Fault Detection For Networked System With Time-varying Packet Dropout Rate

Networked control systems connect distributed sensors,controllers,actuators and plants together by a communications network for the transmission of information to form an intermittent closed-loop feedback system,and complete control tasks of the system.Due to the limited communication capacity of the network media,there may be a nonzero probability that packet dropouts,which can deteriorate the performance of networked control systems,even result in the problem of instability.These dynamics bring new challenges to the analyses and syntheses of networked control systems.This paper mainly studies the designs of H_?filter and fault detection with time-varying packet dropout rate(PDR).This paper will focus on the PDR,which will jump from several different ranges.It can be described as a Markov chain(MC)as well as semi-Markov chain(SMC).PDR in the different cases uses different filter gain,which will greatly reduce the design conservatism.Under modeling error or parameter uncertain,the robust filtering problem is also investigated.We design the robust filter and fault detection filter based on norm-bounded.Main results and contributions are summarized as follows:Chapter 2 introduces two stochastic process: MC and SMC.In the meanwhile,Markov renewal chain(MRC)and embedded Markov chain(EMC)are also defined in this chapter.The differences and relationships between those concepts are illustrated in detail.Taking into account the large number of theorems of the follow-up chapters to prove that in this chapter also introduced Shur complement and H_?control theory.In order to analyze the system performance,this chapter also introduces stability in sense of mean-square,? mean square stability,nonlinear system stochastic stability,and infinite series convergence theorem.In order to make preparation for fault detection filter(FDF)design,this chapter is also concerned with the H_?filtering problem for a class of discretetime networked systems under unreliable communication links.It is assumed that the data transmission between the plants and the filters are subject to randomly occurred packet dropouts satisfying Bernoulli distribution,such that the system is stochastically stable in sense of mean-square and achieves a guaranteed H_?performance.This chapter analyzes the robustness based on such norm bounded uncertainties.Finally,a numerical example is provided to illustrate the performance of the proposed approach.Chapter 3 investigates the H_?fault detection problem for MJSs with time-varying PDR.By using a FDF as residual generator,the FDF design is formulated as an H_?filtering problem.The communication channels between the plant and the FDF are imperfect.LMI-based conditions have been presented for the admissible filters,which ensure that the system is stochastically mean-square stable with a guaranteed H_?index ?.This chapter also analyzes the robustness based on such norm bounded uncertainties.Finally,a tunneldiode circuit example is presented to demonstrate the effectiveness of the filter design methodologies and the necessity of considering the uncertainty and variation of PDR.Chapter 4 focuses on the H_?FDF design problem for a class of networked systems with random packet dropouts,which are governed by a Bernoulli distributed sequence.The PDR is uncertain and variable,and described by a semi-Markov stochastic process.A more general class of Lyapunov functions that depends not only on the system modes,but also on the staying time during the current system mode is introduced.The parameters or the types of sojourn time probability density functions(PDFs)for each mode can be different via semi-Markov kernel(SMK)approach.The numerically testable sufficient conditions on the existence of a desired FDF are presented such that the filtering error system is ?-error mean square stability(?-MSS)with given H_?performance.This chapter also analyzes the robustness based on such norm bounded uncertainties.Finally,an example of the inverted pendulum is presented to demonstrate the effectiveness of the filter design methodologies as well as the superiority of introducing the concept of SMC and the necessity of considering the variation of PDR.