Available online at www.sciencedirect.

com

Journal of the Franklin Institute 359 (2022) 1320–1346

www.elsevier.com/locate/jfranklin

Education Article
Finite-time extended dissipativity control for interval
type-2 fuzzy systems with resilient memory
sampled-data controller
N. Sakthivel a,∗, C.A. Suruthi Sri a, Guisheng Zhai b
a Department of Applied Mathematics, Bharathiar University, Coimbatore 641 046, India
b Department of Mathematical Sciences, Shibaura Institute of Technology, Saitama 337-8570, Japan
Received 30 December 2020; received in revised form 20 September 2021; accepted 18 November 2021
Available online 29 November 2021

Abstract
In this work, the finite-time extended dissipativity of the interval type-2 (IT2) fuzzy systems with
probabilistic time-varying delay is discussed via resilient memory sampled-data control. To enable the
stability analysis and control combination, an IT2 fuzzy model is employed to represent the dynam-
ics of nonlinear systems of which the parameter uncertainties are taken by IT2 membership functions
distinguish by the lower and upper membership functions. The main objective of this paper is to de-
sign a resilient memory sampled-data controller such that the resulting closed-loop system is finite-time
bounded and satisfies extended dissipative performance. Moreover, the solvability of the derived condi-
tions not only depends on the size of the delay but also on the probabilistic distribution of the delay
taking values in some interval, thus probabilistic delay protocol is encountered in the IT2 fuzzy model.
By employing suitable Lyapunov-Krasovskii functional (LKF) along with Wirtinger-based inequality,
a set of sufficient conditions ensuring the finite-time extended dissipative performance for IT2 fuzzy
systems are derived in terms of linear matrix inequalities (LMIs). Finally, two numerical simulations
are presented to reveal the effectiveness of the developed technique.
© 2021 The Franklin Institute. Published by Elsevier Ltd. All rights reserved.

∗ Corresponding author.
E-mail address: [email protected] (N. Sakthivel).

https://doi.org/10.1016/j.jfranklin.2021.11.011
0016-0032/© 2021 The Franklin Institute. Published by Elsevier Ltd. All rights reserved.
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

1. Introduction

In the real world scenario, modelling of physical dynamical systems results in formulation
of complex non-linear systems. Analysis and synthesis process of such complex system poses
great difficulty. Linearization is one of the conventional techniques to handle such non-linear
systems but it is not well recognized in the field due to the fact that it authentics for only
certain operating conditions, assures only the local stability and it is unsustainable when
the range is large. Hence, it relies upon the design of the controller to cope with the high
nonlinearities, discontinuous nature of the plant, structural constraints, etc. Henceforth, the
fuzzy model-based control has been gathering great attention among researchers due to its
capability of handling complexity and non-linearity identification of the system. In particular,
Takagi-Sugeno (T-S) fuzzy model [1–3] is an efficient tool to deal with non-linearities which
are interpreted as mean weighted sum of linear subsystem whose membership functions grades
are characterized by type-1 membership functions. Though, fuzzy model based system has
been successful in handling the complex non-linear system, it may fail when the system is
captured with uncertainties which is inevitable in practical circumstances. Hence, Mendel et al.
[4] implemented type-2 fuzzy logic system which overcomes the demerits of type-1 fuzzy
logic systems. Moreover, the concept of lower and upper membership function is employed
in [5,6] to manipulate the parameter uncertainties presented in complex non-linear system
and also provides the better performances than general fuzzy logic system. In addition, some
fruitful results have been proposed on IT2 fuzzy system and controller. For instance, in [7],
the author discussed the results of ball and beam system through IT2 fuzzy state feedback
control. Xie et al. [8] proposed stabilization criteria for interval type-2 fuzzy network control
system by employing an adaptive event-triggered fuzzy filtering control with H∞ performance.
The problem of filter-based fault detection for IT2 fuzzy networked control systems with
quantization is discussed in [9].
Besides, on a physical dynamical system there may occur a time lag between input and
output of the system. This introduces time delay, which is unavoidable in practical circum-
stances and could degrade the system performance or even cause oscillation and instability.
Since realistic systems namely biological systems, chemical modelling and electrical circuit
systems are time delay systems, it is essential to deal with time-delay whenever we simulate
such type of dynamical system. There exist various kinds of delay namely constant delay
[10], time-varying delay [11], discrete delay [12], distributed delay [13]. Hence, it is manda-
tory to employ appropriate one in order to reflect the real behaviour of practical system. In
general, time-delay in real time systems are random in manner, that is certain values of the
time-varying are huge whereas the distribution of delay taking such values is small which
could result in less conservatism [14]. Also, the stochastic nature of delays such as Bernoulli
distribution, Poisson distribution can be obtained by statistical process. Hence, it would be
more realistic and less conservative when stability condition is obtained with the consideration
of the probabilistic time-varying delay. Due to several application aspects, the concept of the
probabilistic time-varying delay became active research topic among researchers and has been
studied for the various kinds of the models such as linear systems [15], complex network
systems [16], neural network systems [17], high-speed train [18].
It is well known that, the external disturbances are also ubiquitous in practical system
which make unsuspected performances in the system and they lead to further degradation
of the system [19,20]. Several methodologies has been exposed so far in order to deal with
problem of performance analysis of the system with disturbance attenuation. To name a few,

1321
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

H∞ [21], L2 − L∞ [22], passivity [23], dissipativity [24] etc. In concerning with the robustness
of control systems and aspect of designing in engineering problem, many researches have
been made to unify those performance in a single framework for a potentful performance
of the system. Zhang et al. unifies aforementioned performances and addressed as extended
dissipativity in [25,26]. Since extended dissipativity merges all effective performances together,
it enhances the system and makes it much more robust. The authors in [27] discussed the
extended dissipativity analysis and control synthesis through line integral Lyapunov function
approach for interval type-2 fuzzy system. In [28], the authors derived existence condition
for extended dissipativity filter for IT-2 fuzzy system with D-stability constraints as convex
optimization problem.
All the above discussed factors tend to affect the stability and performance of the non-
linear system. For this perspective, it is necessary to construct a suitable controller for such
systems. Hence, the fuzzy control approach was implemented and it has been proved that
fuzzy controllers are successful among other conventional control approaches, since they
easily handle the intricacy of controlling the complex non-linear system. Benefited by the
advantages of the fuzzy controllers, several results have been studied which include fuzzy
feedback [29], fuzzy adaptive control [30], fuzzy sliding mode method [31], fuzzy sampled-
data method [32]. Among those, fuzzy sampled-data control has become research hotspot
among the research community for its prominence where sampled-data control system is
a continuous plant and is connected to the discrete-time controller consists of components
digital-to-analogy and analogy-to-digital. Moreover, this protocol extremely reduces quantity
of information transmitted to the system, only requires samples of the state variable at discrete
instants and also enhances the bandwidth of network communication which make this method
efficient and applicable to the real-world problem [33]. Till now, three kinds of approaches
have been used for sampled-data control system namely (1) discrete-time model, (2) impul-
sive model and (3) input delay model. The input delay model has benefit over the other two,
since the sampling distance need not to be constant. In the view of their advantages, fuzzy
sampled-data controller by input delay method has been studied recently in [34,35]. Moreover,
various studies in the literatures, have shown that constant signal transmitting delay occurs
frequently in memory and zero-order holder (ZOH) while updating signal from sampler to
the controller. The memory capability of memory sampled-data controller is evidenced by the
constant transmission delay in the controller input [36]. Thus, the study of controller with
constant signal transmitting delay, known as memory sampled-data controller, is regarded as
significant control method among others. Recently, the authors in [37] developed memory
sampled-data control for the problem of stabilization of chaotic systems. Wang et al. [38] dis-
cusses the synchronization control of complex networks along with communication delays
by memory sampled-data control technique. In [39], the authors concerned the problem of
exponential memory sampled-data synchronization of delayed complex dynamical networks
with gain perturbations.
In general, control systems are intended to have fast cognizant performance. Specifically,
asymptotic stability cannot assure that system could attain its equilibrium rapidly or in a
desired period. Hence, finite-time stability mechanism is used in the engineering control
problems to seek the convergence in finite-time. Though many results discussed on finite-
time [40–42], finite-time memory sampled-data control design has not been concentrated
until now. On the other hand, coefficients of the controller are generally precise value but
inaccuracies and uncertainties may be encountered in control design which are inevitable in
practical situations, so that controller can’t be implemented exactly. Thus it is important to

1322
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

design a controller which is insensitive to the gain fluctuation. Such a controller is designed
in [43] and is called as non-fragile/resilient controller. On the whole, resilient finite-time
memory sampled-data control is more realistic and guarantees swift convergence and better
performance. These inadequate results of resilient finite-time memory sampled-data control
provokes this research.
Inspired by the aforementioned concepts, some novel criteria for IT2 fuzzy systems with
random delay and external disturbances via memory sampled-data control with perturbation
are developed. Moreover, some sufficient conditions are obtained in a group of LMIs with
less conservatism. The main contribution of this work are summarized as

1. Fact to the point, uncertainties, time-delay, external disturbance may always exist in real-
world problem. Henceforth, the IT2 fuzzy system is constructed with the above and resilient
fuzzy memory sampled-data controller is designed to insure the asymptotic stability of the
system over a finite period.
2. IT2 fuzzy systems and resilient memory sampled-data controller do not share the same
membership function grades. Hence, distinct membership functions are utilized among
considered system and proposed controller which improves the flexibility of the controller
protocol.
3. For the sake of obtaining the stability conditions, appropriate LKF are used and sufficient
conditions to ensure asymptotic stability of the system are retrieved in terms of LMIs
specifically over a finite-time period.
4. Finally, to illustrate the efficacy of the proposed controller over the considered IT-2 fuzzy
system two numerical simulation results with graphical presentation are provided.

Notations: The notation matrix P  (, ≺, ) 0 denotes positive definite (semi-positive,

negative definite, semi-negative definite) matrix. n represents the n-dimensional Euclidean
space. n×m represents the set of all n × m real matrices. κmin (. ) and κmax (. ) denote the
minimum and maximum eigenvalue of the corresponding matrix respectively. ∗ denotes
the Euclidean norm for vectors or the spectral of the matrices. L2 [0, ∞ ) is the space of
square integrable function vector over the interval [0, ∞ ). E (. ) denotes expectation of the
argument. H e(A ) denotes the argument A + AT . diag{...} denotes a block-diagonal matrix.
(. )T and ( ) denote the transpose of the argument and transposed element in the symmetric
position respectively.

2. Problem formulation and preliminaries

Consider the r-fuzzy rule IT2 fuzzy model with time-varying delay to describe dynamics
of non-linear system. The dynamics of the system are described by following ith IF-THEN
rule whose antecedents are associated with some IT-2 fuzzy set,
Rule ‘i’ : IF 1 (x(t )) is F1i and . . . and p (x(t )) is Fpi , THEN
⎫
x˙ (t ) = Ai x(t ) + Adi x(t − y(t )) + Bi u(t ) + Di w(t ),⎬
z(t ) = Ci x(t ), i = 1, 2, . . . , r, (1)
⎭
x(t ) = σ (t ) ∀t ∈ [−y2 0]

where i (x(t )) denotes measurable premise variables; Fmi is IT2 fuzzy set of ith rule;
x(t ) ∈ n denotes state vector; u(t ) ∈ m denotes the control input; w(t ) ∈ L2 [0, ∞ )

1323
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Tf
denotes exogenous disturbances satisfies the constraints wT (t )w(t )dt ≺ δ; z(t ) ∈ q de-
0
notes the measured output; σ (t ) is a continuous vector-valued initial function; Ai , Adi , Bi , Ci ,
and Di are the known matrices and y(t ) denotes the transmission delay which satisfies the
following assumptions.
Assumption 2.1. Considering the probability distribution of the time-varying delay y(t ) which
takes values in the interval [0, y1 ] or (y1 , y2 ] and we establish two sets
ι1 = {t|y(t ) ∈ [0, y1 ]} and ι2 = {t|y(t ) ∈ (y1 , y2 ]}.
Furthermore, we define the following two functions

y(t ), f or t ∈ ι1
y1 (t ) =
0, else
and

y(t ), f or t ∈ ι2
y2 (t ) =
y1 , else
from which it follows that if t ∈ ι1 , the event y(t ) ∈ [0, y1 ] occurs, that is y(t ) = y1 (t ) and if
t ∈ ι2 , the event y(t ) ∈ (y1 , y2 ] occurs, that is y(t ) = y2 (t ). Furthermore, stochastic behavior
of time delay, we need to introduce a stochastic variable μ(t ) which can be defined as

1, t ∈ ι1
μ(t ) =
0, t ∈ ι2 .
From the above analysis, Bernoulli distributed sequence μ(t ) can be derived with Prob{μ(t ) =
1} = Prob{y(t ) ∈ [0, y1 ]} = E{μ(t )} = β0 and Prob{μ(t ) = 0} = Prob{y(t ) ∈ (y1 , y2 ]} =
1 − E{μ(t )} = 1 − β0 . Furthermore, we can see that
E{μ2 (t )} = β0 , E{1 − μ2 (t )} = 1 − β0 , E{μ(t )(1 − μ(t ))} = 0. (2)
Assumption 2.2. There exist two real constants α1 and α2 such that y˙ 1 (t )  α1 ≺ 1; y˙ 2 (t ) 
α2 ≺ 1.
˜
 pIgniting strength ofpthe i rule is the subsequent interval which is of the form φi (x(t )) ∈
th
 
ϕ F i (s (x(t ))) , ϕ Fsi (s (x(t ))) = φiL (x(t )), φiU (x(t )) , where φiL (x(t )) is the lower
s=1 s s=1
grade of the membership, φiU (x(t )) is the upper grade of the membership, ϕ F i (i (x(t ))) ∈
j
[0, 1] and ϕ F ij (i (x(t ))) ∈ [0, 1] is the lower and upper membership fuctions respec-
tively. Further, we define membership grade of the embedded membership function, i (x(t ))
r
as i (x(t )) = φiL (x(t ))υi (x(t )) + φiU (x(t ))(1 − υi (x(t ))) in which i (x(t )) = 1. 0 
i=1
υi (x(t ))  1 is nonlinear function. Now, based on the membership grade and Assumptions 2.1,
2.2, the system (1) can be rewritten as
⎫
x˙ (t ) =
r
⎪
⎪
i (x(t )) ⎪
⎪
i=1 ⎬
[Ai x(t ) + μ(t )Adi x(t − y1 (t )) + (1 − μ(t ))Adi x(t − y2 (t )) + Bi u(t ) + Di w(t )], (3)
⎪
⎪
z(t ) =
r
⎪
⎪
i (x(t ))[Ci x(t )], i = 1, 2, . . . , r. ⎭
i=1

1324
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Now we consider the discrete measurement of state variable x(t ) of the system that is, only
have the measurement x(tk ) at sampling instant k which is a piecewise constant function and
the control signal is assumed to be created by using a zero-order-hold (ZOH) function with
a sequence of holding times
0 = t0 < t1 < t2 < · · · < limk→∞tk = +∞.
The resultant sampling controller for the j th rule is expressed as follows
Rule ‘j’ : IF 1 (x(t )) is Lj1 and . . . and b (x(t )) is Ljb , THEN

u(t ) = K̄1j x(tk ) + K̄2j x(tk − ), tk  t ≺ tk+1 ,

where Ljn denotes the fuzzy set of rule j corresponding to the function n (x(t )), j =
1, 2, . . . , r, n = 1, 2, . . . , b, r  0 ∈ Z,  is a constant
 b signal transmission delay. Igniting
 b
strength of the jth rule is expressed as ψ˜ j (x(t )) ∈ ϕ Lj (s (x(t ))), ϕ Ljs (s (x(t ))) =
s=1 s s=1
θjL (x(t )), θjU (x(t )) . Moreover, we define θj (x(t )) =
r
ζj (x(t ))θjL (x(t ))+(1−ζj (x(t )))θjU (x(t ))
r  , θj (x(t )) = 1, 0  ζj (x(t ))  1, ∀ j.
ζm (x(t ))θmL (x(t ))+(1−ζm (x(t )))θmU (x(t )) j=1
m=1
In addition, the sampling is not required to be periodic, but the distance between any two
consecutive sampling instants is assumed to belong to an interval. Specifically, it is assumed
that tk+1 − tk = hk ≤ h, ∀ k ≥ 0, h > 0.
Moreover, the input delay approach is used to deal with the sampling instant tk . Defining
τ (t ) = t − tk for tk ≤ t < tk+1 . Then tk can be represented as follows:
tk = t − (t − tk ) = t − τ (t ).
The sampling interval is assumed to satisfy 0  τ (t )  hk  h. It has taken that delay
term τ (t ) is a piecewise-linear function with derivative τ˙ (t ) = 1, where h̄ is the maximum
sampling period. Then, the IT2 resilient memory sample-data controller is modelled as,


r
u(t ) = θj (x(t )) K̄1j x(t − τ (t )) + K̄2j x(t − τ (t ) − ) , (4)
j=1

where x(tk ) is the state vector of subsystem at the sampling instant tk , K̄1j = (K1j +
K1j ), K̄2j = (K2j + K2j ); K1j , K2j are the controller gain matrices to be determined,
K1j and K2j are perturbed gain matrices assumed to be

[K1j , K2j ] = Gj (t )[H1j , H2j ], (5)

here known constant matrices Gj , H1j , H2j and any time-varying matrix function satisfies
 T (t )(t )  I .
For convenience, we notate i = i (x(t )); θj = θj (x(t )). Now, employing control law
(4) in fuzzy system (3), the closed-loop of the IT2 fuzzy system is formulated as
⎧ r r
⎪
⎪ x˙ (t ) =
⎪
⎪ i θj [Ai x(t ) + μ(t )Adi x(t − y1 (t ))
⎪
⎪ i=1 j=1
⎨
+ (1 − μ(t ))Adi x(t − y2 (t )) + Bi K̄1j x(t − τ (t ))
⎪
⎪ + Bi K̄2j x(t − τ (t ) − ) + Di w(t ) ,
⎪
⎪
⎪
⎪
r
⎩z(t ) = i [Ci x(t )], i = 1, 2, . . . , r.
i=1

1325
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Applying relation (2) and taking mathematical expectation on the above system, the closed-
loop IT2 fuzzy system is reformulated as
⎧
⎪
r r
⎪
⎪x˙ (t ) = i θj [Ai x(t )
+ β0 Adi x(t − y1 (t )) + (1 − β0 )Adi x(t − y2 (t )) + (μ(t ) − β0 )Adi
⎪
⎨ i=1 j=1  
× x(t − y1 (t )) − x(t − y2 (t ) + Bi K̄1j x(t − τ (t )) + Bi K̄2j x(t − τ (t ) − ) + Di w (t ) ,
⎪
⎪
⎪
⎪
r
⎩z(t ) = i [Ci x(t )], i = 1, 2, 3, . . . , r.
i=1
(6)

Remark 2.1. It is worth to mention that controller meets the some advantages as lower com-
munication channel occupation, fewer actuation of the controller and lesser amount of signal
transmission when sampled-data control experiences the great sampling period. Moreover, in
the controller design K1j x(tk ) + K2j x(tk − ), when K2j = 0 memory sampled-data control
law reduces to sampled-data control.
Assumption 2.3. Matrices 1 , 2 , 3 , 4 satisfy the following conditions:

1. 1 = T1  0, 3 = T3  0, 4 = T4  0

2. ( 1 + 2 )4 = 0

Definition 2.1. [19] Given three positive constants c1 , c2 , T f with c1 ≺ c2 , a positive definite
matrix L, assume that w(t ) = 0, ∀t ∈ [0, T f ], then the closed-loop system (6) is said to be
finite-time bounded with respect to (c1 , c2 , L, T f , δ), if ∀t ∈ [0 T f ]
sup [xT (u)Lx(u), x˙ T (u)L x˙ (u)]  c1 ⇒ xT (t )Lx(t )  c2 . (7)
−y1 u0

Definition 2.2. [19] For given matrices 1 , 2 , 3 and 4 satisfying Assumption 2.3, the
closed-loop system (6) is said to be extended dissipative if the following inequality holds for
any T f  0 and all w(t ) ∈ L2 [0, ∞ )
Tf
J(t )dt − sup zT (t )4 z(t )  0, (8)
0tT f
0

where J(t ) = zT (t )1 z(t ) + 2zT (t )2 w(t ) + wT (t )3 w(t ).

Remark 2.2. The concept of extended dissipativity introduced in [25] generalizes the well-
recognized performances indices namely, L2 − L∞ , H∞ , passivity, (Q,S,R)- dissipativity. The
generalized extended dissipativity performance reduced to particular cases when weighting
matrices are captured as follows
Performance index (1 , 2 , 3 , 4 )
L2 − L∞ (0, 0, γ 2 I, I)

H∞ (-I, 0, γ 2 I, 0)

Passivity (0, I, γ I, 0)

Dissipativity (Q, S, R − γ I, 0)

1326
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Lemma 2.1. [44] For any constant matrix R  0 and continuously differentiable func-
b
tion x ∈ [a, b] → n , the following inequality holds: (b − a) x˙ T (s)Rx˙ (s)ds  [x(b) −
a
b
x(a)]T R[x(b) − x(a)] + 3χ T Rχ , where χ = x(b) + x(a) − 2
b−a
x(s)ds.
a

Lemma 2.2. [33] For any vectors x, y ∈  p , matrices G, H , matrix function (t ) with  T  
I and positive scalar , the following inequality is true 2x T GH y   −1 x T GGT x + yT H T H y.

3. Main results

This section contains the primary results to ensure the robust stability of the addressed IT2
fuzzy system (6) by establishing the gain values Kj , so that proposed IT-2 fuzzy controller
(4) stimulates the considered system to the equilibrium state. In order to achieve that, some
new set of sufficient conditions assuring the asymptotic stability of the system over the finite
period with determined feedback gains are derived and presented in the Theorem 1. Based on
those conditions, an algorithm is designed to determine the feedback gains which guarantee the
finite-time stability with extended dissipative attenuation level and are presented in Theorem 2.
For the sake of our convenience to derive the results, we employ the following vectors and
matrices representation,
⎡ ⎤
P11 0 0
ϑ1 (t ) = xT (t ) xT (t − τ (t )) xT (t − τ (t ) − ) , P1 = ⎣  0 ⎦,
T
P12
  P13


ξ (t ) = xT (t ) x˙ T (t ) xT (t − τ (t )) xT (t − τ (t ) − ) xT (t − y1 (t )) xT (t − y2 (t ))xT (t − y1 ) xT (t − y2 )

 1 (t )
t−y t  2 (t )
t−y t t
x (s)ds
T
x (s)ds
T
x (s)ds
T
x (s)ds
T
xT (s)ds wT (t ) T
.
t−y1 t−y1 (t ) t−y2 t−y2 (t ) t−τ (t )

3.1. Finite-time bounded analysis

Theorem 1. Consider the IT2 fuzzy system (6), for the given positive scalars ρj (j =
1, 2, . . . , r ), y1 , y2 , α1 , α2 , ω˜ , c1 , c2 , T f , δ, β0 , h̄, h̄k , , symmetric matrix L and controller
gain matrices K1j , K2j , if there exist positive definite symmetric matrices P0 , P11 , P12 ,
P13 , Q1 , Q2 , Q3 , Q4 , R1 , R2 , matrices N1 , N2 , with appropriate dimensions and scalar
l (l = 1, 2, . . . , 11), such that
ρi ij − ρi i + i ≺ 0, i = j, (9)

ρj ij − ρj i + i + ρi ji − ρi j + j ≺ 0, i ≺ j, (10)

ij − i ≺ 0, (11)

1327
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

θj − ρj j  0, (12)

˜ 1 + δ(1 − e−ω˜ Tf ) ≺ κ1 c2 e−ω˜ Tf ,

λc (13)
⎡ 1,1 1,2 1,3 1,16 ⎤
ij ij ij ... ij
⎢  2,2 2,3
... 2,16 ⎥
⎢ ij ij ij ⎥
⎢   3,31,1 ... 3,16 ⎥
where [ij ]16×16 = ⎢ ij ij = H e (N1 Ai ) − ω
ij ⎥,
˜ P0 −
⎢ .. .. .. .. .. ⎥
⎣ . . . . . ⎦
16,16
   ... ij
1,2
P11 + Q1 + Q2 + Q3 + Q4 − 4R1 − 4R2 − 4P h̄k
2
; ij = P0 + [ h̄k − τ (t )]P11 − N1 + N2 Ai ;
T

1,3 1,4 1,5 1,6

ij = N1 Bi K1j − h̄k ; ij = N1 Bi K2j ; ij = β0 N1 Adi − 2R1 ; ij = (1 − β0 )N1 Adi −
2P2

1,10 1,12 1,13 1,14 1,15

2R2 ; ij = y1 ;
6R1
ij = y2 ;
6R2
ij = h̄2 ;
6P2
ij = N1 Di ; ij = N1 Bi Gj ; 2,
ij =
2
k

( h̄k − τ (t ))P2 + y21 R1 +y22 R2 − H e(N2 ); ij = N2 Bi K1j ; ij = N2 Bi K2j ; ij = N2 β0 Adi ;

2,3 2,4 2,5
3,3 3,4
ij = N2 (1 − β0 )Adi ; ij = N2 Di ; ij = N2 Bi Gj ; ij = −P12 − h̄k ; ij =
2,6 2,14 2,15 4P2

( h̄k − τ (t ))P12 ; ij = h̄2 ; 3ij,16 = H1j ; 4,

3,13 6P2
ij = −P13 ; ij
4 4,16
= H2j ; 5ij,5 = −(1 −
k

α1 )Q1 − 8R1 ; 5ij,7 = −2R1 ; 5ij,9 = 6R y1

1
; 5ij,10 = 6R
y1
1
; 6ij,6 = −(1 − α2 )Q2 − 8R2 ; 6ij,8 =
−2R2 ; 6ij,11 = 6Ry2
2
; 6ij,12 = 6R
y2
2
; 7ij,7 = −Q3 − 2R1 ; 7ij,9 = 6R y1
1
; 8ij,8 = −Q4 − 2R2 ; 8ij,11 =
6R2
y2
; 9ij,9 = − 12R
y2
1
; 10,
ij
10
= − 12Ry2
1
; 11
ij
,11
= − 12Ry2
2
; 12,
ij
12
= − 12Ry2
2
; 13
ij
,13
= − 12P
h̄2
2
; 14,
ij
14
=
1 1 2 2 k
,15
−ω˜ I ; 15
ij = −;16,16 = −. κ1 = κmin (P̌0 ); κ2 = κmax (P̌0 ); κ3 = κmax (P̌11 ); κ4 =
κmax (P̌12 ); κ5 = κmax (P̌13 ); κ6 = κmax (P̌2 ); κ7 = κmax (Q̌1 ); κ8 = κmax (Q̌2 ); κ9 = κmax (Q̌3 );
κ10 = κmax (Q̌4 ); κ11 = κmax (Ř1 ); κ12 = κmax (Ř2 ); λ ˜ = κ2 + κ3 + κ4 + κ5 + h̄k κ6 + y1 κ7 +
y31 y32
y2 κ8 + y1 κ9 + y2 κ10 + 2 κ11 + 2 κ12 . Then closed-loop IT2 fuzzy system (6) is finite-time
bounded with respect to (c1 , c2 , T f , δ, L).
Proof. Construct the Lyapunov-Krasovskii functional candidate as

4
V (t, x(t )) = Vm (t, x(t )), (14)
m=1

with
V1 (t, x(t )) = xT (t )P0 x(t ),
t
V2 (t, x(t )) = ( h̄k − τ (t )) [ϑ1T (t )P1 ϑ1 (t ) + x˙ T (s)P2 x˙ (s)ds],
t−τ (t )
t t t
V3 (t, x(t )) = x (s)Q1 x(s)ds +
T
x (s)Q2 x(s)ds +
T
xT (s)Q3 x(s)ds
t−y1 (t ) t−y2 (t ) t−y1

t
+ xT (s)Q4 x(s)ds,
t−y2

1328
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

0  t 0  t
V4 (t, x(t )) = y1 x˙ (s)R1 x˙ (s)dsdθ + y2
T
x˙ T (s)R2 x˙ (s)dsdθ .
−y1 t+θ −y2 t+θ

Then, calculating the derivative of (14) along the trajectories of the closed-loop system (6) and
applying mathematical expectation, we get
E[V˙1 (t, x(t ))] = 2xT (t )P0 x˙ (t ), (15)

t
E[V˙2 (t, x(t ))] = −ϑ1T (t )P1 ϑ1 (t ) − x˙ T (s)P2 x˙ (s)ds + ( h̄k − τ (t )) (16)
t−τ (t )

2ϑ1T (t )P1 ϑ˙1 (t ) + x˙ T (t )P2 x˙ (t ) − x˙ T (t − τ (t ))P2 x˙ (t − τ (t )) , (17)

E[V˙3 (t, x(t ))] =

xT (t )[Q1 + Q2 + Q3 + Q4 ]x(t ) − (1 − α1 )xT (t − y1 (t ))Q1 x(t − y1 (t )) − (1 − α2 ) (18)
xT (t − y2 (t ))Q2 x(t − y2 (t )) − xT (t − y1 )Q3 x(t − y1 ) − xT (t − y2 )Q4 x(t − y2 ),

t
E[V˙4 (t, x(t ))] = y21 x˙ T (t )R1 x˙ (t ) − y1 x˙ T (s)R1 x˙ (s)dsdθ + y22 x˙ T (t )R2 x˙ (t )
t−y1
t
−y2 x˙ T (s)R2 x˙ (s)dsdθ . (19)
t−y2

Now, the integral term in (19) can be written as

t  1 (t )
t−y t
− x˙ (s)R1 x˙ (s)ds = −
T
x˙ (s)R1 x˙ (s)ds −
T
x˙ T (s)R1 x˙ (s)ds, (20)
t−y1 t−y1 t−y1 (t )

t  2 (t )
t−y t
− x˙ (s)R2 x˙ (s)ds = −
T
x˙ (s)R2 x˙ (s)ds −
T
x˙ T (s)R2 x˙ (s)ds. (21)
t−y2 t−y2 t−y2 (t )

Applying the Lemma 2.1 to the integral terms in (17), (20) and (21) we get
t
−1
− x˙ T (s)P2 x˙ (s)ds  [x(t ) − x(t − τ (t ))]T (22)
h̄k
t−τ (t )
3 T
P2 [x(t ) − x(t − τ (t ))] − χ P 2 χ0 ,
h̄k 0

1329
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

 1 (t )
t−y
−1
− x˙ T (s)R1 x˙ (s)ds  [x(t − y1 (t )) − x(t − y1 )]T (23)
y1
t−y1
3 T
R1 [x(t − y1 (t )) − x(t − y1 )] − χ R 1 χ1 ,
y1 1

t
−1 3
− x˙ T (s)R1 x˙ (s)ds  [x(t ) − x(t − y1 (t ))]T R1 [x(t ) − x(t − y1 (t ))] − χ2T R1 χ2 ,
y1 y1
t−y1 (t )
(24)

 2 (t )
t−y
−1
− x˙ T (s)R2 x˙ (s)ds  [x(t − y2 (t )) − x(t − y2 )]T
y2
t−y2
3 T
R2 [x(t − y2 (t )) − x(t − y2 )] − χ R 2 χ3 , (25)
y2 3

t
−1 3
− x˙ T (s)R2 x˙ (s)ds  [x(t ) − x(t − y2 (t ))]T R2 [x(t ) − x(t − y2 (t ))] − χ4T R2 χ4 .
y2 y2
t−y2 (t )
(26)
t
where χ0 = x(t ) + x(t − τ (t )) − 2
h̄k
x(s)ds; χ1 = x(t − y1 (t )) + x(t − y1 ) −
t−τ (t )
1 (t )
t−y t
2
y1
x(s)ds; χ2 = x(t ) + x(t − y1 (t )) − 2
y1
x(s)ds; χ3 = x(t − y2 (t )) + x(t − y2 ) −
t−y1 t−y1 (t )
2 (t )
t−y t
2
y2
x(s)ds; χ4 = x(t ) + x(t − y2 (t )) − 2
y2
x(s)ds. According to (6), the following
t−y2 t−y2 (t )
equation holds for any appropriate dimension matrices N1 , N2 :

  r r
2E xT (t )N1 + x˙ T (t )N2 i θj Ai x(t ) + β0 Adi x(t − y1 (t )) + (1 − β0 )
i=1 j=1
×Adi x(t − y2 (t )) + (μ(t ) − β0 )Adi (x(t − y1 (t )) − x(t − y1 (t )) + Bi K̄1j x(t − τ (t ))
+Bi K̄2j x(t − τ (t ) − ) + Di w(t ) − x˙ (t ) = 0.

(27)
Combining the equations from (15) to (27) along with the relation (5), then applying
Lemma 2.2 and Schur complement procedure, we yield

r 
r
V˙ (t, x(t )) − ω˜ V (t, x(t )) − ω˜ wT (t )w(t )  i θj ξ
T
(t )[ij ]16×16 ξ (t ). (28)
i=1 j=1

1330
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

r r
Note that i( j − θj )i = 0, then the following equation holds:
i=1 j=1


r 
r

i θj ij
i=1 j=1


r 
r 
r 
r
= i θj ij + i( j − θj )i
i=1 j=1 i=1 j=1


r 
r 
r 
r
= i (θj − ρj j + ρj j )ij + i( j − θj + ρj j − ρj j )i
i=1 j=1 i=1 j=1


r 
r 
r 
r 
r 
r
= i (θj + ρj j − ρj j )ij + i( j − ρj j )i − i (θj − ρj j )i
i=1 j=1 i=1 j=1 i=1 j=1


r 
r 
r 
r
= i (θj + ρj j − ρj j )ij + i (θj − ρj ) j (ij − i )
i=1 j=1 i=1 j=1


r r−1 
 r
= i (ρi ij
2
− ρi i + i ) + i j (ρj ij − ρj i + i + ρi ji − ρi j + j )
i=1 i=1 j=i+1


r 
r
+ i (θj − ρj j )(ij − i ).
i=1 j=1

From the Eqs. (9)-(12), we obtain V˙ (t, x(t )) − ω˜ V (t, x(t )) − ω˜ wT (t )w(t )  0.

Now, multiplying this inequality by eω˜ t and then integrating from 0 to T f , we obtain
Tf
e−ω˜ Tf V (t, x(t )) − V (0, x(0)) ≺ ω˜ e−ω˜ s wT (s)w(s)ds
0

ω˜ T f
V (t, x(t )) ≺ e V (0, x(0)) + δ(1 − e−ω˜ Tf ) . (29)

By denoting P̌0 = L −1/2 P0 L −1/2 , P̌11 = L −1/2 P11 L −1/2 , P̌12 = L −1/2 P12 L −1/2 , P̌13 =
L −1/2 P13 L −1/2 ,
P̌2 = L −1/2 P2 L −1/2 , Q̌1 = L −1/2 Q1 L −1/2 , Q̌2 = L −1/2 Q2 L −1/2 , Q̌3 = L −1/2 Q3 L −1/2 , Q̌4 =
−1/2
L Q4 L −1/2 , Ř1 = L −1/2 R1 L −1/2 , Ř2 = L −1/2 R2 L −1/2 .
It is computed as
E[V (t, x(t ))]  E[xT (t )P0 x(t )]  E[xT (t )L 1/2 P̌0 L 1/2 x(t]  κmin (P̌0 )

E[xT (t )Lx(t )]  κ1 E[xT (t )Lx(t )]. (30)

Further, V (0, x(0)) is calculated as follows

E[V (0, x(0))]

1331
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

0
= x (0)P0 x(0) + x (0)P11 x(0) + x (0)P12 x(0) + x (0)P13 x(0) +
T T T T
x˙ T (s)P2 x˙ (s)ds
−τ (0)

0 0 0 0
+ x (s)Q1 x(s)ds +
T
x (s)Q2 x(s)ds +
T
x (s)Q3 x(s)ds +
T
xT (s)Q4 x(s)ds
−y1 (0) −y2 (0) −y1 −y2

0 0 0 0
+ y1 x˙ T (s)R1 x˙ (s)dsdθ + y2 x˙ T (s)R2 x˙ (s)dsdθ ,
−y1 θ −y2 θ

 κmax (P̌0 )E[x (0)L x (0)] + κmax (P̌11 )E[xT (0)L x (0)] + κmax (P̌12 )E[xT (0)L x (0)]
T

+ κmax (P̌13 )E[xT (0)L x (0)] + h̄k κmax (P̌2 ) sup E[x˙ T (u)L x˙ (u)]
− h̄k u0

+ y1 κmax (Q̌1 ) sup E[x (u)Lx(u)] + y2 κmax (Q̌2 ) sup E[xT (u)Lx(u)]
T
−y1 u0 −y2 u0

+ y1 κmax (Q̌3 ) sup E[x (u)Lx(u)] + y2 κmax (Q̌4 ) sup E[xT (u)Lx(u)]
T
−y1 u0 −y2 u0

y3 y3
+ 1 κmax (Ř1 ) sup E[x˙ T (u)L x˙ (u)] + 2 κmax (Ř2 ) sup E[x˙ T (u)L x˙ (u)]
2 −y1 u0 2 −y2 u0

y31 y3
 [κ2 + κ3 + κ4 + κ5 + h̄k κ6 + y1 κ7 + y2 κ8 + y1 κ9 + y2 κ10 + κ11 + 2 κ12 ]
2 2
× sup E[xT (u)Lx(u), x˙ T (u)L x˙ (u)],
−y1 u0
˜ 1.
 λc (31)
From the Eqs. (29)-(31), we can get E[x (t )Lx(t )]  c2 , which is equivalent to the inequality
T

(7). According to Definition 2.1, we can conclude that the IT2 fuzzy closed-loop system (6) is
finite-time bounded with respect to (c1 , c2 , T f , δ, L). 
Remark 3.1. Theorem 1 presents the sufficient conditions with known gain for finite-time
stability of the IT2 fuzzy system by utilizing the IT-2 fuzzy memory sampled-data controller
(4). These conditions are derived with the aid of time dependent LKF in the frame of input-
delay approach. The terms tk , tk+1 in V2 (t, x(t )) imply that more info about actual sampling
pattern in this result. Further, the following theorem extends the results of Theorem 1 with
extended dissipativity performance index in order to obtain the gain matrices.

3.2. Finite-time bounded design with extended dissipative performance

Theorem 2. Consider the IT-2 fuzzy system (6), for the given positive scalars ρj (j =
1, 2, . . . , r ), y1 , y2 , α1 , α2 , ω˜ , c1 , c2 , T f , δ, β0 , h̄, , non-singular symmetric matrix L, 1 =
−˜ T1 ˜ 1 , 3  0 and any real matrix 2 , if there exist positive definite symmetric matrices X ,
P̄11 , P̄12 , P̄13 , Q̄1 , Q̄2 , Q̄3 , Q̄4 , R̄1 , R̄2 , matrices N̄1 , N̄2 , Y1j , Y2j with appropriate dimensions
and scalars l (l = 1, 2, . . . , 11), such that
ˆ ij − ρi 
ρi  ˆi+
ˆ i ≺ 0, i = j, (32)

1332
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

ˆ ij − ρj 
ρj  ˆi+
ˆ i + ρi 
ˆ ji − ρi 
ˆj+
ˆ j ≺ 0, i ≺ j, (33)

 ˆ i ≺ 0,
ˆ ij −  (34)

θj − ρj j  0, (35)

P0 − CiT 4 Ci  0, (36)

1 L −1 ≺ X ≺ L −1 , 0 ≺ P̌11 ≺ 2 L −1 , 0 ≺ P̌12 ≺ 3 L −1 , 0 ≺ P̌13 ≺ 4 L −1 ,

0 ≺ P̌2 ≺ 5 L −1 , 0 ≺ Q̌1 ≺ 6 L −1 , 0 ≺ Q̌2 ≺ 7 L −1 , 0 ≺ Q̌3 ≺ 8 L −1 , (37)
0 ≺ Q̌4 ≺ 9 L −1 , 0 ≺ Ř1 ≺ 10 L −1 , 0 ≺ Ř2 ≺ 11 L −1 .

 √
℘11 c1
≺ 0, (38)
 −1
⎡ ⎤
ˆ 1,1 ˆ 1,2 ˆ 1,3 ... ˆ 1,16
⎢ ij ij
ˆ 2,2
ij
ˆ 2,3 ˆ 2,16 ⎥
ij
⎢  ... ij ⎥
⎢ ij ij
ˆ 3,3 ˆ 3,16 ⎥
ˆ ij ]16×16
where [ = ⎢
⎢
  ij ... ⎥ 1,1
ij ⎥, ˆ ij = H e (ν1 Ai X ) − ω
˜ X − P̄11 +
⎢ .. .. .. .. .. ⎥
⎣ . . . . . ⎦
   ... ˆ 16,16
ij
+ X CiT ˜1 ˜1 Ci X ; ˆ 1ij,2 = X + [ h̄k − τ (t )]P̄11 −
T
Q̄1 + Q̄2 + Q̄3 + Q̄4 − 4R̄1 − 4R̄2 − 4P̄2
h̄k
ν1 X + ν2 AT X ; ˆ 1,3 = ν1 Bi Y1j − 2P̄2 ; ˆ 1,4 = ν1 Bi Y2j ; ˆ 1,5 = ν1 β0 Adi X − 2R̄1 ; ˆ 1,6 = ν1 (1 −
i ij h̄k ij ij ij
β0 )Adi X − 2R̄2 ; ˆ 1ij,10 = 6yR̄11 ; ˆ 1ij,12 = 6yR̄22 ; ˆ 1ij,13 = 6h̄P̄22 ; ˆ 1ij,14 = ν1 Di − X CiT 2 ; ˆ 1ij,15 =
k

ν1 Bi Gj ; ˆ 2,
ij = ( h̄k − τ (t ))P̄2 + y1 R̄1 + y2 R̄2 − H e (ν2 X );
2 2 2 ˆ 2,3 = ν2 Bi Y1j ;
ij
ˆ 2,4 =
ij
ν2 Bi Y2j ; ˆ 2,
ij
5
= ν2 β0 A di X ; ˆ 2,6 = ν2 (1 − β0 )Adi X ; ˆ 2,14 = ν2 Di ; ˆ 2,15 = ν2 Bi Gj ; ˆ 3,3 =
ij ij ij ij
−P̄12 − 4P̄22 ; ˆ 3,13 = 6P̄22 ; ˆ 3,16 = X H1j ; ˆ 4,4 = −P̄13 ; ˆ 4,16 = X H2j ; ˆ 5,5 = −(1 −
h̄k ij h̄k ij ij ij ij

α1 )Q̄1 − 8R̄1 ; ˆ 5ij,7 = −2R̄1 ; ˆ 5ij,9 = 6yR̄11 ; ˆ 5ij,10 = 6yR̄11 ; ˆ 6ij,6 = −(1 − α1 )Q̄2 − 8R̄2 ; ˆ 6ij,8 =
−2R̄2 ; ˆ 6ij,11 = 6yR̄22 ; ˆ 6ij,12 = 6yR̄22 ; ˆ 7ij,7 = −Q̄3 − 2R̄1 ; ˆ 7ij,9 = 6yR̄11 ; ˆ 8ij,8 = −Q̄4 − 2R̄2 ; ˆ 8ij,11 =
6R̄2 ˆ 9,9
y2
; = − 12R̄2 1 ; ˆ 10,10 = − 12R̄2 1 ; ˆ 11,11 = − 12R̄2 2 ; ˆ 12,12 = − 12R̄2 2 ; ˆ 13,13 = − 12P̄2 2 ; ˆ 14,14 =
ij y1 ij y1 ij y2 ij y2 ij h̄k ij
−3 I , ˆ 15
ij
,15
= −; ˆ 16
ij
,16
= −; ℘11 = [2 + 3 + 4 + h̄k 5 + y1 (6 + 8 ) + y2 (7 +
y3 y3
9 ) + 21 10 + 22 11 ]c1 + δ − δe−ω˜ Tf − c2 e−ω˜ Tf . Then closed-loop IT2 fuzzy system (6) is
finite-time bounded with extended dissipative with respect to (c1 , c2 , T f , δ, L). Moreover, gain
matrices of proposed controller are calculated as K1j = Y1j X −1 and K2j = Y2j X −1 .

1333
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Proof. Time derivative of the LKF (14) along with closed-loop IT-2 fuzzy system (6), when
combined with the inequalities from (15) to (28), we have

r 
r  
V˙ (t, x(t )) − ω˜ V (t, x(t )) − J(t )  ˜
i θj ξ (t ) ij ξ (t ).
T
(39)
i=1 j=1

where the elements of  ˜ ij are  ˜ 1×1 = H e(N1 Ai ) − ω˜ P0 − P11 + Q1 + Q2 + Q3 +

Q4 − 4R1 − 4R2 − h̄k − Ci 1 Ci ;
4P2 T ˜
1×2 = P0 + [ h̄k − τ (t )]P11 − N1 + N2 ATi ; ˜ 1×3 =
N1 Bi K1j − 2P 2
; ˜ 1×4 = N1 Bi K2j ; ˜ 1×5 = β0 N1 Adi − 2R1 ;
 ˜ 1×6 = (1 − β0 )N1 Adi −

h̄k
2R2 ;  ˜ 1×10 = 1 ;  6R ˜ 1×12 = 2 ; 
6R ˜ 1×13 = 22 ; 
6P ˜ 1×14 = N1 Di − C T 2 ;  ˜ 1×15 =
y1 y2 h̄k i
˜ 2×2 = ( h̄k − τ (t ))P2 + y21 R1 + y22 R2 − H e(N2 );
N1 Bi Gj ; ˜ 2×3 = N2 Bi K1j ;˜ 2×4 =
˜ ˜ ˜ ˜
N2 Bi K2j ;2×5 = N2 β0 Adi ;2×6 = N2 (1 − β0 )Adi ;2×14 = N2 Di ;2×15 = N2 Bi Gj ;3×3 = ˜
−P12 − 4P 2
;˜ 3×13 = 6P22 ; ˜ 3×16 = H1j ; ˜ 4×4 = −P13 ; ˜ 4×16 = H2j ; ˜ 5×5 = −(1 −
h̄k h̄k
α1 )Q1 − 8R1 ; ˜ 5×7 = −2R1 ; ˜ 5×9 = 6R1 ; ˜ 5×10 = 6R1 ;˜ 6×6 = −(1 − α2 )Q2 − 8R2 ; ˜ 6×8 =
y1 y1
−2R2 ; ˜ 6×11 = 6R2 ; ˜ 6×12 = 6R2 ; ˜ 7×7 = −Q3 − 2R1 ; ˜ 7×9 = 6R1 ;
˜ 8×8 = −Q4 −
y2 y2 y1
2R2 ; ˜ 8×11 = 2 ;
6R ˜ 9×9 = − 2 1 ;
12R ˜ 10×10 = − 2 1 ;
12R ˜ 11×11 = − 2 2 ;
12R ˜ 12×12 = − 12R2 2 ;
˜ 13×13 =
y2 y1 y1 y2 y2
− 12P2 2 ;˜ 14×14 = −3 I ; ˜ 15×15 = −; ˜ 16×16 = −.
h̄k
˜ ij ≺ 0, it follows V˙ (t, x(t )) − J(t )  0. Integrating the relation from 0 to t, we
Since 
get that
t
J(t )dt  V (t, x(t )) − V (0, x(0))  xT (t )P0 x(t )  0. (40)
0

Now, we investigate the two cases of 4 = 0 and 4  0.

In consonance with the framework of extended dissipativity, we can present the H∞ ,
strictly (Q, S, R) dissipative and passivity performance condition when 4 = 0 and L2 − L∞
performance condition when 4  0. First, we consider 4 = 0; from (40), we conclude
Tf
J(t )dt  0.
0
Consequentially, we consider the second case 4  0. In this case, we have the matrices
1 = 0, 2 = 0, 3  0 as mentioned in the Definition 2.2. Then, for any 0  T f  t, the
Tf t
inequality (40) leads to the relation J(t )dt  J(t )dt  xT (t )P0 x(t )  0. Therefore, by
0 0
Tf
equation (36), we have J(t )dt  x (t )P0 x(t )  xT (t )C T 4 Cx(t ) = zT (t )4 z(t )  0. which
T
0
immediately follows from the equation (8). Thus, from above the analysis of two cases of
4 = 0 and 4  0, we can conclude that closed-loop IT2 fuzzy system (6) is finite-time
bounded with extended dissipative by Definition 2.2. Now, we notate the following relations
P0−1 = X , N1 = ν1 P0 N2 = ν2 P0 , X PX = P̄, X QX = Q̄, X MX = M̄ , X RX = R̄, Then, pre-
and post- multiplying LMI ij by diag{X , X , X , X , X , X , X , X , X , X , X , X , X , I , I , I },
LMI  ˆ ij is obtained. In addition, from Theorem 1, it is noted that κ1 ≺ L −1/2 P0 L −1/2 ≺
κ2 , 0 ≺ L −1/2 P11 L −1/2 ≺ κ3 , 0 ≺ L −1/2 P12 L −1/2 ≺ κ4 , 0 ≺ L −1/2 P13 L −1/2 ≺ κ5 , 0 ≺
L −1/2 P2 L −1/2 ≺ κ6 , 0 ≺ L −1/2 Q1 L −1/2 ≺ κ7 , 0 ≺ L −1/2 Q2 L −1/2 ≺ κ8 , 0 ≺ L −1/2 Q3 L −1/2 ≺ κ9 ,
0 ≺ L −1/2 Q4 L −1/2 ≺ κ10 , 0 ≺ L −1/2 R1 L −1/2 ≺ κ11 , 0 ≺ L −1/2 R2 L −1/2 ≺ κ12 .

1334
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

By applying congruence transformation, the above relations get remodelled as κ2−1 L −1 ≺

X ≺ L −1 κ1 , 0 ≺ P̄11 ≺ κ1−2 κ3 , 0 ≺ P̄12 ≺ κ1−2 κ4 , 0 ≺ P̄13 ≺ κ1−2 κ5 , 0 ≺ P̄2 ≺ κ1−2 κ6 , 0 ≺ Q̄1 ≺
κ1−2 κ7 , 0 ≺ Q̄2 ≺ κ1−2 κ8 , 0 ≺ Q̄3 ≺ κ1−2 κ9 , 0 ≺ Q̄4 ≺ κ1−2 κ10 , 0 ≺ R̄1 ≺ κ1−2 κ11 , 0 ≺ R̄2 ≺
κ1−2 κ12 .
Now, the relation (37) is easily obtained by denoting the parameters as κ1 = 1, κ2−1 = 1
and κq = q−1 , (q = 2, . . . , 12). By utilizing the above relations, the condition in (13) can
be rewritten as

y3 y3
1−1 + 2 + 3 + 4 + h̄k 5 + y1 (6 + 8 ) + y2 (7 + 9 ) + 1 10 + 2 11
2 2
c1 + δ(1 − e−ω˜ Tf ) ≺ c2 e−ω˜ Tf . (41)
By applying Schur complement procedure to the inequality (41), it is easy to obtain (38).
This completes the proof. 
Remark 3.2. It is noted that, the stability conditions for considered IT2 fuzzy system (1) by
applying IT2 fuzzy control law are derived in terms of LMIs that presented in Theorem 2 with
the remark that the system and control protocol are under imperfect premise matching. In
other words, the lower and upper bound of membership grade does not require to be same
as membership function of the proposed controller. As a result, it provides a high flexibility
to the IT2 fuzzy controller design than the perfect premise matching.
Now, the case of plant without delay is presented in the following corollary (i.e) the
finite-time stability conditions for IT2 fuzzy system without delay are addressed.
Corollary 1. The closed-loop IT-2 fuzzy system (6) is finite-time stability with extended dissi-
pativity performance attenuation level γ with respect to (c1 , c2 , T f , δ, L), for the given positive
scalars ρi , ω˜ , c1 , c2 , T f , δ, h̄, h̄k , , non-singular symmetric matrix L, 1 = −˜ T1 ˜ 1 , 3  0
and any real matrix 2 , if there exist positive definite symmetric matrices X , P̄11 , P̄12 , P̄13 ,
matrices N̄1 , N̄2 , with appropriate dimensions and scalar l (l = 1, 2, . . . , 5), such that the
LMI (32)-(36) and following LMI hold for i, j = 1, 2, . . . , r
 √
℘11 c1
≺ 0, (42)
 −1

1 L −1 ≺ X ≺ L −1 , 0 ≺ P̌11 ≺ 2 L −1 , 0 ≺ P̌12 ≺ 3 L −1 ,
0 ≺ P̌13 ≺ 4 L −1 , 0 ≺ P̌2 ≺ 5 L −1 . (43)
where ⎡ ⎤
π11 π12 ν1 Bi Y1j − 2h̄P̄k2 ν1 Bi Y2j 6P̄2
h̄2k
ν1 Di − X CiT 2 ν1 Bi Gj 0
⎢ ⎥
⎢  π22 ν2 Bi Y1j ν2 Bi Y2j 0 ν2 Di ν2 Bi Gj 0 ⎥
⎢ ⎥
⎢   −P̄12 − 4h̄P̄22 0 6P̄2
0 0 X H1j ⎥
⎢ h̄2k ⎥
⎢   
k
−P̄13 X H2j ⎥
ij = ⎢
ˆ
⎢
0 0 0 ⎥,
⎥
⎢     − 12h̄P̄2 2 0 0 0 ⎥
⎢ ⎥
⎢     
k
−3 0 0 ⎥
⎢ ⎥
⎣       − 0 ⎦
       −
1335
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

π11 = H e(ν1 Ai X ) − ω˜ X − P̄11 − 4h̄P̄k2 + X CiT ˜1 ˜1 Ci X ,

T
π12 = X + [ h̄k − τ (t )]P̄11 −
ν1 X + ν2 Ai X ,
T
π22 = ( h̄k − τ (t ))P̄2 − H e(ν2 X ), ℘11 = [2 + 3 + 4 + h̄k 5 ]c1 +
δ − δe−ω˜ Tf − c2 e−ω˜ Tf . Moreover, gain matrices of proposed controller are calculated as
K1j = Y1j X −1 and K2j = Y2j X −1 .

Remark 3.3. The asymptotic stability criteria of closed-loop IT-2 fuzzy system (6) over the
finite period of time in Theorem 1, 2 and Corollary 1 are presented interms of LMIs where
desired gain matrices of memory sampled-data controller can be computed by solving these
LMI constraints. While solving the LMI based constraints, the only problem arising is that
computational issue. It is noted that, size of LMI or number of decision variable is directly
proportional to the computational burden. Fortunately, all the computations in LMIs are offline
and with the help of existing software technology such that the proposed LMI-based stability
criteria (9)-(13), (32)-(38) and (42)-(43) can be easily solved. Moreover, slack matrices are
introduced to derived the LMI based stability criteria, which can significantly reduces the
conservatism and improves the feasibility region.

4. Numerical examples

In this section, two numerical examples are illustrated to verify the adaptability of our
results derived in the paper. Moreover, the efficiency of our proposed controller design is ex-
hibited in the following examples. To be precise, capability of the IT2 fuzzy controller against
the random delay, perturbations and uncertainties are demonstrated in Example 1 whereas in
Example 2, efficiency of the controller over chaotic system is verified in the presence and
absence of the gain perturbation.

Example 1. Consider 3-rule IT2 fuzzy system of the form (1) with following matrices and
parameters
  
1.51 1.623 1.32 1.415 1.24 1.53
A1 = , A2 = , A3 = ,
−0.2 −0.018 −0.7 −0.570 −0.5 −0.36
  
−1.20 0.150 −1.30 0.110 −1.10 0.14
Ad1 = , Ad2 = , Ad3 = ,
−0.22 0.120 −0.47 0.13 −0.92 0.16
  T  T  T
0 0.214 0.120 0.206
Bi = , i = 1, 2, 3; C1 = , C2 = , C3 = ,
−1 −0.128 −0.120 −0.021
    T  T  T
−0.05 −0.08 −0.07 0.3 0.1 0.3
D1 = , D2 = , D3 = , G1 = , G2 = , G3 = ,
0.21 0.1 0.15 0.2 0.3 0.2
 T  T  T  T  T
−0.5 0.3 −0.5 0.3 −0.2
H11 = , H12 = , H13 = , H21 = , H22 = ,
0.1 −0.2 0.1 −0.2 0.1
 T
0.3
H23 = .
−0.2

In addition, bounds of membership function of the closed-loop IT-2 fuzzy system (6) and
resilient memory sampled data controller (4) are as follows

1336
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Table 1
Various disturbance attenuation approach.

Performance index γ value

L2 − L∞ case 0.9
H∞ case 0.9
Passivity case 0.9
Dissipativity case 0.2

LMF UMF

φ1L (x1 (t )) = 1 − 1
φ1U (x1 (t )) = 1 − 1
1+e−(x1 (t )+5+3) 1+e−(x1 (t )−5+1)

φ3L (x1 (t )) = 1
φ3U (x1 (t )) = 1
1+e−(x1 (t )−5+1) 1+e−(x1 (t )−5+3)

φ2L (x1 (t )) = 1 − φ1U (x1 (t )) − φ3U (x1 (t )) φ2U (x1 (t )) = 1 − φ1L (x1 (t )) − φ3L (x1 (t ))

LMF’s and UMF’s of the plant

LMF UMF

θ1L (x1 (t )) =1− 1

x (t )+5 θ1U (x1 (t )) = 1 − 1
x (t )+4
− 1 2 − 1 2
1+e 1+e

θ2L (x1 (t )) = 1
−x1 (t )−5 θ2U (x1 (t )) = 1
−x1 (t )−4
1+e 2 1+e 2

θ3L (x1 (t )) = 1 − θ1U (x1 (t )) − θ2U (x1 (t )) θ3U (x1 (t )) = 1 − θ1L (x1 (t )) − θ2L (x1 (t ))

LMF’s and UMF’s of the controller

while nonlinear weighting coefficients υi (x(t )) = sin2 (x1 (t )) and ζj = cos2 (x1 (t )) are
employed to determine the grades of the membership function of the IT-2 fuzzy system,
i (x(t )), i = 1, 2, 3 and controller, θj (x(t )), j = 1, 2, 3 respectively. Furthermore, the proba-
bilistic time-varying delay y(t ) follows a bernoulli distribution with parameter β0 = 0.3 in in-
terval range, 0  y1 (t )  1; 1 ≺ y2 (t )  2 and their corresponding delay derivatives are α1 =
0.5, α2 = 0.5. Now, in consonance with structure of IT-2 fuzzy resilient memory sampled-
data controller (4), we obtain maximal sampling interval as hk = 0.3. Also, for simulation
purpose, we take finite-time parameters (c1 , c2 , T f , δ, L) as (0.6, 11, 25, 0.9, I) along with
other scalars  = 0.1, ν1 = 0.3, ν2 = 0.4, ω˜ = 0.05, ρ1 = 0.2, ρ2 = 0.3, ρ3 = 0.4,  = 0.05.
Moreover, the effective performance of the system are verified with the input disturbance
w(t ) = 0.01 ∗ tanh (t/100) ∗ sin(20t ). Next, under a single frame, we designed IT2 fuzzy
memory sampled-data controller (4) to satisfy H∞ , L2 − L∞ , passivity and dissipativity per-
formances whose performances indices are tabulated in Table 1. By utilizing above stated
values, and then solving LMI constraints stated in the Theorem 2, we arrive at the feasibility
solution along with the control gain matrices
K11 = −0.2830 0.1765 ; K12 = −0.9466 0.2214 ; K13 = −0.4776 0.1880 ;

K21 = 0.8112 0.8236 × 10−3 ; K22 = −0.0117 −0.0002 and

−3
K23 = 0.5835 0.7671 × 10 .
Based upon the above stated values and the corresponding initial condition of the state
x(0) = [2.0 0.0]T , the simulation results are exhibited in Figs. 1 - 4. In brief, Fig. 1 shows

1337
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Fig. 1. Response of State.

Fig. 4. Evolution of xT (t )Lx(t ) .

Fig. 2. Output Response.

the trajectory response of the state which reveals that the state response smoothly converges
to zero within the finite-time period even in the presence of input disturbance and time-delay.
This also indicates that our proposed IT-2 fuzzy memory sampled-data controller is much
more efficient to stabilize the system. The corresponding output response of the system is
plotted in Fig. 2. Besides, trajectory of controller response is depicted in Fig. 3 which attests
that the controller reaches its equilibrium state over prescribed period of time. The evolution of
xT (t )Lx(t ) is shown in Fig. 4 from which it can be seen that the initial values are obeying the

1338
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Fig. 3. Controller Trajectory.

Fig. 5. Bernoulli random variable μ(t ).

Fig. 6. Random delay y1 (t ).

condition xT (0)Lx(0) ≺ c1 and trajectories are satisfying xT (t )Lx(t ) ≺ 11 = c2 implying that

IT-2 fuzzy system is finite-time bounded. In Fig. 5, the variation of Bernoulli random variable
μ(t ) with β0 = 0.3 is portrayed. In Fig. 6 and Fig. 7 the probabilistic time-varying delay y(t )
ranges from 0  y1  1 and 1 ≺ y1  2 are exhibited respectively.Fig. 8 and Fig. 9 represent
pictorial view of grades of membership funtion of the plant and controller respectively. From
these simulation results, it is evident that our proposed controller (4) adequately handles the
system and makes the addressed system to perform in a efficient nature.

1339
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Fig. 7. Random delay y2 (t ).

Fig. 8. Membership function of the plant.

Fig. 9. Membership function of the controller.

Example 2. This example is presented to verify adeptness of controller (4) based on the result
presented in Corollary 1, over chaotic system which is represented by the following Rossler’s
equation
x˙ 1 (t ) = −x2 (t ) − x3 (t )
x˙ 2 (t ) = x1 (t ) + px2 (t ) (44)
x˙ 3 (t ) = qx1 (t ) − (r − x1 (t ))x3 (t ) + u(t )

1340
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

where x1 (t ) ∈ [r + s, r − s], p = 0.2, q = 1, s = 12 with r as uncertain parameter r(t ) ∈

[0, 5]. Then, we represent the Rossler’s system (44) by the following 2-rule fuzzy model
Rule ‘i’ : IF x1 (t ) is i (x1 (t )), THEN

x˙ (t ) = Ai x(t ) + Bi u(t ), i = 1, 2,
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
0 −1 −1 0 −1 −1 0
in which matrices are A1 = ⎣1 p 0 ⎦, A 2 = ⎣ 1 p 0 ⎦, B i = ⎣0 ⎦, i = 1 , 2
q 0 −s q 0 s 1
and
1 (x(t )) = 1/2 + (r (t ) − x1 (t ))/2s and 2 (x(t )) = 1/2 − (r (t ) − x1 (t ))/2s.

Since uncertain parameter r(t ) ranges in [0 5], we obtain lowerbound and upper-
bound of the membership function which are tabulated below. Those membership func-
tion grades are determined by choosing nonlinear weighted functions υi (x(t )) = sin2 (x1 (t ))
and ζj = cos2 (x1 (t )), i, j = 1, 2. In addition, output and disturbance coefficient constraint
T
matrices are taken as Ci = 0.01 0.01 0 , Di = 0.02 0.02 0 , i = 1, 2. In asso-
ciate to the resilient controller (5), the perturbed matrices of the controller are considered
as G1 = 0.09 0.08 0.05 , G2 = 0.07 0.09 0.01 ,H11 = 0.07 0.01 0.1 , H12 =
0.06 0.1 0.2 ,H21 = 0.5 0.2 0.5 ,H22 = 0.2 0.1 0.1 . Assinging the values to
the finite-time scalars as c1 = 0.6, c2 = 20, T f = 50, ω˜ = 0.05, L = I . We assume sampling
interval hk = 0.35 and other parameters h̄ = 0.4,  = 0.09, ν1 = 0.1, ν2 = 0.4, ρ1 = 0.1, ρ2 =
0.2, and  = 0.05. We take exogenous disturbance as w(t ) = 0.5 sin (0.5t ) satisfying the con-
Tf
dition wT (t )w(t )dt ≺ 0.9. The extended disspativity mechanism is proposed to reduce effect
0
of the disturbance affecting the system behaviour. Performance attenuation indices of the ex-
tended disspativity are drawn as
Performance L2 − L∞ H∞ Passivity Dissipativity

γ value 0.1 0.1 0.001 0.01

.
With the above parameters, we arrive the feasibility solution of LMI constraints presented
in the Theorem 2 and the corresponding control gains are as follows

K11 = 0.0447 −0.0049 −1.0040 ; K12 = 0.0441 −0.0048 −0.9980 ;

K21 = 0.0005 −0.0001 −0.0122 ; K22 = 0.0004 −0.0000 −0.0082 .

The efficiency of our proposed controller (4) against the chaotic system which is represented in
Rossler’s equation is illustrated in the Figs. 10-16. In Fig. 10 and Fig. 11 the state trajectories
of the plant without gain perturbation and with perturbation are depicted respectively. From
these figures we can see that trajectories converges to zero under influence of controller but
duration of system attaining its equilibrium state is interrupted by gain fluctuation. Controller
response is plotted respectively in Fig. 12 and Fig. 13 under absence and presence of Kj .
The unforced system i.e. the system without controller protocol is shown in Fig. 14. The
output response of the system exhibited in Fig. 15. From Fig. 16, it can be seen that the
evolution of xT (t )Lx(t ) satisifies the Assumption 2.2 that xT (t )Lx(t )  c2 = 20.

1341
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Fig. 10. State trajectory.

Fig. 11. State trajectory with gain fluctuation.

Fig. 12. Controller Response .

The behaviour of the chaotic system represented by Rossler’s equation (44) is depicted in
Fig. 17. It is observed from the above simulation results that IT2 fuzzy memory sampled-data
controller (4) is much capable to make the chaotic system stable, which also implies that our
proposed controller may help the other complex dynamical system to attains its stability.

1342
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Fig. 13. Controller Response with gain fluctuation .

Fig. 14. State trajectory of unforced system.

Fig. 15. Response of Output.

1343
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

Fig. 16. Time history of xT (t )Lx(t ).

Fig. 17. Behaviour of the choatic system.

LMF UMF
   
0−x1 (t ) 5−x1 (t )
Plant φ1L (x1 (t )) = 21 1 + s φ1U (x1 (t )) = 1
2 1+ s
   
5−x1 (t ) 0−x1 (t )
φ2L (x1 (t )) = 1
2 1− s φ2U (x1 (t )) = 1
2 1− s

−x1 (t )2
θ1U (x1 (t )) = e−x1 (t )
2
Controller θ1L (x1 (t )) = e 0.5

−x1 (t )2
θ2L (x1 (t )) = 1 − e−x1 (t )
2
θ2U (x1 (t )) = 1 − e 0.5

Membership Grades of plant and controller

5. Conclusion

In this paper, the problem of finite-time extended dissipative of IT2 fuzzy systems with
resilient memory-sampled data controller is discussed. An IT2 fuzzy systems is studied with
consideration of probabilistic time-varying delay and exogenous disturbance. The resilient IT2
fuzzy memory-sampled data controller is designed to guarantee that the IT2 fuzzy systems is
finite-time extended dissipative. Further, based on extended dissipative performance, L2 − L∞ ,
H∞ , passivity, (Q, S, R)- dissipative performances are studied in a unified framework. Based

1344
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

on the Lyapunov-krasovskii approach, a set of conditions are derived in terms of LMI. Finally,
two simulation results are presented to verify the efficiency of the proposed method. On
the other hand, our proposed methodology could be extended to investigate the problem of
guaranteed cost control for interval-valued fuzzy system using PI controller subject to time-
varying delay which would be our future work.

Author Contribution

All authors read and approved the final manuscript.

Declaration of Competing Interest

The authors declare that they have no conflicts of interest.

References

[1] T. Takagi, M. Sugeno, Fuzzy identification of systems and it’s application to modeling and control, IEEE Trans.
Syst. Man Cybern. 1 (1985) 116–132.
[2] M. Sugeno, G.T. Kang, Structure identification of fuzzy model, Fuzzy Sets Syst. 28 (1) (1988) 15–33.
[3] J. Wang, J. Xia, H. Shen, M. Xing, J.H. Park, H∞ Synchronization for fuzzy markov jump chaotic systems
with piecewise-constant transition probabilities subject to PDT switching rule, IEEE Trans. Fuzzy Syst. (2020),
doi:10.1109/TFUZZ.2020.3012761.
[4] J.M. Mendel, N.N. Karnik, Q. Liang, Type-2 fuzzy logic systems, IEEE Trans. Fuzzy Syst. 7 (6) (1999) 643–658.
[5] H.K. Lam, H. Li, C. Deters, H. Wuerdemann, E. Secco, K. Althoefer, Control design for interval type-2 fuzzy
systems under imperfect premise matching, IEEE Trans. Ind. Electron. 61 (2) (2014) 956–968.
[6] H.K. Lam, L.D. Seneviratne, Stability analysis of interval type-2 fuzzy-model-based control systems, IEEE
Trans. Syst. Man Cybern. 38 (3) (2008) 617–628.
[7] M. Hamid, L.D. Elaheh, Designing a fuzzy type-2 model-based robust controller for ball and beam system,
Procedia Comput. Sci. 105 (2017) 125–130.
[8] X. Xie, S. Li, B. Xu, Adaptive event-triggered H∞ fuzzy filtering for interval type-2 T–S fuzzy-model-based
networked control systems with asynchronously and imperfectly matched membership function, J. Frankl. Inst.
356 (18) (2019) 11760–11791.
[9] C. Wu, H. Li, H.K. Lam, Fault detection for nonlinear networked systems based on quantization and dropout
compensation: an interval type-2 fuzzy-model method, Neurocomputing. 191 (2016) 409–420.
[10] X. Sun, H. Zhang, Q. Shan, Y. Wang, H∞ control of interval type-2 fuzzy logic system with time-delay partition
method, Neurocomputing. 275 (2018) 200–207.
[11] Z. Du, Y. Kao, J.H. Park, Interval type-2 fuzzy sampled-data control of time-delay systems, Inf. Sci. 487 (2019)
193–207.
[12] N.H. Muoi, G. Rajchakit, V.N. Phat, LMI approach to finite-time stability and stabilization of singular linear
discrete delay systems, Acta Appl. Math. 146 (1) (2016) 81–93.
[13] S. Yan, M. Shen, S.K. Nguang, G. Zhang, L. Zhang, A distributed delay method for event-triggered control of
T-S fuzzy networked systems with transmission delay, IEEE Trans. Fuzzy Syst. 27 (10) (2019) 1963–1973.
[14] G. Nagamani, S. Ramasamy, Stochastic dissipativity and passivity analysis for discrete-time neural networks
with probabilistic time-varying delays in the leakage term, Appl. Math. Comput. 289 (2016) 237–257.
[15] R. Samidurai, R. Sriraman, Non-fragile sampled-data stabilization analysis for linear systems with probabilistic
time-varying delays, J. Frankl. Inst. 356 (8) (2019) 4335–4357.
[16] R. Cheng, M. Peng, Adaptive synchronization for complex networks with probabilistic time-varying delays, J.
Frankl. Inst. 353 (18) (2016) 5099–5120.
[17] P. Muthukumar, K. Subramanian, S. Lakshmanan, Robust finite time stabilization analysis for uncertain neural
networks with leakage delay and probabilistic time-varying delays, J. Frankl. Inst. 353 (16) (2016) 4091–4112.
[18] B. Kaviarasan, R. Sakthivel, Y. Shi, Reliable dissipative control of high-speed train with probabilistic time-vary-
ing delays, Int. J. Syst. Sci. 47 (16) (2016) 3940–3951.
[19] H. Gao, J. Xia, G. Zhuang, Robust finite-time extended dissipative control for a class of uncertain switched
delay systems, Int. J. Cont. Autom. Syst. 16 (3) (2018) 1459–1468.

1345
N. Sakthivel, C.A. Suruthi Sri and G. Zhai Journal of the Franklin Institute 359 (2022) 1320–1346

[20] M. Li, J. Zhao, J. Xia, G. Zhuang, W. Zhang, Extended dissipative analysis and synthesis for network control
systems with an event-triggered scheme, Neurocomputing. 312 (2018) 34–40.
[21] H. Li, L. Wu, H.K. Lam, Y. Gao, Sampled-data control of interval type-2 fuzzy-model-based systems. in:
analysis and synthesis for interval type-2 fuzzy-model-based systems, Springer. 4 (2016) 51–72.
[22] H. Ju, H. Shen, Z.G. Wu, Finite-time reliable L2 − L∞ /H∞ control for takagi-sugeno fuzzy systems with
actuator faults, IET Control Theory Appl. 8 (9) (2014) 688–696.
[23] F. Serrano, J.M. Rossell, Hybrid passivity based and fuzzy type-2 controller for chaotic and hyper-chaotic
systems, Acta Mech. Autom. 11 (2) (2017) 96–103.
[24] Z. Zhang, Q. Zhou, C. Wu, H. Li, Dissipativity-based reliable interval type-2 fuzzy filter design for uncertain
nonlinear systems, Int. J. Fuzzy Syst. 20 (2) (2018) 309–402.
[25] B. Zhang, W. Zheng, S. Xu, Filtering of markovian jump delay systems based on a new performance index,
IEEE Trans. Circ. Syst. 60 (5) (2013) 1250–1263.
[26] J. Wang, Z. Huang, Z. Wu, J. Cao, H. Shen, Extended dissipative control for singularly perturbed PDT switched
systems and its application, IEEE Trans. Circuit. Syst. I (2020), doi:10.1109/TCSI.2020.3022729.
[27] S. Zhou, Y. Han, Extended dissipativity and control synthesis of interval type-2 fuzzy systems via line integral
Lyapunov function, IEEE Trans. Fuzzy Syst. (2019), doi:10.1109/TFUZZ.2019.2945258.
[28] H. Li, Y. Pan, Q. Zhou, Filter design for interval type-2 fuzzy systems with D stability constraints under a
unified frame, IEEE Trans. Fuzzy Syst. 23 (3) (2015) 719–725.
[29] X. Tang, L. Deng, J. Yu, H. Qu, Output feedback predictive control of interval type-2 T—S fuzzy systems with
Markovian packet loss, IEEE Trans. Fuzzy Syst. 26 (4) (2018) 2450–2459.
[30] H. Chaoui, M. Khayamy, A.A. Aljarboua, Adaptive interval type-2 fuzzy logic control for PMSM drives with
a modified reference frame, IEEE Trans. Ind. Electron. 64 (5) (2017) 3786–3797.
[31] H. Li, J. Wang, L. Wu, H. Lam, Y. Gao, Optimal guaranteed cost sliding-mode control of interval type-2 fuzzy
time-delay systems, IEEE Trans. Fuzzy Syst. 26 (1) (2018) 246–257.
[32] H. Li, X. Sun, P. Shi, H.K. Lam, Control design of interval type-2 fuzzy systems with actuator fault: sampled-data
control approach, Inf. Sci. 302 (2015) 1–13.
[33] C. Ge, Y. Shi, J.H. Park, C. Hua, Robust H∞ stabilization for T–S fuzzy systems with time-varying delays and
memory sampled-data control, Appl. Math. Comput. 346 (2019) 500–512.
[34] E. Fridman, A refined input delay approach to sampled-data control, Automatica 46 (2) (2010) 421–427.
[35] Y. Wu, H.R. Karimi, R. Lu, Sampled-data control of network systems in industrial manufacturing, IEEE Trans.
Ind. Electron. 65 (11) (2018) 9016–9024.
[36] J. Luo, X. Liu, W. Tian, S. Zhong, K. Shi, J. Cheng, A new approach to generalized dissipativity analysis for
fuzzy systems with coupling memory sampled-data control, Appl. Math. Comput. 368 (2020) 124774.
[37] Y. Liu, J.H. Park, B. Guo, Y. Shu, Further results on stabilization of chaotic systems based on fuzzy memory
sampled-data control, IEEE Trans. Fuzzy Syst. 26 (2) (2018) 1040–1045.
[38] X. Wang, X. Liu, K. She, S. Zhong, Q. Zhong, Extended dissipative memory sampled-data synchronization
control of complex networks with communication delays, Neurocomputing. 347 (2018) 1–12.
[39] Y. Liu, B. Guo, J.H. Park, S. Lee, Nonfragile exponential synchronization of delayed complex dynamical
networks with memory sampled-data control, IEEE Trans. Neural Netw. Learn. Syst. 29 (1) (2018) 118–128.
[40] T. Zhao, L. Jiahao, S. Dian, Finite-time control for interval type-2 fuzzy time-delay systems with norm-bounded
uncertainties and limited communication capacity, Inf. Sci. 483 (2019) 153–173.
[41] Q. Lu, P. Shi, H.K. Lam, Y. Zhao, Interval type-2 fuzzy model predictive control of nonlinear networked control
systems, IEEE Trans. Fuzzy Syst. 23 (6) (2015) 2317–2328.
[42] X. Song, S. Song, L. Liu, B. Inés Tejado, Adaptive interval type-2 fuzzy sliding mode control for fractional-order
systems based on finite-time scheme, J. Intel. Fuzzy Syst. 32 (3) (2017) 1903–1915.
[43] X. Sun, H. Zhang, J. Han, Y. Wang, Non-fragile control for interval type-2 T–S fuzzy logic control systems
with time-delay, J. Frankl. Inst. 354 (18) (2017) 7997–8014.
[44] K. Liu, E. Fridman, Wirtinger’s inequality and Lyapunov-based sampled-data stabilization, Automatica 48 (1)
(2012) 102–108.

1346