Filtering And Control For Discrete-Time Stochastic Systems With Incomplete Information

With the completeness of the computer and network communication technology,network technique is adopted for information transmission by more and more control systems.The combination of computer,network communication and control technolo-gy make data communication exists widely in control systems and the control and filter algorithms highly depend on communicaion network.The involvement of communica-tion network triggers all kinds of network-based random and uncertainty phenomena,such as communication delays,data dropouts,channel fadings and sensor or actuator saturations and so on.These network-induced phenomena together with the uncertain-ties in the system itself and the physical limits of the devices,among other factors,lead to a lot of incomplete information,which may damage the system performance.In order to lower the impact of the incomplete information,it becomes a crucial issue to develop the relevant filtering and control theory.Based on the existed work,this thesis mainly studies the filtering and control prob-lem for several classes of discrete-time stochastic systems with incomplete information and presents some novel filter and controller design scheme.The main content and contribution of this thesis are as follows.(1)This paper aims at designing a full-order non-fragile H∞ fuzzy filtering is s-tudied for a class of discrete Takagi-Sugeno(T-S)stochastic system with gain uncertainties and channel fadings.Through the Lyapunov stability theory and the stochastic analysis method,a sufficient condition is derived to guarantee the fuzzy filtering error system is exponentially mean-square stable with a prespec-ified H∞ disturbance attenuation.Based on this condition,by introducing the slack variables,the filter design problem is translated into a convex optimiza-tion problem with the linear matrix inequality constraints and the designed filter parameters can be obtained by solving this problem.(2)The finite-time dissipative filtering problem is investigated for a class of discrete stochastic system with the randomly occurring model uncertainties and channel fadings.A modified Rice fading model is introduced to describle the reality of the channel fadings,which includes not only the additive external disturbance but also the disturbance-dependent Gaussian white noise.Through the layer by layer analysis and processing,a sufficient condition is obtained which makes the filter-ing error system is finite-time stochastic bounded with the specified exponential dissipativity performance.Meanwhile,the corresponding solution method and the explicit representation of the filter parameters are given.(3)The finite-time dissipative control problem is considered for a class of discrete stochastic system with the randomly occurring model uncertainties and channel fadings.Through the intensive analysis,the sufficient conditions for the existence of the state feedback controller are established,which make the closed system is finite-time stochastic bounded and satisfy the specified exponential dissipativity performance,and the cone complementarity linearization iterative algorithm is given to solve this condition.(4)The distributed model predictive control problem for the uncertain discrete sys-tem subject to actuator saturations and packet dropouts with time-varying prob-abilities in controller-to-actuator(C/A)channels.The global system is decom-posed into several coupled subsystems and a novel augmented distributed con-troller model is proposed,based on which the distributed controllers are designed for each subsystem.Finally,the distributed model predictive controller design problem is converted into the parallel on-line solution of several sets of opti-mization problems involving linear matrix inequalities.Furthermore,an on-line iterative solving algorithm is proposed,then the convergence of the algorithm and the system robust stability are analyzed.Numerical examples are supplied to demonstrate the effectiveness of the results above.Finally,a conclusion is drawn and some following study aspects are presented.