
Genetic algorithm (GA) and ZieglerNichols
(ZN) method
a b s t r a c t
Since the emergence of PID controllers, control system engineers are in pursuit of more and more
sophisticated versions of these controllers to achieve better performance, particularly in situations
where providing a control action to even a minimal degree of satisfaction is a problem. This work is an
attempt to contribute in this eld. Variations in the values of weight, the friction coefcient of the road,
road inclination and other nonlinear dynamics may highly affect the performance of antilock braking
systems (ABS). A self-tuning scheme seems necessary to overcome these effects. Addition of automatic
tuning-tool can track changes in system operation and compensate for drift, due to aging and parameter
uncertainties. The paper develops a self-tuning PID control scheme with an application to ABS via
combinations of fuzzy and genetic algorithms (GAs). The control objective is to minimize the stopping
distance, while keeping the slip ratio of the tires within desired range. Computer simulations are
performed to verify the proposed control scheme. Results are reported and discussed.
& 2010 Elsevier Ltd. All rights reserved.
1. Introduction
Difculties in designing an ABS may be classied as follows.
First, the vehicle braking-dynamics are nonlinear. Second,
there are many unknown environmental parameters, like road
coefcient of friction which may be wet, snowy or dry. Finally,
parameter changes due to mechanical wear and aging. Recently, a
great deal of research has been performed on the antilock brake
system (ABS) (Assadin, 2001; Nouilllant et al., 2002; von Altrock
Nov., 1997; Choi et al., 2002). Currently, most commercial ABSs
are based on look-up tabular approach (Chih-Min Lin and Hsu
March, 2003). These tables are calibrated through iterative
laboratory experiments and engineering eld tests.
The conventional PID controller for automated machines is
widely accepted by industry. According to a survey reported in Yu
(1999), more than 90% of control loops used in an industry use
PID. This is because PID controllers are easy to understand (has
clear physical meanings i.e. present, past and predictive), easy to
explain to others, and easy to implement. Unfortunately, many of
the PID loops that are in operation are in continual need of
monitoring and adjustment since they can easily become
improperly tuned (Passino and Yurkovich, 1998). Generally
speaking, in order to meet the demands of real time operation,
self-tuning is necessary.
Motivated by the success of fuzzy controllers in controlling
nonlinear, complex, time-varying dynamic processes in real
world, there has been steep increase in the research work on
the theoretical aspects of fuzzy logic controller (FLC). The main
reason is that FLCs essentially incorporate human expertise in the
control strategy, exploiting easier understanding of linguistic
interpretation.
Among the different types of FLC structures, PI-type and
PD-type FLCs are very common (Passino and Yurkovich, 1998;
Sharkawy et al., 2003). Other related works have employed
different adaptation policies to improve one or more performance
indices (Sharkawy, 2005; Mudi. and Pal, 1999; Xu et al., 1998).
However, development of PID-type FLCs was not popular because
they need the construction of three dimensional rule-base, which
complicates the design. Moreover to make PID-type FLC adaptive
in nature, the number of free adaptable parameters increases and
their interaction and interdependence further complicates the
situation (Mann et al., 1999).
The area of auto-tuning of PID controller using fuzzy systems
has attracted many authors (Xu et al., 1998; Mann et al., 1999;
S.-Z. et al., 1993; Visioli January, 2001; Bhattacharya et al., 2003;
Visioli, 1999; Macvicar-Whelan, 1976; Tzafestas and Papanikolo-
poulos, 1990; Marsh, 1998). A fuzzy self-tuning incremental PID
controller has been proposed by He et al. S.-Z. et al. (1993). The
controller implements a conventional PID structure, which starts
its operation with values of proportional gain, integral time, and
derivative time, obtained from the well-known ZieglerNichls
(ZN) tuning formula. This scheme implements a supervisory
fuzzy system, which adaptively changes the control parameter
after each sampling instant to improve the control performance.
In another study, a survey for tuning PID controllers with fuzzy
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/engappai
Engineering Applications of Articial Intelligence
0952-1976/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engappai.2010.06.011
E-mail address: ab.shark@aun.edu.eg
Engineering Applications of Articial Intelligence 23 (2010) 10411052
logic has been made by Visioli January (2001). In his work, several
PID control schemes have been simulated for linear systems.
Although in most cases, the controllers showed superior perfor-
mance for linear processes, they could only reduce the peak
overshoot at the expense of increasing rise time, and the
degradation becomes more and more signicant with increasing
time delay. This restricted the overall acceptance as a good
controller mechanism (Bhattacharya et al., 2003).
Referring to aforementioned works, one may conclude that
ZN method has been widely accepted as a base for tuning PID
controllers off- and on-line. This may be referred to the ability of
the method to preserve good load disturbance attenuation. Most of
these studies however have transformed the tuning problem from
using the ZN tuning parameters to other author-dened para-
meters and use fuzzy logic as the tuning-tool for the new
parameters; (S.-Z. et al., 1993; Visioli January, 2001; Bhattacharya
et al., 2003; Visioli, 1999; Macvicar-Whelan, 1976), [others]. This
indicates the lack of a generalized approach which can be followed
for linear and nonlinear plants. Furthermore, the suitability and
application range of the ZN method are very limited (Marsh,
1998; Zhuang and Atherton, 1993). For example, it is not suitable
for plants with a delay to time-constant ratio smaller than 0.15 or
larger than 0.6. This method also yields poor damping and high
sensitivity and does not achieve robustness of the closed-loop
when considerable parameter variations take place.
In this study, a self-tuning PID controller is proposed for the
ABS. Controlling the braking torque of an ABS is necessary to avoid
locking of the wheels, so that the driver can keep control on the
vehicles motion. Parameter variations and uncertainties imply
the need for an auto-tuning operation to achieve the performance
consistency. The article describes a generalized procedure for the
development of a simple, model free fuzzy PID-type structure as
an effective combination of three independent fuzzy systems.
Each PID parameter is tuned via rst order TakagiSugeno (TS)
fuzzy system, whose parameters are optimally determined off-
line using a modied genetic algorithm (GA). The control goal is to
keep the slipping ratio of the tires within the desired range, while
maintaining minimal stopping distance when braking is
Nomenclature
a
x
the vehicle body acceleration, m/s
2
a
i,j,k
i 1,2,3, j 1,2, y, 9, k1,2,3 are the coefcients of
the TS rst order output-model
A area
ABS antilock braking system
B
1
,B
2
fuzzy sets
c center of Gaussian membership function
c
1
, c
2
scaling factors
C
1
the maximum value of friction curve
C
2
the friction curve shape
C
3
the friction curve difference between the maximum
value and the value at l1
C
4
the wetness characteristic value
e(t) error of the closed-loop system
F objective function
F
IAE
objective function that uses the stopping distance and
IAE as the performance measures
F
ITAE
objective function that uses the stopping distance and
ITAE as the performance measures
F
N
the normal force, N
FSW PID controller using fuzzy set-point weighting
G number of generations
GA genetic algorithm
G-PID PID controller in which the three parameters are
determined using GAs
H number of iterations
IFE incremental fuzzy expert PID control
IAE integrated absolute error
ITAE integrated time multiplied by the absolute error
J
ITAE
performance index
J
IAE
performance index which uses IAE and the stopping
distance as the performance measure
J
ITAE
performance index which uses ITAE and the stopping
distance as the performance measure
J
w
the wheel inertia, kg m
2
k
1
,k
2
,k
3
constant parameters
K
d
the derivative gain
K
i
the integral gain
K
p
is the proportional gain
K
u
ultimate gain
m the quarter vehicle mass, kg
n number of rules
N negative
NB negative big
NS negative small
NVB negative very big
P positive
PB positive big
PS positive small
PVB positive very big
PID-Fuzzy PID controller tuned using initially guessed fuzzy
systems
PID-IAE PID controller tuned by GA using F
IAE
as the objective
function
PID-ITAE PID controller tuned by GA using F
ITAE
as the objective
function
R the wheel radius, m
R
e
low reproduction rate
R
h
high reproduction rate
S
x
stop distance, m
SSP PID controller using fuzzy self-tuning of a single
parameter
t
u
oscillation period
TS TakagiSugeno fuzzy systems
T
d
the derivative time constant
T
i
the integral time constant
u the braking torque, N m
V population size
V
x
the speed of the vehicle, m/s
w positive constant
w
i
ring strength of rule i
Z zero
ZN PID controller tuned by the ZieglerNichols tuning
method
a fuzzy tuning parameter
a
w
angular acceleration of the wheel, rad/s
2
g positive constant
Z
1
,Z
2
weighting factors
l the slip ratio
l
d
desired value of the slip ratio
m membership grade
m
r
road coefcient of friction
s slope of Gaussian membership function
o angular velocity of the wheel, rad/s
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1042
requested by the driver. With this approach, the PID controller
can be automatically adjusted to meet the system uncertainties
and achieve satisfactory response. Robustness against road
conditions is examined via numerical tests and results are
compared with previous works.
This paper is organized as follows. Section 2 presents the
mathematical model of an ABS based on a quarter car model.
Section 3 introduces previous works in the area of tuning PID
controllers. Section 4 demonstrates the proposed PID self-tuning
scheme based on rst order TS fuzzy system. Learning of the
fuzzy systems database is achieved via optimal performance
indices (tness functions), using a modied genetic algorithm.
Simulation results and discussions are given in Section 5. Section
6 offers our concluding remarks.
2. Modeling of an antilock braking system
2.1. The quarter vehicle dynamic model
To verify the control performance, this section demonstrates a
simplied model for the quarter vehicle dynamic motion. It was
derived by Assadin and Nouilllant (Assadin, 2001; Nouilllant et al.,
2002); Fig. 1. The nonlinear dynamics can be described as follows.
The force balance in the longitudinal direction
ma
x
m
r
F
N
1
The slip ratio is dened by
l
V
x
oR
V
x
2
Summing torques about the wheel center
J
w
a
w
um
r
RF
N
3
using Eqs. (1) and (2) and rearranging for
_
l yields
_
l
m
r
F
N
V
x
1l
m

R
2
J
w
_ _

R
J
w
V
x
u 4
Eq. (4) clearly shows that, during braking, the slip ratio is
dependent on the input torque u and the vehicle velocity V
x
. In
state space, the system state variables are: x
1
S
x
, x
2
V
x
, x
3
l,
where S
x
is the stopping distance. The state space equations are
_ x
1
x
2
_ x
2

m
r
F
N
m
_ x
3

m
r
F
N
x
2
1x
3
m

R
2
J
w
_ _

R
J
w
x
2
u 5
This state space model in Eq. (5) has been used in the
simulation tests to evaluate the performance of the ABS, using
different PID control strategies.
During braking, it is assumed that the wheel radius is constant.
Also, the vehicle speed V
x
and the wheel angular velocity o are
available signals through transducers mounted on suitable places.
So that the slip ratio l, is an available parameter for the ABS
closed-loop system.
2.2. Braking basics and problem denition
The ability of the ABS to maintain vehicle stability and
steerability, and still produce shorter stopping distances than
those from a locked wheel stop, comes from the relation of the
adhesion coefcient m
r
versus wheel slip ratio l. The friction
coefcient can vary in a very wide range, depending on factors
like:
(a) road surface conditions (dry or wet),
(b) tire side-slip angle,
(c) tire brand (summer tire, winter tire),
(d) vehicle speed, and
(e) the slip ratio between the tire and the road.
In this paper, the tire friction model introduced by Burckhardt
(1993) and adopted in Hari et al. (2008) has been used. It
provides the tire-road coefcient of friction m
r
as a function of the
wheel slip l and the vehicle velocity V
x
.
m
r
l,V
x
bC
1
1e
C
2
l
C
3
lce
C
4
lVx
6
The parameters in Eq. (6) denote the following: C
1
is the
maximum value of friction curve; C
2
the friction curve shape; C
3
the friction curve difference between the maximum value and the
value at l1; and C
4
is the wetness characteristic value and in the
range 0.020.04 s/m. Table 1 shows the friction model parameters
for different road conditions.
Dependence of the road friction coefcient on surface condi-
tions and slip ratio is shown in Fig. 2, (von Altrock Nov., 1997).
The lateral force is essential to the steering of vehicle. It is obvious
when slipping is equal to one; this force is equal to zero, which
explains why the steering ability is lost during wheel lockup. The
effective coefcient of friction between the tire and the road has
an optimum value which differs according to the road type and
the worst performance occurs at l1 (locked wheel). Most
manufacturers use a set point for the slipping ratio l
d
equal to 0.2,
which is a good compromise for all road conditions, (Chih-Min Lin
and Hsu March, 2003). So that the control problem can be
described as a set-point control system that may implement PID
controller; Fig. 3. Next section describes several PID strategies for
this closed-loop system.
Table 1
Friction model parameters (Burckhardt (Burckhardt, 1993)).
Surface conditions C
1
C
2
C
3
Dry asphalt 1.2801 23.99 0.52
Wet asphalt 0.857 33.822 0.347
Dry concrete 1.1973 25.168 0.5373
Snow 0.1946 94.129 0.0646
Ice 0.05 306.39 0
Vehicle Body
N
F
N r
F
u J
w

w
R
ma
x
x
V
Fig. 1. The quarter vehicle dynamic model.
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1043
3. PID tuning with fuzzy logic
In the early 1940s, after extensive manual experimentation by
the way of trial and error, Ziegler and Nichols invented the
well-known ZN formula for off-line tuning (Ziegler and Nichols,
1942). In time domain for example, the method yielded PID
coefcients directly from the three important parameters of a
stable plant to be controlled, namely the plant gain, time-constant
and transport delay. These parameters can be easily obtained
graphically from a step response of the open loop plant. The
concept of fuzzy systems instead may be used to emulate human
expertise and on-line tune the control gains using ZN formula.
The coming sub-sections demonstrate some of the previous works
in this area. They are used in later sections for comparison
purposes.
3.1. Standard and nonlinear PID controllers
There are many types of PID controllers, e.g., PID plus gravity
compensator, PID plus friction compensator, PID plus disturbance
observer, etc. (Astrom, 1996). Here, we shall consider the basic
forms for PID controller which is placed in a unity feedback
control system. A typical PID control law in its standard form is
ut K
p
etT
d
det
dt

1
T
i
_
t
0
etdt
_ _
7
where u(t) is the control variable, e(t)l
d
(t)l(t) is the system
error (difference between the demand input l
d
and the system
output, l), K
p
is the proportional gain, T
d
the derivative time
constant and T
i
is the integral time constant. Eq. (7) can be
rewritten as
ut K
p
et K
i
_
etdt K
d
det
dt
8
K
d
K
p
T
d
the derivative gain and K
i
K
p
/T
i
the integral gain.
In the design and tuning of a PID controller, the P, I and D
actions need to be coordinated. This is not a trivial process, since
coefcients of these three actions interact mutually and cannot be
simply tuned individually in a de-coupled manner. An experiment
is carried out on the process using ZN tuning method, can be
stated as follows (Yu, 1999; Ziegler and Nichols, 1942). First, the
process is controlled using the proportional gain K
p
. The value of
K
p
is slowly increased until continuous oscillations are happened.
At the time of oscillation, the values of the gain K
u
and the
oscillation period t
u
are noted. The method assumes that K
p
is 60%
of the gain at the time of oscillation. The integral time constant T
i
is 50% of the oscillation period t
u
and T
d
the derivative time
constant is 12.5% of the oscillation period. The ZN method is
devised for off-line tuning of continuous systems and can also be
used on discrete cases for a fast sampling time.
Nonlinear PID controllers have been widely considered in
literature, (Yu, 1999; Mann et al., 1999; Visioli January, 2001;
Tzafestas and Papanikolopoulos, 1990). A large class of nonlinear
PID controllers can be generalized as follows:
ut K
p
f e, _ eK
d
det
dt
K
i
_
t
0
etdt 9
where f e, _ e is a nonlinear function that depends on the closed-
loop error and the process delay (Mann et al., 1999). The f e, _ e
may be represented by a fuzzy logic system, in which the inputs
are e(t) and _ et (Tzafestas and Papanikolopoulos, 1990). It comes
out that the proportional gain depends on the current error and
other parameters contained on f e, _ e. Effectiveness of this non-
linear PID has been discussed in Mann et al. (1999) for linear
systems.
3.2. Fuzzy self-tuning of a single parameter (SSP)
Because of its simplicity, this scheme has been widely
considered in literature (S.-Z. et al., 1993; Visioli January, 2001;
Bhattacharya et al., 2003). The method devised by He et al. S.-Z.
et al. (1993), consists of parameterizing the ZieglerNichols
formula by means of a single parameter a, then using an online
fuzzy inference system to self-tune the parameter. In this method,
the three PID parameters can be expressed as
K
p
1:2atK
u
T
i
0:75
1
1at
t
u
T
d
0:25T
i
10
where K
u
and t
u
are the ultimate gain and ultimate period,
respectively. The value of a(t) is determined recursively with the
following equation:
at 1
atght1at for at40:5
atghtat for atr0:5
_
11
where g is a positive constant that has to be chosen in the range
[0.2,0.6] and h(t) is the output of the fuzzy inference system. The
fuzzy system has seven membership functions for each of the two
input (e and _ e) and seven for the output, i.e. the rule-base consists
of 49 rules. The initial value of a(t) is set equal to 0.5, which
corresponds to the ZieglerNichols formula. With respect to the
method however, the tuning of the scaling coefcient of the fuzzy
modules and of the parameter g is left to the user and no rule of
thumb is given for this task.
3.3. Fuzzy set-point weighting (FSW)
This approach proposed by Visioli (1999) consists of fuzzifying
the set-point weight, leaving xed the other three parameters
(again, determined with the ZieglerNichols method). In this way,
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1

Wet Road
Icy Road
Design Region
Lateral Force
Dry Road

r
Fig. 2. Coefcient of road friction versus wheel slip ratio.
d
u
Wheel
Brakes
(ABS)
PID
Fig. 3. The closed-loop control of ABS.
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1044
the control law can be written as
ut K
p
btl
d
tlt
_
K
d
det
dt
K
i
_
t
0
etdt
bt wf t 12
where w is a positive constant less than or equal to 1, and f(t) is
the output of the fuzzy inference system, which consists of ve
triangular membership functions for each of the two inputs e(t)
and _ et and nine triangular membership functions for the output.
The fuzzy rules are based on the Macvicar-Whelan matrix
(Macvicar-Whelan, 1976), as shown in Table 2. The linguistic
variables are negative small NS, negative big NB, negative very big
NVB, zero Z, positive small PS, positive big PB, and positive very
big, PVB.
The method however, is not a straightforward one and large
number of arithmetic operations is needed for on-line tuning.
3.4. Incremental fuzzy expert PID control (IFE)
Tzafestas and Papanikolopoulos (Tzafestas and Papanikolo-
poulos, 1990), introduced a procedure for scaling the values
of the three control parameters (initially determined by the
ZieglerNichols formula) during the transient period based on
the system error and its rate. In other words, the current values of
the proportional, integral and derivative gains are increased or
decreased by means of a fuzzy inference system according to the
following relations:
K
p
K
p
FSc
1
et,c
2
_ et k
1
K
i
K
i
FSc
1
et,c
2
_ et k
2
K
d
K
d
FSc
1
et,c
2
_ et k
3
13
where the basic tuning method is that of the ZieglerNichols;
FSc
1
et,c
2
_ et is the output of fuzzy inference system, based on
Macvicar-Whelan fuzzy rule matrix (Macvicar-Whelan, 1976)
(see Table 2), which reects the typical action of a human control.
For example, the integral action has to be increased at the
beginning of the transient response to decrease the rise time, and
then has to be decreased when the system error is negative to
reduce the overshoot. The range of the input membership
functions is normalized between 1 and 1. Finally, c
1
, c
2
are
scaling factors and k
i
, i 1,2,3 are constant parameters that
determine the range of variation of each term. The whole fuzzy
system involve 14 quantization levels for both e and _ e. Similar
approach has been followed in Bhattacharya et al. (2003), where
the authors have to rely on trial and error procedure in order to
identify parameters of the fuzzy systems.
However, tuning the three parameters k
i
and the two scaling
factors (c
1
, c
2
) that multiply the inputs e and _ e, is left to the
user, and it is not clear how these parameters inuence the
performance of the overall controller.
4. The proposed fuzzy self-tuning PID control scheme
The proposed control scheme does not rely on ZN method nor
on a previous rule-base, especially designed for some plants.
Instead, it uses rst order TS fuzzy systems as the tuning-tool for
each of the PID control modules. A modied genetic algorithm is
used to optimally select parameters of the fuzzy systems.
The main control objectives of an ABS are:
minimization of the stopping distance S
x
, and
maintaining the fastest possible response with no/low over-
shoot and close to zero steady state error.
The rst objective is a key issue related to ABS, while the
second is a common objective for any closed-loop control system.
In this work, optimality of the second objective is based on
conventional cost functions; i.e. integrated absolute error (IAE)
and the integrated time multiplied by the absolute error (ITAE).
For example, in an IAE, the hatched area in Fig. 4 represents the
cost function. A minimum area is achieved at the fastest
physically possible response, no/small overshoot and close to
zero steady state error. It means that decreasing overshoot will
not result in an increase in the rise time or vise versa, as it is the
case with ZN method. The expected tuning result is, simply, the
best physically possible response.
In short, the objective of an ABS design is to minimize a
performance index that contains the stopping distance and an IAE
or ITAE.
4.1. The control system architecture
Motivated by the work of Bhattacharya et al. (2003); Tzafestas
and Papanikolopoulos (1990), three decoupled fuzzy systems
constitute the proposed self-tuning system; each for one para-
meter of the PID controller, i.e. K
p
, K
i
and K
i
; Eq. (8). The error e
and change in error _ e are used as behavior-recognizers of the
closed-loop performance. They are available signals in the closed-
loop system of the ABS and do not require extra hardware. The
self-tuner can be expressed as
K
p
FS
1
et, _ et
K
i
FS
2
et, _ et
K
d
FS
3
et, _ et 14
where a fuzzy P controller, a fuzzy I controller and a fuzzy D
controller are connected in parallel to give the resultant controller
signal. With this structure, independent control actions can be
generated, which should necessarily eliminate the problems
associated with most practical two-terms or three-terms FLCs.
The basic approach is summarized in Fig. 5, where each fuzzy
system is trying to recognize when the corresponding parameter
E
r
r
o
r
,

e
(
t
)

0
) ( dt t e A
0
A
Time, sec
Fig. 4. A typical closed-loop error time-history.
Table 2
Rule table of the FSW (Macvicar-Whelan, 1976).
f(t) Change in error, _ e
NB NS Z PS PL
Error, e NB NVB NB NM NS Z
NS NB NM NS Z PS
Z NM NS Z PS PM
PS NS Z PS PM PB
PB Z PS PM PB PVB
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1045
is not properly tuned and then seeks to adjust it to obtain
improved performance. In such a way, each fuzzy system can be
looked at as gain scheduler (module). TS type fuzzy systems are
used to synthesize each of the self-tuning systems.
The TS fuzzy system (also known as functional fuzzy system
(Passino and Yurkovich, 1998)) was proposed in an effort to
develop a systematic approach to generating fuzzy rules from a
given inputoutput data set; (Takagi and Sugeno, 1985; Jang et al.,
1997). A typical rule has the following form
IF x
1
IS B
1
AND x
2
IS B
2
THEN z f x
1
,x
2

where B
1
and B
2
are fuzzy sets in the antecedent, while zf(x
1
,x
2
)
is a crisp function in the consequent. With this form, the fuzzy
system can be characterized as two input one output fuzzy
systems.
Usually f(x
1
,x
2
) is a polynomial in the input variables x
1
and x
2
,
but it can be any function as long as it can appropriately describe
the output of the model within the fuzzy region specied by
the antecedent of the rule. Although there are no restrictions on
the form of the input membership functions, Gaussians are used
in the premise through out this work. A Gaussian membership
function is specied by two parameters (c,s)
m
B
l
j
x
j
gaussianx
j
;c,s exp
1
2
x
j
c
s
_ _
2
_ _
15
where m is the membership grade, c represents the membership
functions center, s determines its spread; B is the membership
which represents a linguistic variable, i 1,2, y, n is the rule
number, j 1,2 is subscript of the input variables.
In the proposed self-tuner, the inputs e and _ e are normalized
using three Gaussian membership functions; negative N, zero Z,
and positive P. So that nine rules constitute the rule-base for each
module. For simplicity, the consequent part has been chosen to be
a rst order function of e and _ e. So that the rules have the
following form:
Rule
j
: IF e IS B
1
AND e IS B
2
THEN K
i
a
i,j,1
ea
i,j,2
_ e a
i,j,3
16
where both B
1
and B
2
are positive (P), zero (Z) or negative (N),
K
i
f e, _ e is the gain to be tuned, i.e. K
p
, K
i
or K
d
. a
i,1,j
, a
i,2,j
are the
constants, and j 1,2, y, 9 is the rule number.
Fig. 6 shows the fuzzy reasoning procedure for a rst order TS
model. The fuzzy part is only in its antecedent. Each rule has a
crisp output and the overall output is obtained via weighted
average. This fuzzy procedure avoids the time-consuming process
of defuzzication required in a Mamdani fuzzy model. The
weighted sum has been also used in Jang et al. (1997) to further
reduce computations.
With the above structure, the data-base of each module
consists of 27 free parameters (coefcients of the rst order
polynomial) and 12 free parameters for the inputs membership
functions. So that, the total number of free parameters is 117. If an
identical input membership is chosen for the three modules, the
total number of parameters is reduced to 93; i.e. parameters of
the input membership functions c
N1
, s
N1
, c
Z1
, s
Z1
c
P1
, s
P1
for e, c
N2
,
s
N2
, c
Z2
, s
Z2
, c
P2
, s
P2
for _ e, and a
ijk
where i 1,2,3 are the
coefcients of the rst order polynomial, j 1,2, , 9 the rule
number for each tuner and k1,2,3 the number of modules.
Determination of optimal values for these parameters is the
subject of the next sub-section.
4.2. Genetic algorithm-based parameter learning
GAs are optimization stochastic technique mimicking the
natural selection, which consists of three operations, namely,
reproduction, crossover, and mutation (Jang et al., 1997). The
most general considerations about GA can be stated as follows:
(i) The searching procedure of the GA starts from multiple initial
states simultaneously and proceeds in all of the parameter
subspaces simultaneously.
(ii) GA requires almost no prior knowledge of the concerned
system, which enables it to deal with completely unknown
systems that other optimization methods may fail.
(iii) GA cannot evaluate the performance of a system properly at
one step. For this reason, generally, it cannot be used as an
on-line optimization strategy and is more suitable for fuzzy
modeling.
In practice, training data can be obtained by experimentation
or by the establishment of an ideal model. In this work, the ABS
model in Section 2 is used to emulate the behavior of the ABS, in
order to collect training data. Fig. 7 shows the training process for
each fuzzy module involved in the self-tuning system.
d

e
.
dt
d

dt
dt
d

Update Kp, K
i
and K
d
d
K
p
K
i
K
e
e Fuzzy System
1
Fuzzy System
2
Fuzzy System
3
ABS
+
+
+
+
Fig. 5. The self-tuning fuzzy PID controller.
Product inference
P
N
Weighted
Average
......
......
Fig. 6. TakagiSugeno fuzzy system for a module.
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1046
The following two closed-loop performance indices have been
examined. The rst uses the stopping distance and the integrated
absolute error (IAE). The second uses the stopping distance and
the integrated time multiplied by the absolute error (ITAE). These
two performance indices are dened as follows:
J
IAE
Z
1
S
x
Z
2
IAE Z
1
S
x
Z
2
_
T
0
et

dt
Z
1
S
x
Z
2

M
k 0
9l
d
lk9Dt 17
J
ITAE
Z
1
S
x
Z
2
ITAE Z
1
S
x
Z
2
_
T
0
t9et9dt
Z
1
S
x
Z
2

M
k 0
k9l
d
lk9Dt
2
18
where el
d
l is the closed-loop error, l
d
is the desired slip ratio,
l(k) is the current slip ratio and Dt is the time step. M is the
number of training samples. In addition, the coefcients Z
1
and Z
2
are the weighting factors to emphasize the relative importance of
the associated terms.
Because GA endeavors to maximize the tness function, the
tness function of each gene is calculated as follows:
F
1
1J
19
where J is the performance index and 1 is introduced at the
denominator to prevent the tness function from becoming
innitely large. The overall training procedure is shown in Fig. 8.
To simplify the presentation, let us denote F in Eq. (19) as F
IAE
when J J
IAE
and F
ITAE
when J J
ITAE
. With this notation, the
controller is called PID-IAE when the performance index is
calculated by Eq. (17) and PID-ITAE when the performance index
is given by Eq. (18).
Coding of the parameters to be adjusted can be stated as
follows:
c
N1
,c
N2
, , s
N1
,s
N2
, , a
111
,a
121
, , a
339
20
where a certain number of binary bits stands for each parameter.
The combined string composes a gene (possible solution) in a
population. Evaluation of each possible solution is performed via
IAE/ITAE and genes of best solutions are allowed to reproduce.
Although, genetic algorithms were developed a few decades
ago, concrete theoretical analysis of the algorithm have not been
provided until recent years (Golberg, 1989; Rudolph, 1994).
Reference (Rudolph, 1994) concludes that the canonical GA
cannot always nd the optimal solution within the denite time.
Furthermore, the paper pointed out that if the chromosome with
the best performance in each generation is reserved for the next
generation, the algorithm will globally converge. Inspired by these
conclusions, the following two measures have been considered in
this work:
(i) In reproduction, we stochastically introduce a randomly
generated gene at a probability of R
h
to replace one of the
two parents selected for reproduction.
(ii) Select the best performed genes in the current population at a
rate of R
e
and place them directly in the next generation.
If the reproduction is carried out in the traditional way, the
best gene will globally be lost, and thus convergence cannot
be guaranteed. However, if we only adopt the second measure,
the procedure of the evolution is no longer a Markov process, and
thus it does not satisfy the assumption of the convergence theory.
Despite our successful application of these measures, mathema-
tical analysis of them still lacks. The parameters R
h
and R
e
are
adjusted such that at the beginning of the learning, R
h
is relatively
large and R
e
is relatively small, and later, vice versa.
Fuzzy Expert
ABS PID
Genetic Fuzzy
Optimizer
Modify membership
functions
Modify coefficients of
the consequent parts
Update the gains
d

Search for maximum

fitness (performance)
Performance
Index
e e,
IAE/ITAE
Fuzzy
System 1
Fuzzy
system 3
Fuzzy
System 2
p
K
i
K
d
K
Fig. 8. Genetic training of the overall fuzzy PID self-tuning system.
GA
Fuzzy system
Fig. 7. Genetic learning of the data-base for each module.
Table 3
Genetic parameter settings.
Genetic parameter value
Number of generations, G 150
Population size, V 100
Crossover probability R
h
,R
e
0.90, 0.60
Mutation Probability 0.01
Bit number for each variable 32
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1047
Fig. 9. Membership functions of the antecedent part before tuning.
Fig. 10. Membership functions of the antecedent part after tuning using F
ITAE
.
Fig. 11. Membership functions of the antecedent part after tuning using F
IAE
.
Fig. 12. The output surfaces of the fuzzy gain-tuners.
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1048
5. Numerical tests
5.1. Simulation data
Several numerical tests have been performed using the
example data presented in Chih-Min Lin and Hsu March (2003).
They are: R0.33 m, m342 kg, J
w
1.13 kg m
2
, g9.81 m/s
2
and
the desired slip ratio l
d
0.2. Due to the fact that when the wheel
and vehicle velocity are nearly zero at the end of braking time, the
magnitude of slip tends to innity. Therefore, simulations are
conducted up to the point when the vehicle is slowed to 0.5 m/s.
The following case study deals with braking on dry asphalt, then
after 1 s, the road changed to be a snowy one. The initial speed
V
x
27.78 m/s, i.e. 100 km/h. These conditions have been considered,
in order to examine robustness against road conditions.
To verify the full potentialities of the investigated PID
controllers, during genetic training, it is assumed that no saturation
levels are imposed to the control signal. After training, a saturation
level u
s
74000 N m has been imposed to test the controllers in
the presence of typical process nonlinearity. Genetic training has
been performed using the parameter settings listed in Table 3.
The tuning procedure has been initialized using the membership
functions shown in Fig. 9. After training, the obtained membership
functions using F
ITAE
and F
IAE
are depicted in Fig. 10 and 11,
respectively. The corresponding controllers are denoted PID-Fuzzy
(before training), PID-IAE and PID-ITAE, respectively. The output
surfaces of the gain self-tuners, before and after learning, are depicted
in Fig. 12. It should be noted that the shown universe of discourse of e
(from E2.5 to E2.5) is not fully used by the controllers. The aimof
displaying the membership functions within this range is to fully
demonstrate their diversities. This also applies to the shown universe
of discourse of _ e.
Referring to Fig. 10 and 11, it is shown that the negative (N)
membership functions of both e and _ e are widely spread through
the whole universe of discourses. It means that they have the
largest impact in the computed control signal. This may explain
why the PID-IAE and PID-IAE controllers exhibit faster conver-
gence relative to PID-fuzzy controller, as will be shown in Section
5.3. To some extent, this can be attributed to that the closed-loop
cycle (braking) starts with a positive error e (el
d
l) equal to
0.2 leading to a negative overshoot at the transient period.
This negative overshoot (e) invokes large control signal which
counter-attacks this error, resulting in a faster convergence of the
error relative to the PID-fuzzy controller.
For the other investigated methodologies (SSP and FSW
described in Section 3), the control parameters have been
determined using the GA with F
IAE
as the objective function. For
the sake of brevity, only the membership functions of the
proposed self-tuning scheme are reported here. Furthermore,
the genetic algorithm with the parameters listed in Table 3 has
been used to retune the PID parameters. It is intended for the
resulted PID to use the best possible xed parameters. This PID
controller is denoted by G-PID.
5.2. Computational considerations
The initial fuzzy modules were determined after small number of
trials. It is our point of view that the initially guessed modules should
work properly for successful learning process. Dealing with genetic
learning, the total number of iterations H should be chosen so as the
system has enough chance to converge. A suitable choice ensures
correct training and saves computation time since each gene is a
possible solution which has to be evaluated according to Eq. (19). In
literature, two approaches are generally used for selecting the
suitable number of population and generation for optimization
problems similar to the problem considered in this work. The rst
relies on a relatively small number of populations (e.g. 20-30) and
high number of generations (e.g. 700-3000) (Hwang, 1999;
Rahmoun and Benmohamed, 1998; Skarmeta and Jimenez, 1999).
The second uses high number of populations (e.g. 100-200) and low
number of generations (Attia and Horacek, 2001). We have chosen
the former to assign the suitable genetic parameters. Adaptive change
of the crossover probability helps more in speeding up the
convergence. The coming results have been obtained using genetic
parameters listed in Table 3.
With a population of V100 individuals for G150 generations,
the tness function in Eq. (19) is evaluated 15,000 times. Indeed,
Fig. 13. Slip ratio under different control laws.
Table 4
Performance of PID tuning methods.
Controller type Performance measure
S
x
, m IAE ITAE Maximum O.S%
ZN 51.61 0.1171 0.0909 73.24
G-PID 48.77 0.0435 0.0210 13.52
SSP 47.28 0.0526 0.0309 31.61
FSW 48.06 0.0998 0.0494 0.05
PID-fuzzy 45.97 0.0482 0.0297 13.16
PID-ITAE 41.82 0.0442 0.0240 15.35
PID-IAE 39.92 0.0433 0.0291 12.84
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1049
this number (HVG) represents the number of evaluated points
inside the search space, which may be used as a reference for
similar optimization problems. The wining gene (optimal solution)
is the best of stochastically competitive 15,000 genes. Referring to
Eq. (20), with 32 bits for each variable, each possible solution
(gene) has the length of 2976 bits. Performing the learning process
using Matlab-7, M-le under Windows XP on a PC Pentium IV,
2800 Hz speed, requires about one and half hour. Due to the
stochastic nature of GAs, at least two or three trials should be
performed, in order to be sure that convergence has taken place.
5.3. Results and discussion
Fig. 13 shows the ABS response under seven investigated
control schemes. As it can be noticed, PID-IAE gives the best
response compared with other schemes. To provide a more
detailed insight of the results, Table 4 gives the values of the
performance measures S
x
, IAE, ITAE, and maximum attained
overshoot for the different PID controllers. G-PID and PID-fuzzy
show similar performances, while PID-fuzzy exhibits shorter
stopping distance. With respect to constant gain schemes, G-PID
shows better results than ZN, which exhibits the highest
overshoot, longest settling time and stopping distance; i.e. the
worst performance. Nevertheless, the shortest stopping distance
has been taken place by the PID-IAE controller.
The barking torques have signicantly changed their
magnitude after one second to meet the new road condition
(snowy road); Figs. 14(a) and (b). Low braking torque is required
for the vehicle to move on icy road, in order to avoid locking.
Fig. 15 shows different stopping times have been achieved by the
Fig. 14. The input torques.
Fig. 15. Performance of ABS under three PID strategies.
Fig. 16. Time-history of the PID-IAE gains (when input saturation is imposed).
Fig. 17. Slip ratio under different control schemes when saturation on the input is
not imposed.
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1050
investigated controllers. Larger overshoot can be noticed by ZN
and an SSP with respect to PID-IAE.
Time-history of the PID-IAE modules is depicted in Fig. 16.
Unlike constant gain schemes (e.g. ZN), their values are
continuously changing during the braking period, which is
imposed by the fuzzy modules, in order to counter attack the
error and change in error. After one second, smooth transition on
their values has taken place, in order to meet the new road
demands. This behavior resulted in a relatively smoother torque,
faster convergence of the slip ratio to the desired value and
minimum stopping distance.
Because the input saturation level is signicantly high, it may
be interesting for the study to investigate the response when no
saturation is imposed. Fig. 17 shows the response of 4 PID
schemes. It comes out that fuzzy logic is more useful if saturations
are signicant in the process. This remark strengthens the fact
that the performance of the classic PID can be improved using
time-varying parameters (Astrom 1996). Fig. 18 depicts the
PID-IAE gains when saturation is not imposed. Higher variations
can be noticed at the beginning of the braking. Table 5 gives the
performance measures of the investigated seven controllers.
PID-IAE exhibits the shortest stopping distance although it
exhibits much higher overshoot than the case when saturation
is imposed. As a performance measure, the percent overshoot
when input saturation is and is not imposed is depicted in Fig. 19.
It is clear that saturation has little or no effect on the xed gain
schemes, i.e. ZN and G-PID. On the other hand, PID-IAE and
PID-Fuzzy have been greatly inuenced.
Finally, one should consider the case of manual tuning of the
controllers i.e. the fuzzy modules parameters, since in practical
industrial applications; the use of genetically trained controllers is
possible only if a good process model (or experimental data) is
available. It emerges that for the fuzzy-logic-based tuning
methods for which a technique for the selection of the parameters
of the fuzzy logic has not been considered, it might be difcult to
perform this task effectively. For example, for the IFE method
(Sub-Section 3.4), the choice of parameter k
3
is critical since its
value has to be kept very low, otherwise the overall control
system is destabilized. Also, the SSP control structure is very
difcult to set. Otherwise, it is not always easy to improve the ZN
response and, when it succeeds the selected rules is very difcult
to interpret because other tuning parameters interfere (g and a).
Manual tuning of the proposed scheme is straightforward since it
depends on three decoupled modules, which is advantageous over
most of the investigated schemes.
6. Conclusions
In this article, a fuzzy self-tuning scheme has been proposed
for PID controllers. The proposed scheme utilizes three decoupled
modules, each for one of the PID parameters. Each module is two-
input one-output TS type fuzzy system. Optimal selection of the
fuzzy modules has been obtained using a modied genetic
algorithm. The performance has been veried using the auto-
mobiles ABS. In the presented case study, robustness against road
friction characteristics has been considered. The control goal is to
keep the slip ratio to an assigned value (0.2) despite road friction
characteristics, while keeping the shortest possible stopping
distance. Comparison with previous works shows the competi-
tiveness of the proposed scheme.
The salient features can be summarized as follows:
1. The proposed control scheme presents a generalized proce-
dure, which can be followed for linear and nonlinear systems.
2. Improved control action has been obtained by replacing the
ZN tuned PID controllers with fuzzy self-tuning systems. This is
the case in SSP and FSW proposed by earlier investigators, and
further improvement can be obtained by using the proposed
control scheme.
3. The proposed PID-IAE controller resulted in relatively fast
response with low overshoot and short stopping distance. As a
consequence, there is a remarkable improvement with regard
to the examined performance measures.
4. The proposed scheme has the ability to switch on/off any of the
control actions, due to its basic decoupled nature. This
procedure is needed for practical implementation because
of the drift which usually exists between theoretical
establishment and actual experimentation.
5. A probable area of future work is to achieve adaptivity of each/
some of the three control actions, while marinating its simplicity.
Fig. 18. Time history of the PID-IAE gains when saturation is not imposed.
Table 5
Performance of PID tuning methods when saturation is not imposed.
Controller type Performance measure
S
x
, m IAE ITAE Maximum O.S.%
ZN 50.16 0.1142 0.0918 69.60
G-PID 46.62 0.0282 0.0188 18.59
SSP 45.55 0.0477 0.0318 31.34
FSW 46.14 0.0837 0.0588 no O. S.
PID-fuzzy 44.41 0.0661 0.0336 122.82
PID-ITAE 41.85 0.1246 0.0471 265.60
PID-IAE 38.83 0.0784 0.0303 193.50
Fig. 19. Percent overshoot with and without input saturation.
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1051
References
Assadin, F., Mixed H
N
and Fuzzy Logic controllers for the automobile ABS PSA
Peugeot-Citroen, SAE 2001 World congress, March 58, Detroit, Michigan, USA,
2001.
Astrom, K.J., 1996. The Control Handbook. CRC Press/IEEE Press.
Attia, A.A.and Horacek, Adaptation of Genetic Algorithms, In: Proceedings of the
Mendel2001, Brno, Czech Republic, pp. 3641, 2001.
Bhattacharya, S., Chatterjee, A., Munshi, S., 2003. An improved PID-type fuzzy
controller employing individual fuzzy P, fuzzy I and fuzzy PID controllers.
Trans. Inst. Meas. Control 25 (4), 352372.
Burckhardt, M., 1993. Fahrwerktechnik: radschlupf-regelsysteme. Vogel-Verlag,
Wurtzburg.
Choi, S-B., et al., 2002. Sliding mode control for anti-lock brake system of
passenger vehicles featuring electrorheological valves. Proc. Inst. Mech. Eng.,
Vol 216 Part D: J. Automob. Eng.
Chih-Min Lin, Hsu, Chun-Fei, March 2003. Self-learning fuzzy sliding-mode control
for antilock braking systems. IEEE Trans. Control Syst. Technol. 11 (2).
Golberg, D.E., 1989. Genetic Algorithm in Search, Optimization and Machine
Learning. Addison-Wesley, Reading, MA.
Hari, A., Aghagolzadeh, A., Alizadeh, G., Sadeghi, M., 2008. Designing a Sliding
Mode Controller for Slip Control of Antilock Brake Systems. Transp. Res. Part C
16, 731741.
Hwang, H.S., 1999. Automatic design of fuzzy rule base for modeling and control
using evolutionary programming. IEE Proc. Control Theory Appl. 146 (1), 916.
Jang, J.-S.R., Sun, C.-T., Mizutani, E., 1997. Neuro-fuzzy and soft computing: a
computational approach to learning and machine intelligence. Printice-Hall
International, Inc.
Mudi, R.K., Pal, N.R., 1999. A robust self-tuning scheme for PI- and PD-Type fuzzy
controllers. IEEE Trans. Fuzzy Syst. 7, 26.
Mann, G.K.I., Hu, B.G., Gasine, R.G., 1999. Analysis of direct action fuzzy PID controller
structures. IEEE Trans. Syst. Man Cybern.Part B: Cybern. 29, 371388.
Macvicar-Whelan, P.J., 1976. Fuzzy sets for manmachine interaction. Int. J. Man
Mach. Stud. 8, 687697.
Marsh, P., 1998. Turn on, Tune in. New Electron. 31 (4), 3132.
Nouilllant, C., et al., 2002. A cooperative control for car suspension and brake
systems. Int. J. Automot. Technol. 3 (4), 147155.
Passino, K., Yurkovich, S., 1998. Fuzzy Control. Addison-Wesley Longman, Inc.
Rudolph, G., 1994. Convergence analysis of canonical genetic algorithms. IEEE
Trans. Neural Network 5 (1), 96101.
Rahmoun, A., Benmohamed, M., 1998. Genetic algorithm based methodology to
generate automatically optimal fuzzy systems. IEE Proc.-Control Theory Appl.
145 (6), 583586.
Sharkawy, A.B., El-Awady, H., Moustafa, A.F., 2003. Stable fuzzy control for a class
of nonlinear systems. Trans. Inst. Meas. Control 25 (3), 265278.
Sharkawy, A.B., 2005. Fuzzy and adaptive fuzzy control of active suspension
systems. Vehicle Syst. Dyn. 43 (11), 795806.
S.-Z., H.E., Tan, S., Xu, F.-L., 1993. Fuzzy self-tuning of PID controller. Fuzzy Sets
Syst., 3746.
Skarmeta, A.F., Jimenez, 1999. Fuzzy modeling with hybrid systems. Fuzzy Sets
Syst. 104, 199208.
Tzafestas, S.G., Papanikolopoulos, N.P., 1990. Incremental fuzzy expert PID control.
IEEE Trans. Ind. Electron. 37 (5), 365371.
Takagi, T., Sugeno, M., 1985. Fuzzy identicatiof syspems and its applications to
modeling and control. IEEE Trans. Syst. Man Cybern. 15, 116132.
Visioli, A., January 2001. Tuning PID controllers with fuzzy logic. IEE Proc. Control
Theory Appl. 148 (1).
Visioli, A., 1999. Fuzzy logic based set-point weighting for PID controllers. IEEE
Trans. Syst. Man Cybern.Part A 29, 587592.
von Altrock, C., November 1997. Fuzzy logic in automotive engineering. Circuit
Cellar, Comput. Appl. J. 88, 19.
Xu, J-X., Lin, C., Hang, C.C., 1998. Tuning and analysis of a fuzzy PI controller based
on gain and phase margin. IEEE Trans. Syst. Man Cybern.Part A: Syst. Hum.
28, 685694.
Yu, C.C., 1999. Autotuning of PID Controllers: Relay Feedback Approach. Springer.
Zhuang, M., Atherton, D.P., 1993. Automatic tuning of optimum PID controllers.
Proc. Inst. Elect. Eng.Control Theory Appl. 140, 216223.
Ziegler, J.G., Nichols, N.B., 1942. Optimum settings for automatic controllers, 64.
Trans ASME, pp. 759768.
A.B. Sharkawy / Engineering Applications of Articial Intelligence 23 (2010) 10411052 1052