You are on page 1of 14

IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 17, NO.

2, APRIL 2009 329


A Hybrid Approach for Design of Stable Adaptive
Fuzzy Controllers Employing Lyapunov Theory and
Particle Swarm Optimization
Kaushik Das Sharma, Amitava Chatterjee, and Anjan Rakshit
AbstractThis paper proposes a new approach for designing
stable adaptive fuzzy controllers, which employs a hybridization
of a conventional Lyapunov-theory-based approach and a parti-
cle swarm optimization (PSO) based stochastic optimization ap-
proach. The objective is to design a self-adaptive fuzzy controller,
optimizing both its structures and free parameters, such that the
designed controller can guarantee desired stability and can simul-
taneously provide satisfactory performance. The design methodol-
ogy for the controller simultaneously utilizes the good features of
PSO (capable of providing good global search capability, required
to provide a high degree of automation) and Lyapunov-based tun-
ing (providing fast adaptation utilizing a local search method).
Three different variants of the hybrid controller are proposed in
this paper. These variants are implemented for benchmark simu-
lation case studies and real-life experimentation, and their results
demonstrate the usefulness of the proposed approach.
Index TermsAdaptive fuzzy logic controllers (AFLCs), hybrid
approaches, Lyapunov theory, particle swarmoptimization (PSO).
I. INTRODUCTION
F
UZZY logic controllers (FLCs) evolved as a popular do-
main of control system and control engineering more than
two decades ago [17]. In recent times, more and more attention is
being paid to systematic design of FLCs from the point of view
of stable behavior and optimization of performance indexes.
Several such stable and optimized adaptive FLCs have been
proposed utilizing, e.g., neural network [1], Lyapunov meth-
ods [2], [21], genetic algorithm (GA) [3], [4], ant colony opti-
mization [14], etc.
Lyapunov-based strategies have been utilized in conventional
control theory for a long time to design controllers that can
guarantee closed-loop stability of the system [21][24]. In [5]
and [6], two of the earliest design methods of Lyapunov-based
direct adaptive fuzzy controllers were proposed. The main prob-
lem in Lyapunov-based method is that it requires a considerable
amount of time to track the reference input and also the inherent
approximation error in striving to achieve an unknown theoreti-
Manuscript received November 29, 2007; revised September 15, 2008; ac-
cepted December 1, 2008. First published January 6, 2009; current version pub-
lished April 1, 2009. This work was supported by the University Grants Commis-
sion, India, under Major Research Project Scheme [Grant 32-118/2006(SR)].
K. D. Sharma is with the Department of Electrical Engineering, Future Insti-
tute of Engineering and Management, West Bengal University of Technology,
Kolkata 700150, India, and also with Jadavpur University, Kolkata 700032,
India (e-mail: kaushikdassharma@yahoo.com).
A. Chatterjee and A. Rakshit are with the Department of Electrical Engineer-
ing, Jadavpur University, Kolkata 700032, India (e-mail: cha_ami@yahoo.co.in;
anjan_rakshit@yahoo.com).
Digital Object Identier 10.1109/TFUZZ.2008.2012033
cal control law leads the system toward larger integral absolute
error (IAE) [2]. A major drawback of many existing designs
of adaptive FLCs (AFLCs) is that they are developed for sys-
tems with unlimited actuation authority [21]. In [5], one of the
earliest methods to propose Lyapunov-based learning for stable
AFLCs, the controller had a serious drawback that it needed
adaptation of controller parameters for every variation in refer-
ence signal, which implied that for temporally varying reference
signals, the controller parameters would never converge. In [6],
it was demonstrated that a modied form of stable AFLC can
be proposed where the controller parameters would converge
even in the presence of a temporally varying reference signal.
However, the controllers in [6] were proposed based on control
laws adapting a parameter vector containing the positions of the
fuzzy sets of the controller outputs only. The basic structure of
the FLC was assumed to be a static one. A further improvement
of this scheme was proposed in [12], where a GA-based strategy
was hybridized with a Lyapunov-based strategy to adapt both
controller structures and free parameters, containing center lo-
cations of input membership functions (MFs), scaling gains,
learning rate, etc., in addition to the positions of the singletons
describing output fuzzy sets of the FLC. But the problem asso-
ciated with this hybrid model is that this GA-based Lyapunov
method needs several thousands of tness evaluations to achieve
parameter convergence and guarantee closed-loop stability.
This paper proposes a systematic design of stable AFLCs
employing a hybridization of the Lyapunov-strategy-based ap-
proach (LSBA) and particle swarm optimization (PSO) based
approach, called the PSO-based approach (PSOBA) [8], [9],
[11]. It attempts to combine strong points of both methods to
evolve a superior method. The objective of this design strategy
is to perform simultaneous adaptation of both the FLC struc-
ture [15], [22] and its free parameters, so that two competing
requirements can be fullled: 1) to guarantee stability of the con-
troller design and 2) to achieve very high degree of automation
in the process of controller design by employing a global search
method. The LSBAhelps to guarantee closed-loop stability with
improved speed of convergence. On the other hand, the PSOBA
helps to achieve desired automation for the design methodology
of the AFLC by getting rid of many manually tuned parameters
in LSBAdesign procedure. However, a well-known drawback of
these global optimization techniques is that they are essentially
slow in their search procedure to optimize a criterion function.
This motivated us to choose PSOamong these available stochas-
tic methods because of their simple functioning, and generally,
they can provide relatively faster, satisfactory solution.
1063-6706/$25.00 2009 IEEE
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
330 IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 17, NO. 2, APRIL 2009
In this paper, three types of hybridization of locally opera-
tive LSBA and a global version of PSOBA have been proposed
and simulated for benchmark nonlinear plants and also tested
for a real-life experiment. These approaches are called hybrid-
adaptation-strategy-based approaches (HASBA). The rst two
hybrid approaches are the cascade and concurrent combinations
of LSBAand PSOBA. In cascade hybrid approach, the structure
of the controller and the values of the free parameters for LSBA
are determined by the PSOBA, i.e., the tedious manual tuning
part of the conventional LSBA is replaced. So, the drawbacks
of conventional LSBA are not fully taken care of. In concurrent
hybrid approach, the LSBA performs within the supervision of
PSOBA, with the objective of keeping the inherent approxima-
tion error limited and the tracking of the input reference signal,
may be a temporal one, and is expected to be good enough.
In some cases, our experimentations show that PSOBA alone
performs better than the concurrent hybrid approach. Hence, we
have proposed a third hybrid strategy, called preferential hybrid
approach. In this case, the movements of the particles in the
multidimensional global search space are sometimes governed
by the concurrent hybrid method and sometimes by the PSOBA
method, depending on the performance of the corresponding
tness evolutions. Therefore, the best candidate solution in this
preferential hybrid approach evolves from a second level of hy-
bridized process that, according to our knowledge and belief,
is a novel idea to design direct AFLCs. The proposed hybrid
approaches have all been implemented for different benchmark
case studies and also experimented on a real-life problem, and
the three variants of HASBA are compared with controllers
designed by LSBA and PSOBA alone. The obtained results il-
lustrate, on the whole, the superiority of the preferential hybrid
approach over the other approaches.
The rest of the paper is organized as follows. Section II
presents a discussion about the design of stable adaptive fuzzy
controllers. Section III presents the conventional adaptation
strategies and the newly proposed hybrid adaptation strategies,
while in Section IV, simulation and experimental case studies
for both conventional and proposed hybrid adaptation strategies
are demonstrated. Section V concludes the paper.
II. DESIGN OF STABLE ADAPTIVE FUZZY CONTROLLERS
Let us consider that our objective is to design an adaptive strat-
egy for an nth-order single-inputsingle-output (SISO) nonlin-
ear plant given as
x
(n)
= f(x) + bu
y = x
_
(1)
where f() is an unknown continuous function, u R and
y Rare the input and output of the plant, and b is an unknown
positive constant. We assume that the state vector is given as
x = (x
1
, x
2
, . . . , x
n
)
T
= (x, x, . . . , x
(n1)
)
T
R
n
. The sys-
tem under control has a reference model given as
x
(n)
m
= f(x
m
, w)
y
m
= x
m
_
(2)
where w(t) is the excitation signal input to the reference
model, and the state vector of the reference model is x
m
=
(x
m
, x
m
, . . . , x
(n1)
m
)
T
R
n
.
The control objective is to force the plant output y(t) to follow
a given bounded reference signal y
m
(t) under the constraints
that all closed-loop variables involved must be bounded to guar-
antee the closed-loop stability of the system. Thus, the tracking
error e = y
m
y.
A. Control Objective
Our objective is to design a stable adaptive fuzzy controller
for the systemdescribed earlier. AFLCs have evolved as popular
control solutions for those classes of plants whose inputoutput
characteristics are not precisely known. In case of the system in
(1), this is given by the condition when f() and b are not pre-
cisely known, which is widely prevalent in practical situations.
Here, we need to nd the structure of the fuzzy controller as
well as a feedback control strategy u = u( x| ), using a fuzzy
logic system and an adaptive law for adjusting the parameter
vector such that the following conditions are satised.
The closed-loop system must be globally stable in the
sense that all variables, x(t), (t), and u( x| ) must be uni-
formly bounded, i.e., |x(t)| M
x
< , |(t)| M

< ,
and |u( x| )| M
u
< , where M
x
, M

, and M
u
are set by
the designer [5].
The tracking error e(t) should be as small as possible under
the constraints in (1) and the e space should be stable in the
large sense for the system [6].
Now, the ideal control law for the system in (1) and (2) is
given as [2], [12]
u

=
1
b
_
f(x) + y
(n)
m
+ k
T
e
_
(3)
where y
(n)
m
is the nth derivative of the output of the refer-
ence model, e = (e, e, . . . , e
(n1)
)
T
is the error vector, and
k = (k
1
, k
2
, . . . , k
n
)
T
R
n
is the vector describing the desired
closed-loop dynamics for the error.
This denition implies that u

guarantees perfect tracking,


i.e., y(t) y
m
(t) if Lt
t
e(t) = 0. It indicates that as t ap-
proaches innity, e(t) tends to zero, and under that condition,
output y(t) becomes equal to the reference y
m
(t) and perfect
tracking is achieved.
In practical situations, since f and b are not known precisely,
the ideal u

of (3) cannot be implemented in real practice. Thus,


a suitable solution can be to design a fuzzy logic system to
approximate this optimal control law.
Now, to ensure stability, we assume that the control u(t) is
given by the summation of a fuzzy control u
c
( x| ) and an
additional supervisory control u
s
(x) [2], [5], [25]. Thus, u(t) is
given as
u(t) = u
c
( x| ) + u
s
(x). (4)
Let us assume that the AFLC is constructed using a zero-
order TakagiSugeno (TS) fuzzy system. Then, u
c
( x| ) for the
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
DAS SHARMA et al.: HYBRID APPROACH FOR DESIGN OF STABLE ADAPTIVE FUZZY CONTROLLERS USING LYAPUNOV THEORY AND PSO 331
Fig. 1. Flowchart representation of LSBA algorithm. k: plant simulation counter; PST: plant simulation time; t
c
: controller simulation sampling time.
AFLC is given in the form
u
c
( x| ) =

N
l=1

l

l
(x)
_

N
l=1

l
(x)
_ =
T
(x) (5)
where = [
1

2

N
]
T
= the vector of the output singletons,

l
(x) =

r
i=1

l
i
(x
i
) = the ring degree of rule l, N is the
total number of rules,
l
i
(x
i
) is the membership value of the
ith input MF in the activated lth rule, (x) = vector contain-
ing normalized ring strength of all fuzzy IFTHEN rules =
(
1
(x),
2
(x), . . . ,
N
(x))
T
, and

l
(x) =

l
(x)
_

N
l=1

l
(x)
_. (6)
Let us dene a quadratic form of tracking error as V
e
=
1/2e
T
Pe where P is a symmetric positive denite matrix sat-
isfying the Lyapunov equation [2]. With the use of u
s
(x), it
can be shown that

V
e
1/2e
T
Qe 0, where Q is a positive
denite matrix [5], [13].
Thus, as P > 0, boundedness of V
e
implies the boundedness
in x. Hence, the closed-loop stability is guaranteed.
The zero-order TS-type fuzzy control u
c
(x|) can be so
constructed that it will produce a linear weighted combination
of adapted parameter vector . Thus, a simple singleton-based
adaptation law as proposed in [2] and [12] can be given as

= e
T
p
n
(x) (7)
where > 0 is the adaptation gain or learning rate and p
n
is the
last column of P.
It should be noted here that when the reference signal y
m
is the output of a reference model, excited with a given input
signal w, and the reference plant has the same dynamics as that
of the controlled plant and can be expressed by k, then the input
states include w along with x [6], [12].
III. ADAPTATION STRATEGIES FOR THE AFLC
A. Lyapunov-Strategy-Based Approach
Design methodologies for a stable direct adaptive fuzzy con-
troller, based on the Lyapunov method, are given in [5], [6],
and [21]. The owchart representation of LSBA is shown in
Fig. 1. The design methodology can be divided into two parts:
1) ofine calculations and 2) online adaptation.
The ofine calculation deals with the choice of the val-
ues of k. A typical method of specifying k is given in
[6]. The P matrix is chosen with the constraint P > 0.
The adaptation of is carried out using (7), if (|| < M

)
or(|| = M

and e
T
p
n

T
(x) 0). Otherwise, a projection
operator has to be used for adapting [5].
B. PSO-Based Approach (PSOBA)
PSO, originally developed by Kennedy and Eberhart in 1995,
is a population-based swarm algorithm [7], [8], [16]. In the
PSO computational algorithm, population dynamics simulates
bioinspired behavior, i.e., a bird ocks behavior that involves
sharing of information and allows particles to take prot from
their own discoveries and previous experience (pBestpersonal
best) as well as that of all other particles (gBestglobal best)
during the search for food.
To design a direct adaptive fuzzy controller with completely
adaptive controller structure and free parameters, we have to
choose the values of the scaling gains, the settings of controller
structure (i.e., number of MFs in each variable, number of rules,
etc.), the supports of input MFs, the positions of output single-
tons, etc. The PSOBA design methodology not only tunes the
singletons properly but it can also optimize the values of the scal-
ing gains and can determine the optimum controller structure.
In PSOBA design, a particles position in solution space is a
vector containing all required information to construct a fuzzy
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
332 IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 17, NO. 2, APRIL 2009
controller [12] as
X =
[structural ags for MFs | center locations of the MFs |
scaling gains | learning rate | position of the output singletons].
(8)
Each structural ag bears the information about the existence
or nonexistence of an MF in its corresponding variable. It can
only take binary values where 0 indicates the nonexistence and
1 indicates the existence of the corresponding MF. However,
PSOBA is an algorithm where each entry in the particle vector
can take continuous values. Hence, for each structural ag in
the vector, the universe of discourse is set as [0, 1] and the ag
is set to 0 if the continuous value of the variable is less than 0.5
and the ag is set to 1 if the variable is greater than or equal to
0.5. For each MF of an input variable, there is a corresponding
structural ag assigned. A specic MF for a given input vari-
able will only become active when its corresponding structural
ag is set to 1. When this ag becomes 0, in any given itera-
tion, the corresponding MF becomes inactive and all rules, in
the fuzzy rule base, containing that MF in antecedent clauses,
become inactive. The total number of 1s in the structural ags
keeps changing in each iteration, and hence, the total number
of MFs in which an input variable is fuzzied also changes
on each iteration. This changes the structure of the AFLC in
each iteration, thus changing the total number of rules of the
fuzzy rule base, and hence, changes the total number of ac-
tive output singletons. Each input is fuzzied using triangular
MFs and the center locations of the MFs in the particle vector
store the peaks of the MFs. Each input is fuzzied in the range
[1, 1] with two xed triangular MFs having their peaks xed
at 1 and 1, respectively. All intermediate MFs for that input
variable are exible in nature. They can be either active or in-
active during an iteration and their peaks are also adapted by
PSOBA in each iteration. For each MF, the peaks of the imme-
diate adjacent active MFs on either side of its own peak form
the left and right base support of it. Whether the immediate ad-
jacent MF is an active MF or not is determined by the content
of its corresponding structural ag. Hence, some of the center
locations of MFs are ignored while evaluating the AFLC out-
put, because their corresponding entries in the structural ags
are zero. A similar logic holds true for the output singletons in
a particle vector. Here, those singletons become inactive whose
antecedent parts contain one or more input MFs that become
inactive in that iteration.
To design an AFLCutilizing PSOBAalgorithmrst, the pop-
ulation of particles in the swarm is chosen, then the algorithm
will decide the structure of the candidate controller according to
its structural ags. The candidate controller simulation (CCS)
algorithm, as shown in Fig. 3, is carried out on the candidate
controller to calculate the tness function, the IAE, which can
be dened as IAE =

PST
n=0
e(n), where PST = plant simula-
tion time and = step size or sampling time. Then, according
to the value of the tness function IAE of each particle in each
iteration, the pBest and gBest candidate controllers are calcu-
lated and the particles position and velocity in the search space
are updated by the global version PSO method [7], [8], [10], as
shown in the owchart representation of the PSOBA algorithm
in Fig. 2. The PSOBAalgorithmwill stop searching the solution
space when the number of iterations specied by the designer
is reached or a prespecied error is attained by the controller.
C. Hybrid-Adaptation-Strategy-Based Approach
In hybrid design methods, the complementary features of
LSBA and the PSOBA are combined in three different ways
to get a superior systematic design procedure of direct AFLCs
for a class of SISO systems. The proposed hybrid strategies are
described now.
1) HASBA-Concurrent Model: In this design process, the
LSBA and the PSOBA run concurrently or in parallel to op-
timize the: 1) structure of the controller; 2) scaling gains; 3)
learning rate (or adaptation gain); and 4) positions of the output
singletons.
In this method, a particle X is divided into two subgroups,
unlike PSOBA design methodology, and is given as
X = [ | ] (9)
where
= [structural ags for MFs | center locations of
the MFs | scaling gains | learning rate]
= [position of the output singletons]. (10)
The partition of X vector separates the parameters accord-
ingly as has nonlinear inuence and has linear inuence on
u
c
, respectively [12]. In LSBA, the values of the free parameters
in the vector are dened a priori and the value of the corre-
sponding minimizing the tracking error can be obtained by the
adaptation lawas in (7). Here, is set by a hand-tuned trial-and-
error method. In this paper, PSOBAis applied to optimize and
in tandem by updating the particles position, containing both
and , as shown in the owchart representation of PSOBA in
Fig. 2. In addition to that, adaptation law, as in (7), is applied to
every updated candidate controller to adjust the value of only.
So, in this method, explores both the local search space and
global search space simultaneously, in a twofold manner [29].
The owchart representation of HASBA-concurrent model is
shown in Fig. 4.
2) HASBA-Cascade Model: In this model, LSBA and
PSOBA optimizations are used in cascade. Therefore, in this
model, the structure [22] and the free parameters of the con-
troller are optimized by the PSOBA method as well as the po-
sitions of the output singletons are primarily tuned in global
search space by the PSOBA method. So, after a primary global
tuning of by the PSOBA method, LSBA optimizes locally
to further improve the performance of the controller. Thus, in
this hybrid model also, the LSBA method is enhanced by the
PSOBA method. However, the basic difference lies in the fact
that the LSBAalgorithmis applied to the best solution vector de-
termined by the PSOBA algorithm. Hence, the LSBA algorithm
runs only once at the completion of the PSOBA algorithm. It is
in contrast to the HASBA-concurrent model, where the LSBA
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
DAS SHARMA et al.: HYBRID APPROACH FOR DESIGN OF STABLE ADAPTIVE FUZZY CONTROLLERS USING LYAPUNOV THEORY AND PSO 333
Fig. 2. Flowchart representation of PSOBA algorithm. j: particle counter; NOP: number of particles (population size); g: PSO iteration counter; iter
max
:
maximum number of PSO iterations.
Fig. 3. Flowchart representation of candidate controller simulation (CCS) algorithm. k: plant simulation counter; PST: plant simulation time; t
c
: controller
simulation sampling time.
algorithm runs as a submodule of the PSOBA algorithm and at-
tempts to locally ne-tune each candidate solution vector, called
particle. In the cascade model, the learning rate is adjusted
in such a fashion that LSBA adaptation searches only the near
neighborhood of the best produced by the PSOBA method.
The owchart representation of HASBA-cascade is shown in
Fig. 5.
3) HASBA-Preferential Model: In this proposed model, both
the PSOBAmethod and the HASBA-concurrent method are em-
ployed to keep the tracking error as minimum as possible. Here,
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
334 IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 17, NO. 2, APRIL 2009
Fig. 4. Flowchart representation of HASBAconcurrent model algorithm. j: particle counter; NOP: number of particles (population size); g: PSO iteration
counter; iter
max
: maximum number of PSO iterations.
Fig. 5. Flowchart representation of HASBAcascade model algorithm.
each particle in the swarm, i.e., each candidate controller, eval-
uates the CCS algorithm and the LSBA algorithm separately at
the same time. The tness values of the candidate controller,
evaluated by employing the CCS algorithm and the LSBA al-
gorithm separately, are then compared, and then, the algorithm
showing a better performance is taken as the guiding factor for
updating the velocity and the position of that specic particle
in that specic iteration, according to the PSO method. If the
LSBA algorithm shows a better performance for that particle in
that iteration, then the system will follow the algorithm for the
HASBA-concurrent model; otherwise, the system will follow
the PSOBA model in that iteration. Thus, in this process, the
updates for each particle in a swarm switch between two algo-
rithms, even within the same iteration, to search the solution
space. This process of evaluation is performed separately for
each particle in each iteration, so that the same particle may fol-
low different update rules in different iterations. The owchart
representation of HASBA-preferential is shown in Fig. 6.
IV. SIMULATION AND EXPERIMENTAL STUDIES
To evaluate the effectiveness of the proposed schemes, we
considered four benchmark case studies, previously utilized
in several research works and a real-life experiment. The pro-
cess models are simulated each using a xed-step fourth-order
RungeKutta method with sampling time T = 0.01 s.
A. Case Study I
The controlled plant under consideration is a second-order dc
motor containing nonlinear friction characteristics described by
the following model [6], [12]:
x
1
= x
2
x
2
=
f (x
2
)
J
+
C
T
J
u
y = x
1
_

_
(11)
where y = x
1
is the angular position of the rotor (in radians),
x
2
is the angular speed (in radians per second), and u is the
current fed to the motor (in amperes). The plant parameters
are C
T
= 10 Nm/A, J = 0.1 kgm
2
, and the nonlinear friction
torque is dened as
f(x
2
) = 5 tan
1
(5x
2
) N m. (12)
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
DAS SHARMA et al.: HYBRID APPROACH FOR DESIGN OF STABLE ADAPTIVE FUZZY CONTROLLERS USING LYAPUNOV THEORY AND PSO 335
Fig. 6. Flowchart representation of HASBApreferential model algorithm. j: particle counter; NOP: number of particles (population size); g: PSO iteration
counter; iter
max
: maximum number of PSO iteration.
The control objective is to make the angular position y follow
a reference signal given by
y
m
= 400w(t) 400y
m
40 y
m
(13)
(i.e., the reference model has a double pole at s
1,2
= 20),
where w(t) is a square wave of random amplitude, as shown in
Fig. 7(a), and y
m
is the output of w(t) ltered by the second-
order linear lter [given by (13)] as shown in Fig. 7(b).
The control signal obtained as the output from the fuzzy
controller, in response to input signals, is given as follows:
u
c
= u
c
(x, w | X), for PSOBA and HASBA
u
c
= u
c
(x, w | ), for LSBA.
_
(14)
1) Comparison of Design Strategies: In this section, the fol-
lowing design strategies are compared.
a) LSBA: In this case, only is optimized and the static
is manually set a priori. In our simulation, input MFs are
distributed evenly on the corresponding universe of discourse.
The other parameters are chosen as
P =
_
p
0
p
1
p
2
p
2
_
=
_
5 0.01
0.01 0.001
_
k = [ 400 40 ]
such that p
1
k
1
> 0 and p
2
> p
1
/k
2
. The value of adaptation
gain is chosen as = 5, giving a good tradeoff between adap-
tation rate and small amplitudes of parameter oscillations that
are due to inherent approximation error. The other manually
set parameters are b = 100, b
L
= 95, and f
U
= 1.5. The in-
put and output scaling gains of the AFLC are chosen so as to
map the ranges of the corresponding variables into the interval
[1, 1].
The controller is trained by applying a signal w(t) that
changes its value randomly in the interval (1.5, 1.5) in ev-
ery 0.5 s. This signal is chosen as it contains innitely many
frequencies and leads to good exploration of controller input
space [6].
The plant is simulated for 210 s. The adaptation starts at
t = 0 s and continues up to t = 200 s when the adaptation is
switched off. Then, the plant is tested for 10 s with this reference
signal. During this period, adaptation is turned off (i.e., = 0)
and the controller operates with already adapted parameters.
b) PSOBA: In this case, all the parameters of X = [| ]
are optimized by the PSOBA. To performthis simulation, a pop-
ulation size of ten particles is taken. For each simulation, 200
iterations of PSOBAis set. In xed structure controller design, a
xed number of evenly distributed input MFs are used for initial
population of PSOBA. The PSOBAalgorithmfurther tunes only
the free parameters and the positions of the output singletons
to minimize the tracking error. In variable structure controller
design, the structural ags, are utilized to automatically vary
the structure of the candidate controller in each iteration. The
input MFs and the corresponding position of the output single-
tons are set according to the active structural ags, as stated in
Section III-B. IAE between the reference signal y
m
and the
motor output y is taken as the tness function for optimization.
c) HASBA: It consists of the following models.
i) Concurrent hybrid model: In this simulation pro-
cess, the PSOBA-based optimization of parameters
for X = [| ] and the LSBA algorithm, which only
adjusts , run concurrently. The position vector X
now additionally includes the adaptation gain in its
portion. The gain is varied over the range 14 000
16 000. The structural ags are set and reset in the
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
336 IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 17, NO. 2, APRIL 2009
Fig. 7. (a) Input signal w(t). (b) Reference signal y
m
(t).
same manner as is done in PSOBA and utilized to
automatically nd out the structure of the candidate
controller.
ii) Cascade hybrid model: This simulation process starts
with the PSOBAsimulation. The result obtained from
PSOBA is then passed through the Lyapunov-based
adaptation strategy to further adjust the positions of
the output singletons . The adaptation gain during
the LSBAis set to 5 to explore the near neighborhood
of the best as determined by the PSOBA.
iii) Preferential hybrid model: In this hybrid simula-
tion process, both concurrent hybrid model and pure
PSOBAmodel are implemented for the same particle
separately at the same time, and the candidate con-
trollers, i.e., the particles of PSO, are updated accord-
ingly. The same particle can take different searching
algorithms to search the solution space depending on
the value of its tness function IAE during the entire
optimization process.
The results of simulation studies are tabulated in Table I. In
all these simulations, the plant is evaluated for 10 s after the
training is completed by different adaptation strategies so as to
compare the results.
TABLE I
COMPARISON OF SIMULATION RESULTS OF CASE STUDY I
(EVALUATION TIME = 10 S)
The simulations are carried out for three xed structures
of 3 3 3, 5 5 5, and 7 7 7 input MFs and a vari-
able structure with a potentially biggest possible structure of
7 7 7 MFs. Table I presents comparison of three possible
HASBA controllers vis-` a-vis LSBA and PSOBA controllers for
case study I. Fig. 8 shows temporal variations of system re-
sponses for the sample case of xed structure controllers with
5 5 5 input MFs. Fig. 9(a) shows the temporal variations of
system responses of the variable structure HASBA-preferential
model for 100 intermediate iterations. In Fig. 9(b)(d), the sys-
temresponse for 10 s evaluation phase, the control effort, and the
error plot for this AFLC are shown as a sample case. For xed
structure controllers, it can be seen that the performances of
the controllers (demonstrated by the performance index of IAE)
improve with increase in the number of rules from a 3 3 3
to 5 5 5 conguration. However, for a 7 7 7 congu-
ration, results are not better than that for a 5 5 5 congura-
tion. Hence, we can conclude that merely increasing the num-
ber of rules in fuzzy controllers does not necessarily improve
their performance. This is in conformation with the observa-
tions presented in [12]. For a given controller MF conguration
(e.g., 5 5 5 or 7 7 7), among the competing controllers,
the HASBA-preferential scheme shows the best performance.
Even for the variable structure scheme, HASBA-preferential
controller emerges as the best possible solution. Although the
IAE value is higher for the variable structure case, it can be
seen that the best result, in terms of IAE, is obtained with only
100 rules. This is a signicant reduction in the total number of
rules and such a simplied structure of an FLC will be very
useful from the point of view of implementation in real life.
Although the HASBA-concurrent model could converge to an
even smaller number of rules (80, in this case), a comparatively
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
DAS SHARMA et al.: HYBRID APPROACH FOR DESIGN OF STABLE ADAPTIVE FUZZY CONTROLLERS USING LYAPUNOV THEORY AND PSO 337
Fig. 8. Responses of the xed structure (MF: 5 5 5) adaptive fuzzy controllers of case study I. (a) LSBA. (b) PSOBA. (c) HASBA-concurrent.
(d) HASBA-cascade. (e) HASBA-preferential.
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
338 IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 17, NO. 2, APRIL 2009
Fig. 9. Responses of the variable structure HASBApreferential adaptive fuzzy controller of case study I. (a) Training phase response after 100 PSO iterations.
(b) Evaluation phase response. (c) Evaluation phase control effort. (d) Evaluation phase error plot.
much larger IAE prompted us to conclude that the HASBA-
preferential gave the overall superior performance.
B. Case Study II
In this case study, the Dufngs forced oscillation system,
which is itself a chaotic system if unforced, is considered and is
given as
x
1
= x
2
x
2
= 0.1x
2
x
3
1
+ 12cos(t) + u(t)
y = x
1
_

_
. (15)
The control objective is to track the reference signal y
m
,
where y
m
= sin(t). The control signal fed to the fuzzy controller
is
u
c
= u
c
(x|X), for PSOBA and HASBA
u
c
= u
c
(x| ), for LSBA
_
. (16)
1) Comparison of Design Strategies: The choice of param-
eters for this simulation process is:
P =
_
1 0.01
0.01 0.001
_
b = 1
hence, b
L
= 0.95 and f
U
= 12 +

x
3
1

, K = [2 1], [18]. The


LSBA, PSOBA design strategies, and different hybrid meth-
ods are simulated in a similar fashion as in the earlier case
study.
The adaptation gain is set to 2 for LSBA and HASBA-
concurrent, 0.5 for HASBA-cascade, and 1 for HASBA-
preferential. In a similar fashion of case study I, the simulations
are carried out for three xed structures of 3 3, 5 5, and
7 7 input MFs and a variable structure. In this case study also,
the HASBA-preferential design strategy emerges as the best
solution, for both xed structure and variable structure cong-
urations. In fact, for this case study, the best performance result
for HASBA-preferential controller for variable structure cong-
uration is obtained with simultaneous achievement of minimum
number of rules (here, it is 15). Hence, this design strategy
evolves as the most superior solution from both the points
of view of best performance and simplest implementation.
Table II shows the results for sample case of variable struc-
ture controllers. Fig. 10 shows the response of variable-structure
HASBA-preferential controller for case study II as a represen-
tative manner.
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
DAS SHARMA et al.: HYBRID APPROACH FOR DESIGN OF STABLE ADAPTIVE FUZZY CONTROLLERS USING LYAPUNOV THEORY AND PSO 339
TABLE II
COMPARISON OF SIMULATION RESULTS OF CASE STUDY II
(EVALUATION TIME = 10 S)
Fig. 10. Evaluation phase response of the variable structure HASBA
preferential adaptive fuzzy controller of case study II.
C. Case Study III
In this case study, the Dufngs forced oscillation systemwith
disturbance is considered and is given as
x
1
= x
2
x
2
= 0.1x
2
x
3
1
+ 12cos(t) + u(t) + d(t)
y = x
1
_

_
(17)
where the external disturbance d(t) is a square wave of random
amplitude within the range [1, 1] and a period of 0.5 s. The
control objective is to track the reference signal y
m
, where
y
m
= sin(t). The control signal fed to the fuzzy controller is
the same as in (16).
1) Comparison of Design Strategies: Here also, we choose
the same parameters for this simulation process as was cho-
sen in case study II. The design strategies of LSBA, PSOBA,
and different hybrid methods are simulated in a similar fash-
ion as in case study II. The adaptation gain is set to 2 for
LSBA and HASBA-concurrent, 0.5 for HASBA-cascade, and
1 for HASBA-preferential. In this case study, among the xed
structure controllers, the HASBA-concurrent shows least av-
erage IAE value for 7 7 conguration and PSOBA shows
best performance for variable conguration. However, HASBA-
preferential conguration emerges as the second best solution
in each conguration. It also achieves the minimum number of
rules for the variable conguration (along with HASBA-cascade
and PSOBA controllers). This is understandable from the fact
TABLE III
COMPARISON OF SIMULATION RESULTS OF CASE STUDY III
(EVALUATION TIME = 10 S)
Fig. 11. Evaluation phase response of the variable structure HASBA
preferential adaptive fuzzy controller of case study III.
that, very rarely, a single controller design strategy emerges as
the best solution for many types of systems under several types
of input conditions. Table III shows the results for a sample case
of variable structure controllers. Fig. 11 shows the response of
the variable-structure HASBA-preferential controller for case
study III in a representative manner.
D. Case Study IV
All the discussions presented so far are given with reference to
the system in (1), where the control input is given in bu form.
The discussions hold equally for a more general system, where
control input is given in g(x)u, as demonstrated in [6]. Now
we demonstrate the utility of our proposed controllers for such
a system. We consider the inverted pendulum system [26], [28]
as
x
1
= x
2
x
2
= f(x) + g(x)u(t)
y = x
1
_

_
(18)
where
f(x) =
mlx
2
sin x
1
cos x
1
(M + m)g sin x
1
ml cos
2
x
1

4
/
3
l(M + m)
and
g(x) =
cos x
1
ml cos
2
x
1

4
/
3
l(M + m)
where g = 9.8 m/s
2
, m = 0.1 kg, M = 1 kg, and l = 0.5 m
[26], [28]. The control objective is to track the reference signal
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
340 IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 17, NO. 2, APRIL 2009
TABLE IV
COMPARISON OF SIMULATION RESULTS OF CASE STUDY IV
(EVALUATION TIME = 10 S)
Fig. 12. Evaluation phase response of the variable structure HASBA
preferential adaptive fuzzy controller of Case Study IV.
y
m
, where y
m
= sin(t). The control signal fed to the fuzzy
controller is the same as in (16).
1) Comparison of Design Strategies: The choice of param-
eters for this simulation process is
P =
_
1 0.01
0.01 0.001
_
b = g(x)
hence, b
L
= b 0.1 and f
U
= 5, K = [2 1]. The design strate-
gies of LSBA, PSOBA, and different hybrid methods are simu-
lated in a similar fashion as in earlier case studies. The adaptation
gain is set to 5 for LSBA, 0.001 for HASBA-concurrent and
HASBA-preferential, and 0.01 for HASBA-cascade. The main
difference of this case study is that, here, the coefcient of the
control input u(t) is not a constant like in earlier cases, but is
a function of the states of the inverted pendulum system and is
varying with time. Table IV and Fig. 12 show results in tabular
and graphical form for case study IV in a similar manner as in
case studies II and III. It is seen from Table IV that the IAE of
the HASBA-preferential model is close to that of the PSOBA
and HASBA-concurrent model but the number of fuzzy rules is
minimum (15 in this case). So, the performance of the HASBA-
preferential model is good enough to justify it as a superior
control strategy over the other control strategies.
Overall, from our four simulation case studies with different
types of temporal reference input variations, it can be concluded
that the HASBA-preferential controller emerges as a superior,
feasible control solution.
Fig. 13. (a) Real-life experimental arrangements. (b) and (c) Evaluation phase
response of HASBApreferential adaptive fuzzy controller of Case Study IV
with different reference signals.
E. Case Study V
To demonstrate the usefulness of the proposed algorithms,
we have performed a real-life experiment where we attempt to
perform the speed control of a real dc motor. Fig. 13(a) shows
the experimental arrangement where the PC-based controller
controls the dc motor through a parallel port interface in real
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
DAS SHARMA et al.: HYBRID APPROACH FOR DESIGN OF STABLE ADAPTIVE FUZZY CONTROLLERS USING LYAPUNOV THEORY AND PSO 341
TABLE V
COMPARISON OF SIMULATION RESULTS OF CASE STUDY V
(EVALUATION TIME = 240 S)
time. The scheme employs a power-transistor-based chopper-
controlled mechanism to achieve the desired speed in armature
control mode. At rst, the motor parameters, i.e., armature re-
sistance (R
a
), armature inductance (L
a
), inertia constant (J),
damping constant (B), etc., are identied by exciting the motor
in open loop and collecting the inputoutput data. Here, the dc
motor characteristics are modeled as
x
1
= x
2
x
2
=
_
BR
a
J L
a
+
K
b
K
T
J L
a
_
x
1

_
R
a
L
a
+
B
J
_
x
2
+
KK
T
J L
a
u(t)
y = x
1
_

_
(19)
where [x
1
x
2
] = [ ], the output y is the angular speed of the
motor (in radians per second), K is the constant of the chopper
circuit, and K
b
and K
T
are the constants of the dc motor. The
unknown parameters are learned utilizing PSO, where the PSO
algorithm determines the best vector comprising the unknown
parameters for which discrepancy between the model output and
the actual experimental output data, for the same input, is min-
imum, considering the entire set of inputoutput real-life data.
When the state model of this real dc motor is identied, our con-
troller design is carried out ofine for different control strategies
as discussed earlier. Fixed structure controllers with 5 5 in-
put MFs are implemented for this real-life experimentation for
controlling the same motor under temporally varying reference
speed conditions for 240 s. Fig. 13(b) and (c) shows the perfor-
mance of HASBA-preferential controller, as sample cases, in
tracking two different temporally varying reference speed sig-
nals. Table V shows the sample performances of the controllers
in the real implementation phase for the dc motor, for the case
as shown in Fig. 13(b). In each of these experimentations, the
speed measurement is carried out using shaft encoders having
ten holes in the wheel mounted on the shaft. Then, the speed sig-
nal acquired is ltered in the software developed, employing an
exponential averaging technique, to remove noise pickup. It can
be seen that the controllers developed were able to provide sat-
isfactory performance. The interrupt-driven software developed
for this sophisticated motor control scheme has a sampling rate
of 100 s within the interrupt mode, whereas each PWM cycle
generated for control input is of duration 10 ms (i.e., ON-time +
OFF-time). The controller output is used to suitably control this
ON-time. Table V, reveals that, in this real-life experiment also,
the HASBA-preferential controller emerges as the best design
strategy among the LSBA, PSOBA, and other hybrid variants.
V. CONCLUSION
This paper has proposed a new hybrid approach for designing
the AFLCs employing both the conventional Lyapunov theory,
to locally optimize the positions of the output singleton, and
PSO, which is a stochastic optimization technique. In this pro-
posed hybridization process, while LSBA optimizes the posi-
tions of the output singleton, PSOBA is utilized to design the
structure of the AFLC and also evaluate the free parameters of
it. Three variants of the hybrid models have been proposed and
simulated for the benchmark case studies. The preferential hy-
brid model was evolved as a superior technique as compared to
the LSBA, PSOBA, and other hybrid models. The stability of
the closed-loop system and the convergence of the plant output
to a desired reference signal are guaranteed precisely. The main
advantages of these proposed methods are that it requires no a
priori knowledge about the controlled plant and the approxima-
tion error is reduced greatly. Moreover, the preferential hybrid
model is able to avoid the local minima. In this paper, we have
chosen PSO as the representative stochastic global optimization
algorithm for hybridization with Lyapunov theory. The ratio-
nale behind this is that PSO is a relatively simple algorithm
among stochastic optimization algorithms. However, it should
be pointed out that other stochastic global algorithms like GA,
differential evaluation, etc. [12], [19], [20], [27], can also be
potentially employed to design similar AFLCs. The authors in-
tend to design similar AFLCs using other global optimization
algorithms and compare their performances in the near future.
REFERENCES
[1] P. He and S. Jagannathan, Reinforcement learning neural network based
controller for non-linear discrete time systems with input constraints,
IEEE Trans. Syst., Man, Cybern. B, Cybern., vol. 37, no. 2, pp. 425436,
Apr. 2007.
[2] S. H. Zak, Systems and Control. New York: Oxford Univ. Press, 2003.
[3] F. Cupertino, E. Mininno, D. Naso, B. Turchiano, and L. Salvatore, On-
line genetic design of anti-windup unstructured controllers for electric
drives with variable load, IEEE Trans. Evol. Comput., vol. 8, no. 4,
pp. 347364, Aug. 2004.
[4] C. L. Chiang and C. T. Sub, Tracking control of induction motor using
fuzzy phase plane controller with improved genetic algorithm, Electr.
Power Syst. Res., vol. 73, pp. 239247, 2005.
[5] L. X. Wang, Stable adaptive fuzzy control of nonlinear system, IEEE
Trans. Fuzzy Syst., vol. 1, no. 2, pp. 146155, May 1993.
[6] K. Fischle and D. Schroder, An improved stable adaptive fuzzy control
method, IEEE Trans. Fuzzy Syst., vol. 7, no. 1, pp. 2740, Feb. 1999.
[7] R. C. Eberhart and J. Kennedy, A new optimizer using particle swarm
theory, in Proc. 6th Int. Symp. Micromach. Hum. Sci., Nagoya, Japan,
1995, pp. 3943.
[8] J. Kennedy and R. C. Eberhart, Particle swarm optimization, in Proc.
IEEE Int. Conf. Neural Netw., 1995, vol. 4, pp. 19421948.
[9] G. Venter and J. S. Sobieski, Particle swarm optimization, presented at
the 43rd AIAA Conf., Denver, CO, 2002.
[10] Y. Shi and R. C. Eberhart, Amodied particle swarmoptimizer, in Proc.
IEEE Int. Conf. Evol. Comput., May 1998, pp. 6973.
[11] R. C. Eberhart and Y. Shi, Comparing inertia weights and constriction
factors in particle swarm optimization, in Proc. 2000 Congr. Evol. Com-
put., pp. 8489.
[12] V. Giordano, D. Naso, and B. Turchiano, Combining genetic algorithm
and Lyapunov-based adaptation for online design of fuzzy controllers,
IEEETrans. Syst., Man, Cybern. B, Cybern., vol. 36, no. 5, pp. 11181127,
Oct. 2006.
[13] L. X. Wang and J. M. Mendel, Three-dimensional structured networks
for matrix equation solving, IEEE Trans. Comput., vol. 40, no. 12,
pp. 13371346, Dec. 1991.
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.
342 IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 17, NO. 2, APRIL 2009
[14] M. Dorigo, V. Maniezzo, and A. Colorni, Ant system: Optimization by
a colony of cooperating agent, IEEE Trans. Syst., Man, Cybern. B,
Cybern., vol. 26, no. 1, pp. 2941, Feb. 1996.
[15] A. A. A. Esmin, A. R. Aoki, and G. L. Torres, Particle swarmoptimization
for fuzzy membership functions optimization, in Proc. IEEE Int. Conf.
Syst., Man, Cybern., 2002, pp. 108113.
[16] Y. Shi and R. C. Eberhart, Parameter selection in particle swarm opti-
mization, in Evol. Program. VII: Proc. EP1998. NewYork: Springer-
Verlag, pp. 591600.
[17] T. J. Procyk and E. H. Mamdani, A linguistic self-organizing process
controller, Automatica, vol. 15, pp. 1530, 1979.
[18] B. S. Chen, C. H. Lee, and Y. C. Chang,, H

tracking design of uncertain


nonlinear SISO systems: Adaptive fuzzy approach, IEEE Trans. Fuzzy
Syst., vol. 4, no. 1, pp. 3243, Feb. 1996.
[19] R. Storn and K. Price, Differential evolutionA simple and efcient
adaptive scheme for global optimization over continuous spaces, Int.
Comput. Sci. Inst., Berkley, CA, Tech. Rep., 1995.
[20] K. S. Lee and Z. W. Geem, Anewmeta-heuristic algorithmfor continuous
engineering optimization: Harmony search theory and practice, Comput.
Methods Appl. Mech. Eng., vol. 194, no. 3638, pp. 39023933, 2005.
[21] L. X. Wang, A Course in Fuzzy Systems and Control. Englewood Cliffs,
NJ: Prentice-Hall, 1997.
[22] D. L. Tsay, H. Y. Chung, and C. J. Lee, The adaptive control of nonlinear
systems using the Sugeno-type of fuzzy logic, IEEE Trans. Fuzzy Syst.,
vol. 7, no. 2, pp. 225229, Apr. 1999.
[23] R. Chekkouri, L. Romeral, and M. Moreno, Direct adaptive fuzzy con-
troller by changing the output membership functions for nonlinear motion
applications, in Proc. IEEE ICIT 2003, Maribor, Slovenia, 2000, pp. 22
26.
[24] S. S. Chen, J. L. Wu, S. F. Su, J. C. Chang, Y. C. Chang, and T. T. Lee,
On the analysis of direct adaptive fuzzy controllers, in Proc. 3rd Int.
Conf. Mach. Learn. Cybern., Shanghai, China, 2004, pp. 43684373.
[25] S. H. Chin and M. J. Er, Hybrid adaptive fuzzy controllers of robot
manipulators, in Proc. 1998 IEEE/RSJ Int. Conf. Intell. Robots Syst.,
pp. 11321137.
[26] T. Wang and S. Tong, Adaptive fuzzy output feedback control for SISO
nonlinear systems, Int. J. Innovative Comput., Inf. Control, vol. 2, no. 1,
pp. 5160, 2006.
[27] R. Toscano and P. Lyonnet, Stabilization of systems by static output
feedback via heuristic Kalman algorithm, J. Appl. Comput. Math., vol. 5,
no. 2, pp. 154165, 2006.
[28] H. X. Li and S. Tong, A hybrid adaptive fuzzy control for a class of
nonlinear MIMOsystem, IEEE Trans. Fuzzy Syst., vol. 11, no. 1, pp. 24
34, Feb. 2003.
[29] K. D. Sharma, A. Chatterjee, and F. Matsuno, A Lyapunov theory and
stochastic optimization based stable adaptive fuzzy control methodology,
in Proc. SICE Annu. Conf. 2008, Univ. Electro-Commun., Aug. 2022,
pp. 18391844.
Kaushik Das Sharma received the B.Sc. (Hons.)
degree in physics, and the B.Tech. and M.Tech. de-
grees in electrical engineering from the University of
Calcutta, Kolkata, India, in 1998, 2001, and 2004, re-
spectively. He is currently working toward the Ph.D.
degree at Jadavpur University, Kolkata.
He is currently an Assistant Professor of the De-
partment of Electrical Engineering, Future Institute of
Engineering and Management, West Bengal Univer-
sity of Technology, Kolkata. His current research in-
terests include fuzzy control system design, stochas-
tic optimization applications, and embedded system design, among others.
Amitava Chatterjee received the B.E., M.E., and
Ph.D. degrees in electrical engineering fromJadavpur
University, Kolkata, India, in 1991, 1994, and 2002,
respectively.
In 1997, he joined the Department of Electrical En-
gineering, Jadavpur University, where he is currently
a Reader. In 2003, he received the Japanese Gov-
ernment (Monbukagakusho) Scholarship and went
to Saga University, Saga, Japan. In early 2004, he
was invited as a TeacherResearcher in the Uni-
versity of Paris XII, Val de Marne, France. From
November 2004 to November 2005, he was with the University of Electro-
Communications, Tokyo, Japan, on a Japan Society for the Promotion of Science
(JSPS) Post-Doctoral Fellowship for Foreign Researchers. His current research
interests include intelligent instrumentation and control, signal processing, im-
age processing, and robotics. He has authored or coauthored more than 50
technical papers, including 30 international journal papers.
Anjan Rakshit received the M.E. and Ph.D. degrees
in electrical engineering from Jadavpur University,
Kolkata, India, in 1978 and 1987, respectively.
He is currently a Professor in the Department
of Electrical Engineering, Jadavpur University. His
current research interests include digital signal pro-
cessing, Internet-based instrumentation, smart instru-
mentation, intelligent control, and design of real-time
systems. He has authored or coauthored almost 70
technical papers.
Authorized licensed use limited to: B D COLLEGE OF ENGINEERING. Downloaded on January 14, 2010 at 11:01 from IEEE Xplore. Restrictions apply.

You might also like