Circuits, Systems, and Signal Processing (2020) 39:4868–4889

https://doi.org/10.1007/s00034-020-01414-w

An Input–Output Approach to Anti-windup Design for

Sampled-Data Systems with Time-Varying Delay

Ouarda Lamrabet1 · Khadija Naamane1 · El Houssaine Tissir1 ·

Fatima El Haoussi1 · Fernando Tadeo2

Received: 11 June 2019 / Revised: 31 March 2020 / Accepted: 2 April 2020 /

Published online: 22 April 2020
© Springer Science+Business Media, LLC, part of Springer Nature 2020

Abstract
A methodology for designing anti-windup compensators is investigated, for sampled-
data systems with delays and actuator saturation. More precisely, criteria for the
existence of an anti-windup compensator that ensure simultaneously stability and
an H∞ norm bound in closed-loop are developed, thanks to the use of a three-term
approximation of the delays, of the scaled small gain theorem, and of a Wirtinger-
based inequality. The criteria are in the form of a set of linear matrix inequalities:
an optimization algorithm is proposed to maximize the estimated domain of attrac-
tion that can be easily implemented. Some simulation examples are also provided to
demonstrate the superiority of the proposed approach.

Keywords Sampled data control · Anti-windup · Time-varying delay · Small gain

theorem · Actuator saturation

B Fernando Tadeo
[email protected]
Ouarda Lamrabet
[email protected]
Khadija Naamane
[email protected]
El Houssaine Tissir
[email protected]
Fatima El Haoussi
[email protected]

1 LESSI, Department of Physics, Faculty of Sciences Dhar El Mehraz, University Sidi Mohammed
Ben Abdellah, BP 1796, Fes-Atlas, Morocco
2 Department of Systems Engineering and Automatic Control and Institute of Sustainable Processes,
University of Valladolid, 47005 Valladolid, Spain
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4869

1 Introduction

In practical systems, the control signals are implemented by actuators that are subject to
saturation. This issue is receiving significant attention during the last decades (see [38]
and references therein ), as closed-loop performance deteriorates, and stability can be
lost when the actuators saturate. The most frequent strategy to attenuate this problem
is to include an anti-windup compensator in the control system [14], that do not act
until the saturation is reached. Many results are available for the anti-windup design:
for instance, in [24], the authors used the input delay approach for synthesizing an
Anti-windup compensator to AQM in TCP/IP networks. For time delay systems under
input saturation constraint via sampled-data, we can cite [25].
With the development of intelligent instruments and the availability of inexpen-
sive digital sensors, digital controllers are generally used for many real-life systems.
Thus, continuous-time systems are frequently controlled by discrete-time controllers
implemented in digital devices, so the overall system becomes a sampled-data con-
trol system. Therefore, considerable attention is also been paid to these sampled-data
control systems. For instance, in [27], sampled-data stabilization criteria have been
established by using Wirtinger’s integral inequality. A sampled-data strategy was used
to study the problem of robust stabilization of uncertain neutral state delayed-systems
under input saturation in [17]. In [32], the problem of stabilization of switched systems
with actuator faults via the robust reliable sampled-data control was studied.
Time-varying delays arise naturally in most real-world systems. They have been
considered in the literature for instance, for AQM/TCP systems, in [5], for wind tur-
bines, in [16,29], and more general systems as descriptor systems [37] or neutral
systems [18,19]. Moreover, time-delays are often a source of oscillation, poor per-
formance and instability of a control system. Considering these facts, a great deal of
attention has been devoted to stability analysis and controller synthesis for time-delay
systems [1–4,35,36]. Some useful approaches have already been established for time-
delay systems, including the free weighting matrices technique [21], the Wirtinger
inequality approach [11,26,28] and the input–output (IO) approach [20]. When using
the IO approach, the original system is divided into two interconnected subsystems.
Then, by using the scaled small gain (SSG) theorem, we can demonstrate that the sta-
bility condition can considerably be improved. In the literature, several results can be
found concerning the IO approach. In [22] the delayed state x(t −τ (t) is approximated
by its average value 21 (x(t − τ1 ) + x(t − τ1 )) (two-terms approximation) and stabil-
ity criteria have been proposed. Extension of the two-terms approximation method
to study the T-S Fuzzy systems with time-varying delay is considered in [39]. In
[12], a new model transformation was proposed for continuous-time systems by using
three-terms approximation. This approximation has also been successfully used to
investigate the robust stabilization of delta operator systems with time-varying delays
in [13], to design an H∞ filter for discrete time-varying delay systems in [40], and
to synthesize an anti-windup compensator for delta operator systems with actuator
saturation in [31]. The aforementioned consideration shows that the IO approach has
been considered by several authors and has an important role in studies embroiled
with delay systems. To the authors’ best knowledge, this idea has not yet been done
4870 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

in sampled-data systems under input saturation, and this is the motivation behind the
presented work.
In summary, the problem of H∞ sampled-data anti-windup compensator design for
time varying delay systems with input saturation will be investigated in this paper via
the IO approach. The main novelty and contributions of this paper can be summarized
as follows:
1. A methodology to design anti-windup compensator is proposed to mitigate the
effect of saturation in sample-data systems with delays. The designed compensator
ensures asymptotic stability of the closed-loop system and enlarges the domain of
attraction.
2. The methodology is based on transforming the original sampled-data system
with time varying delay and input saturation into two interconnected subsystems.
Then, by using and input/output approach and the scaled small gain theorem and
using Wirtinger’s integral inequality, LMI-based conditions are derived that can
be numerically solved with little conservativeness, as illustrated in the numerical
examples provided.
Notations: G1 ◦ G2 represents the series connection of mapping G1 and G2 . P >
0(≥ 0) means that matrix P is positive (semi) definite. P T and P −1 denote the
transpose and inverse of matrix P, respectively. ∗ stands for the symmetric term of the
diagonal elements of square symmetric matrix. .∞ represents the l2 -induced norm
of a transfer function matrix or a general operator. Ai denotes the ith line of the matrix
A.

2 Problem Formulation and Preliminaries

This paper considers the class of plants described by the following continuous-time
system with delay:

ẋ(t) = Ax(t) + Aτ x(t − τ (t)) + Bsat(u(t)) + Bw w(t)

y(t) = C y x(t)
z(t) = C z x(t) (1)

where x(t) ∈ Rn , u(t) ∈ Rm , w(t) ∈ Rq , y(t) ∈ R p , z(t) ∈ R p are the state vector,
control vector, disturbance, the measured output and the regulated output, respectively,
with A, Aτ , B, Bw , C y and C z known constant real matrices.
This continuous-time plant is controlled with a digital device forming a sampled-
data control system. By taking this into consideration, the control input takes the
following form:

u(t) = u(tk ), tk ≤ t < tk+1 , (k = 0, 1, 2, . . .)

The interval between any two sampling instants is assumed to be bounded by h, which
means that for any k > 0, tk+1 − tk = dk ≤ h, where h is the maximum sampling
interval. By defining d(t) = t − tk with ḋ(t) = 1 for t = tk , the sampling instant can
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4871

be written as tk = t − (t − tk ) = t − d(t). Then, we have d(t) ≤ tk+1 − tk = dk ≤ h;

thus, the control law u(t) can be further written as

u(t) = u(t − d(t)), tk ≤ t < tk+1 , (k = 0, 1, 2, . . .) (2)

Moreover, u(t) (with m components) is bounded as follows

− u 0(i) ≤ u (i) ≤ u 0(i) , u 0(i) > 0, i = 1, . . . , m (3)

τ (t) is the time-varying delay, which satisfies

0 < τ1 ≤ τ (t) ≤ τ2 , τ̇ (t) ≤ μ < 1 (4)

where μ is a constant positive scalar.

The disturbance vector w(t) is assumed to be Lebesgue measurable, that is, w(t) ∈
L2 . Hence, the disturbance w(t) is bounded as follows:
 ∞
w(t)22 = w T (t)w(t)dt < ∞ (5)
0

To stabilize the system (1), we consider the following dynamic stabilizing controller

ẋc (t) = Ac xc (t) + Bc y(t)

yc (t) = Cc xc (t) + Dc y(t) (6)

where xc (t) is the controller state, u c (t) = y(t) is the controller input, and yc (t) is the
controller output. Ac , Bc , Cc and Dc are known matrices of appropriate dimensions.
In the presence of actuator saturation, the control signal of the system can be
described as u(t) = sat(yc (t)), where sat(yc(i) (t)) = sign(yc(i) (t))min{|yc(i)
(t)|, u 0(i) }, i = 1, . . . , m
To reduce the undesirable effects of the windup caused by the saturation, an anti-
windup compensator is added to the controller as follows:

ẋc (t) = Ac xc (t) + Bc y(t) − E c ψ(yc (t − d(t)))

yc (t) = Cc xc (t) + Dc y(t) (7)

Note that, ψ(yc (t − d(t))) corresponds to a decentralized dead-zone nonlinearity,

where ψ(yc (t − d(t))) = yc (t − d(t)) − sat(yc (t − d(t))).
Then, the augmented system can be represented as follows

ξ̇ (t) = Aξ(t) + Aτ ξ(t − τ (t)) + Ad ξ(t − d(t)) − (B + RE c )ψ(Kξ(t − d(t)))

+ Bw w(t)
z(t) = Cz ξ(t) (8)
4872 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

and the augmented system is given by

       
x(t) A 0 B Aτ 0
ξ(t) = , A= , B= , Aτ =
xc (t) Bc C y Ac 0 0 0
     
B Dc C y BCc Bw 0  
Ad = , Bw = , R= , K = Dc C y Cc ,
0 0 0 In c
 
Cz = C z 0 .

The initial condition is given by

ξ(θ ) = φξ , θ ∈ [−τ̄ , 0], τ̄ = max(τ2 , h)

Consider a matrix H ∈ m×(n+n c ) , and define the following polyhedral set

 
S = ξ(t) ∈ n+n c
; |(K(i) − H(i) )ξ(t)| ≤ u 0(i)

The following useful lemmas will be used in this paper.

Lemma 1 [34] If ξ(t) ∈ S, then the following relation

ψ T (Kξ(t))M ψ(Kξ(t)) − H ξ(t) ≤ 0

is verified for any diagonal positive matrix M ∈ m×m .

Lemma 2 [30] For any scalar b > a, positive matrix R and function ξ , the following
inequality holds
 b 1 3
ξ̇ T (s)R ξ̇ (s)ds ≥ (ξ(b) − ξ(a)T R(ξ(b) − ξ(a)) + θ T Rθ
a b−a b−a
b
where θ = ξ(b) + ξ(a) − 2
b−a a ξ(s)ds
Lemma 3 [39] (Scaled small gain theorem) Consider the following interconnected
feedback system

(S1 ) : y (t) = G (t); (S2 ) :  (t) = y (t) (9)

where subsystem S1 is a known LTI system with operator G mapping  (t) to y (t),
and subsystem S2 is an unknown linear time-varying one with operator ∈ D 
{ : ∞ ≤ 1}. Assume that S1 is internally stable. The closed-loop system formed
by S1 and S2 is robustly asymptotically stable for all ∈ D if there exist matrices
{T , Ty } ∈ T with T  {{T , Ty } ∈ R × × R y×y : T , Ty non-singular, T ◦
◦ Ty−1 ∞ ≤ 1} such that the SSG condition holds:

Ty ◦ G ◦ T−1 ∞ < 1 (10)

Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4873

Finally, for a positive scalar κ, the ellipsoid ε(P, κ −1 ) is defined as follows:

ε(P, κ −1 ) = {ξ(t) ∈ n+n c

; ξ T (t)Pξ(t) ≤ κ −1 }

3 Main Results

This section first introduces the transformation of system (8), so then the stability
condition is developed by using the SSG theorem.

3.1 Model Transformation

Consider the system (8): following [12], the delayed state is approximated by the
following equation:

1 τ12
ξ(t − τ (t)) = [ξ(t − τ1 ) + ξ(t − τa ) + ξ(t − τ2 )] + r (t) (11)
3 3

where τ12 = τ2 − τ1 , τa = τ1 +τ
2
2
and τ312 r (t) is the approximation error. From (11),
system (8) can be written as a forward interconnection system (S1 ) and as a feedback
interconnection system (S2 ).

Aτ Aτ Aτ
ξ̇ (t) = Aξ(t) + ξ(t − τ1 ) + ξ(t − τa ) + ξ(t − τ2 )
3 3 3
τ12
(S1 ) : + Aτ r (t) + Ad ξ(t − d(t)) − (B + RE c )ψ(Kξ(t − d(t)))
3
+ Bw w(t)
z(t) = Cz ξ(t)
y (t) = ξ̇ (t)
√
(S2 ):  (t) = y (t) with r = (3/ 2)
√
Remark 1 The equation r = (3/ 2) provides the relation between feedback sys-
tem (S2 ) and forward system (S1 ), to give a representation of subsystem (S1 ) in a
compact form, for the IO approach. If we denote the right-hand side of (S1 ) by f (ξ, w)
dξ
then we can write d(t) = f (ξ, w). It is clear that f (ξ, w) is continuous in some closed
and bounded set around the equilibrium point. It follows that f (ξ, w) is bounded and
dξ
satisfies Lipschitz conditions in a certain domain. Consequently, d(t) = f (ξ, w) has
at least one solution in that domain, for more details see [23]

The approximation error can be written as follows:

Case 1. If τ1 ≤ τ (t) ≤ τa , then

τ12 1
r = ξ(t − τ (t)) − ξ(t − τ1 ) + ξ(t − τa ) + ξ(t − τ2 )
3 3
4874 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

  t−τ1  t−τ (t)  t−τa 

1
= − y (s)ds + 2 y (s)ds + y (s)ds
3 t−τ (t) t−τa t−τ2

Case 2. If τa ≤ τ (t) ≤ τ2 , then

τ12 1
r = x(t − τ (t)) − x(t − τ1 ) + x(t − τa ) + x(t − τ2 )
3 3
  t−τ1  t−τa  t−τ (t) 
1
= − y (s)ds + 2 y (s)ds + y (s)ds
3 t−τa t−τ (t) t−τ2

Lemma 4 Operator : y →  , satisfies the SSG condition N ◦ ◦ N −1 ∞ ≤ 1,

where N is a general invertible matrix.
Proof If we can prove that N ◦ ◦ N −1 ∞ ≤ 1} holds for τ1 ≤ τ (t) ≤ τa and for
τa ≤ τ (t) ≤ τ2 , then N ◦ ◦ N −1 ∞ ≤ 1} is true. Let S = N T N
Case 1. τ1 ≤ τ (t) ≤ τa
We have that by Jensen (Cauchy–Schwartz) inequality, for all t ≥ 0

τ12
2 τ2
rT (t)r (t) = 12 r (t)2
9 9
  t−τ1  t−τ( t)  t−τa 2
1 
=   − y (s)ds + 2 y (s)ds + y (s)ds 

9 t−τ (t) t−τa t−τ2
  t−τ1 2   t−τ (t) 2
3  

 + 2
 

≤  y (s)ds   y (s)ds 
9 t−τ (t) t−τa
  t−τa 2
 
+ y (s)ds 
t−τ2

We continue the proof for each term separately. A function s = p(t) = t − τ (t) is
strongly increasing. Hence, the inverse t = p −1 (s) = q(s) is well-defined such that
2
y (s)ds  between 0 and ∞,
t−τ1
| q(s) − (s + τ1 ) |≤ τ12 /2. Then, integrating  t−τ (t)
changing the order of the integration and then taking into account that y (s) = 0, for
s ≤ 0, we find that
 
∞   t−τ1
2  ∞  t−τ1

 y (s)ds  dt ≤ (τ (t) − τ ) y (s)2 dsdt
  1
0 t−τ (t) 0 t−τ (t)
 ∞  s+τ1 
 
= (τ (q(s)) − τ1 ) y (s) dsdt 
2
0 q(s)
 ∞ 
 
= (τ (q(s)) − τ1 )(s + τ1 − q(s))y (s) dsdt 
2
0

τ12 τ12 ∞
≤ y (s)2 ds
2 2 0
τ2
= 12 y (t)l22
4
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4875

For the other terms, we follow the same process, to obtain

   t−τ (t)
∞
2

 dt ≤ 4 τ12 y (t)2
2
2 y (s)ds
  4 l2
0 t−τa
 ∞   t−τa 2
 
 dt ≤ τ12 y (t)2
2
 y (s)ds
  4 l2
0 t−τ2

Then, the summation of the three terms together gives

 
τ12
2
3 τ12
2 τ2 τ2
r (t)l22 ≤ + 4 12 + 12 y (t)l22
9 9 4 4 4
τ12
2
= y (t)l22
2
√
when substituting r (t) by (3/ 2) (t), we obtain  (t)l22 ≤ y (t)l22 . For case 2,
by using similar proof process, we obtain the same results as in case 1. This completes
the proof. 


Remark 2 {N , N } ∈ T are the matrices in the SSG theorem given in Lemma 3, to

ensure that the system (8) is IO stable, it is necessary to verify that (S1 ) is internally
stable and there exists N such that SSG condition N ◦ G ◦ N −1 ∞ ≤ 1 holds.

3.2 Anti-windup Design

In this subsection, a methodology for computing the anti-windup gain that guarantees
asymptotic stability of system (1) is derived.

Theorem 1 For given scalars h, τ1 , τ2 , μ, the closed-loop system (8) is asymptotically

stable with a prescribed level γ , if there exist symmetric positive definite matrices
X , X j , ( j = 1, 2, 3, 4, 5), Q̂ l , (l = 1, 2, 3, 4) and appropriately sized matrices
Yc , W and L such that
⎡ ⎤
Π τ1 Π1 τa Π1 τ2 Π1 hΠ1 Π1 Π2
⎢∗ −X 1 0 0 0 0 0 ⎥
⎢ ⎥
⎢∗ ∗ −X 2 0 0 0 0 ⎥
⎢ ⎥
Γ =⎢
ˆ
⎢∗ ∗ ∗ −X 3 0 0 0 ⎥⎥ < 0 (12)
⎢∗ ∗ ∗ ∗ −X 4 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ −X 5 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −γ I
 
X X K(i) − W(i)
T T
≥ 0, (13)
∗ κu 20(i)
4876 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

where
 T
Π1 = AX A3τ X A3τ X Aτ
3 X Ad X τ12
3 Aτ X −BL − RFc Bw 04n
 T
Π2 = Cz X 011n
and the elements of Π are defined by
Π11 = AX + X AT + Q̂ 1 + Q̂ 2 + Q̂ 3 + Q̂ 4 + 4(X 1 − 2X )
+ 4(X 2 − 2X ) + 4(X 3 − 2X ) + 4(X 4 − 2X )
Aτ Aτ
Π12 = X + 2(X 1 − 2X ), Π13 = X + 2(X 2 − 2X )
3 3
Aτ
Π14 = X + 2(X 3 − 2X ), Π15 = Ad X + 2(X 4 − 2X )
3
h 12
Π16 = Aτ X , Π17 = BN − RFc , Π18 = Bw , Π19 = 6(2X − X 1 )
3
Π110 = 6(2X − X 2 ), Π111 = 6(2X − X 3 ) Π1112 = 6(2X − X 4 )
(1 − μ) (1 − μ)
Π22 = − Q̂ 1 + 4(X 1 − 2X ) − Q̂ 4 , Π23 = − Q̂ 4
9 9
(1 − μ) h 12
Π24 = − Q̂ 4 , Π26 = − (1 − μ) Q̂ 4 , Π29 = 6(2X − X 1 )
9 9
(1 − μ) (1 − μ)
Π33 = − Q̂ 2 + 4(X 2 − 2X ) − Q̂ 4 , Π34 = − Q̂ 4
9 9
τ12
Π36 = − (1 − μ) Q̂ 4 , Π310 = 6(2X − X 2 )
9
(1 − μ) τ12
Π44 = − Q̂ 3 + 4(X 3 − 2X ) − Q̂ 4 , Π46 = − (1 − μ) Q̂ 4
9 9
Π411 = 6(2X − X 3 ), Π55 = 4(X 4 − 2X ), Π57 = W T
2 h 12
Π512 = 6(2X − X 4 ), Π66 = (X 5 − 2X ) − (1 − μ) Q̂ 4
9 9
Π77 = −L − L T , Π88 = −I , Π99 = 12(X 1 − 2X )
Π1010 = 12(X 2 − 2X ), Π1111 = 12(X 3 − 2X ), Π1212 = 12(X 4 − 2X )

then, the anti-windup gain matrix E c = Fc L −1 ensures that:

1. the trajectories of the system (8) are bounded for every initial condition satisfying
β ≤ κ −1 ;
2. Under the zero initial condition,
z(t)22 ≤ γ w(t)22 ;

3. when w(t) = 0, for all initial conditions belonging to β ≤ κ −1 , the corresponding

trajectories converge asymptotically to the origin, where

β = λ(X −1 ) + τ1 λ̄(X −1 Q̂ 1 X −1 ) + τa λ̄(X −1 Q̂ 2 X −1 ) + τ2 λ̄(X −1 Q̂ 3 X −1 )
 τ3 τ3
+ τ2 λ̄(X −1 Q̂ 4 X −1 ) φξ 2c + 1 λ̄(X 1−1 ) + a λ̄(X 2−1 )
2 2
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4877

τ23 h3 
+ λ̄(X 3−1 ) + λ̄(X 4−1 ) φ̇ξ 2c
2 2

with λ is the maximal eigenvalue.

Proof To prove this theorem, let us consider the following Lyapunov functional
 t  t
V (t) = ξ T (t)Pξ(t) + ξ T (s)Q 1 ξ(s)ds + ξ T (s)Q 2 ξ(s)ds
t−τ1 t−τa
 t  t
+ ξ T (s)Q 3 ξ(s)ds + ξ T (s)Q 4 ξ(s)ds
t−τ2 t−τ (t)
 0  t  0  t
+ τ1 ξ̇ T (s)R1 ξ̇ (s)dsdθ + τa ξ̇ T (s)R2 ξ̇ (s)dsdθ
−τ1 t+θ −τa t+θ
 0  t  0  t
+ τ2 ξ̇ (s)R3 ξ̇ (s)dsdθ + h
T
ξ̇ T (s)R4 ξ̇ (s)dsdθ (14)
−τ2 t+θ −h t+θ

Computing the derivative of the functional (14) along the trajectory of the system (S1 ),
we get

V̇ (t) = 2ξ T (t)P ξ̇ (t) + ξ T (t)[Q 1 + Q 2 + Q 3 + Q 4 ]ξ(t) − ξ T (t − τ1 )Q 1 ξ(t − τ1 )

− ξ T (t − τa )Q 2 ξ(t − τa ) − ξ T (t − τ2 )Q 3 ξ(t − τ2 ) − (1 − μ)ξ T (t − τ (t))
×Q 4 ξ(t − τ (t)) + ξ̇ T (t)[τ12 R1 + τa2 R2 + τ22 R3 + h 2 R4 ]ξ̇ (t)
 t  t
− τ1 ξ̇ T (s)R1 ξ̇ (s)ds − τa ξ̇ T (s)R2 ξ̇ (s)ds
t−τ1 t−τa
 t  t
− τ2 ξ̇ T (s)R3 ξ̇ (s)ds − h ξ̇ T (s)R4 ξ̇ (s)ds (15)
t−τ2 t−h

with the aid of Lemma 2, we have

 t T
− τ1 ξ̇ T (s)R1 ξ̇ (s)ds ≤ − ξ(t) − ξ(t − τ1 ) R1 ξ(t) − ξ(t − τ1 )
t−τ1
− 3Θ1T R1 Θ1 (16)
 t T
−τa ξ̇ T (s)R2 ξ̇ (s)ds ≤ − ξ(t) − ξ(t − τa ) R2 ξ(t) − ξ(t − τa )
t−τa
− 3Θ2T R2 Θ2 (17)
 t T
−τ2 ξ̇ T (s)R3 ξ̇ (s)ds ≤ − ξ(t) − ξ(t − τ2 ) R3 ξ(t) − ξ(t − τ2 )
t−τ2
− 3Θ3T R3 Θ3 (18)

where
4878 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

 t  t
2 2
Θ1 = ξ(t) + ξ(t − τ1 ) − ξ(s)ds, Θ2 = ξ(t) + ξ(t − τa ) − ξ(s)ds
τ1 t−τ1 τa t−τa
 t
2
Θ3 = ξ(t) + ξ(t − τ2 ) − ξ(s)ds
τ2 t−τ2

It is clear that the following is true

 t  t
− ξ̇ T (s)R4 ξ̇ (s)ds ≤ − ξ̇ T (s)R4 ξ̇ (s)ds (19)
t−h t−d(t)

Moreover, by applying Lemma 2 to (19) the following inequality is obtained:

 t T
−h ξ̇ T (s)R4 ξ̇ (s)ds ≤ − ξ(t) − ξ(t − d(t)) R4 ξ(t) − ξ(t − d(t))
t−d(t)
− 3Θ4T R4 Θ4 (20)
t
where Θ4 = x(t) + ξ(t − d(t)) − h2 t−d(t) ξ(s)ds
We define the following vectors:
  t  t  t  t 
1 1 1 1
α (t) =
T
ξ (s)ds T
ξ (s)ds
T
ξ (s)ds
T
ξ (s)ds
T
τ1 t−τ1 τa t−τa τ2 t−τ2 h t−d(t)

ζ T (t) = ξ T (t) ξ T (t − τ1 ) ξ T (t − τa ) ξ T (t − τ2 ) ξ T (t − d(t)) rT (t)

ψ T (Kξ(t − d(t))) , η T = [ζ T (t) w(t)T α T (t)]

And let
 ∞ 1 T

J= y (s)Sy (s) −  (s)S (s) − w (s)w(s) + z (s)z(s) ds (21)
T T T
0 γ

and
 ∞ 2 1

J1 = V̇ (t) + y T (s)Sy (s) − rT (s) Sr (s) − w T (s)w(s) + z T (s)z(s) ds
0 9 γ

By summing Eqs. (15)–(20) and using Lemma 1, we have

 ∞
J1 ≤ η T (t) Λ + Λ1T τ12 R1 Λ1 + Λ1T τa2 R2 Λ1 + Λ1T τ22 R3 Λ1 + Λ1T h 2 R4 Λ1
0
1
+Λ1T SΛ1 η(t) + (Cz ξ(t))T (Cz ξ(t)) (22)
γ

where
 Aτ Aτ Aτ τ12
T
Λ1 = A 3 3 3 Ad 3 Aτ −(B + RE c ) Bw 04n
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4879

and the elements of Λ are defined by

Λ11 = PA + AT P + Q 1 + Q 2 + Q 3 + Q 4 − 4R1 − 4R2 − 4R3 − 4R4

Aτ Aτ Aτ
Λ12 = P − 2R1 , Λ13 = P − 2R2 , Λ14 = P − 2R3
3 3 3
τ12
Λ15 = PAd − 2R4 , Λ16 = P Aτ , Λ17 = −P(B + RE c )
3
Λ18 = PBw , Λ19 = 6R1 , Λ110 = 6R2 , Λ111 = 6R3 , Λ112 = 6R4
(1 − μ) (1 − μ)
Λ22 = −Q 1 − 4R1 − Q 4 , Λ23 = − Q4
9 9
(1 − μ) τ12
Λ24 = − Q 4 , Λ26 = − (1 − μ)Q 4 , Λ29 = 6R1
9 9
(1 − μ) (1 − μ)
Λ33 = −Q 2 − 4R2 − Q 4 , Λ34 = − Q4
9 9
τ12 (1 − μ)
Λ36 = − (1 − μ)Q 4 , Λ310 = 6R2 , Λ44 = −Q 3 − 4R3 − Q4
9 9
τ12
Λ46 = − (1 − μ)Q 4 , Λ411 = 6R3 , Λ55 = −4R4 , Λ57 = H T T0T
9
2 τ12
Λ512 = 6R4 , Λ66 = − S − (1 − μ)Q 4 , Λ77 = −T0 − T0T , Λ88 = −I
9 9
Λ99 = −12R1 , Λ1010 = −12R2 , Λ1111 = −12R3 , Λ1212 = −12R4

By using the Schur complement, it can be easily seen that J1 < 0 if the following
condition holds.

⎡ ⎤
Λ τ1 Λ1 τa Λ1 τ2 Λ1 hΛ1 Λ1 Λ2
⎢∗
⎢ −R1−1 0 0 0 0 0 ⎥ ⎥
⎢∗ ∗ −R2−1 0 0 0 0 ⎥
⎢ ⎥
Ξ =⎢
⎢∗ ∗ ∗ −R3−1 0 0 0 ⎥ ⎥<0 (23)
⎢∗ ∗ ∗ ∗ −R4−1 0 ⎥
⎢ 0 ⎥
⎣∗ ∗ ∗ ∗ ∗ −S −1 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −γ I

where Λ2 = [Cz 0 p×10n ]T .

Under zero initial conditions, V (0) = 0, we have that

J1 = J + V (∞) − V (0) = J + V (∞) < 0

which implies that J < 0

When w(t) = 0, we obtain

 ∞
J1 < [ζ (t) α(t)]T Ξ̃ [ζ (t) α(t)]
0
4880 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

with

⎡ ⎤
Λ̃ τ1 Λ̃1 τa Λ̃1 τ2 Λ̃1 h Λ̃1 Λ̃1
⎢∗
⎢ −R1−1 0 0 0 0 ⎥ ⎥
⎢∗ ∗ −R2−1 0 0 0 ⎥
Ξ̃ = ⎢
⎢∗
⎥
⎢ ∗ ∗ −R3−1 0 0 ⎥ ⎥
⎣∗ ∗ ∗ ∗ −R4−1 0 ⎦
∗ ∗ ∗ ∗ ∗ −S −1

where

⎡ ⎤
Λ11 Λ12 Λ13 Λ14 Λ15 Λ16 Λ17 Λ19 Λ110 Λ111 Λ112
⎢ ∗ Λ22 Λ23 Λ24 0 Λ26 0 Λ29 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ Λ33 Λ34 0 Λ36 0 0 Λ310 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ Λ44 0 Λ46 0 0 0 Λ411 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ Λ55 0 Λ57 0 0 0 Λ512 ⎥
⎢ ⎥
Λ̃ = ⎢
⎢ ∗ ∗ ∗ ∗ ∗ Λ66 0 0 0 0 0 ⎥ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ Λ77 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ Λ99 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ Λ1010 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ Λ1111 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ Λ1212
 T
Λ̃1 = A A3τ Aτ
3
Aτ
3 Ad τ12
3 Aτ −(B + RE c ) 04n

So Ξ̃ < 0 implies that

 ∞  ∞
y T (s)Sy (s)ds <  T (s)S (s)ds (24)
0 0

The inequality (24) guarantees that N ◦ G ◦ N −1  < 1.

From Lemma 4, we have the inequality N ◦  ◦N −1  ≤ 1. Also, from, J1 < 0, it
can be easily seen that V̇ (t) < 0 which implies that the system (S1 ) with w(t) = 0 is
asymptotically stable.
When w(t) = 0, we multiply the both sides of (23) by diag{X , X , X , X , X , X , T0−1 ,
I , X , X , X , X , X 1 , X 2 , X 3 , X 4 , X 5 , I } and by letting X = P −1 , X 1 =
R1−1 , X 2 = R2−1 , X 3 = R3−1 , X 4 = R4−1 , X 5 = S −1 , L = T0−1 , Fc =
E c T0−1 , W = H X T , Q̂ l = X Q l X , (l = 1, 2, 3, 4), and using the relation
−X X −1j X ≤ X j − 2X for j = 1, 2, 3, 4, 5. It is easy to obtain (12). Since (12)
holds, it follows that J < 0. Consequently,

 ∞  ∞
1
z T (s)z(s)ds < ω T (s)ω(s)ds
γ 0 0
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4881

On the other hand, the satisfaction of (13) guarantees that ∀ξ ∈ ε(P, κ −1 ), ξ ∈ S. In

fact, ε(P, κ −1 ) ⊂ S is verified by the following conditions
 
P K(i)
T − HT
(i)
≥0 (25)
∗ κu 20(i)


Pre- and post-multiplying (25) by Δ = diag{P −1 , I }, the LMI (13) is obtained.
Moreover, from the L–K functional defined in (14), we have
 0  0
V (0) = ξ T (0)Pξ(0) + ξ T (s)Q 1 ξ(s)ds + ξ T (s)Q 2 ξ(s)ds
−τ1 −τa
 0  0
+ ξ (s)Q 3 ξ(s)ds +
T
ξ T (s)Q 4 ξ(s)ds
−τ2 −τ (t)
 0  0  0  0
+ τ1 ξ̇ T (s)R1 ξ̇ (s)dsdθ + τa ξ̇ T (s)R2 ξ̇ (s)dsdθ
−τ1 θ −τa θ
 0
 0  0  0
+ τ2 ξ̇ T (s)R3 ξ̇ (s)dsdθ + h ξ̇ T (s)R4 ξ̇ (s)dsdθ
−τ2 θ −h θ
 
≤ λ̄(P) + τ1 λ̄(Q 1 ) + τa λ̄(Q 2 ) + τ2 λ̄(Q 3 + τ2 λ̄(Q 4 ) φ(θ )2
τ3 τa3 τ3 h3 
+ 1
λ̄(R1 ) + λ̄(R2 ) + 2 λ̄(R3 ) + λ̄(R4 ) φ̇(θ )2 = ρ
2 2 2 2

Making the above change of variables, we have x T (t)P x(t) ≤ V (t) ≤ V (0) ≤ β ≤
κ −1 ; that is, for all t ≥ 0, the trajectories of the system do not leave the set ε(P, κ −1 )
for any initial condition φ(θ ) in ε(P, κ −1 ), which ensures that x(t) ∈ S.
The proof is thus completed. 


4 Optimization Problems

4.1 Minimization of 

The problem of minimizing γ can be formulated as follows

min γ
subject to (12), (13)
X > 0, X j > 0, ( j = 1, . . . , 5), Q̂ l > 0, ( j = 1, . . . , 4) (26)

4.2 Maximization of 

The objective now is to provide (in the disturbance-free case, w(t) = 0) a methodology
to estimate the largest possible domain of initial conditions for which the closed-
loop system trajectories remain bounded. This is mathematically complex due to the
4882 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

nonlinearity of β. The solution proposed is based on solving the optimization problem

that is now developed: Let
     
σ1 I I σ2 I X̃ σ3 I X̃
≥ 0, ≥ 0, ≥0
I X X̃ T Q̃ 1 X̃ T Q̃ 2
     
σ4 I X̃ σ5 I X̃ σ6 I I
≥ 0, ≥ 0, ≥0
X̃ T Q̃ 3 X̃ T Q̃ 4 I X1
     
σ7 I I σ8 I I σ9 I I
≥ 0, ≥ 0, ≥0 (27)
I X2 I X3 I X4

where X −1 = X̃ , Q̂ i−1 = Q̃.

It follows that the condition β ≤ κ −1 is satisfied if the following condition holds:
 
τ13 τa3 τ23 h3
σ1 + τ1 σ2 + τa σ3 + τ2 (σ4 + σ5 ) + σ6 + σ7 + σ8 + σ9 ρ 2 ≤ κ −1
2 2 2 2
(28)

where ρ 2 = max(φξ 2 , φ̇ξ 2 ), and the stability radius ρ is a scalar to be determined.

Combining the facts derived above, we can construct a feasibility problem for given
τ1 , τa , τ2 , h, γ , as follows

τ13 τ3 τ3 h3
minimize ϕ = σ1 + τ1 σ2 + τa σ3 + τ2 (σ4 + σ5 ) + σ6 + a σ7 + 2 σ8 + σ9
2 2 2 2
subject to (12), (13), (27), (28), X > 0, X j > 0, ( j = 1, .., 5), Q̂ l > 0, ( j = 1, .., 4)
(29)

This optimization problem can then be solved using off-the-shelf numerical tools. It
must be pointed out that by minimizing ϕ, we are, implicitly, maximizing ρ.

Remark 3 It is worth mentioning that the computational complexity can be handled by

the Matlab LMI toolbox and all designs can be implemented offline, making the LMI
method practicable and effective. In order to obtain a simpler form of LMI, no free-
weighting matrices are adopted, which leads to small number of decision variables
and consequently reduce the computational complexity significantly.

Remark 4 It should be noted that there are few works dealing with sampled-data control
in the presence of actuator saturation, time-varying delay and external disturbance.
We cite [33], where a dissipative-based controller was developed for a class of time-
varying delay systems via sampled-data approach, but no saturation was considered.
The robust H∞ sampled-data control problem for uncertain nonlinear time-varying
delay systems is considered in [10], but the saturation was not considered either. In this
paper, an anti-windup compensator is considered, as it is the most used technique to
counterbalance the saturation effects and improve the closed-loop system performance
during saturations.
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4883

Table 1 Comparison of
Theorem 1 [8] [7] [9]
maximum radius ρ for different
τ2 τ2 = 0.5 10037.8 9152.0 4846.8 4520.0
τ2 = 1 8137.9 5061.3 3722.4 2986.0
τ2 = 2 5100.1 2575.0 2542.7 1772.7

Remark 5 The three-term approximation approach for x(t − τ (t)) that is used here,
which includes lower bound, upper bound and mean value of x(t − τ (t)), was first
proposed in [12] for continuous-time systems. Then, the same approximation was
extended to delta operator systems and discrete time systems in [13] and [40], respec-
tively. It is worth noting that it has been shown in those works that error using the
three-term approximation is smaller than the one using two-term [22] and one-term
approximation [20]. The three-term approximation is used here for the first time in
the context of sampled-data systems subject to input saturation.

Remark 6 From the approaches to handle the saturation nonlinearity, the sector bound
approach is used here as there results can be expressed as LMIs, that are easier to solve
that those obtained using the polytopic modeling approach [18,19].

5 Numerical Examples

In the following, two numerical examples are developed to illustrate the effectiveness
of the proposed methodology.

Example 1 Consider the system (1) with w(t) = 0 and the following parameters [9]:
     
1 1.5 0 −1 10  
A= , Aτ = , B= , Cy = 5 1 .
0.3 −2 0 0 0

The dynamic controller is given as:

   
−20.2042 2.5216 1.9516  
Ac = , Bc = , Cc = −0.9165 0.1091 , Dc = 0
2.1415 −4.4821 −0.0649

In order to compare with previous results in the literature, that assume constant delays,
solving the optimization problem (29), with h = 0.1, κ = 1 and u 0 = 15, the
calculated stability radius ρ for different values of τ2 gives the results in Table 1. It is
clear from Table 1 that the obtained stability radius is larger than those obtained by
existing works.
Applying Theorem 1 with h = 0.15, κ = 1, some bounds of ρ can be obtained
for different value of τ2 . Table 2 shows the comparison between our results and the
results in [8] (τ1 = 0.5 and μ = 0.5). It can be seen from Table 2 that the proposed
approach provides larger bounds ρ than [8].
4884 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

Table 2 Bound ρ for different

Method τ2 = 1 τ2 = 1.5 τ2 = 2
τ2 when τ1 = 0.5 and μ = 0.5
[8] 4327.7 2494.0 1788.1
Theorem 1 4487.1 3010.5 2086.2

5
x
4 1
x2
3

−1

−2

−3

−4

−5
0 1 2 3 4 5 6 7 8 9 10
Time(t)

Fig. 1 State trajectories of the closed-loop system in Example 1

For instance, when τ2 = 2 and u 0 = 0.5 the corresponding anti-windup compen-

sator gain is E c = 17915. Figure 1 presents the state trajectories of the closed-loop
system, and Fig. 2 shows the control input, simulated from the initial condition
x(0) = [−3.5 4.5]T . It can be seen that the state trajectories of closed-loop sys-
tem converge to zero, even when the system saturates, showing the feasibility of the
procedure proposed.

Example 2 In this example, we adopt the Mach Number in a Wind Tunnel model. As it
has been shown in the literature [29], the deviations δ M of the Mach number induced
by small deviations in the guide vane angle actuator δθa in a driving fan are precisely
described at a given operating point (determined by the fan speed, liquid nitrogen
injection rate, and gaseous-nitrogen vent rate) by the following dynamic model [29]:

1
δ Ṁ(t) + δ M(t) = k δθ (t − τ (t)),
a
δ θ̈ (t) + 2ξ wδ θ̇(t) + w 2 δθ (t) = w 2 δθa (t), (30)

where δθ is the guide vane angle, a, k, ξ , w are parameters which, at each working
point, can be assumed constant if the deviations δ M, δθ , δθa are small. The delay
τ (t) represents the time required by the movement of air between the fan and the test
section.
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4885

0.6
u(t)

0.5

0.4

0.3

0.2

0.1

−0.1
0 1 2 3 4 5 6 7 8 9 10
Time(s)

Fig. 2 Evolution of control signal for Example 1

Table 3 The minimum

τ2 γ Ec
allowable γ for various τ2 in
Example 3 12 9.9 × 10−2 54.7003
10 7.0 × 10−2 26.7199
8 1.7 × 10−2 16.3399

Rewriting (30) in state space form yields the system (1). where
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
δ M(t) −a 0 0 0 ak 0
x = ⎣ δθ (t) ⎦ , A = ⎣ 0 0 1 ⎦ , Aτ = ⎣ 0 0 0⎦,
δ θ̇ (t) 0 −w 2 −2ξ w 0 0 0
⎡ ⎤ ⎡ ⎤
0 0  
B = ⎣ 0 ⎦ , Bw = ⎣ 0 ⎦ , y(t) = δθ (t), C y = 0 1 0 ,
w2 10
   
δ M(t) 1 0 0
z(t) = , Cz = , u(t) = δθa (t)
δθ (t) 0 1 0

The parameters of the wind tunnel are borrowed from [29], and are as follows: a1 =
1.964 s, k = −0.0117deg −1 , ξ = 0.8, and w = 6rad/s.
The dynamic controller of Example 1 is considered again here. For the system
with a time-invariant delay with h = 0.9, applying Theorem 1 it is possible to obtain
the optimal H∞ performance γ (and the corresponding anti-windup gain matrix) for
different values of τ2 , as given in Table 3.
4886 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

6
x (t)
1
4 x (t)
2
x (t)
3
2

−2

−4

−6

−8

−10

−12

−14
0 5 10 15 20 25
Time(s)

Fig. 3 The state responses of the closed-loop system

0.025

0.02

0.015

0.01

0.005

−0.005

−0.01

−0.015

−0.02

−0.025
0 5 10 15 20 25
Time(s)

Fig. 4 Disturbance used in the simulation

Figure 3 presents the simulation result for state trajectories of the closed-loop
system (for the anti-windup gain corresponding to τ2 = 10), with the initial condition
x(0) = [−5 5 − 5]T . The disturbance used in simulation is the Gaussian noise
presented in Fig. 4.
Figure 3 shows that the anti-windup compensator works well to stabilize the Mach
number in wind tunnels.
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4887

√
Table 4 Comparison of γ √
Reference γ

[6] 0.6744
[15] 0.00412
[24] 0.0002
[25] 0.000093
Theorem 1 0.000083

Example 3 Consider the following system as in [6]:

   
−0.2644 −0.0044 −0.2644 −0.0044
A= , Aτ = ,
243.9024 −4.065 0 0
   
−480.47 0  
B= , Bw = , C y = C z = 5 1 , τ2 = 0.246
0 1

The dynamic controller is given as:

Ac = 0, Bc = 1, Cc = 9.6426 × 10−6 , Dc = 18.194 × 10−6

Applying (26) with h = 0.6 and τ1 = 0 gives the anti-windup gain E c = 92782. The
optimal H∞ performance γ is listed in Table 4, comparing it with existing results. It is
clearly observed from Table 4 that Theorem 1 gives the best H∞ performance index,
showing that the method in this paper yields better result than existing methods.

6 Conclusion

This paper has presented a design methodology for anti-windup compensators for
a class of systems with time-varying delay and input saturation. By employing the
input–output approach, the scaled small gain theorem and a Lyapunov–Krasovskii
Functional, compensators can be designed to ensure closed-loop system stability and
a given disturbance attenuation. Moreover, the results have been rendered to be poten-
tially less conservative, as illustrated by simulation results.

Acknowledgements Prof. Fernando Tadeo is funded by Conserjeria de Educacion, Junta de Castilla y Leon
with European Regional Development Funds (Grant No. CLU 2017-09 and UIC 233).

References
1. A. Arbi, F. Charif, C. Aouiti, A. Touati, Dynamics of new class of Hopfield neural networks with
time-varying and distributed delays. Acta Math. Sci. 36(3), 891–912 (2016)
2. A. Arbi, J. Cao, Pseudo-almost periodic solution on time-space scales for a novel class of competitive
neutral-type neural networks with mixed time-varying delays and leakage delays. Acta Math. Sci.
46(2), 719–745 (2017)
4888 Circuits, Systems, and Signal Processing (2020) 39:4868–4889

3. A. Arbi, J. Cao, A. Alsaedi, Improved synchronization analysis of competitive neural networks with
time-varying delays. Nonlinear Anal. Model. Control 23(1), 82–102 (2018)
4. A. Arbi, Dynamics of BAM neural networks withmixed delays and leakage time-varying delays in the
weighted pseudo almost periodic on time-space scales. Math. Methods Appl. Sci. 41(3), 1230–1255
(2018)
5. S.A. Belamfedel, E.H. Tissir, N. Chaibi, Active queue management based feedback control for TCP
with successive delays in single and multiple bottleneck topology. Comput. Commun. 117, 58–70
(2018)
6. F.A. Bender, Delay dependent antiwindup synthesis for time delay systems. Int. J. Intell. Control Syst.
18(1), 1–9 (2013)
7. Y.Q. Cao, Y.Y. Sun, X.Y. Wang, Anti-windup compensator gain design for time-delay systems with
constraints. Acta Autom. Sin. 32(1), 1–8 (2006)
8. Y. Chen, Y. Li, S. Fei, Anti-windup design for time-delay systems via generalised delay-dependent
sector conditions. IET Control Theory Appl. 11(10), 1634–1641 (2017)
9. J.M.G. Da Silva, G. Garcia, Anti-windup design for time-delay systems subject to input saturation-an
LMI-based approach. Eur. J. Control 12(6), 622–634 (2006)
10. Z. Du, Z. Qin, H. Ren, Z. Lu, Fuzzy robust H∞ sampled-data control for uncertain nonlinear systems
with time-varying delay. Int. J. Fuzzy Syst. 19(5), 1417–1429 (2016)
11. A. Ech-charqy, M. Ouahi, E.H. Tissir, Delay-dependent robust stability criteria for singular time-delay
systems by delay-partitioning approach. Int. J. Syst. Sci 39(4), 2957–2967 (2018)
12. H. El Aiss, A. Hmamed, A. El Hajjaji, Improved stability and H∞ performance criteria for linear
systems with interval time-varying delays via three terms approximation. Int. J. Syst. Sci. 48(16),
3450–3458 (2017)
13. H. El Aiss, H. Rachid, A. Hmamed, A. El Hajjaji, Approach to delay-dependent robust stability and
stabilisation of delta operator systems with time-varying delays. IET Control Theory Appl. 12(6),
786–792 (2018)
14. N. El Fezazi, F. El Haoussi, E.H. Tissir, F. Tadeo, Delay dependent anti-windup synthesis for time-
varying delay systems with saturating actuators. Int. J. Comput. Appl. 111(1), 1–6 (2015)
15. N. El Fezazi, O. Lamrabet, F. El Haoussi, E.H. Tissir, T. Alvarez, F. Tadeo, Proceedings of the Inter-
national Conference on Systems and Control (ICSC). Robust controller design for congestion control
in TCP/IP routers, Marrakesh, Morocco (2016), p. 243–249
16. N. El Fezazi, E.H. Tissir, F. El Haoussi, T. Alvarez, F. Tadeo, Control Based on Saturated Time-Delay
Systems Theory of Mach Number in Wind Tunnels. Circuits, Systems and Signal Processing (2017)
17. N. El Fezazi, F. El Haoussi, E.H. Tissir, T. Alvarez, F. Tadeo, Robust stabilization using a sampled-
data strategy of uncertain neutral state-delayed systems subject to input limitations. Int. J. Appl. Math.
Comput. Sci. 28, 111–122 (2018)
18. F. El Haoussi, E.H. Tissir, Delay and its time-derivative dependent robust stability of uncertain neutral
systems with saturating actuators. Int. J. Autom. Comput. 7(4), 455–462 (2010)
19. F. El Haoussi, E.H. Tissir, A. Hmamed, F. Tadeo, Stabilization of neutral systems with saturating
actuators. J. Control Sci. Eng. (2012). https://doi.org/10.1155/2012/823290
20. E. Fridman, U. Shaked, Input–output approach to stability and L 2 -gain analysis of systems with time-
varying delays. Syst. Control Lett. 55(12), 1041–1053 (2007)
21. Y. He, Q.G. Wang, L. Xie, C. Lin, Further improvement of free weighting matrices technique for
systems with time-delay. IEEE Trans. Autom. Control 52(2), 293–299 (2007)
22. A. Hmamed, H. El Aiss, A. El Hajjaji, Stability analysis of linear systems with time varying delay:
an input output approach. IEEE 54th Annual Conference Decision and Control (CDC), Osaka, Japan
(2015), p. 15–18
23. J. Hu, W.P. Li, Theory of Ordinary Differential Equations Existence (Department of Mathematics, The
Hong Kong University of Science and Technology, Uniqueness and Stability, 2005)
24. O. Lamrabet, N. El Fezazi, F. El Haoussi, E.H. Tissir, Using input delay approach for synthesizing
an anti-windup compensator to AQM in TCP/IP networks. International Conference on Advanced
Technologies, Signal and Image Processing, Fez, Morocco (2017). https://doi.org/10.1109/ATSIP.
2017.8075573
25. O. Lamrabet, E.H. Tissir, F. El Haoussi, Anti-windup compensator synthesis for sampled-data delay
systems. Circuits Syst. Signal Process (2018). https://doi.org/10.1007/s00034-018-0971-9
26. O. Lamrabet, A. Ech-charqy, E.H. Tissir, F. El Haoussi, Sampled data control for Takagi–Sugeno fuzzy
systems with actuator saturation. Proc. Comput. Sci. 148, 448–454 (2019)
Circuits, Systems, and Signal Processing (2020) 39:4868–4889 4889

27. K. Liu, E. Fridman, Wirtinger’s inequality and Lyapunov-based sampled-data stabilization. Automatica
48(1), 102–108 (2012)
28. Y. Liu, M. Li, Improved robust stabilization method for linear systems with interval time-varying input
delays by busing Wirtinger inequality. ISA Trans. 56, 111–122 (2015). https://doi.org/10.1016/j.isatra.
2014.12.008
29. A. Manitius, Feedback controllers for a wind tunnel model involving a delay: analytical design and
numerical simulation. IEEE Trans. Autom. Control 29(12), 1058–1068 (1984)
30. M.J. Park, O.M. Kwon, J.H. Park, S.M. Lee, E.J. Cha, Stability of time-delay systems via wirtinger-
based double integral inequality. Automatica 55, 204–208 (2015)
31. H. Rachid, O. Lamrabet, E.H. Tissir, Stabilization of delta operator systems with actuator saturation via
anti-windup compensator. Symmetry 11, 1084–1093 (2019). https://doi.org/10.3390/sym11091084
32. R. Sakthivel, M. Joby, P. Shi, K. Mathiyalagan, Robust reliable sampled-data control for switched
systems with application to flight control. Int. J. Syst. Sci. 47(15), 3518–3528 (2016)
33. S. Santra, H.R. Karimi, R. Sakthivel, A.S. Marshal, Dissipative based adaptive reliable sampled-data
control of time-varying delay systems. J Franklin Inst. 14(1), 39–50 (2016)
34. S. Tarbouriech, J.M.G. Da Silva, G. Garcia, Delay-dependent anti-windup loops for enlarging the
stability region of time-delay systems with saturating inputs. J. Dyn. Syst. Meas. Control 125(2),
265–267 (2003)
35. E.H. Tissir, Delay-dependent robust stability of linear systems with non-commensurate time varying
delays. Int. J. Syst. Sci. 38, 749–757 (2007)
36. E.H. Tissir, Delay-dependent robust stabilization for systems with uncertain time varying delay. J. Dyn.
Syst. Meas. Control 132(5), 054504 (2010)
37. Y. Wu, B. Jiang, N. Lu, A descriptor system approach for estimation of incipient faults with application
to high-speed railway traction devices. IEEE Trans. Syst. Man Cybern. Syst. (2017). https://doi.org/
10.1109/TSMC.2017.2757264
38. H. Yang, Z. Li, C. Hua, Z. Liu, Stability analysis of delta operator systems with actuator saturation by
a saturation dependent Lyapunov function. Circuits Syst. Signal Process. (2015). https://doi.org/10.
1007/s00034-014-9876-4
39. L. Zhao, H. Gao, H.R. Karimi, Robust stability and stabilization of uncertain T–S fuzzy systems with
time varying delay: an input–output approach. IEEE Trans. Fuzzy Syst. 21(5), 883–897 (2013)
40. T. Zoulagh, H. El Aiss, A. Hmamed, A. El Hajjaji, H∞ filter design for discrete time-varying delay
systems: three-term approximation approach. IET Control Theory Appl. 12(2), 254–262 (2018)

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.