Ismail Er Rachid, Redouane Chaibi, and Abdelaziz Hmamed

Stability of 2-D Continuous T-S Fuzzy Systems

Based on KYP Lemma
Abstract: This paper focuses on the problem of stability for two-dimensional (2-D) con-
tinuous system described by the Takagi-Sugeno (T-S) fuzzy for Roesser model, with the
aid of the frequency-partitioning approach combined with the Generalized Kalman
Yakubovich Popov (GKYP) lemma. In order to reduce the conservativeness of the ex-
isting simple 2-D Lyapunov inequality, a sufficient stability condition of the 2-D contin-
uous T-S fuzzy systems is proposed via linear matrix inequalities (LMIs) formulation.
Finally, numerical example is given, so that illustrate the effectiveness of the proposed
method.

Keywords: Stability, 2-D Continuous systems, Roesser model, KYP lemma, Frequency-
partitioning, LMIs

1 Introduction
As is well known, stability of two-dimensional (2-D) continuous systems is the major
aim in all research, so as to guarantee the normal operation of systems. For this rea-
son, the stability of 2-D continuous systems has received considerable attention from
many papers, to list some of these, the works [12–14, 20–24], consider the stability and
stabilization of 2-D continuous systems. Furthermore, the 2-D continuous systems are
studied in some papers as well, for instance, the robust state feedback H∞ control and
robust H∞ filtering have been studied in [1, 2, 7, 25].
Recently, attention has been devoted towards the Kalman−Yakubovich−Popov
(KYP) lemma [8], this lemma makes equivalence between frequency domain inequal-
ity (FDI) characterizing a class of properties of a transfer function, and a linear matrix
inequality (LMI) [10], for its state space realization. Therefore, authors in [9] proposed
an extension of the KYP lemma, which is known as Generalized KYP (GKYP) lemma
for the case of finite frequency domain. The 2-D GKYP lemma is obtained for 2-D dis-
crete systems, of Fornasini-Marchesini (FM) and of Roesser models in [30] and [29],
respectively, and for Roesser model of 2-D continuous systems in [11], and for 2-D
continuous-discrete systems of Roesser model in [28]. The finite frequency H∞ filter-
ing for uncertain 2-D systems, of FM model and of Roesser model, was considered
in [27] and [26], respectively. The GKYP combined with the frequency-partitioning
approach to stability analysis, were obtained in [15] for 2-D discrete system, and for

Ismail Er Rachid, Redouane Chaibi, Abdelaziz Hmamed, LESSI, Department of Physics, Faculty of
Sciences Dhar El Mehraz B.P. 1796 Fes-Atlas, Morocco, e-mails: ismail.errachid@gmail.com,
c.redouane.chaibi@gmail.com, hmamed_abdelaziz@yahoo.fr

De Gruyter Oldenbourg, ASSD – Advances in Systems, Signals and Devices, Volume 9, 2019, pp. 267–280.
https://doi.org/10.1515/9783110591729-015
Brought to you by | Cambridge University Library
Authenticated
Download Date | 12/23/19 1:38 PM
268 | I. Er Rachid et al.

hybrid systems in [16] and [17], and for linear time-delay systems in [18]. However, the
afore-mentioned results were only for linear 2-D systems. As well known, most of the
actual 2-D systems are nonlinear.
On the other hand, Takagi-Sugeno (T-S) fuzzy systems have been used for nonlin-
ear control systems, because the T-S fuzzy model is a good approximation to nonlinear
systems [33]. Therefore, Takagi-Sugeno fuzzy systems have played an important role
in designing stabilizing controllers for nonlinear systems. Hence, the T-S fuzzy sys-
tems have received a great important interest these last two decades, to mention a few,
the problem of stability condition of the T-S fuzzy control systems is proposed in [31],
robust stabilization of T-S fuzzy systems with parametric uncertainties is presented
in [32]. By using the basis-dependent Lyapunov function approach and adding slack
matrix variable, a new condition for H∞ performance analysis is proposed in [36] for a
class of 2-D nonlinear systems. In [4, 6], the authors studied the H∞ control problems
for Takagi-Sugeno (T-S) fuzzy systems. Stability and stabilization for 2-D discrete T-S
fuzzy systems have been solved lately in [35, 37, 38, 34], stabilization and stability for
2-D continuous T-S fuzzy systems were considered in [39, 40].
Motivated by the Previous research, in this paper, we suggest the stability con-
ditions of 2-D continuous T-S fuzzy systems for Roesser model, via GKYP lemma and
frequency-partitioning approach. Generally, in order to realize a series of novel stabil-
ity conditions for the system, the GKYP lemma is applied on each one of the N inter-
vals divided of the whole frequency domain. Finally, numerical example is given to
demonstrate the effectiveness of the proposed approach.
The rest of this paper is organized as follows. In section 2 the problem description
and preliminaries are given. The main results of this paper are given in Section 3. In
Section 4, numerical example is given to show the effectiveness of the results. Section 5
concludes the paper.
Notation: The superscript T, ∗, −1 stand for matrix transpose, matrix complex
conjugate transpose, matrix inverse, respectively. I denotes an identity matrix with
appropriate dimension. The notation P > 0 (P < 0) means that matrix P is positive
(negative). diag{.} stands for the block diagonal matrix, Reλ(.) is the real of eigenvalue
of a square matrix. Matrices, if their dimensions are not explicitly stated, are assumed
to be compatible for algebraic operations.

2 Problem Formulation and Preliminaries

Consider a Roesser type continuous time 2-D T-S fuzzy system described by:
IF z1 (t1 , t2 ) is Mλ1 and... and zl (t1 , t2 ) is Mλl
THEN
𝜕x h (t1 ,t2 ) r h
[ 𝜕t1 ] = ∑ hλ (z(t1 , t2 ))Aλ [ x (t1 , t2 ) ] (1)
𝜕x v (t1 ,t2 ) xv (t1 , t2 )
[ 𝜕t2 ] λ=1

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
 Stability of 2-D Continuous T-S Fuzzy Systems Based on KYP Lemma | 269

with
Aλ1 Aλ2
Aλ = [ ], xh (0, t2 ) = f (t2 ); xv (t1 , 0) = g(t1 )
Aλ3 Aλ4

where, xh (t1 , t2 ) ∈ ℝnh , xv (t1 , t2 ) ∈ ℝnv are the horizontal state and vertical state of
system, respectively, Aλ are real matrices with appropriate dimensions, Mλl is the fuzzy
set, zp (t1 , t2 ), for p = 1, . . . , l are the premise variables, r is the number of IF-THEN rules,
and t1 , t2 ∈ ℝ+ are the two variables of the systems,
r
A1 (h) A2 (h)
A(h) = ∑ hλ (z(t1 , t2 ))Aλ , A(h) = [ ] (2)
λ=1
A3 (h) A4 (h)
l
αλ (z(t1 , t2 )) = ∏ Mrp (zp (t1 , t2 )) (3)
p=1

and the normalized membership functions fulfill

αλ (z(t1 , t2 ))
hλ (z(t1 , t2 )) = r (4)
∑λ=1 αλ (z(t1 , t2 ))
0 ≤ hλ (z(t1 , t2 )) ≤ 1 (5)
r
∑ hλ (z(t1 , t2 )) = 1 (6)
λ=1

Lemma 1 ([5]). Simple necessary conditions for asymptotic stability of the 2-D continu-
ous T-S fuzzy system (1) are as follows:
i) A1 (h) are Hurwitz (i. e. Reλi (A1 (h)) < 0, i = 1, .., nh )
ii) A4 (h) are Hurwitz.

Proof. From (1) for A2 (h) = A3 (h) = A4 (h) = 0, we obtains the state equation of the
continuous system (for the fixed 0 ≤ t2 ∈ ℝ)

𝜕x h (t1 , t2 )
= A1 (h)xh (t1 , t2 ) (7)
𝜕t1
the system (7) is asymptotically stable if the matrix A1 (h) is Hurwitz.
Similarly, we can proof ii).

Therefore, we assume the following throughout the paper.

Assumption 1. The matrices A1 (h) and A4 (h) are Hurwitz.

Lemma 2 ([3]). Let the Assumption 1 be satisfied, the 2-D continuous T-S fuzzy system
(1) is asymptotically stable if

S(s, h) = A3 (h)(sI − A1 (h))−1 A2 (h) + A4 (h), s = jω (8)

is Hurwitz matrix for ω ∈ ℝ.

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
270 | I. Er Rachid et al.

Proof. Taking the Laplace transformation of system (1) for t1 only, and under zero ini-
tial condition, we get
sX h (s,t2 )
A1 (h) A2 (h) X h (s, t2 )
[ v ]=[ ][ v ] (9)
𝜕X (s,t2 )
𝜕t2
A3 (h) A4 (h) X (s, t2 )

Solving (9), we obtain

𝜕X v (s, t2 )
= [ A3 (h)(sI − A1 (h))−1 A2 (h) + A4 (h) ] X v (s, t2 ) (10)
𝜕t2
System (10) can be regarded as a 1-D continuous system with complex variable s, and
we notice that the variable t2 of the system does not depend on the variable s.
So the 2-D continuous T-S fuzzy system (10) is asymptotically stable if [A3 (h)(sI −
A1 (h))−1 A2 (h) + A4 (h)] is Hurwitz matrix for Re(s) = 0. Hence, the 2-D continuous T-S
fuzzy system (1) is asymptotically stable if [A3 (h)(jωI −A1 (h))−1 A2 (h)+A4 (h)] is Hurwitz
matrix for s = jω, ∀ω ∈ ℝ.

Remark 1. Notice that when interchanging A1 (h) with A4 (h) and A2 (h) with A3 (h) the
2-D continuous T-S fuzzy system (1) is asymptotically stable if A2 (h)(jωI−A4 (h))−1 A3 (h)+
A1 (h) is Hurwitz matrix for s = jω, ∀ω ∈ ℝ.

We will use the following lemmas, known as the KYP lemma, the GKYP and the
Projection lemma, respectively.

Lemma 3 ([8]). Let matrices A, B, and Θ = ΘT be given, if det(jωI − A) ≠ 0 ∀ω ∈ ℝ. Then

the following two statements are equivalent.
(i) For any ω ∈ ℝ ∪ ∞,
∗
(jωI − A)−1 B (jωI − A)−1 B
[ ] Θ[ ]<0 (11)
I I
(ii) There exists a symmetric matrix P such that
∗
A B 0 P A B
[ ] [ ][ ]+Θ<0 (12)
I 0 P 0 I 0

Lemma 4 ([9]). Let the matrices Θ, F, Φ and Ψ be given, and denote by Nω the null space
of Tω F, where Tω = [ I −jωI ]. The inequality

Nω∗ ΘNω < 0, with ω ∈ [ω1 , ω2 ], (13)

holds if and only if there exist Q > 0 and a symmetric matrix P, such that

F ∗ (Φ ⊗ P + Ψ ⊗ Q)F + Θ < 0 (14)

where
0 1 −1 jωc (ω1 + ω2 )
Φ=[ ], Ψ=[ ], ωc = .
1 0 −jωc −ω1 ω2 2

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
 Stability of 2-D Continuous T-S Fuzzy Systems Based on KYP Lemma | 271

Lemma 5 ([10]). Given a symmetric matrix Γ ∈ ℝp×p and two matrices X, Z of column
dimension p, there exists a matrix Y such that the LMI

Γ + symX T YZ < 0 (15)

holds if and only if the following two projection inequalities with respect to Y are satis-
fied:
T T
X ⊥ ΓX ⊥ < 0, Z ⊥ ΓZ ⊥ < 0, (16)

where X ⊥ and Z ⊥ are arbitrary matrices whose columns form a basis of the null spaces
of X and Z, respectively.

3 Stability Analysis
We are now in a position to present a new condition for checking the stability of 2-D
continuous T-S fuzzy Roesser model.

Lemma 6. The 2-D continuous T-S fuzzy system (1) is asymptotically stable if there exist
P1 > 0 and P2 > 0 such that the LMI

A(h)T P + PA(h) < 0, is feasible (17)

where
A1 (h) A2 (h)
P = diag{P1 , P2 }, and A(h) = [ ]
A3 (h) A4 (h)

Proof. LMI (17) can be rewritten as

P1 A1 (h) + A1 (h)T P1 A3 (h)T P2 + P1 A2 (h)

[ ]<0 (18)
∗ P2 A4 (h) + A4 (h)T P2

and this latter LMI can be rewritten as

T
A1 (h) A2 (h) 0 P1 A (h) A2 (h)
[ ] [ ][ 1 ] + Θ(h) < 0 (19)
I 0 P1 0 I 0

where
T
A3 (h) A4 (h) 0 P2 A (h) A4 (h)
Θ(h) = [ ] [ ][ 3 ] (20)
0 I P2 0 0 I

by Lemma 3, (19) is equivalent to

∗
(jωI − A1 (h))−1 A2 (h) (jωI − A1 (h))−1 A2 (h)
[ ] Θ(h) [ ] < 0,
I I

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
272 | I. Er Rachid et al.

or
∗
S(jω, h) 0 P2 S(jω, h)
[ ] [ ][ ]<0 (21)
I P2 0 I

where S(jω, h) is the frequency response matrix obtained from S(s, h) of Lemma 4,
moreover, (21) can be written as

S(jω, h)∗ P2 + P2 S(jω, h) < 0, (22)

and the existence of a P2 > 0 satisfying this last condition immediately implies that
all eigenvalues of S(jω, h) must have strictly negative real parts, ∀ω ∈ ℝ ∪ ∞, that is,
feasibility of (17) guarantees that condition of Lemma 2 holds. Moreover, feasibility of
(17) implies that

P1 A1 (h) + A1 (h)T P1 < 0,

P2 A4 (h) + A4 (h)T P2 < 0,

and, since P1 > 0 and P2 > 0, all eigenvalues of the matrices A1 (h) and A4 (h) must
have strictly negative real parts, and feasibility of (17) guarantees that Lemma 1 and
Lemma 2 are satisfied.

Remark 2. There is some conservativeness with Lemma 6 although it proposes an con-

dition for the asymptotical stability of the system (1) in term of LMI, this conservative-
ness due to the existence of some 2-D continuous T-S fuzzy systems that have Hur-
witz parameters A1 (h) and A4 (h) and Hurwitz function S(jω, h) but the application of
Lemma 6 does not decide the feasibility (as the system we have in the example).

Main result: In order to reduce the conservativeness of the Lemma 6 we can choose
instead of P2 (which is restrictive ∀ω ∈ ℝ ∪ ∞) a piecewise constant matrices P2 (jω)
over the whole frequency field via a frequency-partitioning approach, the existence of
P2 (jω) > 0 such that:

S(jω, h)∗ P2 (jω) + P2 (jω)S(jω, h) < 0, ∀ω ∈ ℝ ∪ ∞.

Denote Ω = ℝ ∪ ∞ and due to S(−jω, h) = S∗ (jω, h), the following identities hold:

sup Reλ(S(jω, h)) = sup Reλ(S(jω, h)) = sup Reλ(S(jω, h)) (23)
ω∈Ω ω∈Ω+ ω∈Ω−

where Ω+ = [0, ∞], Ω− = [∞, 0]. Hence, without loss of generality, it suffices to con-
sider the half frequency ranges Ω+ . Now, for a given positive integer N, the frequency
domain Ω+ is divided into N sub-intervals, such that
N
Ω+ = ⋃ Ωi , Ωi = [ωi−1 , ωi ], ω0 = 0, ωN = ∞, (24)
i=1

and then applying the result of Lemma 4 to each interval, we obtain the following
theorem.

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
 Stability of 2-D Continuous T-S Fuzzy Systems Based on KYP Lemma | 273

Theorem 1. For a given positive integer N, define frequency intervals Ω+ as in (24). Sys-
tem (1) is asymptotically stable if there exist P1i > 0, P2i > 0, Qi > 0, i = 1, 2, . . . , N,

ATλ Pi + Pi Aλ + FλT (Ψi ⊗ Qi )Fλ < 0, λ = 1, . . . , r (25)

where

A1λ A2λ A1λ A2λ

Aλ = [ ], Fλ = [ ], and Pi = diag{P1i , P2i }.
A3λ A4λ I 0

For i = 2, 3, . . . , N − 1,

−1 jωci (ωi−1 + ωi )
Ψi = [ ], with ωci = . (26)
−jωci −ωi−1 ωi 2

For i = 1 and i = N,

−1 0 1 0
Ψ1 = [ ] and ΨN = [ ] (27)
0 ω21 0 −ω2N−1

respectively.

Proof. For i = 2, . . . , N − 1, i = 1 and i = N, according to [9] the matrix Ψi should be

taking as (26) and (27), respectively.
The condition in (25), can be written as

F T (h)(Φ ⊗ P1i + Ψi ⊗ Qi )F(h) + Θi (h) < 0 (28)

where

A1 (h) A2 (h) 0 1
F(h) = [ ], Φ=[ ] (29)
I 0 1 0

and
T
A3 (h) A4 (h) A (h) A4 (h)
Θi (h) = [ ] (Φ ⊗ P2i ) [ 3 ] (30)
0 I 0 I

Denote G(jω, h) = (jωI − A1 (h))−1 A2 (h), by the lemma 4, the following inequality fol-
lows:
∗
G(jω, h) G(jω, h)
[ ] Θi (h) [ ] < 0, ∀ω ∈ Ω+ (31)
I I

note that

S(jω, h) = A4 (h) + A3 (h)(jωI − A1 (h))−1 A2 (h) = A4 (h) + A3 (h)G(jω, h).

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
274 | I. Er Rachid et al.

Thus, substituting the expression of Θi (h) into (31) results in

∗
S(jω, h) S(jω, h)
[ ] (Φ ⊗ P2i ) [ ] < 0, ∀ω ∈ Ω+ (32)
I I

or in a more compact form,

S(jω, h)∗ P2i + P2i S(jω, h) < 0, P2i > 0, ∀ω ∈ Ω+ (33)

So Reλ(S(jω, h)) < 0 is finally guaranteed for all ω ∈ Ω+ . Combining Reλ(A1 (h)) < 0,
Reλ(A4 (h)) < 0 and Reλ(S(jω, h)) < 0, we conclude that system (1) is asymptotically
stable based on Lemma 2. The proof is completed.

Remark 3. When N = 1, if letting Qi = 0, P1i > 0 and P2i > 0 be real, then (25) reduces
to (17), that is, Lemma 6 is a special case of Theorem 1.

Theorem 2. For a given positive integer N, define frequency intervals Ω+ as in (24). Sys-
tem (1) is asymptotically stable if there exist P1i > 0, P2i > 0, Qi > 0, Y1i , Y2i , i =
2, 3, . . . , N − 1, such that
ϒ11i ϒ12iλ 0 ϒ14iλ
[ ∗ ϒ22iλ ϒ23iλ ϒ24iλ ]
[ ]
Ξiλ = [ ] < 0, λ = 1, . . . , r (34)
[ ∗ ∗ ϒ33iλ ϒ34iλ ]
[ ∗ ∗ ∗ ϒ44iλ ]
ϒ11i = −Qi − Y1i − Y1iT ,
ϒ12iλ = P1i + jωci Qi − Y1iT + Y1i A1λ ,
ϒ14iλ = Y1i A2λ ,
ϒ22iλ = −ωi−1 ωi Qi + AT1λ Y1iT + Y1i A1λ ,
ϒ23iλ = AT3λ Y2iT ,
ϒ24iλ = AT3λ Y2iT + Y1i A2λ ,
ϒ33i = −Y2i − Y2iT ,
ϒ34iλ = P2i − Y2iT + Y2i A4λ ,
ϒ44iλ = Y2i A4λ + AT4λ Y2iT .
For i = 1, we replace ϒ12i , ϒ22iλ in (34) by
T
ϒ121λ = P11 − Y11 + Y11 A1λ ,
2
ϒ221λ = ω1 Q1 + AT1λ Y11
T
+ Y11 A1λ , respectively.
For i = N, we replace ϒ11i , ϒ12iλ and ϒ22iλ in (34) by
T
ϒ11N = QN − Y1N − Y1N ,
T
ϒ12Nλ = P1N − Y1N + Y1N A1λ ,
ϒ22Nλ = −ω2N−1 QN + AT1λ Y1N
T
+ Y1N A1λ , respectively.

Proof. From Theorem 1, let

Φ ⊗ P1i + Ψi ⊗ Qi 0
Γ=[ ]. (35)
∗ Φ ⊗ P2i

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
 Stability of 2-D Continuous T-S Fuzzy Systems Based on KYP Lemma | 275

According to [9], for i = 2, . . . , N − 1, i = 1 and i = N, Ψi as in (26) and (27) respectively,

and Φ as in (29).
Let
Y1i 0
[ Y 0 ] −I A1 (h) 0 A2 (h)
[ ]
Y = [ 1i ], Z(h) = [ ], X = I,
[ 0 Y2i ] 0 A3 (h) −I A4 (h)
[ 0 Y2i ]
(34) is equivalent to

sym(X T YZ(h)) + Γ < 0 (36)

since one can choose X ⊥ = 0, the first inequality in (16) vanishes, and then by lemma 5,
T
(36) hold for some Y if and only if Z(h)⊥ ΓZ(h)⊥ < 0. Note that Z(h)⊥ can be selected
as
A1 (h) A2 (h)
[ I 0 ]
[ ]
Z(h)⊥ = [ ],
[ A3 (h) A4 (h) ]
[ 0 I ]
T
and then by calculation, we can obtain the equivalence between Z(h)⊥ ΓZ(h)⊥ < 0
and (25). Consequently (25) is equivalent to (34).

4 Numerical Example
In this part, we provide an example to demonstrate the effectiveness of the proposed
method.
Consider system (1), where matrices are obtained by a [19]:

A1 A2
Ac = [ ], Ac = (Ad − 1)(Ad + 1)−1
A3 A4

The matrices in the original problem are the following [15]:

Ad1 Ad2 0.5 0.5 0.4 1.1

Ad = [ ], Ad1 = [ ], Ad2 = [ ],
Ad3 Ad4 0.1 −0.1 0.6 0.1
−0.1 −0.1 −0.5 −0.5
Ad3 = [ ], Ad4 = [ ]
−0.2 0.6 −0.1 −0.7

we obtain
5.2979 −16.0426 20.5957 23.9149
A1 = [ ], A2 = [ ],
4.2128 −12.7447 16.4255 16.5106

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
276 | I. Er Rachid et al.

−5.7872 16.2553 −22.5745 −23.4894

A3 = [ ], A4 = [ ].
−7.4894 22.2128 −26.9787 −30.5745

Theorem 2 with N = 1 don’t decide the stability of the system. By increasing N,

it is found that Theorem 2 with N = 2, 4, 8 succeeds, note that, the above system is
asymptotically stable only for i = 2, . . . , N. But for i = 1, whatever N, and whatever the
way of partitioning the entire interval (fig. 1), always system is not stable. This is due
to the quick variation of the curve of Reλmax (S(jω, h)) in the vicinity of ω0 = 0.

Figure 1: Reλmax (S(jω)) and γi∗ for Ω = [0, 1].

Denote γi∗ as the minimum value of −γi that satisfies

sup Reλmax (S(jω, h)) < −γi < 0

ω∈Ωi

0 P
γi∗ could be computed from (34) by replacing Φ ⊗ P2i in (35) by [ P2i γ2ii ] and minimiz-
ing −γi .
Fig. 1 shows Reλmax (S(jω, h)) and the executed γi∗ by Theorem 2 with N = 1, 2, 4, 8.
The stability of the above system is obtained, since Reλmax (S(jω, h)) < 0 is evident.
With N growing, it is further shown that −γi tends to the value of Reλmax (S(jω, h))
over Ω+ .

Remark 4. In the literature, the authors of [15] and [16] get almost the similar situ-
ation for the discrete and the continuous-discrete cases, respectively, in which they
note that the way of partitioning the entire frequency domain play a role in the con-
servativeness of the used method. Means that there is no systematic way to reduce this
conservativeness.

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
 Stability of 2-D Continuous T-S Fuzzy Systems Based on KYP Lemma | 277

Remark 5. In the field [16, +∞], ℜe[λmax (S(jω, h))] remains relatively stationary to the
value Reλmax (S(j∞)) = ℜe[λmax (A4 (h))] = −1.0850, then γi∗ also tends to this value
throughout the domain. Even if it decomposed, we find very similar values to −1.0850.
That’s why we worked on just the domain [0, 16], (Fig. 1).

5 Conclusions
In this paper, the stability problem of 2-D continuous T-S fuzzy systems in Roesser
model has been studied. Sufficient condition of stability for 2-D continuous T-S fuzzy
system is proposed via LMIs formulation. By combining the GKYP lemma with the
frequency-partitioning approach and introducing a piecewise constant matrix func-
tion, the conservativeness of the problem has been reduced.
Finally, the advantage of the proposed approach have been demonstrated through
numerical example.

Bibliography
[1] I. Ghous, Z. Xiang, Robust state feedback H∞ control for uncertain 2-D continuous state
delayed systems in the Roesser model, Multidimensional Systems and Signal Processing,
27(2):297–319, 2016.
[2] I. Ghous, Z. Xiang, H∞ control of a class of 2-D continuous switched delayed systems via
state-dependent switching. Int. Journal of Systems Science, 47(2):300–313, 2016.
[3] I. Er Rachid, A. Hmamed, B. El Haiek, Stability of 2-D continuous systems in roesser model
based on KYP lemma. Systems, Signals & Devices (SSD), 2017 14th IEEE Int. Multi-Conf., 2017.
[4] R. Chaibi, I. Er Rachid, E. H. Tissir, A. Hmamed, Finite Frequency Static Output Feedback H∞
Control of Continuous-Time T-S Fuzzy Systems. Journal of Circuits, Systems and Computers,
1950023, 2018.
[5] I. Errachid, A. Hmamed, Stability and Robust Stabilization of 2-D Continuous Systems in
Roesser Model Based on GKYP Lemma. Int. Journal of Power Electronics and Drive Systems
(IJPEDS), 8(3):990–1001, 2017.
[6] I. Er Rachid, R. Chaibi, E. H. Tissir, A. Hmamed. Observer-Based H∞ Control with Finite
Frequency Specifications for Discrete-time T-S Fuzzy Systems. Int. Journal of Systems Science,
2018.
[7] A. Hmamed, C. E. Kasri, E. H. Tissir, T. Alvarez, F. Tadeo, Robust H∞ filtering for uncertain
2-D continuous systems with time-varying delays, Int. Journal of Innovative Computing,
Information and Control, 9(5):2167–2183, 2016.
[8] A. Rantzer, On the Kalman–Yakubovich–Popov lemma. Systems and Control Letters,
1(28):7–10, 1996.
[9] T. Iwasaki, S. Hara, Generalized KYP lemma: Unified frequency domain inequalities with design
applications, IEEE Trans. Autom. Control, 50(1):41–59, 2005.
[10] P. Gahinet, P. Apkarian, A linear matrix inequality approach to H∞ control, Int. Journal of
Robust and Nonlinear Control 4:421–448, 1994.
[11] Z. Duan, Z. Xiang, Finite frequency H∞ control of 2-D continuous systems in Roesser model,
Multidimensional Systems and Signal Processing 2016.

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
278 | I. Er Rachid et al.

[12] E. I. Jury, P. Bauer, On the stability of two-dimensional continuous systems, IEEE Trans. Circuits
Systems, 35:1487–1500, 1988.
[13] N. E. Mastorakis, M. Swamy, A new method for computing the stability margin of
two-dimensional continuous systems, IEEE Trans. Circuits Systems, 49(1):869–872 2002.
[14] K. Galkowski, LMI based stability analysis for 2−D continuous systems, 9th IEEE Int. Conf. on
Electronics, Circuits and Systems-ICECS, 923–926, Dubrovnik, Croatia, 2002.
[15] X. Li, J. Lam, H. Gao, Y. Gu, A frequency-partitioning approach to stability analysis of
two-dimensional discrete systems. Multidimensional Systems and Signal Processing, 26(1),
67–93, 2013.
[16] L. Wang, W. Wang, J. Gao, W. Chen, Stability and robust stabilization of 2−D
continuous−discrete systems in Roesser model based on KYP lemma, Multidimensional
Systems and Signal Processing, 2015.
[17] W. Paszke, E. Rogers, K. Galkowski, KYP lemma based stability and control law design for
differential linear repetitive processes with applications. Systems and Control Letters,
62(7):560–566, 2013.
[18] X. Li, H. Gao, K. Gub, Delay-independent stability analysis of linear time-delay systems based
on frequency discretization, Automatica, 70:288–294, 2016.
[19] J. B. Lodge, M. M. Fahmy, The Bilinear Transformation of Two-Dimensional State-Space
Systems, IEEE Trans. on Acoustics, Speech and Signal Processing, 30(3), 1982.
[20] M. Alfidi, A. Hmamed, Robust Stability Analysis for 2−D Continuous-Time Systems Via
Parameter-Dependent Lyapunov Functions, Wseas Trans. on systems and control, 2(11), 2007.
[21] S. Kririm, A. Hmamed, Stability Analysis for Uncertain 2-D Continuous Systems using
Polynomially Parameter-Dependent Lyapunov Functions, 5th Workshop on Information
Technologies and Communication, FEZ, Morocco, 2013.
[22] A. Hmamed, S. Kririm, A. Benzaouia, F. Tadeo, Delay-dependent stability and stabilization of
continuous 2-D delayed systems with saturating control, Int. Journal of Systems Science, 2015.
[23] S. Xu, J. Lam, Y. Zou, Z. Lin, W. Paszke, Robust positive real synthesis for 2D continuous
systems via state and output feedback, Circuits systems signal processing, 24(2):183–199,
2005.
[24] J. Lam, S. Xu, Y. Zou, Z. Lin, and K. Galkowski, Robust output feedback stabilization for
two-dimensional continuous systems in Roesser form, Applied Mathematical Letters,
17(12):1331–1341, 2004.
[25] C. El-kasri, A. Hmamed, T. Alvarez and F. Tadeo, Robust H∞ Filtering for Uncertain 2-D
Continuous Systems, based on a Polynomially Parameter-Dependent Lyapunov Function,
Multidimensional (nD) Systems (nDs), 7th Int. Workshop on. IEEE, 2011.
[26] X. Li, H. Gao, Robust finite frequency H∞ filtering for uncertain 2-D Roesser systems,
Automatica 48:1163–1170, 2012.
[27] X. Li, H. Gao, Robust finite frequency H∞ filtering for uncertain 2-D systems: The FM model
case, Automatica 49:2446–2452, 2013.
[28] L. Wang, W. Wang, G. Zhang, W. Chen, Generalised KYP lemma with its application in finite
frequency positive realness control for 2−D continuous-discrete systems in the Roesser model
form, IET Control Theory and Applications, 2015.
[29] R. Yang, L. Xie, C. Zhang, Generalized two-dimensional Kalman-Yakubovich-Popov
lemma for discrete Roesser model. IEEE Trans. on Circuits and Systems I: Regular Papers,
55(10):3223–3233, 2008.
[30] X. Li, H. Gao, C. Wang, Generalized Kalman–Yakubovich–Popov lemma for 2-D FM LSS model.
IEEE Trans. on Automatic Control, 57(12):3090–3103, 2012.
[31] E. Kim and H. Lee, New approaches to relaxed quadratic stability condition of fuzzy control
systems, IEEE Trans. Fuzzy Systems, 8(5):523–534, 2000.

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
 Stability of 2-D Continuous T-S Fuzzy Systems Based on KYP Lemma | 279

[32] S. Tong, H.-Hiong Li, Observer-based robust fuzzy control of nonlinear systems with parametric
uncertainties, Fuzzy Sets and Systems 131(2):165–184, 2002.
[33] T. Takagi, M. Sugeno, Fuzzy Identification of Systems and its Applications to Modeling and
Control, IEEE Trans. on Systems, Man and Cybernetics, 15(1):116–132, 1985.
[34] B. Boukili, A. Hmamed, A. Benzaouia, A. El hajjaji, H∞ State Control for 2D Fuzzy FM Systems
with Stochastic Perturbation, Circuits Syst Signal Process, DOI 10.1007/s00034-014-9889-z,
2014.
[35] A. Hmamed, A. El Hajjaji, A. Benzaouia, Stabilization of discrete-time 2D T-S Fuzzy systems
by state Feedback Control, 17th Mediterranean Conf. on Control and Automation Makedonia
Palace, Thessaloniki, Greece, (June 24–26, 2009).
[36] B. Boukili, A. Hmamed, A. Benzaouia, A. El Hajjaji, H∞ Filtering of Two-Dimensional T-S Fuzzy
Systems, Circuits Syst Signal Process, DOI 10.1007/s00034-013-9720-2, 2013.
[37] M. Ghamgui, D. Mehdi, O. Bachelier, M. Chaabane, Stability of 2D Discrete Takagi-Sugeno
Systems, Circuits Syst Signal Process DOI 10.1007/s00034-016-0425-1, 2016.
[38] X. Chen, J. Lam, H. Gao, S. Zhou, Stability analysis and control design for 2-D fuzzy systems
via basis-dependent Lyapunov functions, Multidimensional Systems and Signal Processing
24:395–415, 2013.
[39] M. Benhayoun, A. Benzaouia, F. Mesquine, F. Tadeo, Stabilization of 2D Continuous
Takagi-Sugeno Systems with non-PDC State Feedback Control, 12th Int. Conf. on Sciences
and Techniques of Automatic control and computer engineering, Sousse, Tunisia, (December
18–20, 2011).
[40] J. P. Emelianova, P. V. Pakshin, K. Galkowski, E. Rogers, Stability of Nonlinear 2D Systems
Described by the Continuous-Time Roesser Model, Automation and Remote Control,
75(5):845–858, 2014.

Biographies
Ismail Er Rachid Was born in Fes, Morocco, in 1989. He received his Master
degree in signals systems and informatics from University Sidi Mohammed
Ben Abdellah, faculty of sciences, Fez, Morocco in 2013, he is currently a
Ph. D. student in the same faculty. His current research interests include
stability theory and performance of systems in the frequency domain.

Redouane Chaibi received the Master in Signals Systems and Computing

from University of Sidi Mohammed Ben Abdellah, Faculty of Sciences, Fez,
Morocco in 2014. His research interests include stability and stabilization
of T-S fuzzy system.

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM
280 | I. Er Rachid et al.

Abdelaziz Hmamed was born in Sefrou, Morocco, in 1951. He received his

Doctorate of State degree in Electrical Engineering from the Faculty of Sci-
ences, Rabat, Morocco, in 1985. Since 1986, he has been with the Depart-
ment of Physics, Faculty of Sciences, Dhar El Mehraz at Fes, where he is
currently a Full Professor. His research interests are delay systems, stability
theory, systems with constraints, 2D systems and applications.

Brought to you by | Cambridge University Library

Authenticated
Download Date | 12/23/19 1:38 PM