Stability Analysis And Fault Detection For Nonlinear Stochastic Networked Systems

Networked control systems (NCSs) are feedback control systems, in which information (input and output) is exchanged via network among control systems components (plant, sensors, controller, actuators, etc.). Recently, there are increasing interests on fault detection (FD) of NCSs. However, compared with the rich results in control and estimation of NCSs, only a limited number of contributions to FD of linear NCSs with delay or data loss, and network only exist in transmission channel from sensor to control. This dissertation studies limited transmission nonlinear NCSs with both packet dropout and delay, and network exists in both from sensor to controller and from controller to actuator. Based on the obtained stochastic model, by employing Lyapunov functional theory and linear matrix inequality technique, corresponding conclusions of stability and fault detection are achieved. The main contributions of this dissertation are listed as follows:Firstly, we give a review of previous results on NCSs. Specificly, delay and data loss are two main attention; their mathematics model may be stochastic or deterministic and also be linear or nonlinear; and the progress of NCSs in control theory and fault detection.Secondly, robust stability of nonlinear NCSs with two successive delays is considered. Especially, for nonlinear NCSs with non-uniformly distributed delay, observer-based fault detection or robust fault detection and filter-based fault detection are focused. Corresponding conclusions are delay-interval-dependent, which reveal the effect of non-uniformly distributed networked-induced delay and nonlinear level on fault detection of nonlinear NCSs.Thirdly, stability and fault detrection of nonlinear NCSs with random packet dropout and delay distribution character in both from sensor to controller and from controller to actuator are disscussed. By choosing appropriate Lyapunov functional, the conclusions of stability and fault detection are obtained, which depend on not only delay-interval but also on delay-interval-occurrence-rate and joint data transmission rate. Especially, the trade-off phenomenon between delay bound and joint data transmission rate is found, which typically resuls from the limited network bandwidth. Finally, stability and fault detrection for nonlinear limited transmission NCSs with Markovian data loss and delay is investigated. Novel nonlinear Markovian jump system models are proposed to account for limited transmission NCSs with data loss and time-varying delay in both from sensors to controller and from controller to actuators. Based on the obtained nonlinear stochastic model, employing observer-based or filter-based fault detection filter as residual generator, the addressed fault detection problem is converted into an auxiliary nonlinear H∞control and H∞filtering problem, respectively. Then, with the help of Lyapunov-Krasovskii functional approach, sufficient conditions for the desired fault detection filter are constructed in terms of certain linear matrix inequalities, which depend on not only the delay interval but also on the two irrelevant Markov chains.In order to demonstrate the effectiveness and feasibility of proposed design techniques, simulation examples are presented by employing Matlab LMI Toolbox.