ISRN.APPLIED.MATHEMATICS ISRN Applied Mathematics 2090-5572 Hindawi Publishing Corporation 230158 10.1155/2013/230158 230158 Research Article Fault Detection Filter Design for Stochastic Systems with Mixed Time-Delays and Parameter Uncertainties http://orcid.org/0000-0002-9082-9554 Hou Liyuan 1 Zhong Shouming 2, 3 Zhu Hong 1 Zeng Yong 1 http://orcid.org/0000-0002-9383-3516 Shi Lin 1 Bellouquid A. Kearsley A. J. Lien C.-H. 1 School of Automation Engineering University of Electronic Science and Technology of China Chengdu 611731 China uestc.edu.cn 2 School of Mathematical Sciences University of Electronic Science and Technology of China Chengdu 611731 China uestc.edu.cn 3 Key Laboratory for Neuroinformation of Ministry of Education University of Electronic Science and Technology of China Chengdu 611731 China uestc.edu.cn 2013 13 11 2013 2013 29 05 2013 05 08 2013 2013 Copyright © 2013 Liyuan Hou et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

This paper purposes the design of a fault detection filter for stochastic systems with mixed time-delays and parameter uncertainties. The main idea is to construct some new Lyapunov functional for the fault detection dynamics. A new robustly asymptotically stable criterion for the systems is derived through linear matrix inequality (LMI) by introducing a comprehensive different Lyapunov-Krasovskii functional. Then, the fault detection filter is designed in terms of linear matrix inequalities (LMIs) which can be easily checked in practice. At the same time, the error between the residual signal and the fault signal is made as small as possible. Finally, an example is given to illustrate the effectiveness and advantages of the proposed results.

1. Introduction

Stochastic systems have strong practical relevance in mechanical systems, economics, systems with human operators, and other engineering areas. Meanwhile, the filtering problem has many applications in the areas of signal processing, signal estimation, pattern recognition, and many practical control systems. Therefore, the filter design problems and the stochastic systems have become important areas of research and received great attention over the past few years .

With the rising demand for higher safety and increasing demand for higher performance in the modern industries, the research on fault detection for dynamic systems has received more and more attention during the past two decades, like model-based schemes , knowledge-based approaches , and signal-based methods , and so forth. Particularly, the problem of model-based fault detection has been an active research area among them. The basic objective of model-based fault detection is to construct the residual generator, and to determine the residual evaluation function and the threshold. An alarm of fault will be generated, when the value of the evaluation function is greater than the stated threshold. However, randomly occurring nonlinearities and the existence of unknown inputs may seriously affect the performance of model-based fault detection systems. Therefore, robustness issue plays an important role in the application of model-based fault detection schemes. Increasing the robustness of residual to unknown inputs and modelling errors and enhancing the sensitivity to faults are of prime importance in designing a model-based fault detection system. Among different methods for fault detection, the model-based approach has been widely used in recent years. So far, the problem of fault detection has been thoroughly investigated for a variety of systems including uncertain systems [2, 1014], time-delay systems , and Markovian jump linear systems .

Besides, time-delay is one of the major sources of instability and poor performance of a practical control system. And many results on stochastic systems with time delays have been reported in the literature  and the references therein. There is a need to discuss the distributed delays that occur very often in practical systems. The engineering significance of distributed delays has been widely recognized, and a number of corresponding results have been published; see, for example, . The distributed delays in the discrete-time setting, on the other hand, have received little attention. It is well known that nonlinearities exist universally in practical systems, and therefore nonlinear control has been an ever hot topic in the past few decades.

As far as we know, the delay-dependent criteria on fault detection filter design for delayed stochastic systems with parameter uncertainties have not been fully studied, which is still open. Motivated by the above discussion and in order to obtain less conservative results, we choose an appropriate new Lyapunov functional and establish a new integral inequality in the stochastic setting. What is more, because we have carefully considered the ranges for the time-varying delays, our criteria can be applicable to both fast and slow time-varying delays. The stability criteria obtained are in terms of linear matrix inequalities (LMIs) which can be checked efficiently via the LMI toolbox. Finally, a numerical example is also given to demonstrate the effectiveness and advantages of our theoretical results.

Notations. The notations used throughout the paper are fairly standard. The superscript “T” stands for matrix transposition; Rn denotes the n-dimensional Euclidean space; Rn×m is the set of all n×m real matrices; the notation P>0 means that P is a positive definite matrix; I and 0 represent identity matrix and zero matrix, respectively; and diag(·) denotes the diagonal matrix. The vector norm is taken to be Euclidean, and the matrix norm is the corresponding induced one. E{·} denotes the expectation operator. 2[0,+) is the space of square-integrable vector function over [0,+). In symmetric block matrices, we use an asterisk (*) to represent a term that is induced by symmetry. Matrices, if their dimensions are not explicitly stated, are assumed to be compatible for algebraic operations.

2. Problem Formulation and Some Preliminaries

The simplest and most fundamental case considered is the problem of quadratic stabilization for the following system: (1)x(k+1)=Ax(k)+Bu(k)y(k)=Cx(k), where x(k)Rn is the state and u(k) is the control input. Then, we add discrete time-delays τi(k), (i=1,2,,q), while d(d=1,2,,) describe the distributed time-delays. In addition, we are motivated to study the fault detection problem for a class of discrete-time systems involving stochastic time-delay, such that we consider the following stochastic time-delay system: (2)x(k+1)=Ax(k)+A1i=1qαi(k)x(k-τi(k))+β(k)A2d=1μdx(k-d)+γ(k)g(k,x(k))+D1w(k)+Gf(k)y(k)=Cx(k)+D2w(k)+Hf(k)x(k)=ϕ(k),k-, where x(k)Rn is the state; y(k)Rm is the process output; w(k)Rp is the unknown input which belongs to 2[0,); f(k)Rl is the fault to be detected; A, A1, A2, C, D1, D2, and G are real known constant matrices with compatible dimensions. The discrete time-delay τi(k)(i=1,2,,n) is a time-varying differentiable function satisfying τmτi(k)τM, while d(d=1,2,,) denotes the discrete time-delays, and the constant μd0 satisfies the following convergence conditions μ-:=d=1μdd=1dμd<, μ^:=d=1μd2<.

Assumption 1.

The nonlinear function g(k,x(k))Rn satisfies the following condition: for any x, yR, (xy), (3)g^i-gi(x)-gi(y)x-yg^i+, where g^i-, g^i+(i=1,2,,n) are constants.

Assumption 2.

The stochastic variables αi(k)(i=1,2,n1), β(k), and γ(k) and mutually uncorrelated Bernoulli distributed white sequences that account for, respectively, the phenomena of randomly occurring discrete time-delays, distribute time-delays, and nonlinearities. A natural assumption on the sequences αi(k)(i=1,2,n1), β(k), and γ(k) is made as follows: (4)Prob{αi(k)=1}=𝔼{αi(k)}=α-i,Prob{αi(k)=0}=1-α-i,Prob{β(k)=1}=𝔼{β(k)}=β-,Prob{β(k)=0}=1-β-,Prob{γ(k)=1}=𝔼{γ(k)}=γ-,Prob{γ(k)=0}=1-γ-.

In a networked environment, it is quite common that the measurements y(k) of the system are quantized during the signal transmission. Let us denote the quantizer as h(·)=[h1(·),h2(·),,hm(·)]T which is symmetric, that is, hj(-ν)=-hj(ν), j=1,,m.

The map of the quantization process is y^(k)=h(y(k))=[h1(y(1)(k)),h2(y(2)(k)),,hm(y(m)(k))]T. In this paper, we are interested in the logarithmic static and time-invariant quantizer. For each, the set of quantization levels is described by 𝒰j={±μ^i(j),μ^i(j)=χjjμ^0(j),i=0,±1,±2,}{0}, 0<χj<1, μ^0(j)>0, and each of the quantization levels corresponds to a segment, such that the quantizer maps the whole segment to this quantization level. According to , the quantizer is given by (5)hj(y(j)(k))={μ^i(j),11+δjμ^i(j)y(j)(k)11-δjμ^i(j)0,y(j)(k)=0-hj(-y(j)(k)),y(j)(k)<0, where δj=(1-χj)(1+χj), such that hj(y(j)(k))=(I+Δk)y(j)(k) with |Δk(j)|δj; the measurements with quantization effect can be expressed as (6)y^(k)=(I+Δk)y(k)=(I+Δk)Cx(k)+(I+Δk)D2ω(k)+(I+Δk)Hf(k), with Δk=diag(Δk(1),,Δk(m)).

Consider the following full-order fault detection filter for system (2): (7)x^(k+1)=Afx^(k)+Bfy^(k),r(k)=Cfx^(k)+Dfy^(k), where x^(k)Rn is the filter state vector, r(k)l is the so called residual that is compatible with the fault vector f(k), and Af, Bf, and Cf are appropriately dimensioned filter matrices to be designed.

From (2), (5), and (7), we can obtain the following filtering error system: (8)x~(k+1)=(A~+ΔA)x~(k)+i=1q(A~1i+A~1i(k))x~(k-τi(k))+(A~2+A~2(k))d=1μdx~(k-d)+(γ-+γ~(k))I~g(k,x(k))+(D~+ΔD)v(k)r~(k)=(C~+ΔC)x~(k)+(D~f+ΔDf)v(k)x~(k)=ϕ~(k),k[-τM,0], with (9)x~(k)=[xT(k),x^T(k)]T,ϕ~T(k)=[ϕT(k),0T],r~(k)=r(k)-f(k),v(k)=[ωT(k),fT(k)]T,A~=(A0BfCAf),ΔA=(00BfΔkC0),A~1i=(α-iA1000),A~1i(k)=(α~i(k)A1000),A~2=(β-A2000),  A~2(k)=(β~(k)A2000),I~=(I0),D~=(D1GBfD2BfH),ΔD=(00BfΔkD2BfΔkH),C~=(DfCCf),ΔC=(DfΔkC0),D~f=(DfD2DfH-I),ΔDf=(DfΔkD2DfΔkH),α~i(k)=αi(k)-α-i,      β~(k)=β(k)-β-,γ~(k)=γ(k)-γ-.

In order to conduct the stability analysis for the above systems, it is necessary to make the following definitions and lemmas.

Definition 3.

Consider the system (8) with v(k)=0; it is said to be mean-square exponentially stable if for any initial conditions, there exist a δ>0 and 0<α<1, such that E{|x~(k)|2}δαksupiZ-E{|ϕ~(i)|2}, k0. Moreover, if limtE{|x(k)|2}=0, then the system is said to be robust asymptotically mean-square stable.

Definition 4.

Given a scalar γ>0, the error system (8) is said to be mean-square robustly exponentially stable with the attenuation level γ if the error system (8) with v(k)=0 is mean-square robustly exponentially stable, and under zero initial condition, the following is true for all nonzero v(k): (10)k=0+E{r~T(k)r~(k)}γ2k=0E{vT(k)v(k)}(i.e.,eEγv2,w(k)2[0,+)). Correspondingly, system (7) is said to be stochastic 2- filter of system (2).

We further adopt a residual evaluation stage including an evaluation function J(k) and a threshold Jth of the following form: J(k)=h=0kr(h)r(h)1/2 where L denotes the maximum time step of the evaluation function. Based on (16), the occurrence of faults can be detected by comparing J(k) with Jth according to the following rule: J(k)>Jth (with faults) alarm; J(k)<=Jth no faults.

Remark 5.

The parametric uncertainties ΔA, ΔD, ΔC, and ΔDf satisfy (11)[ΔAΔDΔCΔDf]=[0BfDf]F(k)[ΛC,0,ΛD2,ΛH], where Λ=diag{δ1,δ2,,δm} and F(k)=ΔkΛ-, and F(k) satisfies FT(k)F(k)I.

Lemma 6.

Let M, N, and F be real matrices of appropriate dimensions with FTFI. Then, for any scalar ϵ>0, one has (12)MFN+(MFN)Tϵ-1MMT+ϵNNT.

Lemma 7 (Schur complement).

Given constant matrices S1, S2, and S3, where S1=S1T and S2=S2T, then S1+S3TS2-1S3<0 if and only if (13)[S1S3TS3-S2]<0,or[-S2S3S3TS1]<0.

Lemma 8 (Liu et al. [<xref ref-type="bibr" rid="B9">5</xref>]).

For any constant matrix MRn×n, M=MT>0, xiRn, and constants ai>0(i=1,2,). If the series concerned is convergent, then one has (14)-(i=1ai)i=1aixiTMaixiT-(i=1aixi)TM(i=1aixi).

3. Main Results

In this section, we will investigate a sufficient condition on the performance analysis for the filter error system (8). And then, the solution to the fault detection filter design for the system (2) is listed as follows.

Theorem 9.

For nominal system of (8) with given filter parameters and the index γ>0, the fault detection dynamics is robustly exponentially stable in mean square, if there exist constant l>0, positive matrices P, Q, Qi(i=1,2,,q), and any matrices U with appropriate dimensions, such that the following LMIs hold:

(15) Γ = [ Γ 11 0 0 0 Γ 14 0 C ~ T A ~ T P * α - 1 ( 1 - α - 1 ) A ^ 1 T P A ^ 1 - Q 1 0 0 0 0 0 A ~ 11 T P * * * * * α - q ( 1 - α - q ) A ^ 1 T P A ^ 1 - Q q 0 0 0 0 A ~ 1 q T P * * * * - Q μ - 0 0 0 A ~ 2 T P * * * * * Γ 44 0 0 γ - I ~ T P * * * * * * Γ 55 D ~ f T D ~ T P * * * * * * * - I 0 * * * * * * * * - P ] < 0 , Γ 11 = - P + i = 1 q ( τ M - τ m + 1 ) Q i + μ - Q + μ ^ β - ( 1 - β - ) A ^ 2 T P A ^ 2 - [ U G 1 0 0 0 ] , Γ 14 = [ U G 2 0 ] , Γ 44 = γ - ( 1 - γ - ) I ~ T P I ~ - U , Γ 55 = - γ 2 I , A ^ 1 = ( A 1 0 0 0 ) , A ^ 2 = ( A 2 0 0 0 ) , G 1 = diag { g ^ 1 - g ^ 1 + , , g ^ n - g ^ n + } , G 2 = diag { g ^ 1 - + g ^ 1 + 2 , , g ^ n - + g ^ n + 2 } .

Proof.

For simplicity, let (16)h1(k)=A~x~(k)+i=1qA~1ix~(k-τi(k))+A~2d=1μdx~(k-d)+γ-I~g(k,x(k))+D~v(k)h2(k)=i=1qA~1i(k)x~(k-τi(k))+A~2(k)d=1μdx~(k-d)+γ~(k)I~g(k,x(k)). Then, we have the following equation: (17)x~(k+1)=h1(k)+h2(k). For this system, we construct the following Lyapunov functional: (18)V(k)=V1(k)+V2(k)+V3(k)+V4(k), with (19)V1(k)=x~T(k)Px~(k),V2(k)=i=1qj=k-τi(k)k-1x~T(j)Qix~(j)V3(k)=i=1qj=-τM+1-τml=k+jk-1x~T(l)Qix~(l)V4(k)=d=1i=k-dk-1x~T(i){μdQ+μd2β-(1-β-)A^2TPA^2}x~(i).

Taking the difference of the functional along the solution of the system, we obtain (20)𝔼{ΔV(k)}=i=14𝔼{ΔVi(k)}𝔼{ΔV1(k)}=𝔼{𝔼{V1(k+1)x~k}-V1(k)}=𝔼{x~T(k+1)Px~(k+1)-x~T(k)Px~(k)}=h1T(k)Ph1(k)+𝔼{2h1T(k)Ph2(k)}+𝔼{i=1qj=1qx~T(k-τi(k))×A~1iT(k)PA~1j(k)x~(k-τj(k))+d1=1[d2=1μd1μd2x~T(k-d1)0000000×A~2T(k)PA~2(k)x~(k-d2)d2=1]+γ~(k)2gT(k,x(k))I~TPI~g(k,x(k))i=1qj=1q}-x~T(k)Px~(k).

It is obvious that (21)𝔼{2h1T(k)Ph2(k)}=0,𝔼{A~1iT(k)PA~1j(k)}={0ijα-i(1-α-i)A^1TPA^1i=j,𝔼{i=1qj=1qx~T(k-τi(k))A~1i(k)PA~1j(k)x~(k-τj(k))+d1=1d2=1μd1μd2x~T(k-d1)A~2T(k)PA~2(k)x~(k-d2)+γ~(k)2gT(k,x(k))I~TPI~g(k,x(k))i=1qj=1q}=i=1qx~T(k-τi(k))α-i(1-α-i)A^1TPA^1x~(k-τi(k))+d=1x~T(k-d)β-(1-β~)μd2A^2TPA^2x~(k-d)+gT(k,x(k))γ-(1-γ-)I~TPI~g(k,x(k)). Then, we have the following equations: (22)𝔼{ΔV1(k)}=h1T(k)Ph1(k)-x~T(k)Px~(k)+i=1qx~T(k-τi(k))α-i(1-α-i)00000×A^1TPA^1x~(k-τi(k))+d=1x~T(k-d)β-(1-β~)μd2A^2TPA^2x~(k-d)00000+gT(k,x(k))γ-(1-γ-)I~TPI~g(k,x(k))𝔼{ΔV2(k)}=𝔼{𝔼{V2(k+1)}-V2(k)}=𝔼{i=1qj=k+1-τi(k+1)kx~T(j)Qix~(j)-i=1qj=k-τi(k)k-1x~T(j)Qix~(j)}=i=1q𝔼{j=k-τi(k)+1k-1x~T(k)Qix~(k)000000-x~T(k-τi(k))Qig(k-τi(k))000000+j=k-τi(k+1)+1k-1x~T(j)Qix~(j)000000-j=k-τi(k)+1k-1x~T(j)Qix~(j)}i=1q𝔼{j=k-τM+1k-τmx~T(k)Qix~(k)00000-x~T(k-τi(k))Qix~(k-τi(k))00000+j=k-τM+1k-τmx~T(j)Qix~(j)},𝔼{ΔV3(k)}=𝔼{𝔼{V3(k+1)}-V3(k)}=i=1qj=-τM+1-τml=k+j+1kx~T(l)Qix~(l)-i=1qj=-τM+1-τml=k+jk-1x~T(l)Qix~(l)=i=1q(j=k-τM-1k-τm(τM-τm)x~T(k)Qix~(k)-j=k-τM-1k-τmx~T(j)Qix~(j)),𝔼{ΔV4(k)}=𝔼{𝔼{V4(k+1)}-V4(k)}=d=1𝔼{i=k-d+1kx~T(i)00000×[μdQ+μd2β-(1-β-)A^2TPA^2]x~(i)00000-i=k-dk-1x~T(i)00000×[μdQ+μd2β-(1-β-)A^2TPA^2x~(i)]i=k-d+1k}=x~T(k)(μ-Q+μ^β-(1-β-)A^2TPA^2)x~(k)-d=1x~T(k-d)μd2β-(1-β-)A^2TPA^2x~×(k-d)-d=1x~T(k-d)μdQx~(k-d). By using Lemma 8, we have (23)-d=1x~T(k-d)μdQx~(k-d)-1μ-(d=1x~(k-d))TQ(d=1x~(k-d)). From Assumption 1, g^i-(gi(x)-gi(y))/(x-y)g^i+, we have (g(k,x(k))-g^i-x(k))(g(k,x(k))-g^i+x(k))0.

It can be deduced that there exists U=diag[u1,u2,,un]>0 such that (24)i=1nui[x(k)g(k)]T[g^i-g^i+eieiT-g^i-+g^i+2eieiT-g^i-+g^i+2eieiTeieiT]×[x(k)g(k)]=[x(k)g(k)]T[UG1-UG2-UG2U][x(k)g(k)]0, where ei denotes the unit column vector having one element on its rth row and zeros elsewhere.

Now, we are ready to prove the exponential stability of the system (8) with v(k)=0. Combining (16)–(24), we can easily see that (25)𝔼{ΔV(k)}𝔼{ζ1T(k)(Γ1+ΔTPΔ)ζ1(k)}, where(26)ζ1(k)=[x~T(k)x~T(k-τ1(k))x~T(k-τ2(k))x~T(t-τq(k))d=1μdx(k-d)gT(k)]T,Γ1=[ϕ110000*α-1(1-α-1)A^1TPA^1-Q1000*****α-q(1-α-q)A^1TPA^1-Qq00****-Qμ-0*****γ-(1-γ-)I~TPI~-lI]Δ=[A~A~11A~1qA~2γ-I~].

If Γ1+ΔTPΔ<0, we can obtain that Γ2<0 by Schur complements (Lemma 7), where(27)Γ2=[ϕ110000A~TP*α-1(1-α-1)A^1TPA^1-Q1000A~11TP*****α-q(1-α-q)A^1TPA^1-Qq00A~1qTP****-Qμ-0A~2TP*****γ-(1-γ-)I~TPI~-lIγ-I~TP******-P].Then, the system (8) with v(k)=0 is mean-square robustly asymptotically stable. Furthermore, along the same line of the proof for Theorem 1 in , the exponential stability of system (8) can be confirmed on the mean-square sense.

Next, we will establish the performance of the filtering error system (8) under the zero initial condition. To this end, we introduce (28)J(k)=Ek=0{r~T(k)r~(k)}-γ2k=0E{vT(k)v(k)}. And, it is obvious to see that (29)J(n)=Ek=0n{r~T(k)r~(k)}-γ2k=0nE{vT(k)v(k)}+Ek=0n{V(k+1)-V(k)}-V(n+1)+V(0)Ek=0n{r~T(k)r~(k)}-γ2k=0nE{vT(k)v(k)}+Ek=0n{ΔV(k)}. Along the same line as the proof of the stability of system (7), J(n)Ek=0n{ζ2T(k)Γζ2(k)}, where(30)ζ2(k)=[x~T(k)x~T(k-τ1(k))x~T(k-τ2(k))x~T(t-τq(k))d=1μdxT(k-d)gT(k)vT(k)]T,we can show that J(n)0.

Letting n, we obtain (31)k=0+E{r~T(k)r~(k)}γ2k=0E{vT(k)v(k)}.

This completes the proof.

Next, we are in a position to deal with the design of the filter output feedback controller for the system (8) and obtain the main result of this paper in the following theorem.

Theorem 10.

For given constants 0τm<τM, ε>0, and a scalar λ>0, the fault detection filter (7) exists such that error system (8) is robustly asymptotically exponentially stable in mean square, if there exist constant l>0, positive matrices P, R, and Ri(i=1,2,,q), and any matrices A^f, B^f, C^f, and D^f with appropriate dimensions, respectively,

(32) Σ = [ Σ 11 0 0 0 Σ 14 0 Σ 16 Σ 17 * α - 1 ( 1 - α - 1 ) A - 1 - R 1 0 0 0 0 0 Σ 2,17 * * * * * α - q ( 1 - α - q ) A - 1 - R q 0 0 0 0 Σ 2 , q 7 * * * * - R μ - 0 0 0 Σ 37 * * * * * Σ 44 0 0 Σ 47 * * * * * * Σ 55 Σ 56 Σ 57 * * * * * * * - I 0 * * * * * * * * Σ 7 ] < 0 , (33) Σ 11 = - ( X Y Y Y ) + i = 1 q ( τ M - τ m + 1 ) R i + μ - R + μ ^ β - ( 1 - β - ) A - 2 T - U ~ , U ~ = ( U G 1 U G 1 U G 1 U G 1 ) , A - 1 = ( A 1 T X A 1 A 1 T X A 1 A 1 T X A 1 A 1 T X A 1 ) , A - 2 = ( A 2 T X A 2 A 2 T X A 2 A 2 T X A 2 A 2 T X A 2 ) , Σ 14 = [ U G 2 U G 2 ] , Σ 16 = [ C T D ^ f T C T D ^ f T + C ^ f T ] , Σ 17 = [ A T X + C T B ^ f T A T Y A T X + C T B ^ f T + A ^ f T A T Y ] , Σ 2 , i 7 = [ α - i A 1 T X α - i A 1 T Y α - i A 1 T X α - i A 1 T Y ] ( i = 1,2 , , q ) , Σ 37 = [ β - A 2 X β - A 2 Y β - A 2 X β - A 2 Y ] , Σ 44 = γ - ( 1 - γ - ) X - U , Σ 55 = - γ 2 I , Σ 47 = [ γ X γ Y ] , Σ 56 = [ D 2 T D ~ f T H T D ~ f T - I ] , Σ 57 = [ D 1 T X + D 2 T B ^ f T D 1 T Y G T X + H T B ^ f T G T Y ] , Σ 77 = - [ X Y Y Y ] .

If (X,Y,Q,Qi) is a feasible solution of (15), there exist nonsingular matrices U and V satisfying UV=I-XY-1. The fault detection filter parameters in the form of (7) are given as follows: (34)Af=U-1A^fY-1V-1,Bf=U-1B^f,Cf=C^fY-1V-1,Df=D^f.

Proof.

The condition in (31) pledges that (XY*Y)>0.

This inequality implies that I-XY-1 and X-Y are nonsingular by Schur complement. If the condition in (32) is feasible, there exist nonsingular matrices U, V satisfying UV=I-XY-1.

Then, we introduce the following matrices: (35)Ω1=(XYUT0),Ω2=(II0VY).

Obviously Ω1, Ω2 are nonsingular.

Let (36)PΩ1Ω2-1=(XUUT-UTY-1V-1); then Ω2TPΩ2=(XYYTY).

Pre- and postmultiplying inequality (32) by diag{Ω2-1,Ω2-1,,Ω2-1q,Ω2-1,I,I,I,Ω2-1},  and  letting Ω2TQiΩ2=Ri, Ω2TQΩ2=R, we can obtain (8). This completes the proof.

When there are parameters uncertainties, we have the following corollary.

Corollary 11.

For given constants 0τm<τM, ε>0 and a scalar λ>0, the fault detection filter (7) exists such that error system (8) is robustly asymptotically exponentially stable in mean square, if there exist constant l>0, ϵ>0, positive matrices P, R, and Ri(i=1,2,,q), and any matrices A^f, B^f, C^f, and D^f with appropriate dimensions, respectively,(37)Σ¯=[Σ11000Σ140Σ16Σ170Σ19*α-1(1-α-1)A-1-R100000Σ2,1700***α-q(1-α-q)A-1-Rq0000Σ2,q700****-Rμ-000Σ3700*****Σ4400Σ4700******Σ55Σ56Σ570Σ59*******-I0Σ680********Σ7Σ780*********-ϵI0**********-ϵI]<0,where, Σ11, Σ16, Σ17, Σ2,17,, Σ2,q7, A-1, Σ37, Σ44, Σ47, Σ55, Σ56, Σ57, and Σ7 have been defined in Theorem 9: (38)Σ19=[CTΛTCTΛT],Σ59=[D2TΛTHTΛT],Σ68=[ϵD^f],Σ78=-[ϵB^f0]. If (X,Y,Q,Qi) is a feasible solution of (15), there exist nonsingular matrices U and V satisfying UV=I-XY-1. The fault detection filter parameters in the form of (7) are given as follows: (39)Af=U-1A^fY-1V-1,Bf=U-1B^f,Cf=C^fY-1V-1,Df=D^f.

Proof.

When considering the system with uncertain parameters, we need to replace A~, D~, C~, D~f with A~+ΔA, D~+ΔD, C~+ΔC, D~f+ΔDf in (32). Then, (32) changes into Σ+MF(k)N+NTFT(k)MT<0, where M=(0,0,,0,D^fT,B^fT,0)T; N=(ΛC,ΛC,0,,0,ΛD2,ΛH,0,0,0).

Using Schur complement and lemma, we can derive (37). This completes the proof.

4. Illustrative Example

In this section, a numerical example will be presented to illustrate the effectiveness of our results.

Example 1.

Consider the system (2) with the parameters listed as follows : (40)A=[0.9719-0.0013-0.03400.8628],A1=[0.140.200.2],A2=,D1=[0.1000.3],D2=[00.1],C=[10.1],G=[-0.08390.0761]T,B=0,H=0,α-1=0.9,α-2=0.7,γ-=0.8,g1(k,x(k))=0.5x1(k)sin(x2(k)),g2(k,x(k))=0.5x2(k)sin(x1(k)),ε(k)=1,G1=[0.25000.26],G2=[0.45000.45].

Let the time-varying communication delays satisfy 1τi(k)3(i=1,2). For the measurement quantization, the parameters of the logarithmic quantizer are set as μ^0=2 and χ=0.8. Then, the fault detection filter parameters can be obtained from Corollary 11 as follows.

For the parameters listed previously, by Corollary 11 in our paper, we can obtain the following feasible filtering parameters: (41)Af=[-0.2524-0.0325-0.0325-0.1975],Bf=[-3.4252-0.4137],Cf=[0.0065-0.0045],Df=[0.0122], and the performance index between the robustness and sensitivity is γ=3.8382.

Example 2.

To show the usefulness and effectiveness of the designed fault detection filter, let the external disturbance be w(k)=0. Figures 1 and 2 show the residual signal r(k) and evolution of residual function J(k), respectively, when the fault signal f(k) is given as (42)f(k)={1,40k800,else. It indicates that the designed filter can detect the fault effectively. In , the fault can be detected in 6 time steps after its occurrence. However, from Figure 2, we can see that the fault can be detected in 4 time steps after its occurrence. Therefore, it can be seen that the residual can not only reflect the fault in time but also detect the fault without confusing it with the disturbance w(k).

Residual signal without w(k).

Evolution of residual function J(k) without w(k).

5. Conclusion

The fault detection filter design for stochastic systems with time-varying delays has been investigated in this research. Based on the Lyapunov functional method, sufficient conditions are obtained to ensure that the error systems are mean-square robustly asymptotically stable, and then the filters are designed in terms of LMIs. Numerical example has been given to illustrate the effectiveness of the proposed main results. The foregoing results have the potential to be useful for the study of stochastic systems.

Acknowledgments

This research was supported by the National Basic Research Program of China (2010 CB732501), the Fund of Sichuan Provincial Key Laboratory of Signal and Information Processing (SZJJ2009-002), and the Fund of Sichuan Provincial Key Laboratory of Signal and Information Processing (SGXZD0101-10-1).

Zhou S. Feng G. H filtering for discrete-time systems with randomly varying sensor delays Automatica 2008 44 7 1918 1922 2-s2.0-45849112158 10.1016/j.automatica.2007.10.026 Gao H. Wang C. Delay-dependent robust H and 2- filtering for a class of uncertain nonlinear time-delay systems IEEE Transactions on Automatic Control 2003 48 9 1661 1665 2-s2.0-0141638540 10.1109/TAC.2003.817012 Fridman E. Shaked U. Xie L. Robust H filtering of linear systems with time-varying delay IEEE Transactions on Automatic Control 2003 48 1 159 165 2-s2.0-0037247024 10.1109/TAC.2002.806674 Li H. Fu M. Linear matrix inequality approach to robust H filtering IEEE Transactions on Signal Processing 1997 45 9 2338 2350 2-s2.0-0031232943 10.1109/78.622956 Liu Y. Wang Z. Liang J. Liu X. Synchronization and state estimation for discrete-time complex networks with distributed delays IEEE Transactions on Systems, Man, and Cybernetics B 2008 38 5 1314 1325 2-s2.0-52349098268 10.1109/TSMCB.2008.925745 Grigoriadis K. M. Watson J. T. Jr. Reduced-order H and 2- filtering via linear matrix inequalities IEEE Transactions on Aerospace and Electronic Systems 1997 33 4 1326 1338 2-s2.0-0031258540 10.1109/7.625133 Frank P. M. Analytical and qualitative model-based fault diagnosis-a survey and some new results European Journal of Control 1996 2 1 6 28 2-s2.0-0345929238 10.1016/S0947-3580(96)70024-9 ZBL0857.93015 Frank P. M. Fault diagnosis in dynamic systems using analytical and knowledge-based redundancy. A survey and some new results Automatica 1990 26 3 459 474 2-s2.0-0025421966 10.1016/0005-1098(90)90018-D ZBL0713.93052 Daubechies I. Wavelet transform, time-frequency localization and signal analysis IEEE Transactions on Information Theory 1990 36 5 961 1005 2-s2.0-0025482241 10.1109/18.57199 ZBL0738.94004 Wang Z. Ho D. W. C. Filtering on nonlinear time-delay stochastic systems Automatica 2003 39 1 101 109 2-s2.0-0037210862 10.1016/S0005-1098(02)00178-4 ZBL1010.93099 Zhang X. Han Q. Robust H filtering for a class of uncertain linear systems with time-varying delay Automatica 2008 44 1 157 166 2-s2.0-37549025769 10.1016/j.automatica.2007.04.024 Xu S. Lam J. Gao H. Zou Y. Robust H filtering for uncertain discrete stochastic systems with time delays Circuits, Systems, and Signal Processing 2005 24 6 753 770 2-s2.0-31144469461 10.1007/s00034-005-0921-1 Dong H. Wang Z. Gao H. On design of quantized fault detection filters with randomly occurring nonlinearities and mixed time-delays Signal Processing 2012 92 4 1117 1125 2-s2.0-83955161121 10.1016/j.sigpro.2011.11.002 Xu S. Chen T. Robust H filtering for uncertain stochastic time-delay systems Asian Journal of Control 2003 5 3 364 373 2-s2.0-1442352218 Wang Z. Wei G. Feng G. Reliable H control for discrete-time piecewise linear systems with infinite distributed delays Automatica 2009 45 12 2991 2994 2-s2.0-70449632295 10.1016/j.automatica.2009.09.012 Chen Y. Xue A.-K. Improved stability criterion for uncertain stochastic delay systems with nonlinear uncertainties Electronics Letters 2008 44 7 458 459 2-s2.0-41149087377 10.1049/el:20083334 Chen W. Guan Z. Lu X. Delay-dependent exponential stability of uncertain stochastic systems with multiple delays: an LMI approach Systems and Control Letters 2005 54 6 547 555 2-s2.0-17844365843 10.1016/j.sysconle.2004.10.005 ZBL1129.93547 Luo M. Liu G. Zhong S. Robust fault detection of Markovian jump systems with different system modes Applied Mathematical Modelling 2013 37 7 5001 5012 10.1016/j.apm.2012.10.023 Fu M. Xie L. The sector bound approach to quantized feedback control IEEE Transactions on Automatic Control 2005 50 11 1698 1711 2-s2.0-28644435994 10.1109/TAC.2005.858689