
Phase-type distributions (PHDs), which are defined as the distribution of the lifetime up to the absorption in an absorbent Markov chain, are an appropriate candidate to model the lifetime of any system, since any non-negative probability distribution can be approximated by a PHD with sufficient precision. Despite PHD potential, friendly statistical programs do not have a module implemented in their interfaces to handle PHD. Thus, researchers must consider others statistical software such as R, Matlab or Python that work with the compilation of code chunks and functions. This fact might be an important handicap for those researchers who do not have sufficient knowledge in programming environments. In this paper, a new interactive web application developed with shiny is introduced in order to adjust PHD to an experimental dataset. This open access app does not require any kind of knowledge about programming or major mathematical concepts. Users can easily compare the graphic fit of several PHDs while estimating their parameters and assess the goodness of fit with just several clicks. All these functionalities are exhibited by means of a numerical simulation and modeling the time to live since the diagnostic in primary breast cancer patients.
Citation: Christian Acal, Elena Contreras, Ismael Montero, Juan Eloy Ruiz-Castro. A shiny app for modeling the lifetime in primary breast cancer patients through phase-type distributions[J]. Mathematical Biosciences and Engineering, 2024, 21(1): 1508-1526. doi: 10.3934/mbe.2024065
[1] | Piermarco Cannarsa, Rossana Capuani, Pierre Cardaliaguet . C1;1-smoothness of constrained solutions in the calculus of variations with application to mean field games. Mathematics in Engineering, 2019, 1(1): 174-203. doi: 10.3934/Mine.2018.1.174 |
[2] | Francesco Maddalena, Danilo Percivale, Franco Tomarelli . Signorini problem as a variational limit of obstacle problems in nonlinear elasticity. Mathematics in Engineering, 2024, 6(2): 261-304. doi: 10.3934/mine.2024012 |
[3] | Zeljko Kereta, Valeriya Naumova . On an unsupervised method for parameter selection for the elastic net. Mathematics in Engineering, 2022, 4(6): 1-36. doi: 10.3934/mine.2022053 |
[4] | Federico Cluni, Vittorio Gusella, Dimitri Mugnai, Edoardo Proietti Lippi, Patrizia Pucci . A mixed operator approach to peridynamics. Mathematics in Engineering, 2023, 5(5): 1-22. doi: 10.3934/mine.2023082 |
[5] | Fernando Farroni, Gioconda Moscariello, Gabriella Zecca . Lewy-Stampacchia inequality for noncoercive parabolic obstacle problems. Mathematics in Engineering, 2023, 5(4): 1-23. doi: 10.3934/mine.2023071 |
[6] | Emilio N. M. Cirillo, Giuseppe Saccomandi, Giulio Sciarra . Compact structures as true non-linear phenomena. Mathematics in Engineering, 2019, 1(3): 434-446. doi: 10.3934/mine.2019.3.434 |
[7] | L. De Luca, M. Ponsiglione . Variational models in elasticity. Mathematics in Engineering, 2021, 3(2): 1-4. doi: 10.3934/mine.2021015 |
[8] | Qiang Du, Tadele Mengesha, Xiaochuan Tian . Lp compactness criteria with an application to variational convergence of some nonlocal energy functionals. Mathematics in Engineering, 2023, 5(6): 1-31. doi: 10.3934/mine.2023097 |
[9] | Michiaki Onodera . Linear stability analysis of overdetermined problems with non-constant data. Mathematics in Engineering, 2023, 5(3): 1-18. doi: 10.3934/mine.2023048 |
[10] | Yan Li, Gang Tian, Xiaohua Zhu . Singular Kähler-Einstein metrics on Q-Fano compactifications of Lie groups. Mathematics in Engineering, 2023, 5(2): 1-43. doi: 10.3934/mine.2023028 |
Phase-type distributions (PHDs), which are defined as the distribution of the lifetime up to the absorption in an absorbent Markov chain, are an appropriate candidate to model the lifetime of any system, since any non-negative probability distribution can be approximated by a PHD with sufficient precision. Despite PHD potential, friendly statistical programs do not have a module implemented in their interfaces to handle PHD. Thus, researchers must consider others statistical software such as R, Matlab or Python that work with the compilation of code chunks and functions. This fact might be an important handicap for those researchers who do not have sufficient knowledge in programming environments. In this paper, a new interactive web application developed with shiny is introduced in order to adjust PHD to an experimental dataset. This open access app does not require any kind of knowledge about programming or major mathematical concepts. Users can easily compare the graphic fit of several PHDs while estimating their parameters and assess the goodness of fit with just several clicks. All these functionalities are exhibited by means of a numerical simulation and modeling the time to live since the diagnostic in primary breast cancer patients.
Along with the rapid development of biology and biotechnology, the regulation and interaction of mRNAs and the produced proteins in genetic regulatory networks (GRNs) have attracted great research attention, and various related applications have been available to address the burgeoning areas [1,2,3]. Furthermore, much effort has been devoted to the biochemical dynamics of GRNs, which can better describe the expression mechanisms of genes from a dynamical system perspective [4,5]. The dynamical behaviors of GRNs can considerably provide deeper insights into the processing conditions. Generally speaking, differential equation and Boolean network models are employed and investigated to characterize complex properties of mRNA and protein concentrations in GRNs [6,7,8]. In particular, biological evidences have proved the existences of jumping mechanisms of GRNs. This means that the dynamical parameters of GRNs are varying according to different modes rather than kept constant owing to random abrupt changes during gene expressions. This means that the parameters of GRNs should be modified according to certain modes. With this context, mathematical models of Markovian jumping systems with Markov chains have been widely employed to mimic the intrinsic variate of GRNs [9,10,11]. As a result, many remarkable results on Markovian jumping GRNs have been reported in the literature, which includes stochastic state estimation [12], stability [13] and synchronization issues [14].
In addition, the accessibility of true jumping mode information of GRNs has also drawn much interest, especially for some actual applications. It is noteworthy that the exact mode information of GRNs in practical applications is always difficult or expensive to acquire, which is mostly because of detection delays or unpredictable disturbances during the transcription and translation processes [15,16]. Especially, the mode information characterized by Markov chains for state estimation or control issues is also challenging to main synchronous design. On the other hand, nonsynchronous analysis and synthesis methods are more meaningful and practical for Markovian jumping GRNs with mode-dependent designs. Lately, hidden Markovian jumping systems have been initially investigated by external mode observations and encouraging results of the so-called nonsynchronous designs have been reported accordingly. More precisely, on the basis of conditional probability between true modes and observed modes, the observed models for mode information are effectively utilized without synchronous mode operations [17,18,19].
On another active research line, distributed strategies for control systems have been extensively developed due to their increasing requirements for efficiency and robustness in engineering areas. Compared with traditional control techniques, the distributed architecture can process and fuse the information via multiple sensors with each domain data. With respect to its advances, many remarkable analysis and synthesis of distributed control systems can be found in recent works [20,21,22]. To name a few, in [23], a novel distributed-observer-based distributed control design is developed for affine nonlinear systems and is effectively applied to interconnected cruise control of intelligent vehicles. In [24], the distributed observer and controller design issue is discussed for spatially interconnected systems. In [25], the distributed state observers and disturbance observers are proposed for cooperative output regulation of LTI system to ensure the reference tracking properly. In [26], a distributed Nash equilibrium seeking scheme is designed with nonlinear dynamical system. In [27], the leader-following output consensus problem is concerned with distributed nonlinear observers. These notable works have proven that distributed frameworks can effective improve the system performance for complex dynamical systems. However, the problems of network communication burden must be faced among the nodes. It should be emphasized that traditional time-triggered periodic communication schedules have certain drawbacks in computation and network resources. Fortunately, the emerging event-triggered schemes are developed aiming at necessary transmission only when speechified triggering conditions are satisfied [28,29,30]. Hence, the event-triggered mechanism could lead to less network consumptions and could be more reliable in line with practical network implementation. Furthermore, in view of distributed control systems, distinguishing advantages can also been made by adopting event-triggered communications [31,32,33]. Nevertheless, there are challenging problems for distributed event-triggered control in corresponding algorithm complexity. For the consideration of GRNs with HMJPs, it is reasonable and significant to apply the distributed event-triggered design. Unfortunately, until now, distributed event-triggered state estimation problem of GRNs has not been entirely studied, not to mention the case of HMJPs with NSE. This motivates the authors for current study to shorten such a gap.
To deal with the aforementioned issue, the key idea in this paper is to investigate the distributed event-triggered state estimation problem for GRNs with HMJPs based on the NSE scheme. Compared with most existing literature, the primary novelties of our work are highlighted as follows:
1) The attempt to deal with the distributed NSE design of GRNs with HMJPs is made for the first time, which can better model GRNs dynamics and mode jumping phenomena during the state estimation procedures.
2) The distributed nonsynchronous state estimation framework is also introduced for GRNs with nonsynchronous event-triggered communications, such that more estimation efficiency for concentrations of mRNAs and proteins can be achieved with effective mode information utilization.
3) The sufficient mode-dependent analysis criteria are established by novel Lyapunov-Krasovski functions, and then the desired nonsynchronous mode-dependent estimator gains can be derived with convex optimization.
The remainder of this paper is stated by the following lines. Section 2 introduces some background information regarding GRNs with HMJPs and formulates the state estimation problem. Subsequent session of Section 3 presents the main theoretical developments with details. Section 4 approves the correctness of our developed results through a numerical example. Section 5 concludes the paper and addresses future challenges.
All referred notations are standardly utilized. Rn stands for n-Euclidean space. Matrix P>0 means that P is symmetric positive definite, and vise versa. (O,F,P) corresponds a probability space, where O denotes sample space, F represents δ-algebra of subsets within O and P stands for probability measurement. E means the mathematical expectation and ∗ implies a induced term in symmetry matrices, respectively.
Consider the equilibrium points of GRNs model shifted to the origin and the following GRNs model with HMJPs can be described by:
(ΣM):{˙xm(t)=−A(r(t))xm(t)+B(r(t))g(xp(t−dp(t)))+Em(r(t))ω1(t),˙xp(t)=−C(r(t))xp(t)+D(r(t))xm(t−dm(t))+Ep(r(t))ω2(t),zm(t)=Um(r(t))xm(t),zp(t)=Up(r(t))xp(t),xm(t)=φm(t),xp(t)=φp(t),∀t∈[−max(ˉdp,ˉdm),0), | (2.1) |
where xm(t)=[xm1(t),xm2(t),…,xmn(t)]T∈Rn, xp(t)=[xp1(t),xp2(t),…,xpn(t)]T∈Rn with xmi(t) and xpi(t), i=1,2,…,n representing the concentrations of mRNAs and proteins, respectively; zm(t)=[zm1(t),zm2(t),…,zml(t)]T∈Rl, zp(t)=[zp1(t),zp2(t),…,zpl(t)]T∈Rl with zmk(t) and zpk(t), k=1,2,…,l being the concentrations of certain mRNAs and proteins to be estimated through expression level; g(xp(t−dp(t)))=[g1(xp1(t−dp(t))),g2(xp2(t−dp(t))),…,gn(xpn(t−dp(t)))]T∈Rn stands for the nonlinear feedback regulation function of the protein on transcription, which satisfies the following assumption
0≤gi(xi)xi≤ςi,xi≠0;gi(0)=0,i=1,2,…,n,ς=diag{ς1,ς2,…,ςn}>0. |
i.e.,
gT(x)(g(x)−ςx)≤0; |
0≤dp(t)≤ˉdp and 0≤dm(t)≤ˉdm are the translation delay and the feedback regulation delay, respectively; ω1(t) and ω2(t) refers to the external disturbances that belongs to L2[0,∞).
The degradation or dilution rates of mRNAs and proteins can be described by known matrices A(r(t))=diag{a1(r(t)),a2(r(t)),…,an(r(t))}, C(r(t))=diag{c1(r(t)),c2(r(t)),…,cn(r(t))}, D(r(t))=diag{d1(r(t)),d2(r(t)),…,dn(r(t))}, and the coupling matrix B(r(t))=(bij(r(t)))∈Rn can be defined by the following
bij(r(t))={ˉbij(r(t))0−ˉbij(r(t))iftranscriptionfactorjisanactivatorofgenei,ifthereisnolinkfromnodejtoi,iftranscriptionfactorjisanrepressorofgenei. |
Furthermore, Um(r(t)), Up(r(t)), Em(r(t)) and Ep(r(t)) are also known matrices for a certain mode r(t).
Moreover, the continuous-time discrete-state Markov process r(t)∈S={1,…,N} on (O,F,P) is given to model the HMJPs, which can be defined by Π=(πij)N×N, with
Pr(r(t+Δ)=j|r(t)=i)={πijΔ+o(Δ), i≠j,1+πiiΔ+o(Δ), i=j,, | (2.2) |
and πii=−N∑j=1,i≠jπij.
In this paper, a direct graph G={V,E,A} is given to describe the communication topology of N sensor nodes, where V={v1,v2,⋯,vN} and E⊆V×V stand for the set of nodes and edges, respectively, A=[aij]∈RN×N represent the adjacency matrix with aii=0 for any i. A is associated with the edges of G are positive, i.e., aij>0⟺εij∈E. In addition, the corresponding Laplacian matrix of G is defined as L.
Moreover, the distributed measurement is achieved by N mode-dependent sensor nodes as follows:
(ΣS):{ylm(t)=Hlm(r(t))xm(t)+Glm(r(t))ω3(t),ylp(t)=Hlp(r(t))xp(t)+Glp(r(t))ω4(t), | (2.3) |
where ylm(t) and ylp(t), l=1,2,…,N, represent the measurements of mRNAs and proteins on sensor l, respectively; ω3(t) and ω4(t) are measurement disturbances; Hlm(r(t)), Hlp(r(t)), Glm(r(t)), Glp(r(t)) are known constant matrices.
Then, define the sensor estimation errors ˜ylm(t) and ˜ylp(t) as follows
{˜ylm(t)=ylm(t)−Hlm(r(t))ˆxlm(t),˜ylp(t)=ylp(t)−Hlp(r(t))ˆxlp(t), | (2.4) |
where ˆxlm(t) and ˆxlp(t) denote the filter state for sensor l.
By considering the event-triggered strategy for data transmission with ZOHs, it is assumed that the sampling period hk with sampling instants tk is assumed to be hk=tk+1−tk≤ˉh, ˉh>0. Furthermore, the transmission instant tlδ for sensor l is determined by the event generation functions as follows:
tlδ+1=mintk>tlδ{tk‖˜yl(tk)−˜yl(tlδ)‖≥κ‖˜yl(tlδ)‖}, | (2.5) |
where ˜yl(tk)=[(˜ylm(t))T,(˜ylp(t))T]T and 0<κ<1 is a scalar parameter.
Consequently, the following distributed mode-dependent state estimation filter (ΣF) can be designed for system (ΣM):
(ΣF):{˙ˆxlm(t)=−A(r(t))ˆxm(t)+B(r(t))g(ˆxp(t−dp(t)))+Flm(σ(t))˜ylm(tlδ)+Klm(σ(t))N∑s=1als[˜ysm(tsδ∗)−˜ylm(tlδ)],˙ˆxlp(t)=−C(r(t))ˆxp(t)+D(r(t))ˆxm(t−dm(t))+Flp(σ(t))˜ylp(tlδ)+Klp(σ(t))N∑s=1als[˜ysp(tsδ∗)−˜ylp(tlδ)],ˆzlm(t)=Um(r(t))ˆxlm(t),ˆzlp(t)=Up(r(t))ˆxlp(t), | (2.6) |
where t∈[tlδ,tlδ+1), δ∗=argminμ{t−tsμ|t≥tsμ}, ˆzlm(t) and ˆzlp(t) stand for estimation of zm(t) and zp(t), respectively; Flm(σ(t)),Flp(σ(t)) and Klm(σ(t)),Klp(σ(t)) represent the nonsynchronous mode-dependent filter gains to be determined later. More precisely, σ(t)∈F={1,…,F} represents another Markov process corresponds to r(t) with
Pr{σ(t)=ρ|r(t)=i}=λiρ, |
and ∑Fρ=1λiρ=1.
Remark 1. As event-triggered control strategy for networked control systems becomes more important, it is reasonable to investigate the analysis and synthesis issues for GRNs over sensor networks with event-triggered schemes. In comparison with common time-triggered strategies with extensive data transmissions, the event-triggered framework can considerably reduce the communication burden by generated events. Moreover, our developed event-triggered mechanism is related with both measurements of mRNAs and proteins, which is more efficient for potential applications.
Then, it can be deduced by dividing [tlδ,tlδ+1) into ∪tlδ+1−hk+1ιk=tlδ[ιk,ιk+1) that
{˙ˆxlm(t)=−A(r(t))ˆxlm(t)+B(r(t))g(ˆxlp(t−dp(t)))+Flm(σ(t))˜ylm(ιk)−Flm(σ(t))εlm(ιk)+Klm(σ(t))N∑s=1als[˜ysm(ιk)−˜ylm(ιk)]−Klm(σ(t))N∑s=1als[εsm(ιk)−εlm(ιk)],˙ˆxlp(t)=−C(r(t))ˆxlp(t)+D(r(t))ˆxlm(t−dm(t))+Flp(σ(t))˜ylp(ιk)−Flp(σ(t))εlp(ιk)+Klp(σ(t))N∑s=1als[˜ysp(ιk)−˜ylp(ιk)]−Klp(σ(t))N∑s=1als[εsp(ιk)−εlp(ιk)],ˆzlm(t)=Um(r(t))ˆxlm(t),ˆzlp(t)=Up(r(t))ˆxlp(t), | (2.7) |
where εlm(ιk)=˜ylm(ιk)−˜ylm(tlδ) and εlp(ιk)=˜ylp(ιk)−˜ylp(tlδ), tlδ≤ιk<tlδ+1, represent the event-triggered measurement errors, respectively.
As a result, by denoting elm(t)=x(t)−ˆxlm(t), elp(t)=x(t)−ˆxlp(t), ˜zlm(t)=z(t)−ˆzlm(t) and ˜zlp(t)=z(t)−ˆzlp(t), one can deduce that
{˙elm(t)=−A(r(t))elm(t)+B(r(t))g(xp(t−dp(t)))−B(r(t))g(ˆxlp(t−dp(t)))−Flm(σ(t))Hlm(r(t))elm(ιk)+Flm(σ(t))εlm(ιk)−Klm(σ(t))N∑s=1als[Hsm(r(t))esm(ιk)−Hlm(r(t))elm(ιk)]+Km(σ(t))N∑s=1als[εsm(ιk)−εlm(ιk)]+Em(r(t))ω1(t)−Flm(σ(t))Glm(r(t))ω3(ιk)−Klm(σ(t))N∑s=1als[Gsm(r(t))ω3(ιk)−Glm(r(t))ω3(ιk)],˙elp(t)=−C(r(t))elp(t)+D(r(t))elm(t−dm(t))−Flp(σ(t))Hlp(r(t))elp(ιk)+Flp(σ(t))εlp(ιk)−Klp(σ(t))N∑s=1als[Hsp(r(t))esp(ιk)−Hlp(r(t))elp(ιk)]+Klp(σ(t))N∑s=1als[εsp(ιk)−εlp(ιk)]+Ep(r(t))ω2(t)−Flp(σ(t))Glp(r(t))ω4(ιk)−Klp(σ(t))N∑s=1als[Gsp(r(t))ω4(ιk)−Glp(r(t))ω4(ιk)],˜zlm(t)=Um(r(t))elm(t),˜zlp(t)=Up(r(t))elp(t), | (2.8) |
For simplicity, the resulting closed-loop filter system can be rewritten as follows:
{˙em(t)=−ˉA(i)em(t)+ˉB(i)ˉg(ep(t−dp(t)))−(ˉFm(ρ)−ˉKm(ρ)ˉL)ˉHm(i)em(ιk)+(ˉFm(ρ)−ˉKm(ρ)ˉL)εm(ιk)+ˉEm(i)ω1(t)−(ˉFm(ρ)−ˉKm(ρ)ˉL)ˉGm(i)ω3(ιk),˙ep(t)=−ˉC(i)ep(t)+ˉD(i)em(t−dm(t))−(ˉFp(ρ)−ˉKp(ρ)ˉL)ˉHp(i)ep(ιk)+(ˉFp(ρ)−ˉKp(ρ)ˉL)εp(ιk)+ˉEp(i)ω2(t)−(ˉFp(ρ)−ˉKp(ρ)ˉL)ˉGp(i)ω4(ιk),˜zm(t)=ˉUm(i)em(t),˜zp(t)=ˉUp(i)ep(t), t∈[ιk,ιk+1), | (2.9) |
where
em(t)=[eTm1(t),eTm2(t),…,eTmN(t)]T,ep(t)=[eTp1(t),eTp2(t),…,eTpN(t)]T,˜zm(t)=[˜zTm1(t),˜zTm2(t),…,˜zTmN(t)]T,˜zp(t)=[˜zTp1(t),˜zTp2(t),…,˜zTpN(t)]T,εm(t)=[εTm1(t),εTm2(t),…,εTmN(t)]T,εp(t)=[εTp1(t),εTp2(t),…,εTpN(t)]T,ˉA(i)=IN⊗A(i),ˉB(i)=IN⊗B(i),g(ep(t))=[(g(ep(t))−g(ˆep1(t)))T,(g(ep(t))−g(ˆep2(t)))T,…,(g(ep(t))−g(ˆep2(t)))T]T,ˉFm(ρ)=diag{F1m(ρ),F2m(ρ),…,FNm(ρ)},ˉFp(ρ)=diag{F1p(ρ),F2p(ρ),…,FNp(ρ)},ˉKm(ρ)=diag{K1m(ρ),K2m(ρ),…,KNm(ρ)},ˉKp(ρ)=diag{K1p(ρ),K2p(ρ),…,KNp(ρ)},ˉHm(i)=diag{H1m(i),H2m(i),…,HNm(i)},ˉHp(i)=diag{H1p(i),H2p(i),…,HNp(i)},ˉEm(i)=1N⊗Em(i),ˉEp(i)=1N⊗Ep(i),ˉUm(i)=1N⊗Um(i),ˉUp(i)=1N⊗Up(i),ˉGm(i)=[(G1m(i))T,(G2m(i))T,…,(GNm(i))T]T,ˉGp(i)=[(G1p(i))T,(G2p(i))T,…,(GNp(i))T]T,ˉL=L⊗In. |
Remark 2. It is noticed that the adoption of conditional probability has been proved to be effective to cope with the Markov or semi-Markov jump systems with limited mode information. For the GRNs with hidden Markovian jumping parameters, lack of true operation modes of Markov chains would lead to control performance degradation or even unstable with mode-dependent designs. In this paper, the nonsynchronous mode information corresponds to the distributed sensor mode information, which implies that mode information mismatch between true GRNs model and our developed distributed state estimators is considered during the design. As a consequence, we have developed a novel distributed nonsynchronous filter framework, where the observed Markovian jumping modes are utilized for more practical distributed event-triggered state estimators.
Consequently, by applying the input-delay approach, it can be obtained that
{˙em(t)=−ˉA(i)em(t)+ˉB(i)ˉg(ep(t−dp(t)))−(ˉFm(ρ)−ˉKm(ρ)ˉL)ˉHm(i)em(t−τ(t))+(ˉFm(ρ)−ˉKm(ρ)ˉL)εm(t−τ(t))+ˉEm(i)ω1(t)−(ˉFm(ρ)−ˉKm(ρ)ˉL)ˉGm(i)ω3(t−τ(t)),˙ep(t)=−ˉC(i)ep(t)+ˉD(i)em(t−dm(t))−(ˉFp(ρ)−ˉKp(ρ)ˉL)ˉHp(i)ep(t−τ(t))+(ˉFp(ρ)−ˉKp(ρ)ˉL)εp(t−τ(t))+ˉEp(i)ω2(t)−(ˉFp(ρ)−ˉKp(ρ)ˉL)ˉGp(i)ω4(t−τ(t)),˜zm(t)=ˉUm(i)em(t),˜zp(t)=ˉUp(i)ep(t), t∈[ιk,ιk+1), | (2.10) |
where τ(t):=t−ιk with 0≤τ(t)≤ˉτ.
The aim of our paper is to design the desired nonsynchronous state estimator gains, such that the modified H∞ performance γ can be satisfied in the mean-square sense for augmented state estimation error system with zero initial conditions, if it can hold that
E{∫∞0(˜zTm(t)˜zm(t)+˜zTp(t)˜zp(t))dt}≤γ∫∞0(ωT1(t)ω1(t)+ωT2(t)ω2(t)+ωT3(t−τ(t))ω3(t−τ(t))+ωT4(t−τ(t))ω4(t−τ(t))dt. | (2.11) |
In the sequel, the following lemmas are employed for later theoretical derivations.
Lemma 1. [34] For any matrix M>0, scalars ˉξ>0, ξ(t) satisfying 0≤ξ(t)≤ˉξ, vector function ˙x(t):[−ˉξ,0]→Rn such that the concerned integrations are well defined in (2.12), then
−τ∫tt−ˉξ˙xT(s)M˙x(s)ds≤ζT(t)Uζ(t), | (2.12) |
where
ζ(t)=[xT(t),xT(t−ξ(t)),xT(t−ˉξ)]T,U=[−MM0∗−2MM∗∗−M]. |
Lemma 2. [35] If there exists a parameter ε and real matrices X, Y, Z, W, and it holds that
[XZ+εWT∗−εYT−εY]<0, | (2.13) |
then one has
X+ZTε−TWT+Wε−1Z<0. | (2.14) |
In this section, sufficient modified H∞ performance conditions are first established for the augmented state estimation error system and the distributed state estimator gains are further designed in terms of convex optimization.
Theorem 1. For given ˉdp, ˉdm, ˉτ and mode-dependent filter gains Flm(ρ),Flp(ρ) and Klm(ρ),Klp(ρ), the modified H∞ filtering can be satisfied over distributed sensor networks according to Definition 1, if there exist mode-dependent matrices Pm(i)>0, Pp(i)>0, and matrices Qm>0, Qp>0, Qτ>0, Rm>0, Rp>0, Rτ>0, such that Ξi<0, for all i∈N, k=1,2,…,N and ρ∈F, k=1,2,…,F, where
Ξ(i)=[Ξ1(i)Ξ2(i)∗Ξ3(i)],Ξ1(i)=[Ξ11(i)Ξ12(i)∗Ξ13(i)],Ξ11(i)=[Ξ111(i)Ξ112(i)00Rm00∗Ξ113(i)Rτ0000∗∗−Qτ−Rτ0000∗∗∗−Qm−RmRTm00∗∗∗∗−2Rm00∗∗∗∗∗−Qp−RpRp∗∗∗∗∗∗−2Rp],Ξ111(i)=−2Pm(i)ˉA(i)+Qm+Qτ−Rm−Rτ+ˉUTm(i)ˉUm(i)+∑j∈NπijPm(j),Ξ112(i)=−F∑ρ=1λiρPm(i)ˉFm(ρ)ˉHm(i)+F∑ρ=1λiρPm(i)ˉKm(ρ)ˉLˉHm(i)+Rτ,Ξ113(i)=−2Rτ+κ2ˉHTm(i)ˉHm(i),Ξ12(i)=[000Pm(i)ˉB(i)Ξ111(i)0000−κ2ˉHTm(i)0000000000ˉDT(i)Pp(i)000000000Rp00ς0],Ξ121(i)=F∑ρ=1λiρPm(i)ˉFm(ρ)−F∑ρ=1λiρPm(i)ˉKm(ρ)ˉL,Ξ13(i)=[Ξ131(i)Ξ132(i)000∗Ξ133(i)Rτ00∗∗−Rτ00∗∗∗−2I0∗∗∗∗κ2I−I],Ξ131(i)=−2Pp(i)ˉC(i)+Qp+Qτ−Rp−Rτ+ˉUTp(i)ˉUp(i)+∑j∈NπijPp(i),Ξ132(i)=−F∑ρ=1λiρPp(i)ˉFp(ρ)ˉHp(i)+F∑ρ=1λiρPp(i)ˉKp(ρ)ˉLˉHp(i)+Rτ,Ξ133(i)=−Qτ−2Rτ+κ2ˉHTp(i)ˉHp(i),Ξ2(i)=[Ξ21(i),Ξ22(i)],Ξ21(i)=[0Pm(i)ˉEm(i)0Ξ211(i)0000κ2ˉHTm(i)ˉGm(i)00000000000000000000000000],Ξ22(i)=[−ˉdmˉAT(i)0−ˉτˉAT(i)0Ξ212(i)0Ξ213(i)0000000000ˉdpˉDT(i)0ˉτˉDT(i)00000000],Ξ211(i)=−F∑ρ=1λiρPm(i)ˉFm(ρ)ˉGm(i)+F∑ρ=1λiρPm(i)ˉKm(ρ)ˉLˉGm(i),Ξ212(i)=−ˉdmF∑ρ=1λiρˉHTm(i)ˉFTm(ρ)+ˉdmF∑ρ=1λiρˉHTm(i)ˉLTˉKTm(ρ),Ξ213(i)=−ˉτF∑ρ=1λiρˉHTm(i)ˉFTm(ρ)+ˉτF∑ρ=1λiρˉHTm(i)ˉLTˉKTm(ρ), |
and
Ξ3(i)=[Ξ31(i)Ξ32(i)∗Ξ33(i)],Ξ31(i)=[Ξ311(i)0Pp(i)ˉEp(i)0Ξ312(i)−κ2ˉHTp(i)000κ2ˉHTp(i)ˉGp(i)0000000000000−ˉGm(i)0κ2I−I000−ˉGp(i)∗−γI000∗∗−γI00∗∗∗κ2ˉGTm(i)ˉGm(i)−γI0∗∗∗∗κ2ˉGTp(i)ˉGp(i)−γI],Ξ311(i)=F∑ρ=1λiρPp(i)ˉFp(ρ)−F∑ρ=1λiρPp(i)ˉKp(ρ)ˉL,Ξ312(i)=−F∑ρ=1λiρPp(i)ˉFp(ρ)ˉGp(i)+F∑ρ=1λiρPp(i)ˉKp(ρ)ˉLˉGp(i),Ξ32(i)=[0−ˉdpˉCT(i)0−ˉτˉCT(i)0Ξ321(i)0Ξ322(i)0000ˉdmˉBT(i)0ˉτˉBT(i)0Ξ323(i)0Ξ324(i)00Ξ325(i)0Ξ326(i)ˉdmˉEm(i)0ˉτˉEm(i)00ˉdpˉETp(i)0ˉτˉETp(i)],Ξ321(i)=−ˉdpF∑ρ=1λiρˉHTp(i)ˉFTp(ρ)+ˉdpF∑ρ=1λiρˉHTp(i)ˉLTˉKTp(ρ),Ξ322(i)=−ˉτF∑ρ=1λiρˉHTp(i)ˉFTp(ρ)+ˉτF∑ρ=1λiρˉHTp(i)ˉLTˉKTp(ρ),Ξ323(i)=ˉdmF∑ρ=1λiρˉFTm(ρ)−ˉdmF∑ρ=1λiρˉLTˉKTm(ρ),Ξ324(i)=ˉτF∑ρ=1λiρˉFTm(ρ)−ˉτF∑ρ=1λiρˉLTˉKTm(ρ),Ξ325(i)=ˉdpF∑ρ=1λiρˉFTp(ρ)−ˉdpF∑ρ=1λiρˉLTˉKTp(ρ),Ξ326(i)=ˉτF∑ρ=1λiρˉFTp(ρ)−ˉτF∑ρ=1λiρˉLTˉKTp(ρ),Ξ33(i)=[Ξ331(i)0Ξ332(i)00Ξ333(i)0Ξ334(i)−R−1m000∗−R−1p00∗∗−R−1τ0∗∗∗−R−1τ],Ξ331(i)=−ˉdmF∑ρ=1λiρˉGTm(i)ˉFTm(ρ)+ˉdmF∑ρ=1λiρˉGTm(i)ˉLTˉKTm(ρ),Ξ332(i)=−ˉτF∑ρ=1λiρˉGTm(i)ˉFTm(ρ)+ˉτF∑ρ=1λiρˉGTm(i)ˉLTˉKTm(ρ),Ξ333(i)=−ˉdpF∑ρ=1λiρˉGTp(i)ˉFTp(ρ)+ˉdpF∑ρ=1λiρˉGTp(i)ˉLTˉKTp(ρ),Ξ334(i)=−ˉτF∑ρ=1λiρˉGTp(i)ˉFTp(ρ)+ˉτF∑ρ=1λiρˉGTp(i)ˉLTˉKTp(ρ). |
Proof. Construct the following mode-dependent Lyapunov-Krasovski functions:
V(i,t)=V1(i,t)+V2(i,t)+V3(i,t), | (3.1) |
where
V1(i,t)=eTm(t)Pm(i)em(t)+eTp(t)Pp(i)ep(t),V2(i,t)=∫tt−ˉdmeTm(s)Qmem(s)ds+∫tt−ˉdpeTp(s)Qpep(s)ds+∫tt−ˉτ(eTm(s)Qτem(s)+eTp(s)Qτep(s))ds,V3(i,t)=ˉdm∫0−ˉdm∫tt+φeTm(φ)Rmem(φ)dφds+ˉdp∫0−ˉdp∫tt+φeTp(φ)Rpep(φ)dφds+ˉτ∫0−ˉτ∫tt+φ(eTm(φ)Rτem(φ)+eTp(φ)Rτep(φ))dφds. |
Then, for all r(t)=i and σ(t)=ρ, by defining the weak infinitesimal operator L of V(i,t) as follows
LV(i,t)=limΔ→0+1Δ{E{V(r(t+Δ),t+Δ)|r(t)=i,t}−V(i,t)}, | (3.2) |
it can be deduced that
LV1(i,t)=E{2eTm(t)Pm(i)˙em(t)+2eTp(t)Pp(i)˙ep(t)+∑j∈NπijeTm(t)Pm(j)em(t)+∑j∈NπijeTp(t)Pp(i)ep(t)}=2eTm(t)[−Pm(i)ˉA(i)]em(t)+2eTm(t)[Pm(i)ˉB(i)]ˉg(ep(t−dp(t)))+2F∑ρ=1λiρeTm(t)[−Pm(i)(ˉFm(ρ)+ˉKm(ρ)ˉL)ˉHm(i)]em(t−τ(t))+2F∑ρ=1λiρeTm(t)[Pm(i)(ˉFm(ρ)−ˉKm(ρ)ˉL)]εm(t−τ(t))+2eTm(t)[Pm(i)ˉEm(i)]ω1(t)+2F∑ρ=1λiρeTm(t)[−Pm(i)(ˉFm(ρ)−ˉKm(ρ)ˉL)ˉGm(i)]ω3(t−τ(t))+2eTp(t)[−Pp(i)ˉC(i)]ep(t)+2eTp(t)[Pp(i)ˉD(i)]em(t−dm(t))+2eTp(t)[Pp(i)ˉEp(i)]ω2(t)+2F∑ρ=1λiρeTp(t)[−Pp(i)(ˉFp(ρ)−ˉKp(ρ)ˉL)ˉHp(i)]ep(t−τ(t))+2F∑ρ=1λiρeTp(t)[Pp(i)(ˉFp(ρ)−ˉKp(ρ)ˉL)]εp(t−τ(t))+2F∑ρ=1λiρeTp(t)[−Pp(i)(ˉFp(ρ)−ˉKp(ρ)ˉL)ˉGp(i)]ω4(t−τ(t))+∑j∈NπijeTm(t)Pm(j)em(t)+∑j∈NπijeTp(t)Pp(i)ep(t). | (3.3) |
Similarly, it can be deduced that
LV2(i,t)=eTm(t)Qmem(t)−eTm(t−ˉdm)Qmem(t−ˉdm)+eTp(t)Qpep(t)−eTp(t−ˉdp)Qpep(t−ˉdp)+eTm(t)Qτem(t)+eTp(t)Qτep(t)−eTm(t−ˉτ)Qτem(t−ˉτ)−eTp(t−ˉτ)Qτep(t−ˉτ)=eTm(t)(Qm+Qτ)em(t)+eTp(t)(Qp+Qτ)ep(t)−eTm(t−ˉdm)Qmem(t−ˉdm)−eTp(t−ˉdp)Qmep(t−ˉdp)−eTm(t−ˉτ)Qτem(t−ˉτ)−eTp(t−ˉτ)Qτep(t−ˉτ), | (3.4) |
and
LV3(i,t)=E{ˉd2m˙eTm(t)Rm˙em(t)+ˉd2p˙eTp(t)Rp˙ep(t)+ˉτ2˙eTm(t)Rτ˙em(t)+ˉτ2˙eTp(t)Rτ˙ep(t)−ˉdm∫tt−ˉdm˙eTm(φ)Rm˙em(φ)dφ−ˉdp∫tt−ˉdp˙eTp(φ)Rp˙ep(φ)dφ−ˉτ∫tt−ˉτ˙eTm(φ)Rτ˙em(φ)dφ−ˉτ∫tt−ˉτ˙eTp(φ)Rτ˙ep(φ)dφ}=ψT(t)(ϝmRmϝTm+ϝpRpϝTp+ϝτRτϝTτ)ψ(t)−ˉdm∫tt−ˉdm˙eTm(φ)Rm˙em(φ)dφ−ˉdp∫tt−ˉdp˙eTp(φ)Rp˙ep(φ)dφ−ˉτ∫tt−ˉτ˙eTm(φ)Rτ˙em(φ)dφ−ˉτ∫tt−ˉτ˙eTp(φ)Rτ˙ep(φ)dφ, | (3.5) |
where ψT(t)=[eTm(t),eTm(t−τ(t)),eTm(t−ˉτ),eTm(t−ˉdm),eTm(t−dm(t)),eTp(t−ˉdp),eTp(t−dp(t)),eTp(t),eTp(t−τ(t)),eTp(t−ˉτ),ˉgT(ep(t−dp(t))),εTm(t−τ(t)),εTp(t−τ(t)),ωT1(t),ωT2(t),ωT3(t−τ(t)),ωT4(t−τ(t))]T and
ϝm=[−ˉdmˉAT(i)−ˉdm∑Fρ=1λiρˉHTm(i)(ˉFTm(ρ)−ˉLTˉKTm(ρ))00000000ˉdmˉBT(i)ˉdm∑Fρ=1λiρ(ˉFTm(ρ)−ˉLTˉKTm(ρ))0ˉdmˉEm(i)0−ˉdm∑Fρ=1λiρˉGTm(i)(ˉFTm(ρ)−ˉLTˉKTm(ρ))0],ϝp=[0000ˉdpˉDT(i)00−ˉdpˉCT(i)−ˉdp∑Fρ=1λiρˉHTp(i)(ˉFTp(ρ)−ˉLTˉKTp(ρ))000ˉdp∑Fρ=1λiρ(ˉFTp(ρ)−ˉLTˉKTp(ρ))0ˉdpˉETp(i)0−ˉdp∑Fρ=1λiρˉGTp(i)(ˉFTp(ρ)−ˉLTˉKTp(ρ))], |
ϝτ=[−ˉτˉAT(i)−ˉτ∑Fρ=1λiρˉHTm(i)(ˉFTm(ρ)−ˉLTˉKTm(ρ))00000000ˉτˉBT(i)ˉτ∑Fρ=1λiρ(ˉFTm(ρ)−ˉLTˉKTm(ρ))0ˉτˉEm(i)0−ˉτ∑Fρ=1λiρˉGTm(i)(ˉFTm(ρ)−ˉLTˉKTm(ρ))0]+[0000ˉτˉDT(i)00−ˉτˉCT(i)−ˉτ∑Fρ=1λiρˉHTp(i)(ˉFTp(ρ)−ˉLTˉKTp(ρ))000ˉτ∑Fρ=1λiρ(ˉFTp(ρ)−ˉLTˉKTp(ρ))0ˉτˉETp(i)0−ˉτ∑Fρ=1λiρˉGTp(i)(ˉFTp(ρ)−ˉLTˉKTp(ρ))]. |
In the sequel, by virtue of Lemma 1, one can obtain that
−ˉdm∫tt−ˉdm˙eTm(φ)Rm˙em(φ)dφ≤[eTm(t)eTm(t−dm(t))eTm(t−ˉdm)]T[−RmRm0∗−2RmRm∗∗−Rm][em(t)em(t−dm(t))em(t−ˉdm)], | (3.6) |
\begin{align} &-\bar{d}_{p}\int_{t-\bar{d}_{p}}^{t}\dot{e}_{p}^{T}\left( \varphi \right) R_{p}\dot{e}_{p}\left( \varphi \right) d\varphi \\ \leq &\left[ \begin{array}{c} e_{p}^{T}(t) \\ e_{p}^{T}(t-d_{p}(t)) \\ e_{p}^{T}(t-\bar{d}_{p}) \end{array} \right] ^{T}\left[ \begin{array}{ccc} -R_{p} & R_{p} & 0 \\ \ast & -2R_{p} & R_{p} \\ \ast & \ast & -R_{p} \end{array} \right] \left[ \begin{array}{c} e_{p}(t) \\ e_{p}(t-d_{p}(t)) \\ e_{p}(t-\bar{d}_{p}) \end{array} \right] \end{align} | (3.7) |
\begin{align} &-\bar{\tau}\int_{t-\bar{\tau}}^{t}\dot{e}_{m}^{T}\left( \varphi \right) R_{\tau }\dot{e}_{m}\left( \varphi \right) d\varphi \\ \leq &\left[ \begin{array}{c} e_{m}^{T}(t) \\ e_{m}^{T}(t-\tau (t)) \\ e_{m}^{T}(t-\bar{\tau}) \end{array} \right] ^{T}\left[ \begin{array}{ccc} -R_{\tau } & R_{\tau } & 0 \\ \ast & -2R_{\tau } & R_{\tau } \\ \ast & \ast & -R_{\tau } \end{array} \right] \left[ \begin{array}{c} e_{m}(t) \\ e_{m}(t-\tau (t)) \\ e_{m}(t-\bar{\tau}) \end{array} \right] , \end{align} | (3.8) |
\begin{align} &-\bar{\tau}\int_{t-\bar{\tau}}^{t}\dot{e}_{p}^{T}\left( \varphi \right) R_{\tau }\dot{e}_{p}\left( \varphi \right) d\varphi \\ \leq &\left[ \begin{array}{c} e_{p}^{T}(t) \\ e_{p}^{T}(t-\tau (t)) \\ e_{p}^{T}(t-\bar{\tau}) \end{array} \right] ^{T}\left[ \begin{array}{ccc} -R_{\tau } & R_{\tau } & 0 \\ \ast & -2R_{\tau } & R_{\tau } \\ \ast & \ast & -R_{\tau } \end{array} \right] \left[ \begin{array}{c} e_{p}(t) \\ e_{p}(t-\tau (t)) \\ e_{p}(t-\bar{\tau}) \end{array} \right] . \end{align} | (3.9) |
Then, it can be deduced that
\begin{align*} \mathcal{L}V(i,t)\leq \eta ^{T}\left( t\right)\Lambda \eta \left( t\right) \end{align*} |
where \eta ^{T}\left(t\right) = [\eta ^{T}_{1}\left(t\right), \eta ^{T} _{2}\left(t\right), \eta ^{T} _{3}\left(t\right), \eta ^{T} _{4}\left(t\right) ] with
\begin{align*} \eta ^{T} _{1}\left( t\right) = &\left[ \begin{array}{lllll} e ^{T}_{m}(t) & e ^{T}_{m}(t-\tau (t)) & e ^{T}_{m}(t-\bar{\tau}) & e ^{T}_{m}\left( t-\bar{d} _{m}\right) & e ^{T}_{m}\left( t-d_{m}(t)\right) \end{array} \right] , \\ \eta ^{T}_{2}\left( t\right) = &\left[ \begin{array}{lllll} e ^{T}_{p}\left( t-\bar{d}_{p}\right) & e ^{T}_{p}\left( t-d_{p}(t)\right) & e ^{T}_{p}(t) & e ^{T}_{p}(t-\tau (t)) & e ^{T}_{p}(t-\bar{\tau}) \end{array} \right] , \\ \eta ^{T}_{3}\left( t\right) = &\left[ \begin{array}{lll} \bar{g}^{T}\left( e _{p}\left( t-d_{p}(t)\right) \right) & \varepsilon ^{T} _{m}(t-\tau (t)) & \varepsilon ^{T} _{p}(t-\tau (t)) \end{array} \right] , \\ \eta ^{T} _{4}\left( t\right) = &\left[ \begin{array}{llll} \omega ^{T} _{1}(t) & \omega ^{T} _{2}(t) & \omega ^{T} _{3}(t-\tau (t)) & \omega ^{T} _{4}(t-\tau (t)) \end{array} \right] , \end{align*} |
and \Lambda = \Lambda _{1}+\Lambda _{2}R_{m}\Lambda _{2}^{T}+\Lambda _{3}R_{p}\Lambda _{3}^{T}+\Lambda _{4}R_{\tau }\Lambda _{4}^{T}+\Lambda _{5}R_{\tau }\Lambda _{5}^{T} with
\begin{align*} \Lambda _{1} = &\left[ \begin{array}{cc} \Lambda _{11} & \Lambda _{12} \\ \ast & \Lambda _{22} \end{array} \right] , \\ \Lambda _{11} = &\left[ \begin{array}{ccccccc} \Lambda _{111} & \Lambda _{112} & 0 & 0 & R_{m} & 0 & 0 \\ \ast & -2R_{\tau } & R_{\tau } & 0 & 0 & 0 & 0 \\ \ast & \ast & -Q_{\tau }-R_{\tau } & 0 & 0 & 0 & 0 \\ \ast & \ast & \ast & -Q_{m}-R_{m} & R_{m}^{T} & 0 & 0 \\ \ast & \ast & \ast & \ast & -2R_{m} & 0 & 0 \\ \ast & \ast & \ast & \ast & \ast & -Q_{m}-R_{p} & R_{p} \\ \ast & \ast & \ast & \ast & \ast & \ast & -2R_{p} \end{array} \right] , \\ \Lambda _{111} = &-2P_{m}(i)\bar{A}\left( i\right) +Q_{m}+Q_{\tau }-R_{m}-R_{\tau }+\sum\limits_{j\in \mathcal{N}}\pi _{ij}P_{m}(j), \\ \Lambda _{112} = &-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{m}(i)\bar{F} _{m}(\rho )\bar{H}_{m}(i)+\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{m}(i)\bar{K} _{m}(\rho )\bar{L}\bar{H}_{m}(i)+R_{\tau } \\ \Lambda _{12} = &\left[ \begin{array}{cccccccccc} 0 & 0 & 0 & P_{m}(i)\bar{B}\left( i\right) & \Lambda _{121} & 0 & P_{m}(i) \bar{E}_{m}\left( i\right) & 0 & \Lambda _{122} & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ \bar{D}^{T}\left( i\right) P_{p}(i) & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ R_{p} & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \end{array} \right] , \\ \Lambda _{121} = &\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{m}(i)\bar{F} _{m}(\rho )-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{m}(i)\bar{K}_{m}(\rho ) \bar{L} \\ \Lambda _{122} = &-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{m}(i)\bar{F} _{m}(\rho )\bar{G}_{m}(i)+\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{m}(i)\bar{K} _{m}(\rho )\bar{L}\bar{G}_{m}(i) \\ \Lambda _{22} = &\left[ \begin{array}{cccccccccc} \Lambda _{221} & \Lambda _{222} & 0 & 0 & 0 & \Lambda _{223} & 0 & P_{p}(i) \bar{E}_{p}\left( i\right) & 0 & \Lambda _{224} \\ \ast & -Q_{\tau }-2R_{\tau } & R_{\tau } & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ \ast & \ast & -R_{\tau } & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ \ast & \ast & \ast & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ \ast & \ast & \ast & \ast & 0 & 0 & 0 & 0 & 0 & 0 \\ \ast & \ast & \ast & \ast & \ast & 0 & 0 & 0 & 0 & 0 \\ \ast & \ast & \ast & \ast & \ast & \ast & 0 & 0 & 0 & 0 \\ \ast & \ast & \ast & \ast & \ast & \ast & \ast & 0 & 0 & 0 \\ \ast & \ast & \ast & \ast & \ast & \ast & \ast & \ast & 0 & 0 \\ \ast & \ast & \ast & \ast & \ast & \ast & \ast & \ast & \ast & 0 \end{array} \right] , \\ \Lambda _{221} = &-2P_{p}(i)\bar{C}\left( i\right) +Q_{p}+Q_{\tau }-R_{p}-R_{\tau }+\sum\limits_{j\in \mathcal{N}}\pi _{ij}P_{p}(i), \\ \Lambda _{222} = &-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{p}(i)\bar{F} _{p}(\rho )\bar{H}_{p}(i)+\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{p}(i)\bar{K} _{p}(\rho )\bar{L}\bar{H}_{p}(i)+R_{\tau } \\ \Lambda _{223} = &\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{p}(i)\bar{F} _{p}(\rho )-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{p}(i)\bar{K}_{p}(\rho ) \bar{L} \\ \Lambda _{224} = &-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{p}(i)\bar{F} _{p}(\rho )\bar{G}_{p}(i)+\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }P_{p}(i)\bar{K} _{p}(\rho )\bar{L}\bar{G}_{p}(i) \end{align*} |
\begin{align*} \Lambda _{2} = &\left[ \begin{array}{c} -\bar{d}_{m}\bar{A}^{T}\left( i\right) \\ -\bar{d}_{m}\sum_{\rho = 1}^{F}\lambda _{i\rho }\bar{H}_{m}^{T}(i)(\bar{F} _{m}^{T}(\rho )-\bar{L}^{T}\bar{K}_{m}^{T}(\rho )) \\ 0 \\ 0 \\ 0 \\ 0 \\ 0 \\ 0 \\ 0 \\ 0 \\ \bar{d}_{m}\bar{B}^{T}\left( i\right) \\ \bar{d}_{m}\sum_{\rho = 1}^{F}\lambda _{i\rho }(\bar{F}_{m}^{T}(\rho )-\bar{L} ^{T}\bar{K}_{m}^{T}(\rho )) \\ 0 \\ \bar{d}_{m}\bar{E}_{m}\left( i\right) \\ 0 \\ -\bar{d}_{m}\sum_{\rho = 1}^{F}\lambda _{i\rho }\bar{G}_{m}^{T}(i)(\bar{F} _{m}^{T}(\rho )-\bar{L}^{T}\bar{K}_{m}^{T}(\rho )) \\ 0 \end{array} \right] , \\ \Lambda _{3} = &\left[ \begin{array}{c} 0 \\ 0 \\ 0 \\ 0 \\ \bar{d}_{p}\bar{D}^{T}\left( i\right) \\ 0 \\ 0 \\ -\bar{d}_{p}\bar{C}^{T}\left( i\right) \\ -\bar{d}_{p}\sum_{\rho = 1}^{F}\lambda _{i\rho }\bar{H}_{p}^{T}(i)(\bar{F} _{p}^{T}(\rho )-\bar{L}^{T}\bar{K}_{p}^{T}(\rho )) \\ 0 \\ 0 \\ 0 \\ \bar{d}_{p}\sum_{\rho = 1}^{F}\lambda _{i\rho }(\bar{F}_{p}^{T}(\rho )-\bar{L} ^{T}\bar{K}_{p}^{T}(\rho )) \\ 0 \\ \bar{d}_{p}\bar{E}_{p}^{T}\left( i\right) \\ 0 \\ -\bar{d}_{p}\sum_{\rho = 1}^{F}\lambda _{i\rho }\bar{G}_{p}^{T}(i)(\bar{F} _{p}^{T}(\rho )-\bar{L}^{T}\bar{K}_{p}^{T}(\rho )) \end{array} \right] , \\ \Lambda _{4} = &\left[ \begin{array}{c} -\bar{\tau}\bar{A}^{T}\left( i\right) \\ -\bar{\tau}\sum_{\rho = 1}^{F}\lambda _{i\rho }\bar{H}_{m}^{T}(i)(\bar{F} _{m}^{T}(\rho )-\bar{L}^{T}\bar{K}_{m}^{T}(\rho )) \\ 0 \\ 0 \\ 0 \\ 0 \\ 0 \\ 0 \\ 0 \\ \bar{\tau}\bar{B}^{T}\left( i\right) \\ \bar{\tau}\sum_{\rho = 1}^{F}\lambda _{i\rho }(\bar{F}_{m}^{T}(\rho )-\bar{L} ^{T}\bar{K}_{m}^{T}(\rho )) \\ 0 \\ \bar{\tau}\bar{E}_{m}\left( i\right) \\ 0 \\ -\bar{\tau}\sum_{\rho = 1}^{F}\lambda _{i\rho }\bar{G}_{m}^{T}(i)(\bar{F} _{m}^{T}(\rho )-\bar{L}^{T}\bar{K}_{m}^{T}(\rho )) \\ 0 \end{array} \right] , \\ \Lambda _{5} = &\left[ \begin{array}{c} 0 \\ 0 \\ 0 \\ 0 \\ \bar{\tau}\bar{D}^{T}\left( i\right) \\ 0 \\ -\bar{\tau}\bar{C}^{T}\left( i\right) \\ -\bar{\tau}\sum_{\rho = 1}^{F}\lambda _{i\rho }\bar{H}_{p}^{T}(i)(\bar{F} _{p}^{T}(\rho )-\bar{L}^{T}\bar{K}_{p}^{T}(\rho )) \\ 0 \\ 0 \\ 0 \\ \bar{\tau}\sum_{\rho = 1}^{F}\lambda _{i\rho }(\bar{F}_{p}^{T}(\rho )-\bar{L} ^{T}\bar{K}_{p}^{T}(\rho )) \\ 0 \\ \bar{\tau}\bar{E}_{p}^{T}\left( i\right) \\ 0 \\ -\bar{\tau}\sum_{\rho = 1}^{F}\lambda _{i\rho }\bar{G}_{p}^{T}(i)(\bar{F} _{p}^{T}(\rho )-\bar{L}^{T}\bar{K}_{p}^{T}(\rho )) \end{array} \right] . \end{align*} |
Furthermore, considering the nonlinear properties for the GRNs, it yields that
\begin{align} \left[ \begin{array}{c} \bar{g}\left( e_{p}\left( t-d_{p}(t)\right) \right) \\ e_{p}\left( t-d_{p}(t)\right) \end{array} \right] ^{T}\left[ \begin{array}{ll} -2I & \varsigma \\ \varsigma & 0 \end{array} \right] \left[ \begin{array}{c} \bar{g}\left( e_{p}\left( t-d_{p}(t)\right) \right) \\ e_{p}\left( t-d_{p}(t)\right) \end{array} \right] \geq 0. \end{align} | (3.10) |
In addition, by recalling the event-triggering conditions, one has
\begin{align} \sum\limits_{l = 1}^{N}\kappa ^{2}[\tilde{y}^{l}(\iota _{k})-\varepsilon ^{l}(\iota _{k})]^{T}[\tilde{y}^{l}(\iota _{k})-\varepsilon ^{l}(\iota _{k})]-\sum\limits_{l = 1}^{N}[\varepsilon ^{l}(\iota _{k})]^{T}\varepsilon ^{l}(\iota _{k})\geq 0, \end{align} | (3.11) |
where \tilde{y}^{l}(\iota _{k}) = [(\tilde{y}_{m}^{l}(t))^{T}, (\tilde{y} _{p}^{l}(t))^{T}]^{T} , \varepsilon ^{l}(\iota _{k}) = [(\varepsilon _{m}^{l}(\iota _{k}))^{T}, (\varepsilon _{p}^{l}(\iota _{k}))^{T}]^{T} and it can be further verified that
\begin{align} & \sum\limits_{l = 1}^{N}\kappa ^{2}[H_{m}^{l}(i)e_{m}^{l}(\iota _{k})+G_{m}^{l}(i)\omega _{3}(\iota _{k})-\varepsilon _{m}^{l}(\iota _{k})]^{T}[H_{m}^{l}(i)e_{m}^{l}(\iota _{k})+G_{m}^{l}(i)\omega _{3}(\iota _{k})-\varepsilon _{m}^{l}(\iota _{k})] \\ & +\sum\limits_{l = 1}^{N}\kappa ^{2}[H_{p}^{l}(i)e_{p}^{l}(\iota _{k})+G_{p}^{l}(i)\omega _{4}(\iota _{k})-\varepsilon _{p}^{l}(\iota _{k})]^{T}[H_{p}^{l}(i)e_{p}^{l}(\iota _{k})+G_{p}^{l}(i)\omega _{4}(\iota _{k})-\varepsilon _{p}^{l}(\iota _{k})] \\ & -(\varepsilon _{m}(t-\tau (t)))^{T}\varepsilon _{m}(t-\tau (t))-(\varepsilon _{p}(t-\tau (t)))^{T}\varepsilon _{p}(t-\tau (t))\geq 0. \end{align} | (3.12) |
As a consequence, it can be derived by Schur complement that the modified H_{\infty} performance can be achieved under the mean-square sense when \Xi(i) < 0 is satisfied according to Theorem 1, which implies that
\begin{align} &\mathcal{L}V(i,t)+e_{m}^{T}(t)\bar{U}_{m}^{T}\left( i\right) \bar{U} _{m}\left( i\right) e_{m}(t)+e_{p}^{T}(t)\bar{U}_{p}^{T}\left( i\right) \bar{ U}_{p}\left( i\right) e_{p}(t)-\gamma \omega _{1}^{T}(t)\omega _{1}(t) \\ &-\gamma \omega _{2}^{T}\left( t\right) \omega _{2}\left( t\right) -\gamma \omega _{3}^{T}(t-\tau (t))\omega _{3}(t-\tau (t))-\gamma \omega _{4}^{T}(t-\tau (t))\omega _{4}(t-\tau (t)) < 0 \end{align} | (3.13) |
and therefore completes the proof.
Remark 3. The above established sufficient convex optimization conditions are derived with aid of linear matrix inequality technique, which can be easily solved by Matlab LMI toolbox or other mathematical softwares. It should be pointed out that the matrix dimension dependents on both system mode information and the sensor networks structure, which means that the more system modes and sensors are involved, the more computational consumption is needed. As a result, the distributed state estimators should be well designed for practical system implements with considerable computational efficiency.
Theorem 2. For given \bar{d}_{p} , \bar{d}_{m} , \bar{\tau} and the modified H_{\infty } filtering can be satisfied over distributed sensor networks according to Definition 1, if there exist mode-dependent matrices P_{m}(i) > 0 , P_{p}(i) > 0 , X_{m}(i) > 0 , X_{p}(i) > 0 , Y_{F, m}\left(\rho \right) , Y_{F, p}\left(\rho \right) , Y_{K, m}\left(\rho \right) , Y_{K, p}\left(\rho \right), and matrices Q_{m} > 0 , Q_{p} > 0 , Q_{\tau } > 0 , R_{m} > 0 , R_{p} > 0 , R_{\tau } > 0 , such that \tilde{\Xi} _{i} < 0 , for all i\in \mathcal{N} , k = 1, 2, \ldots, N and \rho \in \mathcal{F} , k = 1, 2, \ldots, F , where
\begin{align*} \tilde{\Xi}(i) = &\left[ \begin{array}{ll} \tilde{\Xi}_{1}(i) & \tilde{\Xi}_{2}(i) \\ \ast & \tilde{\Xi}_{3}(i) \end{array} \right] , \\ \tilde{\Xi}_{1}(i) = &\left[ \begin{array}{ll} \tilde{\Xi}_{11}(i) & \tilde{\Xi}_{12}(i) \\ \ast & \tilde{\Xi}_{13}(i) \end{array} \right] , \\ \tilde{\Xi}_{11}(i) = &\left[ \begin{array}{ll} \tilde{\Xi}_{111}(i) & \tilde{\Xi}_{112}(i) \\ \ast & \tilde{\Xi}_{113}(i) \end{array} \right] , \\ \tilde{\Xi}_{111}(i) = &\left[ \begin{array}{ccccccc} \tilde{\Xi}_{1111}(i) & \tilde{\Xi}_{1112}(i) & 0 & 0 & R_{m} & 0 & 0 \\ \ast & \tilde{\Xi}_{1113}(i) & R_{\tau } & 0 & 0 & 0 & 0 \\ \ast & \ast & -Q_{\tau }-R_{\tau } & 0 & 0 & 0 & 0 \\ \ast & \ast & \ast & -Q_{m}-R_{m} & R_{m}^{T} & 0 & 0 \\ \ast & \ast & \ast & \ast & -2R_{m} & 0 & 0 \\ \ast & \ast & \ast & \ast & \ast & -Q_{p}-R_{p} & R_{p} \\ \ast & \ast & \ast & \ast & \ast & \ast & -2R_{p} \end{array} \right] , \\ \tilde{\Xi}_{112}(i) = &\left[ \begin{array}{cccccc} 0 & 0 & 0 & P_{m}(i)\bar{B}\left( i\right) & \tilde{\Xi}_{1121}(i) & 0 \\ 0 & 0 & 0 & 0 & -\kappa ^{2}\bar{H}_{m}^{T}\left( i\right) & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ \bar{D}^{T}\left( i\right) P_{p}(i) & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ R_{p} & 0 & 0 & \varsigma & 0 & 0 \end{array} \right] , \\ \tilde{\Xi}_{113}(i) = &\left[ \begin{array}{cccccc} \tilde{\Xi}_{1131}(i) & \tilde{\Xi}_{1132}(i) & 0 & 0 & 0 & \tilde{\Xi} _{1133}(i) \\ \ast & \tilde{\Xi}_{1134}(i) & R_{\tau } & 0 & 0 & -\kappa ^{2}\bar{H} _{p}^{T}\left( i\right) \\ \ast & \ast & -R_{\tau } & 0 & 0 & 0 \\ \ast & \ast & \ast & -2I & 0 & 0 \\ \ast & \ast & \ast & \ast & \kappa ^{2}I-I & 0 \\ \ast & \ast & \ast & \ast & \ast & \kappa ^{2}I-I \end{array} \right] , \\ \tilde{\Xi}_{1111}(i) = &-2P_{m}(i)\bar{A}\left( i\right) +Q_{m}+Q_{\tau }-R_{m}-R_{\tau } \\ &+\bar{U}_{m}^{T}\left( i\right) \bar{U}_{m}\left( i\right) +\sum\limits_{j\in \mathcal{N}}\pi _{ij}P_{m}(j), \\ \tilde{\Xi}_{1112}(i) = &-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,m}(\rho ) \bar{H}_{m}(i) \\ &+\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{K,m}(\rho )\bar{L}\bar{H} _{m}(i)+R_{\tau }, \\ \tilde{\Xi}_{1113}(i) = &-2R_{\tau }+\kappa ^{2}\bar{H}_{m}^{T}\left( i\right) \bar{H}_{m}\left( i\right) , \\ \tilde{\Xi}_{1121}(i) = &\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,m}(\rho )-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{K,m}(\rho )\bar{L}, \\ \tilde{\Xi}_{1131}(i) = &-2P_{p}(i)\bar{C}\left( i\right) +Q_{p }+Q_{\tau }-R_{p}-R_{\tau } \\ &+\bar{U}_{p}^{T}\left( i\right) \bar{U}_{p}\left( i\right) +\sum\limits_{j\in \mathcal{N}}\pi _{ij}P_{p}(i), \\ \tilde{\Xi}_{1132}(i) = &-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,p}(\rho ) \bar{H}_{p}(i) \\ &+\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{K,p}(\rho )\bar{L}\bar{H} _{p}(i)+R_{\tau }, \\ \tilde{\Xi}_{1133}(i) = &\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,p}(\rho )-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{K,p}(\rho )\bar{L}, \\ \tilde{\Xi}_{1134}(i) = &-Q_{\tau }-2R_{\tau }+\kappa ^{2}\bar{H} _{p}^{T}\left( i\right) \bar{H}_{p}\left( i\right) , \end{align*} |
\begin{align*} \tilde{\Xi}_{12}(i) = &\left[ \begin{array}{cc} \tilde{\Xi}_{121}(i) & \tilde{\Xi}_{122}(i) \\ \tilde{\Xi}_{123}(i) & \tilde{\Xi}_{124}(i) \end{array} \right] , \\ \tilde{\Xi}_{121}(i) = &\left[ \begin{array}{cccc} P_{m}(i)\bar{E}_{m}\left( i\right) & 0 & \tilde{\Xi}_{1211}(i) & 0 \\ 0 & 0 & \kappa ^{2}\bar{H}_{m}^{T}\left( i\right) \bar{G}_{m}(i) & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & P_{p}(i)\bar{E}_{p}\left( i\right) & 0 & \tilde{\Xi}_{1212}(i) \end{array} \right] , \\ \tilde{\Xi}_{1211}(i) = &-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,m}(\rho ) \bar{G}_{m}(i) \\ &+\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{K,m}(\rho )\bar{L}\bar{G}_{m}(i), \\ \tilde{\Xi}_{1212}(i) = &-\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,p}(\rho ) \bar{G}_{p}(i) \\ &+\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{K,p}(\rho )\bar{L}\bar{G}_{p}(i), \\ \tilde{\Xi}_{122}(i) = &\left[ \begin{array}{cccc} -\bar{d}_{m}\bar{A}^{T}\left( i\right) P_{m}\left( i\right) & 0 & -\bar{\tau}\bar{A}^{T}\left( i\right) P_{m}\left( i\right) & 0 \\ \tilde{\Xi}_{1221}(i) & 0 & \tilde{\Xi}_{1222}(i) & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & \bar{d}_{p}\bar{D}^{T}\left( i\right) P_{p}\left( i\right) & 0 & \bar{\tau}\bar{D}^{T}\left( i\right) P_{p}\left( i\right) \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & -\bar{d}_{p}\bar{C}^{T}\left( i\right) P_{p}\left( i\right) & 0 & -\bar{\tau}\bar{C}^{T}\left( i\right) P_{p}\left( i\right) \end{array} \right] , \\ \tilde{\Xi}_{1221}(i) = &-\bar{d}_{m}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{ H}_{m}^{T}(i)Y_{F,m}^{T}(\rho ) \\ &+\bar{d}_{m}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{H}_{m}^{T}(i)\bar{L} ^{T}Y_{K,m}^{T}(\rho ), \\ \tilde{\Xi}_{1222}(i) = &-\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{ H}_{m}^{T}(i)Y_{F,m}^{T}(\rho ) \\ &+\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{H}_{m}^{T}(i)\bar{L} ^{T}Y_{K,m}^{T}(\rho ), \\ \tilde{\Xi}_{123}(i) = &\left[ \begin{array}{ccccc} 0 & 0 & 0 & \kappa ^{2}\bar{H}_{p}^{T}\left( i\right) \bar{G}_{p}(i) & 0 \\ 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & \bar{d}_{m}\bar{B}^{T}\left( i\right) P_{m}\left( i\right) \\ 0 & 0 & -\bar{G}_{m}(i) & 0 & \tilde{\Xi}_{1231}(i) \\ 0 & 0 & 0 & -\bar{G}_{p}(i) & 0 \end{array} \right] , \\ \tilde{\Xi}_{1231}(i) = &\bar{d}_{m}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,m}^{T}(\rho ) \\ &-\bar{d}_{m}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{L}^{T}Y_{K,m}^{T}(\rho ), \\ \tilde{\Xi}_{124}(i) = &\left[ \begin{array}{ccc} \tilde{\Xi}_{1241}(i) & 0 & \tilde{\Xi}_{1242}(i) \\ 0 & 0 & 0 \\ 0 & \bar{\tau}\bar{B}^{T}\left( i\right) P_{m}\left( i\right) & 0 \\ 0 & \tilde{\Xi}_{1243}(i) & 0 \\ \tilde{\Xi}_{1244}(i) & 0 & \tilde{\Xi}_{1245}(i) \end{array} \right] , \\ \tilde{\Xi}_{1241}(i) = &-\bar{d}_{p}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{ H}_{p}^{T}(i)Y_{F,p}^{T}(\rho ) +\bar{d}_{p}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{H}_{p}^{T}(i)\bar{L} ^{T}Y_{K,p}^{T}(\rho ), \\ \tilde{\Xi}_{1242}(i) = &-\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{ H}_{p}^{T}(i)Y_{F,p}^{T}(\rho ) +\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{H}_{p}^{T}(i)\bar{L} ^{T}Y_{K,p}^{T}(\rho ), \\ \tilde{\Xi}_{1243}(i) = &\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,m}^{T}(\rho )-\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{L}^{T}Y_{K,m}^{T}(\rho ), \\ \tilde{\Xi}_{1244}(i) = &\bar{d}_{p}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,p}^{T}(\rho ) -\bar{d}_{p}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{L}^{T}Y_{K,p}^{T}(\rho ), \\ \tilde{\Xi}_{1245}(i) = &\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }Y_{F,p}^{T}(\rho ) -\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{L}^{T}Y_{K,p}^{T}(\rho ), \end{align*} |
\begin{align*} \tilde{\Xi}_{13}(i) = &\left[ \begin{array}{cc} \tilde{\Xi}_{131}(i) & \tilde{\Xi}_{132}(i) \\ \ast & \tilde{\Xi}_{133}(i) \end{array} \right] , \\ \tilde{\Xi}_{131}(i) = &\left[ \begin{array}{cccc} -\gamma I & 0 & 0 & 0 \\ \ast & -\gamma I & 0 & 0 \\ \ast & \ast & \kappa ^{2}\bar{G}_{m}^{T}(i)\bar{G}_{m}(i)-\gamma I & 0 \\ \ast & \ast & \ast & \kappa ^{2}\bar{G}_{p}^{T}(i)\bar{G}_{p}(i)-\gamma I \end{array} \right] , \\ \tilde{\Xi}_{132}(i) = &\left[ \begin{array}{cccc} \bar{d}_{m}\bar{E}_{m}\left( i\right) P_{m}\left( i\right) & 0 & \bar{\tau} \bar{E}_{m}\left( i\right) P_{m}\left( i\right) & 0 \\ 0 & \bar{d}_{p}\bar{E}_{p}^{T}\left( i\right) P_{p}\left( i\right) & 0 & \bar{\tau}\bar{E}_{p}^{T}\left( i\right) P_{p}\left( i\right) \\ \tilde{\Xi}_{1311}(i) & 0 & \tilde{\Xi}_{1312}(i) & 0 \\ 0 & \tilde{\Xi}_{1313}(i) & 0 & \tilde{\Xi}_{1314}(i) \end{array} \right] , \\ \tilde{\Xi}_{133}(i) = &\left[ \begin{array}{cccc} R_{m}-2P_{m}\left( i\right) & 0 & 0 & 0 \\ \ast & R_{m}-2P_{p}\left( i\right) & 0 & 0 \\ \ast & \ast & R_{\tau }-2P_{m}\left( i\right) & 0 \\ \ast & \ast & \ast & R_{\tau }-2P_{p}\left( i\right) \end{array} \right] , \\ \tilde{\Xi}_{1311}(i) = &-\bar{d}_{m}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{ G}_{m}^{T}(i)Y_{F,m}^{T}(\rho )+\bar{d}_{m}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{G}_{m}^{T}(i)\bar{L}^{T}Y_{K,m}^{T}(\rho ), \\ \tilde{\Xi}_{1312}(i) = &-\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{ G}_{m}^{T}(i)Y_{F,m}^{T}(\rho )+\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{G}_{m}^{T}(i)\bar{L}^{T}Y_{K,m}^{T}(\rho ), \\ \tilde{\Xi}_{1313}(i) = &-\bar{d}_{p}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{ G}_{p}^{T}(i)Y_{F,p}^{T}(\rho )+\bar{d}_{p}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{G}_{p}^{T}(i)\bar{L}^{T}Y_{K,p}^{T}(\rho ), \\ \tilde{\Xi}_{1314}(i) = &-\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{ G}_{p}^{T}(i)Y_{F,p}^{T}(\rho )+\bar{\tau}\sum\limits_{\rho = 1}^{F}\lambda _{i\rho }\bar{G}_{p}^{T}(i)\bar{L}^{T}Y_{K,p}^{T}(\rho ),\\ \tilde{\Xi}_{2}(i) = &\left[ \begin{array}{cc} \tilde{\Xi}_{21}(i) & \tilde{\Xi}_{22}(i) \\ \tilde{\Xi}_{23}(i) & \tilde{\Xi}_{24}(i) \end{array} \right] , \\ \tilde{\Xi}_{21}(i) = &\left[ \tilde{\Xi}_{211}(i),\tilde{\Xi}_{212}(i) \right] , \\ \tilde{\Xi}_{211}(i) = &\left[ \begin{array}{c} \lambda _{i1}(X_{m}\left( 1\right) -P_{m}(i)) \\ \epsilon _{m}\bar{H}_{m}^{T}(i)(Y_{F,m}^{T}(1)-\bar{L}^{T}Y_{K,m}^{T}(1)) \\ 0 \end{array} \right] , \\ \tilde{\Xi}_{212}(i) = &\left[ \begin{array}{ccc} \lambda _{i2}(X_{m}\left( 2\right) -P_{m}(i)) & \cdots & \lambda _{iF}(X_{m}\left( F\right) -P_{m}(i)) \\ \epsilon _{m}\bar{H}_{m}^{T}(i)(Y_{F,m}^{T}(2)-\bar{L}^{T}Y_{K,m}^{T}(2)) & \cdots & \epsilon _{m}\bar{H}_{m}^{T}(i)(Y_{F,m}^{T}(F)-\bar{L} ^{T}Y_{K,m}^{T}(F)) \\ 0 & \cdots & 0 \end{array} \right] , \\ \tilde{\Xi}_{22}(i) = &\left[ \tilde{\Xi}_{221}(i),\tilde{\Xi}_{222}(i) \right] , \\ \tilde{\Xi}_{221}(i) = &\left[ \begin{array}{c} 0 \\ \lambda _{i1}(X_{p}\left( 1\right) -P_{p}(i)) \\ \epsilon _{p}\bar{H}_{p}^{T}(i)(Y_{F,p}^{T}(1)-\bar{L}^{T}Y_{K,p}^{T}(1)) \\ 0 \end{array} \right] , \\ \tilde{\Xi}_{222}(i) = &\left[ \begin{array}{ccc} 0 & \cdots & 0 \\ \lambda _{i2}(X_{p}\left( 2\right) -P_{p}(i)) & \cdots & \lambda _{iF}(X_{p}\left( F\right) -P_{p}(i)) \\ \epsilon _{p}\bar{H}_{p}^{T}(i)(Y_{F,p}^{T}(2)-\bar{L}^{T}Y_{K,p}^{T}(2)) & \cdots & \epsilon _{p}\bar{H}_{p}^{T}(i)(Y_{F,p}^{T}(F)-\bar{L} ^{T}Y_{K,p}^{T}(F)) \\ 0 & \cdots & 0 \end{array} \right] , \\ \tilde{\Xi}_{23}(i) = &\left[ \tilde{\Xi}_{231}(i),\tilde{\Xi}_{232}(i) \right] , \\ \tilde{\Xi}_{231}(i) = &\left[ \begin{array}{c} \epsilon _{m}(\bar{L}^{T}Y_{K,m}^{T}(1)-Y_{F,m}^{T}(1)) \\ 0 \\ 0 \\ 0 \\ \epsilon _{m}\bar{G}_{m}^{T}(i)(Y_{F,m}^{T}(1)-\bar{L}^{T}Y_{K,m}^{T}(1)) \\ 0 \\ \bar{d}_{m}\lambda _{i1}(X_{m}\left( 1\right) -P_{m}(i)) \\ 0 \\ \bar{d}_{m}\lambda _{i1}(X_{m}\left( 1\right) -P_{m}(i)) \\ 0 \end{array} \right] , \\ \tilde{\Xi}_{232}(i) = &\left[ \begin{array}{ccc} \epsilon _{m}(\bar{L}^{T}Y_{K,m}^{T}(2)-Y_{F,m}^{T}(2)) & \cdots & \epsilon _{m}(\bar{L}^{T}Y_{K,m}^{T}(F)-Y_{F,m}^{T}(F)) \\ 0 & \cdots & 0 \\ 0 & \cdots & 0 \\ 0 & \cdots & 0 \\ \epsilon _{m}\bar{G}_{m}^{T}(i)(Y_{F,m}^{T}(2)-\bar{L}^{T}Y_{K,m}^{T}(2)) & \cdots & \epsilon _{m}^{T}\bar{G}_{m}^{T}(i)(Y_{F,m}^{T}(F)-\bar{L} ^{T}Y_{K,m}^{T}(F)) \\ 0 & \cdots & 0 \\ \bar{d}_{m}\lambda _{i2}(X_{m}\left( 2\right) -P_{m}(i)) & \cdots & \bar{d} _{m}\lambda _{iF}(X_{m}\left( F\right) -P_{m}(i)) \\ 0 & \cdots & 0 \\ \bar{d}_{m}\lambda _{i2}(X_{m}\left( 2\right) -P_{m}(i)) & \cdots & \bar{d} _{m}\lambda _{iF}(X_{m}\left( F\right) -P_{m}(i)) \\ 0 & \cdots & 0 \end{array} \right] , \\ \tilde{\Xi}_{24}(i) = &\left[ \tilde{\Xi}_{241}(i),\tilde{\Xi}_{242}(i) \right] , \\ \tilde{\Xi}_{241}(i) = &\left[ \begin{array}{c} 0 \\ \epsilon _{p}(\bar{L}^{T}Y_{K,p}^{T}(1)-Y_{F,p}^{T}(1)) \\ 0 \\ 0 \\ 0 \\ \epsilon _{p}\bar{G}_{p}^{T}(i)(Y_{F,p}^{T}(1)-\bar{L}^{T}Y_{K,p}^{T}(1)) \\ 0 \\ \bar{d}_{p}\lambda _{i1}(X_{p}\left( 1\right) -P_{p}(i)) \\ 0 \\ \bar{d}_{p}\lambda _{i1}(X_{p}\left( 1\right) -P_{p}(i)) \end{array} \right] , \\ \tilde{\Xi}_{242}(i) = &\left[ \begin{array}{ccc} 0 & \cdots & 0 \\ \epsilon _{p}(\bar{L}^{T}Y_{K,p}^{T}(2)-Y_{F,p}^{T}(2)) & \cdots & \epsilon _{p}(\bar{L}^{T}Y_{K,p}^{T}(F)-Y_{F,p}^{T}(F)) \\ 0 & \cdots & 0 \\ 0 & \cdots & 0 \\ 0 & \cdots & 0 \\ \epsilon _{p}\bar{G}_{p}^{T}(i)(Y_{F,p}^{T}(2)-\bar{L}^{T}Y_{K,p}^{T}(2)) & \cdots & \epsilon _{p}^{T}\bar{G}_{p}^{T}(i)(Y_{F,p}^{T}(F)-\bar{L} ^{T}Y_{K,p}^{T}(F)) \\ 0 & \cdots & 0 \\ \bar{d}_{p}\lambda _{i2}(X_{p}\left( 2\right) -P_{p}(i)) & \cdots & \bar{d} _{p}\lambda _{iF}(X_{p}\left( F\right) -P_{p}(i)) \\ 0 & \cdots & 0 \\ \bar{d}_{p}\lambda _{i2}(X_{p}\left( 2\right) -P_{p}(i)) & \cdots & \bar{d} _{p}\lambda _{iF}(X_{p}\left( F\right) -P_{p}(i)) \end{array} \right] , \end{align*} |
and
\begin{align} \tilde{\Xi}_{3}(i) = &\left[ \begin{array}{ll} \tilde{\Xi}_{31}(i) & \tilde{\Xi}_{32}(i) \\ \ast & \tilde{\Xi}_{33}(i) \end{array} \right] , \end{align} | (3.14) |
\begin{align} \tilde{\Xi}_{31}(i) = &\left[ \begin{array}{cccc} -2\epsilon _{m}X_{m}(1) & 0 & \cdots & 0 \\ \ast & -2\epsilon _{m}X_{m}(2) & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ \ast & \ast & \cdots & -2\epsilon _{m}X_{m}(F) \end{array} \right] \end{align} | (3.15) |
\begin{align} \tilde{\Xi}_{32}(i) = &0, \end{align} | (3.16) |
\begin{align} \tilde{\Xi}_{33}(i) = &\left[ \begin{array}{cccc} -2\epsilon _{p}X_{p}(1) & 0 & \cdots & 0 \\ \ast & -2\epsilon _{p}X_{p}(2) & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ \ast & \ast & \ast & -2\epsilon _{p}X_{p}(F) \end{array} \right] . \end{align} | (3.17) |
With the aforementioned feasible solutions, the desired mode-dependent filter gains F_{m}^{l}(\sigma (t)), F_{p}^{l}(\sigma (t)) and K_{m}^{l}(\sigma (t)), K_{p}^{l}(\sigma (t)) can be obtained by the following calculation:
\begin{align*} Y_{F,m}\left( \rho \right) X_{m}^{-1}\left( \rho \right) = &\bar{F} _{m}(\rho ), \\ Y_{F,p}\left( \rho \right) X_{p}^{-1}\left( \rho \right) = &\bar{F} _{p}(\rho ), \\ Y_{K,m}\left( \rho \right) X_{m}^{-1}\left( \rho \right) = &\bar{K} _{m}(\rho ), \\ Y_{K,p}\left( \rho \right) X_{p}^{-1}\left( \rho \right) = &\bar{K} _{p}(\rho ). \end{align*} |
Proof. By pre- and post-multiplying diag\{I, I, I, I, I, I, I, I, I, I, I, P_{m}\left(i\right), P_{p}\left(i\right), P_{m}\left(i\right), P_{p}\left(i\right) \} to \Xi(i) < 0 , Lemma 2 is employed accordingly.
Eventually, define Y_{F, m}\left(\rho \right) = X_{m}\left(\rho \right) \bar{F}_{m}(\rho) , Y_{F, p}\left(\rho \right) = X_{p}\left(\rho \right) \bar{F}_{p}(\rho) , Y_{K, m}\left(\rho \right) = X_{m}\left(\rho \right) \bar{K}_{m}(\rho) , Y_{K, p}\left(\rho \right) = X_{p}\left(\rho \right) \bar{K}_{p}(\rho) , then the remainder of proof will follow directly from Theorem 1.
Remark 4. The proposed modified H_{\infty} performance is more applicable for the distributed sensor networks design. In addition, the performance index \gamma can be further optimized by solving the following convex optimization problem:
\begin{align} &{\rm{minimize }}\; \gamma , \\ & {\rm{subject\; to } }\;\tilde{\Xi}_{i} < 0, i\in \mathcal{N}, \rho \in \mathcal{F}. \end{align} | (3.18) |
This section is dedicated to validate the applicability of our proposed distributed nonsynchronous designs, while illustrative studies have been performed with simulation results.
Consider the following HMJP GRNs model with a set of biological simulated parameters as
\begin{align*} A\left( 1\right) = &\left[ \begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array} \right] ,A\left( 2\right) = \left[ \begin{array}{cc} 1.3 & 0 \\ 0 & 1.1 \end{array} \right] , \\ B\left( 1\right) = &\left[ \begin{array}{cc} 0.2 & 0.1 \\ 0.8 & 0.3 \end{array} \right] ,B\left( 2\right) = \left[ \begin{array}{cc} 0.3 & 0.2 \\ 0.1 & 0.2 \end{array} \right] , \\ C\left( 1\right) = &\left[ \begin{array}{cc} 1 & 0 \\ 0 & 1.2 \end{array} \right] ,C\left( 2\right) = \left[ \begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array} \right] , \\ D\left( 1\right) = &\left[ \begin{array}{cc} 0.8 & 0 \\ 0 & 0.8 \end{array} \right] ,D\left( 2\right) = \left[ \begin{array}{cc} 1.1 & 0 \\ 0 & 1.3 \end{array} \right] , \\ E_{m}\left( 1\right) = &\left[ \begin{array}{l} 1.1 \\ 1 \end{array} \right] ,E_{m}\left( 2\right) = \left[ \begin{array}{l} 1.2 \\ 1.1 \end{array} \right] , \\ E_{p}\left( 1\right) = &\left[ \begin{array}{l} 1 \\ 1.1 \end{array} \right] ,E_{p}\left( 2\right) = \left[ \begin{array}{l} 1.2 \\ 1 \end{array} \right] , \\ U_{m}\left( 1\right) = &\left[ \begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array} \right] ,U_{m}\left( 2\right) = \left[ \begin{array}{cc} 1.02 & 0 \\ 0 & 1.02 \end{array} \right] , \\ U_{p}\left( 1\right) = &\left[ \begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array} \right] ,U_{p}\left( 2\right) = \left[ \begin{array}{cc} 1.01 & 0 \\ 0 & 1.01 \end{array} \right] , \end{align*} |
where the transition rate matrix with two modes of hidden Marko chain are assumed by
\begin{equation*} \Pi = \left[ \begin{array}{cc} -0.2 & 0.2 \\ 0.8 & -0.8 \end{array} \right] . \end{equation*} |
For the sensor networks, it is assumed that two networked sensors are deployed for mRNAs as well as proteins, such that the communication topology matrix for each network is given by
\begin{equation*} \mathbb{L} = \left[ \begin{array}{cc} 1 & -1 \\ -1 & 1 \end{array} \right] . \end{equation*} |
Moreover, the mode-dependent sensor parameters are set as follows:
\begin{align*} H_{m}^{1}\left( 1\right) & = \left[ \begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array} \right] ,H_{m}^{1}\left( 2\right) = \left[ \begin{array}{cc} 1.1 & 0 \\ 0 & 1.1 \end{array} \right] , \\ H_{m}^{2}\left( 1\right) & = \left[ \begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array} \right] ,H_{m}^{2}\left( 2\right) = \left[ \begin{array}{cc} 1.2 & 0 \\ 0 & 1.1 \end{array} \right] , \\ H_{p}^{1}\left( 1\right) & = \left[ \begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array} \right] ,H_{p}^{1}\left( 2\right) = \left[ \begin{array}{cc} 0.9 & 0 \\ 0 & 1.1 \end{array} \right] , \\ H_{p}^{2}\left( 1\right) & = \left[ \begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array} \right] ,H_{p}^{2}\left( 2\right) = \left[ \begin{array}{cc} 1 & 0 \\ 0 & 1.1 \end{array} \right] , \\ G_{m}^{1}\left( 1\right) & = \left[ \begin{array}{l} 1 \\ 1 \end{array} \right] ,G_{m}^{1}\left( 2\right) = \left[ \begin{array}{l} 1 \\ 1.1 \end{array} \right] , \\ G_{m}^{2}\left( 1\right) & = \left[ \begin{array}{l} 1 \\ 1 \end{array} \right] ,G_{m}^{2}\left( 2\right) = \left[ \begin{array}{l} 1.1 \\ 1 \end{array} \right] , \\ G_{p}^{1}\left( 1\right) & = \left[ \begin{array}{l} 1 \\ 1 \end{array} \right] ,G_{p}^{1}\left( 2\right) = \left[ \begin{array}{l} 0.9 \\ 1 \end{array} \right] , \\ G_{p}^{2}\left( 1\right) & = \left[ \begin{array}{l} 1 \\ 1 \end{array} \right] ,G_{p}^{2}\left( 2\right) = \left[ \begin{array}{l} 1 \\ 1.2 \end{array} \right] . \end{align*} |
In the simulation example, the sampling interval of networked sensors is set by h = 0.1 s. Furthermore, the probability matrix of observed modes is supposed to be
\begin{equation*} \Lambda = \left[ \begin{array}{cc} 0.7 & 0.3 \\ 0.2 & 0.8 \end{array} \right] , \end{equation*} |
which indicates that the nonsynchronous mode information is utilized instead of true system modes.
Consequently, with these simulation parameters, the optimized nonsynchronous mode-dependent filter gains are solved by Theorem 2 as follows:
\begin{align*} F_{m}^{1}\left( 1\right) & = \left[ \begin{array}{ll} 0.0071 & 0.0153 \\ -0.0353 & 0.0983 \end{array} \right] ,F_{m}^{1}\left( 2\right) = \left[ \begin{array}{ll} 0.0063 & 0.0147\\ -0.0387 & 0.1017 \end{array} \right] , \\ F_{m}^{2}\left( 1\right) & = \left[ \begin{array}{ll} 0.0130 & 0.0075\\ -0.0256 & 0.0718 \end{array} \right] ,F_{m}^{2}\left( 2\right) = \left[ \begin{array}{ll} 0.0141 & 0.0061\\ -0.0295 & 0.0782 \end{array} \right] , \\ F_{p}^{1}\left( 1\right) & = \left[ \begin{array}{ll} 0.0116 & 0.0051\\ -0.0122 & 0.0406 \end{array} \right] ,F_{p}^{1}\left( 2\right) = \left[ \begin{array}{ll} 0.0107 & 0.0088\\ -0.0133 & 0.0464 \end{array} \right] , \\ F_{p}^{2}\left( 1\right) & = \left[ \begin{array}{ll} 0.0106 & 0.0050\\ -0.0151 & 0.0461 \end{array} \right] ,F_{p}^{2}\left( 2\right) = \left[ \begin{array}{ll} 0.0089 & 0.0044\\ -0.0208 & 0.0444 \end{array} \right] , \\ K_{m}^{1}\left( 1\right) & = \left[ \begin{array}{ll} -0.0134 & -0.0113 \\ -0.0387 & -0.0323 \end{array} \right] ,K_{m}^{1}\left( 2\right) = \left[ \begin{array}{ll} -0.0135 & -0.0127\\ -0.0378 & -0.0351 \end{array} \right] , \\ K_{m}^{2}\left( 1\right) & = \left[ \begin{array}{ll} -0.0118 & -0.0082\\ -0.0263 & -0.0177 \end{array} \right] ,K_{m}^{2}\left( 2\right) = \left[ \begin{array}{ll} -0.0135 & -0.0115\\ -0.0283 & -0.0232 \end{array} \right] , \\ K_{p}^{1}\left( 1\right) & = \left[ \begin{array}{ll} -0.0080 & -0.0079\\ -0.0134 & -0.0149 \end{array} \right] ,K_{p}^{1}\left( 2\right) = \left[ \begin{array}{ll} -0.0079 & -0.0075\\ -0.0145 & -0.0153 \end{array} \right] , \\ K_{p}^{2}\left( 1\right) & = \left[ \begin{array}{ll} -0.0079 & -0.0066\\ -0.0145 & -0.0097 \end{array} \right] ,K_{p}^{2}\left( 2\right) = \left[ \begin{array}{ll} -0.0086 & -0.0061\\ -0.0170 & -0.0102 \end{array} \right] . \end{align*} |
As a result, the concentrations of mRNAs and proteins estimation errors over sensor networks are depicted in Figures 1–4, where the jumping modes of GRNs and sensors are illustrated in Figure 5, respectively. It can be seen that all the sensors can display effective estimations without true GRNs system mode information and thus provides an improved nonsynchronous distributed state estimation approach. Furthermore, the triggering events of the sensors are shown in Figures 6 and 7, respectively. Compared with common time-triggered strategies, one can clearly found that the event-triggered scheme can achieve a considerable improvement of communication efficiency. Therefore, these simulation results well demonstrate our distributed state estimation approach and supports our theoretical findings.
This paper is centered on the state estimation for GRNs with HMJPs over distributed sensor networks. More precisely, a novel event-triggered sensing scheme is developed for more effective estimation. In addition, since the modes of HMJPs might be nonsynchronous with the sensors, the nonsynchronous distributed state estimation is developed with aid of observed information, which is more practical for the implementations. On the basis of model transformation, sufficient mode-dependent conditions for ensuring the modified H_{\infty} performance are established and the desired nonsynchronous state estimation gains are designed by matrix manipulation techniques. In the end, numerical simulations are performed to show the usefulness and advantages of our theoretical results. It is noted that one interesting research issue of our future work is extending the current method to the cases with more complicated mode jumping features.
This work is supported by the National Natural Science Foundation of China under Grant 62173326 and the National Key Research and Development Plan of China under Grant 2020AAA0105900, and supported by Beijing Natural Science Foundation under Grant L211023.
The authors declare there is no conflict of interest.
[1] | F. P. Coolen, Parametric probability distributions in reliability, in Encyclopedia of Quantitative Risk Analysis and Assessment, John Wiley & Sons: Chichester, (2008), 1255–1260. |
[2] | J. W. McPherson, Reliability physics and engineering: Time-to-failure modelling, Springer: Heidelberg, 2013. |
[3] |
A. D. Hutson, An accelerated life model analog for discrete survival and count data, Comput. Meth. Prog. Bio., 210 (2021), 106337. https://doi.org/10.1016/j.cmpb.2021.106337 doi: 10.1016/j.cmpb.2021.106337
![]() |
[4] |
M. C. Aguilera-Morillo, A. M. Aguilera, F. Jiménez-Molinos, J. B. Roldán, Stochastic modeling of random access memories reset transitions, Math. Comput. Simulat., 159 (2019), 197–209. https://doi.org/10.1016/j.matcom.2018.11.016 doi: 10.1016/j.matcom.2018.11.016
![]() |
[5] |
R. Kollu, S. R. Rayapudi, S. Narasimham, K. M. Pakkurthi, Mixture probability distribution functions to model wind speed distributions, Int. J. Energ. Environ. Eng., 3 (2012), 27. https://doi.org/10.1186/2251-6832-3-27 doi: 10.1186/2251-6832-3-27
![]() |
[6] |
F. J. Marques, C. A. Coelho, M. de Carvalho, On the distribution of linear combinations of independent Gumbel random variables, Stat. Comput., 25 (2015), 683–701. https://doi.org/10.1007/s11222-014-9453-5 doi: 10.1007/s11222-014-9453-5
![]() |
[7] | M. F. Neuts, Probability distributions of phase type, Liber Amicorum Prof. Emeritus H. Florin, 1975. |
[8] | M. F. Neuts, Matrix-Geometric Solutions in Stochastic Models: An Algorithmic Approach, John Hopkins University Press: Baltimore, 1981. |
[9] | M. Kijima, Markov processes for stochastic modelling, Springer: New York, 2013. |
[10] | V. G. Kulkarni, Modeling and analysis of stochastic systems, Crc Press, 2016. |
[11] | Q. M. He, Fundamentals of matrix-analytic methods, Springer: New York, 2014. |
[12] | S. Asmussen, Ruin probabilities, World Scientific, 2000. |
[13] |
S. Mahmoodi, S. H. Ranjkesh, Y. Q. Zhao, Condition-based maintenance policies for a multi-unit deteriorating system subject to shocks in a semi-Markov operating environment, Qual. Eng., 32 (2020), 286–297. https://doi.org/10.1080/08982112.2020.1731754 doi: 10.1080/08982112.2020.1731754
![]() |
[14] |
E. Pérez, D. Maldonado, C. Acal, J. E. Ruiz-Castro, A. M. Aguilera, F. Jiménez-Molinos, et al., Advanced temperature dependent statistical analysis of forming voltage distributions for three different HfO2-based RRAM technologies, Solid State Electron., 176 (2021), 107961. https://doi.org/10.1016/j.sse.2021.107961 doi: 10.1016/j.sse.2021.107961
![]() |
[15] |
J. E. Ruiz-Castro, C. Acal, A. M. Aguilera, J. B. Roldán, A complex model via phase-type distributions to study random telegraph noise in resistive memories, Mathematics, 9 (2021), 390. https://doi.org/10.3390/math9040390 doi: 10.3390/math9040390
![]() |
[16] |
S. Gordon, A.H. Marshall, M. Zenga, Predicting elderly patient length of stay in hospital and community care using a series of conditional coxian phase-type distributions, further conditioned on a survival tree, Health Care Manag. Sc., 21 (2018), 269–280. https://doi.org/10.1007/s10729-017-9411-9 doi: 10.1007/s10729-017-9411-9
![]() |
[17] |
M. Bladt, A review on phase-type distributions and their use in risk theory, ASTIN Bull. J. IAA, 35 (2005), 145–161. https://doi.org/10.1017/s0515036100014100 doi: 10.1017/s0515036100014100
![]() |
[18] |
J. E. Ruiz-Castro, C. Acal, A. M. Aguilera, M. C. Aguilera-Morillo, J. B. Roldán, Linear-phase-type probability modelling of functional PCA with applications to resistive memories, Math. Comput. Simulat., 186 (2021), 71–79. https://doi.org/10.1016/j.matcom.2020.07.006 doi: 10.1016/j.matcom.2020.07.006
![]() |
[19] | W. Chang, Joe Cheng, J. J. Allaire, C. Sievert, B. Schloerke, Y. H. Xie, et al., R package shiny (2022). Available from: https://CRAN.R-project.org/package = shiny |
[20] |
M. G. Genton, S. Castruccio, P. Crippa, S. Dutta, R. Huser, Y. Sun, et al., Visuanimation in statistics, Stat, 4 (2015), 81–96. https://doi.org/10.1002/sta4.77 doi: 10.1002/sta4.77
![]() |
[21] |
J. Wrobel, S. Y. Park, A. M. Staicu, J. Goldsmith, Interactive graphics for functional data analyses, Stat, 5 (2016), 108–118. https://doi.org/10.1002/sta4.109 doi: 10.1002/sta4.109
![]() |
[22] |
J. P. Fortin, E. Fertig, K. Hansen, shinyMethyl: Interactive quality control of Illumina 450k DNA methylation arrays in R, F1000research, 3 (2014) 175. https://doi.org/10.12688/f1000research.4680.2 doi: 10.12688/f1000research.4680.2
![]() |
[23] |
C. Tebé, J. Valls, P. Satorra, A. Tobías, COVID19-world: A shiny application to perform comprehensive country-specific data visualization for SARS-CoV-2 epidemic, BMC Med. Res. Methodol., 20 (2020), 235. https://doi.org/10.1186/s12874-020-01121-9 doi: 10.1186/s12874-020-01121-9
![]() |
[24] | J. Gabry et al., R package shinystan: Interactive visual and numerical diagnostics and posterior analysis for Bayesian models, (2015). Available from: https://CRAN.R-project.org/package = shinystan |
[25] |
N. T. Stevens, L. Lu, Comparing Kaplan-Meier curves with the probability of agreement, Stat. Med., 39 (2020), 4621–4635. https://doi.org/10.1002/sim.8744 doi: 10.1002/sim.8744
![]() |
[26] |
T. C. Wang, Developing a flexible and efficient dual sampling system for food quality and safety validation, Food Control, 145 (2023), 109483. https://doi.org/10.1016/j.foodcont.2022.109483 doi: 10.1016/j.foodcont.2022.109483
![]() |
[27] |
T. C. Wang, Generalized variable quick-switch sampling as a novel method for improving sampling efficiency of food products, Food Control, 135 (2022), 108841. https://doi.org/10.1016/j.foodcont.2022.108841 doi: 10.1016/j.foodcont.2022.108841
![]() |
[28] |
M. H. Shu, T. C. Wang, B. M. Hsu, Integrated green-and-quality inspection schemes for green product quality with six-sigma yield assurance and risk management, Qual. Reliab. Eng. Int., 39 (2023), 2720–2735. https://doi.org/10.1002/qre.3381 doi: 10.1002/qre.3381
![]() |
[29] |
T. C. Wang, B. M. Hsu, M. H. Shu, Quick-switch inspection scheme based on the overall process capability index for modern industrial web-based processing environment, Appl. Stoch. Model. Bus., 38 (2022), 847–861. https://doi.org/10.1002/asmb.2667 doi: 10.1002/asmb.2667
![]() |
[30] | H. Okamura, T. Dohi, mapfit: An R-based Tool for PH/MAP parameter estimation, in Quantitative Evaluation of Systems, QEST 2015, Lecture Notes in Computer Science (vol. 9259), Springer, (2015), 105–112. https://doi.org/10.1007/978-3-319-22264-6_7 |
[31] |
C. Acal, J. E. Ruiz-Castro, D. Maldonado, J. B. Roldán, One cut-point phase-type distributions in reliability, an application to resistive random access memories, Mathematics, 9 (2021), 2734. https://doi.org/10.3390/math9212734 doi: 10.3390/math9212734
![]() |
[32] | N. Belgorodski, M. Greiner, K. Tolksdorf, K. Schueller, R package risk distributions: Fitting distributions to given data or known quantiles, R package version, (2017). https://CRAN.R-project.org/package = rriskDistributions |
[33] | J. F. Lawless, Statistical models and methods for lifetime data (2º ed.), John Wiley & Sons, 2003. |
[34] | S. Asmussen, O. Nerman, M. Olsson, Fitting phase-type distributions via the EM algorithm, Scand. J. Stat., 23 (1996), 419–441. http://www.jstor.org/stable/4616418 |
[35] | P. Buchholz, J. Kriege, I. Felko, Input Modeling with Phase-Type Distributions and Markov Models, Theory and Applications, Cham: Springer, 2014. https://doi.org/10.1007/978-3-319-06674-5 |
[36] |
K. Choi, S. M. Park, S. Han, D. S. Yim, A partial imputation EM-algorithm to adjust the overestimated shape parameter of the Weibull distribution fitted to the clinical time-to-event data, Comput. Meth. Prog. Bio., 197 (2020), 105697. https://doi.org/10.1016/j.cmpb.2020.105697 doi: 10.1016/j.cmpb.2020.105697
![]() |
[37] | A. Thummler, P. Buchholz, M. Telek, A novel approach for phase-type fitting with the EM algorithm, IEEE T. Depend. Secure, 3 (2006), 245–258. |
[38] |
A. Panchenko, A. Thummler, Efficient phase-type fitting with aggregated traffic traces, Perform. Evaluat., 64 (2007), 629–645. https://doi.org/10.1016/j.peva.2006.09.002 doi: 10.1016/j.peva.2006.09.002
![]() |
[39] |
H. Okamura, T. Dohi, K. S. Trivedi, Improvement of EM algorithm for phase-type distributions with grouped and truncated data, Appl. Stoch. Model. Bus., 29 (2013), 141–156. https://doi.org/10.1002/asmb.1919 doi: 10.1002/asmb.1919
![]() |
[40] |
P. Royston, D. G. Altman, External validation of a Cox prognostic model: Principles and methods, BMC Med. Res. Methodol., 13 (2013), 1–15. https://doi.org/10.1186/1471-2288-13-33 doi: 10.1186/1471-2288-13-33
![]() |
[41] | J. E. Ruiz-Castro, C. Acal, J. B. Roldán, An approach to non-homogenous phase-type distributions through multiple cut-points, Qual. Eng., 35 (2023), 619–638. |
[42] |
A. Bobbio, A. Horvath, M. Telek, Matching three moments with minimal acyclic phase type distributions, Stoch. Models, 21 (2005), 303–326. https://doi.org/10.1081/STM-200056210 doi: 10.1081/STM-200056210
![]() |
[43] |
T. Osogami, M. Harchol-Balter, Closed form solutions for mapping general distributions to minimal PH distributions, Perform. Evaluat., 63 (2006), 524–552. https://doi.org/10.1016/j.peva.2005.06.002 doi: 10.1016/j.peva.2005.06.002
![]() |
[44] | G. Horváth, M. Telek, Markovian performance evaluation with BuTools, in Systems Modeling: Methodologies and Tools, Springer, Cham, 2019. https://doi.org/10.1007/978-3-319-92378-9_16 |
[45] |
A. Alkaff, M. N. Qomarudin, Modeling and analysis of system reliability using phase‐type distribution closure properties, Appl. Stoch. Model. Bus., 36 (2020), 548–569. https://doi.org/10.1002/asmb.2509 doi: 10.1002/asmb.2509
![]() |
[46] |
M. Langer, Y. Zhang, D. Figueirinhas, J.-B. Forien, K. Mom, C. Mouton, et al., PyPhase—A Python package for X-ray phase imaging, J. Synchrotron Radiat., 28 (2021), 1261–1266. https://doi.org/10.1107/S1600577521004951 doi: 10.1107/S1600577521004951
![]() |