Next Article in Journal
One-Layer Real-Time Optimization Using Reinforcement Learning: A Review with Guidelines
Previous Article in Journal
A Novel Two-Step Electrochemical Deposition Method for Sn-Pd Electrocatalyst Synthesis for a Potential Application in Direct Ethanol Fuel Cells
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

New Relaxed Static Output Feedback Stabilization of T–S Fuzzy Systems with Time-Varying Delays

1
College of Modern Science and Technology, China Jiliang University, Hangzhou 310018, China
2
College of Mechanical and Electrical Engineering, China Jiliang University, Hangzhou 310018, China
3
State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University, Hangzhou 310027, China
*
Author to whom correspondence should be addressed.
Processes 2023, 11(1), 121; https://doi.org/10.3390/pr11010121
Submission received: 26 November 2022 / Revised: 27 December 2022 / Accepted: 28 December 2022 / Published: 1 January 2023
(This article belongs to the Section Automation Control Systems)

Abstract

:
This paper investigates the imperfect premise matching-based (IPMB) static output feedback (SOF) controller for T–S fuzzy systems with time-varying delays. Firstly, by employing integral inequality techniques, the membership-function-independent fuzzy SOF design methods in terms of LMIs are presented based on imperfect premise matching (IPM) strategy. The obtained stabilization conditions do not contain equality constraints, and the output matrices do not have rank constraints. Secondly, some suitable relaxation variables are employed, and more relaxed membership-function-dependent (MFD) stabilization conditions are obtained by considering the local boundary information of membership functions (MFs). Finally, two simulation examples are given to show the progressiveness of the proposed methods in this paper.

1. Introduction

Benefiting from the excellent characteristics, T–S fuzzy models can introduce the perfect linear systems analysis and synthesis methods into the control problem of nonlinear systems, which provide a significant method for the control of nonlinear systems [1,2,3,4,5,6,7,8,9]. Generally, time delays are often encountered in many dynamic systems, and may even lead to the performance degradation, oscillation, and instability of the systems [10,11,12]. Therefore, the issue of T–S fuzzy systems with time delays is widely considered, and many stability and stabilization results have been reported in [13,14,15,16,17,18,19,20,21,22,23,24]. In the above results, the state feedback laws were employed to stabilize the T–S fuzzy systems.
However, in many actual systems, the measurement of state variables is very difficult, sometimes even unmeasurable. Therefore, output feedback control approaches are usually employed. The SOF controller has the advantages of simple implementation and convenient application. In addition, many dynamic output feedback controllers can be redesigned as SOF controllers. Therefore, the research on SOF controllers has important theoretical and practical significance. However, when compared with the state feedback control, the output feedback control is more difficult to solve [25]. In the design of SOF controllers, due to the existence of cross terms between system matrices and control gain matrices, it is difficult to directly obtain less conservative stabilization conditions of time-delay systems in terms of LMIs. This is mainly because the necessary and sufficient conditions for SOF stabilization based on the Lyapunov method are essentially bilinear matrix inequality (BMI) problems. The SOF controllers were proposed for the T-S fuzzy system in [26,27,28]; however, the stabilization results were proposed in terms of BMIs. Equality constraints were introduced into the design conditions of SOF control in [29], but it inevitably brought great conservatism. A SOF controller was designed by using the cone complementarity linearization (CCL) algorithm in [30]. For a class of discrete generalized T–S fuzzy systems with time delays, a novel improved CCL algorithm was proposed in [31] to solve the fuzzy SOF controllers. In fact, the CCL method is based on iterative LMI. By equivalent transformation, the BMIs in the conditions of SOF controllers are transformed into LMIs containing a positive definite matrix and its inverse. It is worth noting that the iteration stop conditions of the CCL algorithm are relatively strict, which cannot guarantee the SOF controller satisfying the conditions, and the convergence speed is slow in the later stage. Based on some appropriate conservative treatment techniques, the design methods of SOF controllers for T–S fuzzy time-delay systems were presented based on LMIs technology in [32,33,34,35,36]. Further research is needed to obtain more relaxed stabilization results for T–S fuzzy time-delay systems through effective processing, especially the design methods of SOF controllers based on LMIs technology.
On the other hand, based on parallel distributed compensation (PDC) technology, the fuzzy SOF controllers were proposed in [26,27,28,29,30,31,32,33,34,35,36]. However, the PDC method may complicate the structure of fuzzy controllers [37]. To solve the above issue, IPM strategy was presented in [38], in which the MFs of fuzzy controllers can be selected arbitrarily. According to our best knowledge, the SOF control of T–S fuzzy time-delay systems based on IPM technology is almost in the blank field; only the fuzzy SOF controllers were designed based on IPM technology in [39,40,41]. In effect, the IPMB fuzzy SOF controllers proposed in [39,40] only considered the case of mismatched premise MFs, but did not consider the case of a different number of fuzzy rules. This does not fully reflect the design advantages of IPM strategy. In addition, the case of a constant time delay, rather than a time-varying delay, was considered in [40,41].
In this paper, we investigated the design of an IPMB fuzzy SOF controller for T–S fuzzy time-delay systems. The main contributions are summarized as follows:
(1) LMIs-based design methods of fuzzy SOF were presented based on the integral inequality technique. The obtained stabilization conditions did not contain equality constraints, and the output matrices did not have rank constraints.
(2) The local boundary information of MFs were considered, and the further relaxed MFD stabilization results were obtained.
(3) The premise MFs and number of fuzzy rules of the SOF controllers were allowed to be selected arbitrarily, which enhanced the flexibility of the controllers design.

2. Preliminaries

2.1. Fuzzy Time-Delay Models

Consider the following nonlinear system that can be represented by a T–S fuzzy model with time-varying delays, and the system has p plant rules:
Rule i : IF θ 1 ( x ( t ) ) is M 1 i and θ 2 ( x ( t ) ) is M 2 i and and θ χ ( x ( t ) ) is M χ i , THEN
{ x ˙ ( t ) = A i x ( t ) + A d i x ( t τ ( t ) ) + B i u ( t ) y ( t ) = C i x ( t ) x ( t ) = ϕ ( t ) , t [ τ M , 0 ]
where M α i denote the fuzzy sets, α = 1 , 2 , , χ , i = 1 , 2 , , p ; x ( t ) n denotes the state vector; u ( t ) m is the control input vector; y ( t ) q is the output vector; A i , A d i , B i , C i are constant real matrices; θ 1 ( x ( t ) ) , , θ χ ( x ( t ) ) are the premise variables. The value ϕ ( t ) is the vector valued initial condition on [ τ M , 0 ] . The value τ ( t ) is time-varying delay and satisfies:
{ 0 τ ( t ) τ M τ ˙ ( t ) μ
where τ M and μ are constants.
By fuzzy blending, the overall T–S fuzzy time-delay system can be expressed as in the following equations:
{ x ˙ ( t ) = i = 1 p m i ( x ( t ) ) ( A i x ( t ) + A d i x ( t τ ( t ) ) + B i u ( t ) ) y ( t ) = i = 1 p m i ( x ( t ) ) C i x ( t )
where m i ( x ( t ) ) = ω i ( x ( t ) ) i = 1 p ω i ( x ( t ) ) , and fuzzy weighting functions satisfy i = 1 p ω i ( x ( t ) ) > 0 , i = 1 p m i ( x ( t ) ) = 1 .

2.2. Fuzzy SOF Controller under IPM

Based on the IPM technique, a fuzzy SOF controller with c rules can be organized as:
Rule j : IF ξ 1 ( x ( t ) ) is N 1 j and ξ 2 ( x ( t ) ) is N 2 j and and ξ γ ( x ( t ) ) is N γ j , THEN
u ( t ) = K j y ( t )
Therefore, the overall output of the fuzzy SOF controller can be represented as:
u ( t ) = j = 1 c h j ( x ( t ) ) K j y ( t )
where h j ( x ( t ) ) = ϖ j ( x ( t ) ) j = 1 c ϖ j ( x ( t ) ) , and fuzzy weighting functions satisfy j = 1 c ϖ j ( x ( t ) ) > 0 , j = 1 c h j ( x ( t ) ) = 1 . K j m × q , j = 1 , 2 , , c are control gain matrices.
By substituting (5) into (3), the T–S fuzzy closed-loop system can be obtained:
x ˙ ( t ) = i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) ( ( A i + B i K j C k ) x ( t ) + A d i x ( t τ ( t ) ) )
wherein, the closed-loop system can be represented as:
x ˙ ( t ) = ( A ( t ) + B ( t ) K ( t ) C ( t ) ) x ( t ) + A d ( t ) x ( t τ ( t ) )
where
A ( t ) = i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) A i
B ( t ) = i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) B i
C ( t ) = i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) C k
K ( t ) = i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) K j
A d ( t ) = i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) A d i
In this paper, the main focus is to obtain the fuzzy SOF gain K j based on the IPM technique so that the closed-loop system (6) is asymptotically stable.

2.3. Useful Lemmas

In order to obtain the main SOF stabilization results of this paper, we firstly give the following useful lemmas.
Lemma 1. 
Ref. [42]. For a positive definite matrix R > 0 , and a differentiable function x : [ a , b ] n , the following inequality holds:
a b x ˙ T ( s ) R x ˙ ( s ) d s 1 b a Ω T R Ω  
where
Ω = x ( b ) x ( a )
Lemma 2. 
Ref. [43]. Let f 1 , f 2 , f N : m have positive values in an open subset D of m . Then, the reciprocally convex combination of f i over D satisfies:
{ g i , j : m , g j , i ( t ) = g i , j ( t ) , [ f i ( t ) g i , j ( t ) g i , j ( t ) f j ( t ) ] 0 }  
subject to
min { α i | α i > 0 , i α i = 1 } i 1 α i f i ( t ) = i f i ( t ) + max g i , j ( t ) i j g i , j ( t )  
Lemma 3. 
Ref. [44]. For any vectors X , Y m , and a positive definite matrix Q n × n , the following inequality holds:
2 X T Y X T Q X + Y T Q 1 Y  
Lemma 4. 
Ref. [45]. The following two inequalities are equivalent:
(1)
There exists a symmetric and positive-definite matrix P satisfying
[ P A T A P 1 ] < 0
(2)
There exists a symmetric and positive-definite matrix P and matrix Y satisfying
[ P ( Y A ) T Y A S y m ( Y ) + P ] < 0

3. Main Results

3.1. Membership-Function-Independent Stabilization Conditions

In this subsection, based on IPM technology, the design methods of fuzzy the SOF controller in terms of LMIs are presented by combining with a reciprocally convex combination inequality.
Theorem 1. 
For given scalars τ M > 0 and μ , the system (6) is asymptotically stable if there exist matrices 0 < P n × n , 0 < Q n × n , 0 < S n × n , 0 < R n × n , 0 < J 2 n × 2 n , and any matrices M n × n , X m × m , G j m × n , such that the following LMIs hold:
[ R M M T R ] > 0  
Σ i j k ( μ ) < 0 , i = 1 , 2 , , p ; j = 1 , 2 , , c ; k = 1 , 2 , , p
The corresponding fuzzy SOF control gains are proposed by K j = X 1 G j , j = 1 , 2 , , c . where
Σ i j k ( τ ˙ ( t ) ) = [ Γ 11 ( τ ˙ ( t ) ) Γ 12 i 0 0 θ 22 i + J θ 23 i j k 0 S y m { B i T B i X } θ 34 i J ]
Γ 11 ( τ ˙ ( t ) ) = [ S R R M T ( 1 τ ˙ ( t ) ) ( S Q ) + S y m { M } 2 R ]
Γ 12 i = [ M T 0 S y m { P A i } + R M T τ M ( R A d i ) T ]
θ 22 i = [ S y m { P A i } + Q R τ M ( R A d i ) T R ]
θ 23 i j k = [ ( B i T B i G j C k ) T 0 ]
θ 34 i = [ B i T P B i T R ]
Proof. 
Consider the following Lyapunov–Krasovskii functional candidate:
V ( t ) = x T ( t ) P x ( t ) + t τ ( t ) t x T ( s ) Q x ( s ) d s + t τ M t τ ( t ) x T ( s ) S x ( s ) d s + τ M t τ M t θ t x ˙ T ( s ) R x ˙ ( s ) d s d θ
where 0 < P n × n , 0 < Q n × n , 0 < S n × n , 0 < R n × n are the matrices to be solved.
Differentiating the derivative of V ( t ) along the trajectory of the system (6) leads to
V ˙ ( t ) = 2 x T ( t ) P x ˙ ( t ) + x T ( t ) Q x ( t ) ( 1 τ ˙ ( t ) ) x T ( t τ ( t ) ) Q x ( t τ ( t ) ) + ( 1 τ ˙ ( t ) ) x T ( t τ ( t ) ) S x ( t τ ( t ) ) x T ( t τ M ) S x ( t τ M ) + τ M 2 x ˙ T ( t ) R x ˙ ( t ) τ M t τ M t x ˙ T ( s ) R x ˙ ( s ) d s = i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) ( 2 x T ( t ) P ( ( A i + B i K j C k ) x ( t ) + A d i x ( t τ ( t ) ) ) + x T ( t ) Q x ( t ) ( 1 τ ˙ ( t ) ) x T ( t τ ( t ) ) Q x ( t τ ( t ) ) + ( 1 τ ˙ ( t ) ) x T ( t τ ( t ) ) S x ( t τ ( t ) ) x T ( t τ M ) S x ( t τ M ) + τ M 2 ( ( A i + B i K j C k ) x ( t ) + A d i x ( t τ ( t ) ) ) T R ( ( A i + B i K j C k ) x ( t ) + A d i x ( t τ ( t ) ) ) τ M t τ ( t ) t x ˙ T ( s ) R x ˙ ( s ) d s τ M t τ M t τ ( t ) x ˙ T ( s ) R x ˙ ( s ) d s
The integral terms in (10) are estimated by employing the Jensen integral inequality in Lemma 1, and the following inequalities are obtained:
τ M t τ ( t ) t x ˙ T ( s ) R x ˙ ( s ) d s τ M τ ( t ) ( x ( t ) x ( t τ ( t ) ) ) T R ( x ( t ) x ( t τ ( t ) ) ) = τ M τ ( t ) i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) ξ T ( t ) ( e 1 e 2 ) T R ( e 1 e 2 ) ξ ( t )
τ M t τ M t τ ( t ) x ˙ T ( s ) R x ˙ ( s ) d s τ M τ M τ ( t ) ( x ( t τ ( t ) ) x ( t τ M ) ) T R ( x ( t τ ( t ) ) x ( t τ M ) ) = τ M τ M τ ( t ) i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) ξ T ( t ) ( e 2 e 3 ) T R ( e 2 e 3 ) ξ ( t )
where
ξ ( t ) = [ x T ( t ) x T ( t τ ( t ) ) x T ( t τ M ) ] T
e σ = [ 0 n × ( σ 1 ) n I n 0 n × ( 3 σ ) n ] , σ = 1 , 2 , 3
According to Lemma 2, if there exists a matrix M n × n satisfying matrix inequality (7), then we can obtain
τ M h τ M t x ˙ T ( s ) R x ˙ ( s ) d s = τ M t τ ( t ) t x ˙ T ( s ) R x ˙ ( s ) d s τ M t τ M t τ ( t ) x ˙ T ( s ) R x ˙ ( s ) d s i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) ξ T ( t ) ( e 1 e 2 e 2 e 3 ) T [ R M M T R ] ( e 1 e 2 e 2 e 3 ) ξ ( t )
Combining Formulas (10) and (13), and we can obtain
V ˙ ( t ) i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) ξ T ( t ) [ Ξ i j k ( τ ˙ ( t ) ) + γ i j k T R 1 γ i j k ] ξ ( t )
where
Ξ i j k ( τ ˙ ( t ) ) = [ S y m { P ( A i + B i K j C k ) } + Q R S y m { P A i } + R M M ( 1 τ ˙ ( t ) ) ( S Q ) + S y m { M } 2 R R M S R ]
γ i j k = [ τ M R ( A i + B i K j C k ) τ M R A d i 0 ]
At this time, if inequality Ξ i j k ( τ ˙ ( t ) ) + γ i j k T R 1 γ i j k < 0 holds, the closed-loop system (6) is asymptotically stable. According to the Schur complement lemma, inequality Ξ i j k ( τ ˙ ( t ) ) + γ i j k T R 1 γ i j k < 0 is equivalent to the following inequality:
[ S y m { P ( A i + B i K j C k ) } + Q R S y m { P A i } + R M M ( 1 τ ˙ ( t ) ) ( S Q ) + S y m { M } 2 R R M S R τ M ( R ( A i + B i K j C k ) ) T τ M ( R A d i ) T 0 R ] < 0
Let Δ = [ e ˜ 3 e ˜ 2 e ˜ 1 e ˜ 4 ] T , where e ˜ σ = [ 0 n × ( σ 1 ) n I n 0 n × ( 4 σ ) n ] , σ = 1 , 2 , 3 , 4 . By pre-multiplying the side of (15) with Δ and post-multiplying the side of (15) with Δ T , we can obtain
[ S R R M T M T ( 1 τ ˙ ( t ) ) ( S Q ) + S y m { M } 2 R S y m { P A i } + R M T S y m { P ( A i + B i K j C k ) } + Q R 0 τ M ( R A d i ) T τ M ( R ( A i + B i K j C k ) ) T R ] < 0
For the convenience of proof, inequality (16) is rewritten as follows:
[ Γ 11 ( τ ˙ ( t ) ) Γ 12 i Γ 22 i j k ] < 0
where
Γ 11 ( τ ˙ ( t ) ) = [ S R R M T ( 1 τ ˙ ( t ) ) ( S Q ) + S y m { M } 2 R ]
Γ 12 i = [ M T 0 S y m { P A i } + R M T τ M ( R A d i ) T ]
Γ 22 i j k = [ S y m { P ( A i + B i K j C k ) } + Q R τ M ( R ( A i + B i K j C k ) ) T R ]
Rewrite Γ 22 i j k as follows:
Γ 22 i j k = [ S y m { P A i } + Q R τ M A i T R R ] + [ P B i K j C k 0 R B i K j C k 0 ] + [ ( P B i K j C k ) T ( P B i K j C k ) T 0 0 ]
Let K j = X 1 G j , where X m × m is an arbitrary invertible matrix. G j m × q is an arbitrary matrix, and the following equation can be obtained:
[ P B i K j C k 0 R B i K j C k 0 ] = [ P B i 0 R B i 0 ] [ X 1 G j C k 0 0 0 ]
Defining W i = [ P B i R B i ] , T j k = [ X 1 G j C k 0 ] , and we rewrite matrix Γ 22 i j k as:
Γ 22 i j k = θ 22 i + S y m { W i T j k }
From (17) and (19), one has
[ Γ 11 ( τ ˙ ( t ) ) Γ 12 i Γ 22 i j k ] = [ Γ 11 ( τ ˙ ( t ) ) Γ 12 i θ 22 i ] + S y m { [ 0 2 n × 2 n I 2 n ] W i T j k [ 0 2 n × 2 n I 2 n ] } < 0
According to Lemma 3, there exists a positive definite matrix J 2 n × 2 n , such that the following inequality holds:
S y m { [ 0 2 n × 2 n I 2 n ] W i T j k [ 0 2 n × 2 n I 2 n ] } [ 0 2 n × 2 n I 2 n ] [ X 1 G j C k 0 ] T [ P B i R B i ] T J 1 [ P B i R B i ] [ X 1 G j C k 0 ] [ 0 2 n × 2 n I 2 n ] + [ 0 2 n × 2 n I 2 n ] J [ 0 2 n × 2 n I 2 n ]
Combining (20) and (21), we can obtain:
[ Γ 11 ( τ ˙ ( t ) ) Γ 12 i Γ 22 i j k ] [ Γ 11 ( τ ˙ ( t ) ) Γ 12 i θ 22 i ] + [ 0 2 n × 2 n I 2 n ] [ X 1 G j C k 0 ] T [ P B i R B i ] T × J 1 [ P B i R B i ] [ X 1 G j C k 0 ] [ 0 2 n × 2 n I 2 n ] + [ 0 2 n × 2 n I 2 n ] J [ 0 2 n × 2 n I 2 n ] < 0
According to the Schur complement lemma, inequality (22) is equivalent to the following inequality:
[ Γ 11 ( τ ˙ ( t ) ) Γ 12 i 0 θ 22 i + J Δ 23 j k θ 33 i ] < 0
where
θ 22 i = [ S y m { P A i } + Q R τ M ( R A d i ) T R ]
Δ 23 j k = [ ( X 1 G j C k ) T 0 ]
θ 33 i = ( [ P B i R B i ] T J 1 [ P B i R B i ] ) 1
According to Lemma 4, inequality (23) is equivalent to the following inequality:
[ Γ 11 ( τ ˙ ( t ) ) Γ 12 i 0 θ 22 i + J Δ 23 j k X T B i T B i S y m { B i T B i X } θ 33 i 1 ] < 0
Employing Schur complement lemma for inequality (24) again, and inequality (24) is equivalent to the following inequality:
Σ i j k ( τ ˙ ( t ) ) = [ Γ 11 ( τ ˙ ( t ) ) Γ 12 i 0 0 θ 22 i + J Δ 23 j k X T B i T B i 0 S y m { B i T B i X } θ 34 i J ] < 0
where
θ 34 i = [ B i T P B i T R ]
Moreover, we can obtain:
V ˙ ( t ) i = 1 p j = 1 c k = 1 p m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) ξ T ( t ) Σ i j k ( τ ˙ ( t ) ) ξ ( t ) < 0
Therefore, from (7) and (8), we can obtain V ˙ ( t ) < 0 . This proof is completed. □

3.2. MFD Stabilization Conditions

To obtain the less conservative stabilization conditions, the piecewise membership function (PMF) concept is introduced firstly based on the results in [46].
Suppose Γ l , l = 1 , 2 , , ϑ represent the connected state subspaces of the operating domain Γ , such that Γ = l = 1 ϑ Γ l . Defining a vector Χ = [ x 1 x η ] and considering Χ Γ l , we have the infimum and supremum of x r as x _ r i r l and x ¯ r i r l , r = 1 , 2 , , η , i r = 1 , 2 , respectively, such that x _ r i r l x r ( t ) x ¯ r i r l . The vertices of the subspaces Γ l are denoted as x i 1 i η l = [ x 1 i 1 l x η i η l ] , and the PMF is defined as:
ω ^ ( x ( t ) ) = l = 1 ϑ i 1 = 1 2 i η = 1 2 r = 1 η ν r i r l ( x r ( t ) ) ω ^ ( x i 1 i η l )
and the function ν r i r l ( x r ( t ) ) has the following characteristics:
(1) 0 ν r i r l ( x r ( t ) ) 1 , i r = 1 2 ν r i r l ( x r ( t ) ) = 1 for all r , l , and x Γ l , otherwise, ν r i r l ( x r ( t ) ) = 0 ;
(2) l = 1 ϑ i 1 = 1 2 i η = 1 2 r = 1 η ν r i r l ( x r ( t ) ) = 1 .
Theorem 2. 
For given scalars τ M > 0 and μ , the system (6) is asymptotically stable if there exist matrices 0 < P n × n , 0 < Q n × n , 0 < S n × n , 0 < R n × n , 0 < F i j k ( 6 n + m ) × ( 6 n + m ) , E = E T ( 6 n + m ) × ( 6 n + m ) , 0 < J 2 n × 2 n and any matrices M n × n , X m × m , G j m × n , such that the following LMIs hold:
[ R M M T R ] > 0  
Σ i j k ( μ ) + E F i j k < 0 ; i = 1 , 2 , , p ; j = 1 , 2 , , c ; k = 1 , 2 , , p
i = 1 p j = 1 c k = 1 p ( ω ^ i j k ( x i 1 i 2 i n l ) Σ i j k ( μ ) + Δ ω _ i j k ( Σ i j k ( μ ) + E ) + ( Δ ω ¯ i j k Δ ω _ i j k ) F i j k ) < 0 i 1 , i 2 , , i n { 1 , 2 } ; l = 1 , 2 , , ϑ
The fuzzy SOF control gains are proposed by K j = X 1 G j , j = 1 , 2 , , c .
Wherein the PMF ω ^ i j k ( x ( t ) ) is defined in (31), x i 1 i 2 i n l , l = 1 , 2 , , ϑ denote the apexes of Γ l , ϑ and represents the number of partitioned state subspaces of PMF. Δ ω ¯ i j k , Δ ω _ i j k are the supremum and infimum of Δ ω i j k ( x ( t ) ) , respectively. Moreover, some other matrices are defined as Theorem 1.
Proof. 
Construct the following PMF according to the definition (28):
ω ^ i j k ( x ( t ) ) = l = 1 ϑ i 1 = 1 2 i η = 1 2 r = 1 η ν r i r l ( x r ( t ) ) ω ^ i j k ( x i 1 i η l )
In the following proof, PMF ω ^ i j k ( x ( t ) ) is introduced to approximate the MFs ω i j k ( x ( t ) ) = m i ( x ( t ) ) h j ( x ( t ) ) m k ( x ( t ) ) , and we can obtain some further relaxed MFD stabilization conditions.
Let Δ ω i j k ( x ( t ) ) = ω i j k ( x ( t ) ) ω ^ i j k ( x ( t ) ) , wherein, Δ ω ¯ i j k and Δ ω _ i j k represent the supremum and infimum of Δ ω i j k ( x ( t ) ) , respectively. Meanwhile, the equation i = 1 p j = 1 c k = 1 p Δ ω i j k ( x ( t ) ) = 0 is obtained by employing the characteristics of the PMF. Furthermore, a slack matrix E = E T ( 6 n + m ) × ( 6 n + m ) is introduced to reduce conservatism, and we can obtain:
i = 1 p j = 1 c k = 1 p Δ ω i j k ( x ( t ) ) ξ T ( t ) E ξ ( t ) = 0
Combining (26) and (32), we can obtain:
V ˙ ( t ) i = 1 p j = 1 c k = 1 p ω i j k ( x ( t ) ) ξ T ( t ) Σ i j k ( τ ˙ ( t ) ) ξ ( t ) = i = 1 p j = 1 c k = 1 p ω ^ i j k ( x ( t ) ) ξ T ( t ) Σ i j k ( τ ˙ ( t ) ) ξ ( t ) + i = 1 p j = 1 c k = 1 p Δ ω i j k ( x ( t ) ) ξ T ( t ) Σ i j k ( τ ˙ ( t ) ) ξ ( t ) + i = 1 p j = 1 c k = 1 p Δ ω i j k ( x ( t ) ) ξ T ( t ) E ξ ( t ) = i = 1 p j = 1 c k = 1 p ω ^ i j k ( x ( t ) ) ξ T ( t ) Σ i j k ( τ ˙ ( t ) ) ξ ( t ) + i = 1 p j = 1 c k = 1 p ( Δ ω i j k ( x ( t ) ) Δ ω _ i j k + Δ ω _ i j k ) ξ T ( t ) ( Σ i j k ( τ ˙ ( t ) ) + E ) ξ ( t ) = i = 1 p j = 1 c k = 1 p ω ^ i j k ( x ( t ) ) ξ T ( t ) Σ i j k ( τ ˙ ( t ) ) ξ ( t ) + i = 1 p j = 1 c k = 1 p ( Δ ω i j k ( x ( t ) ) Δ ω _ i j k ) ξ T ( t ) ( Σ i j k ( τ ˙ ( t ) ) + E ) ξ ( t ) + i = 1 p j = 1 c k = 1 p Δ ω _ i j k ξ T ( t ) ( Σ i j k ( τ ˙ ( t ) ) + E ) ξ ( t )
Introducing another slack matrix 0 < F i j k ( 6 n + m ) × ( 6 n + m ) , i = 1 , 2 , , p , j = 1 , 2 , , c , k = 1 , 2 , , p , and we can obtain the following inequality based on (33):
V ˙ ( t ) i = 1 p j = 1 c k = 1 p ω ^ i j k ( x ( t ) ) ξ T ( t ) Σ i j k ( τ ˙ ( t ) ) ξ ( t ) + i = 1 p j = 1 c k = 1 p Δ ω _ i j k ξ T ( t ) ( Σ i j k ( τ ˙ ( t ) ) + E ) ξ ( t ) + i = 1 p j = 1 c k = 1 p ( Δ ω i j k ( x ( t ) ) Δ ω _ i j k ) ξ T ( t ) ( Σ i j k ( τ ˙ ( t ) ) + E F i j k ) ξ ( t ) + i = 1 p j = 1 c k = 1 p ( Δ ω i j k ( x ) Δ ω _ i j k ) ξ T ( t ) F i j k ξ ( t )
Expanding the PMF ω ^ i j k ( x ( t ) ) in (34) according to the definition (31), and we have:
V ˙ ( t ) l = 1 ϑ i 1 = 1 2 i η = 1 2 r = 1 η ν r i r l ( x r ( t ) ) ξ T ( t ) [ i = 1 p j = 1 c k = 1 p ( ω ^ i j k ( x i 1 i 2 i n l ) Σ i j k ( τ ˙ ( t ) ) + Δ ω _ i j k ( Σ i j k ( τ ˙ ( t ) ) + E ) + ( Δ ω ¯ i j k Δ ω _ i j k ) F i j k ] ξ ( t ) + i = 1 p j = 1 c k = 1 p ( Δ ω i j k ( x ( t ) ) Δ ω _ i j k ) ξ T ( t ) ( Σ i j k ( τ ˙ ( t ) ) + E F i j k ) ξ ( t )
If LMIs (28)–(30) hold, we have V ˙ ( t ) < 0 ; then, the closed-loop system (6) is asymptotically stable. This proof is completed. □
Remark 1. 
Different from the results in [29], the MFD stabilization conditions based on LMIs are presented in Theorem 2, in which the obtained stabilization conditions do not contain equality constraints, and the input and output matrices do not have rank constraints. Therefore, the proposed design methods of fuzzy SOF controllers reduce the conservatism of stabilization results and have better application prospects. In addition, the stabilization conditions of the fuzzy SOF controller given in [30] are nonconvex and cannot be solved directly by MATLAB LMIs toolbox.

4. Simulation Examples

In this subsection, two simulation examples are given to illustrate the progressiveness and effectiveness of the proposed IPMB fuzzy SOF controllers.
Example 1. 
Consider the following complex nonlinear system given in [47]:
{ x ˙ 1 ( t ) = x 1 ( t ) + x 2 ( t ) + sin x 3 ( t ) 0.1 x 4 ( t ) + ( x 1 2 ( t ) + 1 ) u ( t ) x ˙ 2 ( t ) = x 1 ( t ) 2 x 2 ( t ) x ˙ 3 ( t ) = x 1 ( t ) + x 1 2 ( t ) x 2 ( t ) 0.3 x 3 ( t ) x ˙ 4 ( t ) = sin x 3 ( t ) x 4 ( t ) y 1 ( t ) = x 2 ( t ) + ( x 1 2 ( t ) + 1 ) x 4 ( t ) y 2 ( t ) = x 1 ( t )  
wherein a , b are positive real numbers, and x 1 ( t ) [ a , a ] , x 3 ( t ) [ b , b ] .
The above nonlinear time-delay system (36) can be approximated by the following T–S fuzzy models with four plant rules:
  • Rule 1: IF x 1 ( t ) is M 1 1 and x 3 ( t ) is M 3 1 , THEN { x ˙ ( t ) = A 1 x ( t ) + B 1 u ( t ) y ( t ) = C 1 x ( t )
  • Rule 2: IF x 1 ( t ) is M 1 1 and x 3 ( t ) is M 3 2 , THEN { x ˙ ( t ) = A 2 x ( t ) + B 2 u ( t ) y ( t ) = C 2 x ( t )
  • Rule 3: IF x 1 ( t ) is M 1 2 and x 3 ( t ) is M 3 1 , THEN { x ˙ ( t ) = A 3 x ( t ) + B 3 u ( t ) y ( t ) = C 3 x ( t )
  • Rule 4: IF x 1 ( t ) is M 1 2 and x 3 ( t ) is M 3 2 , THEN { x ˙ ( t ) = A 4 x ( t ) + B 4 u ( t ) y ( t ) = C 4 x ( t )
where
A 1 = [ 1 1 1 0.1 1 2 0 0 1 a 2 0.3 0 0 0 1 1 ] ,   A 2 = [ 1 1 sin b b 0.1 1 2 0 0 1 a 2 0.3 0 0 0 sin b b 1 ] ,   A 3 = [ 1 1 1 0.1 1 2 0 0 1 0 0.3 0 0 0 1 1 ]
A 4 = [ 1 1 sin b b 0.1 1 2 0 0 1 0 0.3 0 0 0 sin b b 1 ] ,   B 1 = B 2 = [ 1 + a 2 0 0 0 ] ,   B 3 = B 4 = [ 1 0 0 0 ]
C 1 = C 2 = [ 0 1 0 1 + a 2 1 0 0 0 ] ,   C 3 = C 4 = [ 0 1 0 1 1 0 0 0 ]
In order to prove the effectiveness of the methods proposed in this paper, the other system matrices are selected in [48]:
A d 1 = A d 2 = A d 3 = A d 4 = [ 0.1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ]
The premise MFs of fuzzy models are as the same as those defined in [48]:
m 1 ( x ( t ) ) = M 1 1 M 3 1 ,   m 2 ( x ( t ) ) = M 1 1 M 3 2 ,   m 3 ( x ( t ) ) = M 1 2 M 3 1 ,   m 4 ( x ( t ) ) = M 1 2 M 3 2
where
M 1 1 = x 1 2 a 2 ,   M 1 2 = 1 M 1 1 ,   M 3 1 = { b sin x 3 x 3 sin b x 3 ( b sin b ) , x 3 0 1 , x 3 0 ,   M 3 2 = 1 M 3 1
Based on IPM technology, fuzzy SOF controllers with two plant rules are designed as follows:
  • Rule 1: IF ξ 1 ( x ( t ) ) is N 1 1 , THEN   u ( t ) = K 1 y ( t )
  • Rule 2: IF ξ 2 ( x ( t ) ) is N 1 2 , THEN   u ( t ) = K 2 y ( t )
Based on IPM technology, the following simpler MFs are selected for the fuzzy SOF controllers:
h 1 ( x 1 ) = 1 1 + exp ( 2 x 1 ) ,   h 2 ( x 1 ) = 1 h 1 ( x 1 )
Since the product of MFs ω i j k ( x ) , i = 1 , 2 , 3 , 4 , j = 1 , 2 , k = 1 , 2 , 3 , 4 depends on x 1 and x 3 , and we construct the following PMF ω ^ i j k ( x ) = l = 1 ϑ i 1 = 1 2 i 3 = 1 2 ν 1 i 1 l ( x 1 ) ν 3 i 1 l ( x 3 ) ω ^ i j k ( x i 1 i 3 l ) . Firstly, we divide the state spaces x 1 ( t ) [ g , h ] , x 3 ( t ) [ g , h ] into ϑ connected state subspaces, and denote the l-th subspace as ( g + h ϑ ) ( l ϑ 2 1 ) x s ( g + h ϑ ) ( l ϑ 2 ) , s = 1 , 3 , l = 1 , 2 , , ϑ , g , h R + . In addition, ν 11 l ( x 1 ) = 1 x 1 x 2 l x 1 l x 2 l , ν 12 l ( x 1 ) = 1 ν 11 l ( x 1 ) , ν 31 l ( x 3 ) = 1 x 3 x 2 l x 1 l x 2 l and ν 32 l ( x 3 ) = 1 ν 31 l ( x 3 ) are selected, where x 1 l = ( g + h ϑ ) ( l ϑ 2 1 ) , x 2 l = ( g + h ϑ ) ( l ϑ 2 ) . Especially, g = h = 10 , ϑ = 20 are selected, respectively. In Table 1, we can obtain the infimum and supremum of Δ ω i j k ( x ) based on the formula Δ ω i j k ( x ) = ω i j k ( x ) ω ^ i j k ( x ) .
For the purpose of comparison, we cannot obtain a feasibility region for this example by employing Theorem 1, which demonstrates that the MFD analysis approach can effectively reduce conservatism. Therefore, further relaxed MFD stabilization results can be obtained by considering the information of MFs. Compared with the membership-function-independent method, the MFD approach can obtain less conservative results.
As in [47,48], we selected a = 1.4 , b = 0.7 . Let τ M = 2 , μ = 0.2 , and by employing MATLAB LMIs toolbox, the IPMB fuzzy SOF controllers are given according to Theorem 2 as follows:
K 1 = [ 1.5285 2 . 5338 ] ,   K 2 = [ 1.9137 3 . 3069 ]
We selected time-varying delay τ ( t ) = 1 + sin ( t 5 ) and initial condition ϕ ( 0 ) = [ 1.2 0.5 0.7 0.6 ] T . Figure 1 depicts the state response of the closed-loop system. Obviously, the fuzzy SOF controllers designed in Theorem 2 can guarantee that the system (36) is asymptotically stable.
Example 2. 
Consider the following truck-trailer system, an actual control system with time-varying delays given in [49,50]:
{ x ˙ 1 ( t ) = a v t ¯ L t 0 x 1 ( t ) ( 1 a ) v t ¯ L t 0 x 1 ( t τ ( t ) ) + v t ¯ l t 0 u ( t ) x ˙ 2 ( t ) = a v t ¯ L t 0 x 1 ( t ) + ( 1 a ) v t ¯ L t 0 x 1 ( t τ ( t ) ) x ˙ 3 ( t ) = v t ¯ t 0 sin ( x 2 ( t ) + a v t ¯ 2 L x 1 ( t ) + ( 1 a ) v t ¯ 2 L x 1 ( t τ ( t ) ) )
where x 1 ( t ) is the angular difference between the truck and trailer, x 2 ( t ) is the angle of the trailer relative to the horizontal position, and x 3 ( t ) is the vertical position of the rear end of the trailer; l = 2.8 is the length of truck, L = 5.5 is the length of trailer, and v = 1.0 is constant speed of the tractor in the reverse direction. The model parameters are given as a = 0.7 , t ¯ = 2.0 , t 0 = 0.5 .
Let θ ( x ( t ) ) = x 2 ( t ) + a v t ¯ 2 L x 1 ( t ) + ( 1 a ) v t ¯ 2 L x 1 ( t τ ( t ) ) , and x ( t ) = [ x 1 ( t ) x 2 ( t ) x 3 ( t ) ] T . For the above nonlinear system with time-varying delays, we can employ the following fuzzy model with two fuzzy rules to approximate, and the specific expressions are as follows:
  • Rule 1: IF θ ( x ( t ) ) is 0 rad, THEN
    { x ˙ ( t ) = A 1 x ( t ) + A d 1 x ( t τ ( t ) ) + B 1 u ( t ) y ( t ) = C 1 x ( t )
  • Rule 2: IF θ ( x ( t ) ) is π rad or π rad, THEN
    { x ˙ ( t ) = A 2 x ( t ) + A d 2 x ( t τ ( t ) ) + B 2 u ( t ) y ( t ) = C 2 x ( t )
where
A 1 = [ a v t ¯ L t 0 0 0 a v t ¯ L t 0 0 0 a v 2 t ¯ 2 2 L t 0 v t ¯ t 0 0 ] ,   A d 1 = [ ( 1 a ) v t ¯ L t 0 0 0 ( 1 a ) v t ¯ L t 0 0 0 ( 1 a ) v 2 t ¯ 2 2 L t 0 0 0 ] ,   B 1 = B 2 = [ v t ¯ l t 0 0 0 ] T
A 2 = [ a v t ¯ L t 0 0 0 a v t ¯ L t 0 0 0 a d v 2 t ¯ 2 2 L t 0 d v t ¯ t 0 0 ] ,   A d 2 = [ ( 1 a ) v t ¯ L t 0 0 0 ( 1 a ) v t ¯ L t 0 0 0 ( 1 a ) d v 2 t ¯ 2 2 L t 0 0 0 ] ,   C 1 = C 2 = [ 7 2 0.03 ]
As in [50], take d = 10 t 0 π , and the following MFs are employed:
m 1 ( x ) = 1 ( 1 1 1 + exp ( 3 ( x 1 1 2 π ) ) ) ( 1 1 + exp ( 3 ( x 1 + 1 2 π ) ) )
Based on IPM technology, the fuzzy SOF controllers with two plant rules are designed as follows:
  • Rule 1: IF ξ 1 ( x ( t ) ) is N 1 1 , THEN   u ( t ) = K 1 y ( t )
  • Rule 2: IF ξ 2 ( x ( t ) ) is N 1 2 , THEN   u ( t ) = K 2 y ( t )
Based on IPM technology, the following simpler MFs are selected for the fuzzy SOF controllers:
h 1 ( x 1 ) = 1 1 + exp ( 2 x 1 ) ,   h 2 ( x 1 ) = 1 h 1 ( x 1 )
Similar to example 1, the infimum and supremum of Δ ω i j k ( x ) , i = 1 , 2 , j = 1 , 2 , k = 1 , 2 with ϑ = 20 are obtained as shown in Table 2.
As in [49], we selected τ M = 2.2 , μ = 0.2 , employing MATLAB LMIs toolbox, the IPMB fuzzy SOF controllers are given according to Theorem 2 as follows:
K 1 = 0.7156 ,   K 2 = 1.2584
In addition, as in [50], we selected time-varying delay τ ( t ) = 2 + 0.2 sin t and initial condition ϕ ( 0 ) = [ 0.5 π 0.75 π 5 ] T . Figure 2 depicts the state response of the closed-loop system. Obviously, the fuzzy SOF controllers designed in Theorem 2 can guarantee that the system (37) is asymptotically stable.
Furthermore, as shown in Figure 2, when the same time-delay terms are selected, the fuzzy SOF controllers we designed can ensure that the system (37) is asymptotically stable at t = 40 s , while the system (37) is asymptotically stable at t = 140 s by applying the methods in [49]. Therefore, under the same conditions, the fuzzy SOF controllers designed in this paper can make the system reach a stable state in a shorter time. It further shows that the methods in this paper are less conservative.

5. Conclusions

For a class of T–S fuzzy systems with time-varying delays, the design methods of SOF controllers are presented based on IPM technology. Firstly, based on the integral inequality technique, new LMI-ed stabilization conditions are proposed by selecting an appropriate Lyapunov–Krasovskii functional. The premise MFs and number of fuzzy rules of the IPMB SOF controllers can be selected arbitrarily. Secondly, further relaxed MFD stabilization results are obtained by considering the information of MFs. Finally, two simulation examples were given to demonstrate the progressiveness of the proposed IPMB fuzzy SOF controllers.

Author Contributions

S.Q. contributed to the conception of the study and manuscript preparation, S.X. performed the experiments, Y.G. performed the data analyses, K.Z. wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the Zhejiang Provincial Natural Science Foundation of China, grant number LGC21F030001, and the Public Welfare Technology Application Research Project of the Zhejiang Province Science and Technology Department, grant number LGG20E050013.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

  1. Tian, Y.F.; Wang, Z.S. Asynchronous extended dissipative filtering for T–S fuzzy Markov jump systems. IEEE Trans. Syst. Man Cybern. Syst. 2022, 52, 3915–3925. [Google Scholar] [CrossRef]
  2. Sun, S.X.; Zhang, H.G.; Liu, C.; Liu, Y. Dissipativity-based intermittent fault detection and tolerant control for multiple delayed uncertain switched fuzzy stochastic systems with unmeasurable premise variables. IEEE Trans. Cybern. 2022, 52, 8766–8780. [Google Scholar] [CrossRef] [PubMed]
  3. Chang, W.J.; Lin, Y.W.; Lin, Y.H.; Pen, C.L.; Tsai, M.H. Actuator saturated fuzzy controller design for interval type-2 Takagi-Sugeno fuzzy models with multiplicative noises. Processes 2021, 9, 823. [Google Scholar] [CrossRef]
  4. Sheng, Y.; Huang, T.W.; Zeng, Z.G. Exponential Stabilization of Fuzzy Memristive Neural Networks With Multiple Time Delays Via Intermittent Control. IEEE Trans. Syst. Man Cybern. Syst. 2022, 52, 3092–3101. [Google Scholar] [CrossRef]
  5. Zhou, K.; Wang, B.R.; Qi, S.N. New relaxed stability and stabilization conditions for T-S fuzzy systems with time-varying delays. IET Control Theory Appl. 2021, 15, 1849–1867. [Google Scholar] [CrossRef]
  6. Chadli, M.; Didier, M.; José, R. Static Output Feedback for Takagi-Sugeno Systems: An LMI approach. In Proceedings of the 10th Mediterranean Conference on Control and Automation, Lisbon, Portugal, 9–12 July 2002. [Google Scholar]
  7. Islam, S.I.; Lim, C.C.; Shi, P. Robust fault detection of TS fuzzy systems with time-delay using fuzzy functional observer. Fuzzy Sets Syst. 2020, 392, 1–23. [Google Scholar] [CrossRef]
  8. El Hajjaji, A.; Chadli, M.; Oudghiri, M.; Pagès, O. Observer-Based Robust Fuzzy Control for Vehicle Lateral Dynamics. In Proceedings of the 2006 American Control Conference, Minneapolis, MN, USA, 14–16 June 2006. [Google Scholar]
  9. Wang, Y.Q.; Ren, W.J.; Liu, Z.Q.; Li, J.; Zhang, D. T-S Fuzzy Model-Based Fault Detection for Continuous Stirring Tank Reactor. Processes 2021, 9, 2127. [Google Scholar] [CrossRef]
  10. Li, X.D.; Li, P. Stability of time-delay systems with impulsive control involving stabilizing delays. Automatica 2021, 124, 109336. [Google Scholar] [CrossRef]
  11. Nguyen, M.C.; Tan, P.; Trinh, H. Sliding mode observer for estimating states and faults of linear time-delay systems with outputs subject to delays. Automatica 2021, 124, 109274. [Google Scholar] [CrossRef]
  12. Kong, M.R.; Liu, L. Exponential Stabilization for a Class of Strict-Feedback Nonlinear Time Delay Systems via State Feedback Control Scheme. Processes 2022, 10, 1259. [Google Scholar] [CrossRef]
  13. Li, G.L.; Peng, C.; Xie, X.P.; Xie, S.R. On Stability and Stabilization of T-S Fuzzy Systems With Time-Varying Delays via Quadratic Fuzzy Lyapunov Matrix. IEEE Trans. Fuzzy Syst. 2022, 30, 3762–3773. [Google Scholar] [CrossRef]
  14. Lian, Z.; He, Y.; Wu, M. Stability and stabilization for delayed fuzzy systems via reciprocally convex matrix inequality. Fuzzy Sets Syst. 2021, 402, 124–141. [Google Scholar] [CrossRef]
  15. Liu, C.; Guo, Y.R.; Rao, H.X.; Lin, M.; Xu, Y. Finite-time synchronization for periodic T-S fuzzy master-slave neural networks with distributed delays. J. Frankl. Inst. 2021, 358, 2367–2381. [Google Scholar] [CrossRef]
  16. Datta, R.; Saravanakumar, R.; Dey, R.; Bhattacharya, B.; Ahn, C.-K. Improved stabilization criteria for Takagi-Sugeno fuzzy systems with variable delays. Inf. Sci. 2021, 579, 591–606. [Google Scholar] [CrossRef]
  17. Pan, X.J.; Yang, B.; Cao, J.J.; Zhao, X.-D. Improved stability analysis of Takagi-Sugeno fuzzy systems with time-varying delays via an extended delay-dependent reciprocally convex inequality. Inf. Sci. 2021, 571, 24–37. [Google Scholar] [CrossRef]
  18. Sheng, Z.L.; Lin, C.; Chen, B.; Wang, Q.G. An asymmetric Lyapunov-Krasovskii functional method on stability and stabilization for T-S fuzzy systems with time delay. IEEE Trans. Fuzzy Syst. 2022, 30, 2135–2140. [Google Scholar] [CrossRef]
  19. Wang, X.; Park, J.H.; Yang, H.L.; Zhao, G.Z.; Zhong, S.-M. An improved fuzzy sampled-data control to stabilization of T-S fuzzy systems with state delays. IEEE Trans. Cybern. 2020, 50, 3125–3135. [Google Scholar] [CrossRef] [PubMed]
  20. Wang, L.K.; Liu, J.J.; Lam, H.K. Further study on stabilization for continuous-time Takagi-Sugeno fuzzy systems with time delay. IEEE Trans. Cybern. 2021, 51, 5637–5643. [Google Scholar] [CrossRef]
  21. Li, X.M.; Mehran, K.; Bao, Z.Y. Stability Analysis of Discrete-Time Polynomial Fuzzy-Model-Based Control Systems With Time Delay and Positivity Constraints Through Piecewise Taylor Series Membership Functions. IEEE Trans. Syst. Man Cybern. Syst. 2021, 51, 7517–7529. [Google Scholar] [CrossRef]
  22. Vadivel, R.; Saravanan, S.; Unyong, B.; Hammachukiattikul, P.; Hong, K.-S.; Lee, G.-M. Stabilization of Delayed Fuzzy Neutral-type Systems Under Intermittent Control. Int. J. Control Autom. Syst. 2021, 19, 1408–1425. [Google Scholar] [CrossRef]
  23. Lian, Z.; He, Y.; Zhang, C.K.; Wu, M. Stability and stabilization of T-S fuzzy systems with time-varying delays via delay-product-type functional method. IEEE Trans. Cybern. 2020, 50, 2580–2589. [Google Scholar] [CrossRef] [PubMed]
  24. Wang, G.; Jia, L.; Zhang, H.G. Stability and stabilization of T-S fuzzy time-delay system via relaxed integral inequality and dynamic delay partition. IEEE Trans. Fuzzy Syst. 2021, 29, 2829–2843. [Google Scholar] [CrossRef]
  25. Crusius, C.A.R.; Trofino, A. Sufficient LMI conditions for output feedback control problems. IEEE Trans. Autom. Control 1999, 44, 1053–1057. [Google Scholar] [CrossRef]
  26. Wu, H.N. An ILMI approach to robust H2 static output feedback fuzzy control for uncertain discrete-time nonlinear systems. Automatica 2008, 44, 2333–2339. [Google Scholar] [CrossRef]
  27. Huang, D.; Nguang, S.K. Robust H∞ Static Output Feedback Control of Fuzzy Systems: An ILMI Approach. IEEE Trans. Syst. Man Cybern. Cybern. 2006, 36, 216–222. [Google Scholar] [CrossRef]
  28. Huang, D.; Nguang, S.K. Static output feedback controller design for fuzzy systems: An ILMI approach. Inf. Sci. 2007, 177, 3005–3015. [Google Scholar] [CrossRef]
  29. Saifia, D.; Chadli, M.; Labiod, S.; Guerra, T.-M. Robust H∞ static output feedback stabilization of T-S fuzzy systems subject to actuator saturation. Int. J. Control Autom. Syst. 2012, 10, 613–622. [Google Scholar] [CrossRef]
  30. Jiang, S.; Fang, H.J. H∞ static output feedback control for nonlinear networked control systems with time delays and packet dropouts. ISA Trans. 2013, 2, 215–222. [Google Scholar] [CrossRef]
  31. Chen, J.; Lin, C.; Chen, B.; Wang, G.Q. Fuzzy-model-based admissibility analysis and output feedback control for nonlinear discrete-time systems with time-varying delay. Inf. Sci. 2017, 412, 116–131. [Google Scholar] [CrossRef]
  32. Ma, Y.C.; Yang, P.J.; Zhang, Q.L. Robust H∞ control for uncertain singular discrete T-S fuzzy time-delay systems with actuator saturation. J. Frankl. Inst. 2016, 13, 3290–3311. [Google Scholar] [CrossRef]
  33. Ma, Y.C.; Fu, L. Finite-time H∞ control for discrete-time switched singular time-delay systems subject to actuator saturation via static output feedback. Int. J. Syst. Sci. 2016, 47, 3394–3408. [Google Scholar] [CrossRef]
  34. Pham, T.N.; Nahavandi, S.; Hien, L.V.; Trinh, H.; Wong, K.P. Static output feedback frequency stabilization of time-delay power systems with coordinated electric vehicles state of charge control. IEEE Trans. Power Syst. 2016, 32, 3862–3874. [Google Scholar] [CrossRef]
  35. Wei, Y.L.; Qiu, J.B.; Shi, P.; Wu, L.G. A piecewise-Markovian Lyapunov approach to reliable output feedback control for fuzzy-affine systems with time-delays and actuator faults. IEEE Trans. Cybern. 2018, 48, 2723–2735. [Google Scholar] [CrossRef] [PubMed]
  36. Li, W.F.; Xie, Z.C.; Zhao, J.; Wong, P.K.; Wang, H.; Wang, X.-W. Static-output-feedback based robust fuzzy wheelbase preview control for uncertain active suspensions with time delay and finite frequency constraint. IEEE/CAA J. Autom. Sin. 2021, 8, 664–678. [Google Scholar] [CrossRef]
  37. Zhou, K.; Huang, T.M.; Zhao, T.; Gao, F.H. Membership-function-dependent stability and stabilization conditions for T-S fuzzy time-delay systems. IETE J. Res. 2019, 65, 351–364. [Google Scholar] [CrossRef]
  38. Lam, H.K.; Narimani, M. Stability analysis and performance design for fuzzy-model-based control system under imperfect premise matching. IEEE Trans. Fuzzy Syst. 2009, 17, 949–961. [Google Scholar] [CrossRef]
  39. Zhao, T.; Liu, J.H.; Dian, S.Y. Finite-time control for interval type-2 fuzzy time-delay systems with norm-bounded uncertainties and limited communication capacity. Inf. Sci. 2019, 483, 153–173. [Google Scholar] [CrossRef]
  40. Che, C.; Peng, J.Y.; Xiao, J.; Zhao, T.; Zhou, J. Membership-function-dependent stabilization conditions for interval type-2 fuzzy time-delay systems via static output feedback scheme. Int. J. Fuzzy Syst. 2018, 20, 1439–1450. [Google Scholar] [CrossRef]
  41. Li, X.M.; Merhan, K.; Bao, Z.Y. Membership Function, Time Delay-Dependent η-Exponential Stabilization of Positive Discrete Time Polynomial Fuzzy Model Control System. IEEE Trans. Fuzzy Syst. 2022, 30, 2197–2209. [Google Scholar] [CrossRef]
  42. Gu, K. An Integral Inequality in the Stability Problem of Time-Delay Systems. In Proceedings of the 39th IEEE Conference on Decision and Control, Sydney, NSW, Australia, 12–15 December 2000; pp. 2805–2810. [Google Scholar]
  43. Park, P.G.; Ko, J.W.; Jeong, C. Reciprocally convex approach to stability of systems with time-varying delays. Automatica 2011, 47, 235–238. [Google Scholar] [CrossRef]
  44. Wang, Y.Y.; Xie, L.H.; de Souza, C.E. Robust control of a class of uncertain nonlinear systems. Syst. Control Lett. 1992, 19, 139–149. [Google Scholar] [CrossRef]
  45. De Souza, C.E. Robust stability and stabilization of uncertain discrete-time Markovian jump linear systems. IEEE Trans. Autom. Control 2006, 51, 836–841. [Google Scholar] [CrossRef]
  46. Xie, W.B.; Lim, C.C.; Zhang, J.; Huang, L. New approaches to observer design and stability analysis for T-S fuzzy system with multiplicative noise. J. Frankl. Inst. 2017, 354, 887–901. [Google Scholar] [CrossRef]
  47. Fang, C.H.; Liu, Y.S.; Kau, S.W.; Hong, L.; Lee, C.H. A new LMI-based approach to relaxed quadratic stabilization of T-S fuzzy control systems. IEEE Trans. Fuzzy Syst. 2006, 14, 386–397. [Google Scholar] [CrossRef]
  48. Che, C.; Peng, J.; Zhao, T.; Xiao, J.; Zhou, J. Improved stabilization conditions for nonlinear systems with input and state delays via T-S fuzzy model. Math. Probl. Eng. 2018, 2018, 3542352. [Google Scholar] [CrossRef] [Green Version]
  49. Tao, H.F.; Hu, S.S.; Cai, J.W. L∞ Static Output Feedback Control for T-S Fuzzy Systems with Time-Varying Delays. Acta Autom. Sin. 2008, 34, 453–459. [Google Scholar] [CrossRef]
  50. Chen, B.; Liu, X.P.; Tong, S.C. Delay-dependent stability analysis and control synthesis of fuzzy dynamic systems with time delay. Fuzzy Sets Syst. 2006, 157, 2224–2240. [Google Scholar] [CrossRef]
Figure 1. State response of the closed−loop system for Example 1.
Figure 1. State response of the closed−loop system for Example 1.
Processes 11 00121 g001
Figure 2. State response of the closed−loop system for Example 2.
Figure 2. State response of the closed−loop system for Example 2.
Processes 11 00121 g002
Table 1. The infimum and supremum of Δ ω i j k ( x ) with ϑ = 20 .
Table 1. The infimum and supremum of Δ ω i j k ( x ) with ϑ = 20 .
The   Infimum   of   Δ ω i j k ( x )
The   Supremum   of   Δ ω i j k ( x )
Δ ω _ 111 = 7.7231 × 10 2 Δ ω _ 112 = 6.7881 × 10 2 Δ ω ¯ 111 = 7.9312 × 10 2 Δ ω ¯ 112 = 6.5367 × 10 2
Δ ω _ 113 = 3.7058 × 10 2 Δ ω _ 114 = 4.5041 × 10 2 Δ ω ¯ 113 = 2.0982 × 10 2 Δ ω ¯ 114 = 9.3016 × 10 2
Δ ω _ 121 = 4.5582 × 10 2 Δ ω _ 122 = 8.4183 × 10 2 Δ ω ¯ 121 = 2.6978 × 10 2 Δ ω ¯ 122 = 9.7881 × 10 2
Δ ω _ 123 = 6.3980 × 10 2 Δ ω _ 124 = 1.6824 × 10 2 Δ ω ¯ 123 = 5.3882 × 10 3 Δ ω ¯ 124 = 6.5743 × 10 2
Δ ω _ 211 = 3.6015 × 10 2 Δ ω _ 212 = 5.5514 × 10 2 Δ ω ¯ 211 = 4.5137 × 10 2 Δ ω ¯ 212 = 2.5983 × 10 2
Δ ω _ 213 = 3.6342 × 10 2 Δ ω _ 214 = 9.7648 × 10 2 Δ ω ¯ 213 = 6.5739 × 10 2 Δ ω ¯ 214 = 5.1385 × 10 2
Δ ω _ 221 = 8.4417 × 10 4 Δ ω _ 222 = 7.8991 × 10 2 Δ ω ¯ 221 = 1.5766 × 10 2 Δ ω ¯ 222 = 5.1464 × 10 2
Δ ω _ 223 = 8.6359 × 10 2 Δ ω _ 224 = 5.9734 × 10 2 Δ ω ¯ 223 = 8.1625 × 10 4 Δ ω ¯ 224 = 8.6592 × 10 2
Δ ω _ 311 = 3.6658 × 10 2 Δ ω _ 312 = 2.6971 × 10 2 Δ ω ¯ 311 = 4.5122 × 10 2 Δ ω ¯ 312 = 9.2563 × 10 2
Δ ω _ 313 = 7.4128 × 10 2 Δ ω _ 314 = 9.7881 × 10 2 Δ ω ¯ 313 = 3.7558 × 10 2 Δ ω ¯ 314 = 2.5619 × 10 2
Δ ω _ 321 = 4.5613 × 10 2 Δ ω _ 322 = 8.7516 × 10 2 Δ ω ¯ 321 = 1.0562 × 10 2 Δ ω ¯ 322 = 7.2648 × 10 2
Δ ω _ 323 = 5.8815 × 10 2 Δ ω _ 324 = 2.5097 × 10 2 Δ ω ¯ 323 = 8.5473 × 10 2 Δ ω ¯ 324 = 4.2669 × 10 2
Δ ω _ 411 = 6.3984 × 10 2 Δ ω _ 412 = 1.8547 × 10 2 Δ ω ¯ 411 = 9.5618 × 10 2 Δ ω ¯ 412 = 7.2651 × 10 2
Δ ω _ 413 = 7.6635 × 10 2 Δ ω _ 414 = 7.1564 × 10 2 Δ ω ¯ 413 = 6.5871 × 10 2 Δ ω ¯ 414 = 9.5134 × 10 2
Δ ω _ 421 = 4.6238 × 10 2 Δ ω _ 422 = 2.1657 × 10 4 Δ ω ¯ 421 = 5.6736 × 10 2 Δ ω ¯ 422 = 8.6951 × 10 2
Δ ω _ 423 = 3.8749 × 10 4 Δ ω _ 424 = 8.3346 × 10 2 Δ ω ¯ 423 = 8.2561 × 10 2 Δ ω ¯ 424 = 3.6182 × 10 2
Table 2. The infimum and supremum of Δ ω i j k ( x ) with ϑ = 20 .
Table 2. The infimum and supremum of Δ ω i j k ( x ) with ϑ = 20 .
The   Infimum   of   Δ ω i j k ( x )
The   Supremum   of   Δ ω i j k ( x )
Δ ω _ 111 = 5.3678 × 10 3 Δ ω _ 112 = 3.5846 × 10 3 Δ ω ¯ 111 = 3 .5628 × 10 3 Δ ω ¯ 112 = 2 . 9583 × 10 3
Δ ω _ 121 = 8.1629 × 10 3 Δ ω _ 122 = 4.8617 × 10 3 Δ ω ¯ 121 = 5 .6592 × 10 3 Δ ω ¯ 122 = 6 . 2369 × 10 3
Δ ω _ 211 = 5.6659 × 10 3 Δ ω _ 212 = 2.8631 × 10 3 Δ ω ¯ 211 = 8 .2683 × 10 3 Δ ω ¯ 212 = 7 .2856 × 10 3
Δ ω _ 221 = 1.6852 × 10 3 Δ ω _ 222 = 3.1537 × 10 3 Δ ω ¯ 221 = 8 . 1227 × 10 3 Δ ω ¯ 222 = 7 .3329 × 10 3
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Qi, S.; Zhou, K.; Xu, S.; Gao, Y. New Relaxed Static Output Feedback Stabilization of T–S Fuzzy Systems with Time-Varying Delays. Processes 2023, 11, 121. https://doi.org/10.3390/pr11010121

AMA Style

Qi S, Zhou K, Xu S, Gao Y. New Relaxed Static Output Feedback Stabilization of T–S Fuzzy Systems with Time-Varying Delays. Processes. 2023; 11(1):121. https://doi.org/10.3390/pr11010121

Chicago/Turabian Style

Qi, Shunan, Kun Zhou, Suan Xu, and Yanfeng Gao. 2023. "New Relaxed Static Output Feedback Stabilization of T–S Fuzzy Systems with Time-Varying Delays" Processes 11, no. 1: 121. https://doi.org/10.3390/pr11010121

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop