Along with development of network and computer technology, networked control systems (NCSs) have been widely applied to many complex control systems. NCSs are distributed control systems, where the information of control system components (sensors, controllers, actuators, etc.) is exchanged via wired/wireless communication networks. Compared with the traditional control systems, NCSs have many advantages, such as, low cost of installation and maintenance, ease of diagnosis, and flexible archi-tectures. However, because of the complexity introduced by networks, NCSs are more vulnerable to faults. Therefore, it is very significant to guarantee security and reliability of NCSs. In order to improve the safety and reliability of NCSs, this thesis focuses on the robust fault detection and fault tolerant control for several kinds of NCSs with network-induced delay and data packet dropout. The main contents and contributions are as follows.1. Robust fault detection (RFD) problem for two classes of linear systems is dis-cussed. The fault detection filter is designed to generate the residual signal, which can be used to detect the fault signal. One of key issues related to a model-based fault detection for NCSs is concerned with its robustness against modeling errors, external disturbances, network-induced uncertainties and sensitivity to faults.The first class of plant is a discrete-time system with model error. The data packet dropout occurs in the sensor-to-controller (S-C) link, and is governed by a Bernoulli distribution. The NCS with stochastic variable is modeled, and a robust fault detection filter (RFDF) is designed based on the observer method, then the sufficiency of asymp-totically mean-square stable is achieved. In the residual evaluation part, a threshold involved false alarm rate is presented, and the problem of estimating false alarm rate is analyzed.The second class of plant is a continuous system. Network-induced uncertainties including time-varying unknown network-induced delay and data packet dropout are assumed to be existing in both S-C link and controller-to-actuator (C-A) link, where the total network-induced delay is transformed into the uncertainty of system model, and the data packet dropout is described as a two-state Markov chain. Then, the closed NCS with network-induced uncertainties is modeled as a Markovian jump system with model uncertainty. An observer-based fault detection filter is presented as a residual generator. Sufficient condition for asymptotically mean-square stable of residual dy-namics is derived, and the desired RFDF is obtained, which is constructed in terms of certain linear matrix inequality (LMI). In this part, the model uncertainty approach is adopted to deal with the time-varying unknown network-induced delay, compared with the methods of eigendecomposition, approximate decoupling and Taylor expansion, the model uncertainty approach can avoid on-line computing the parameters of fault detec-tion filter.2. The RFD problem for a nonlinear state-delay NCS with model uncertainty and data packet dropout is considered. In both S-C link and C-A link, the random data packet dropouts are described by two variables, which obey the Bernoulli distribution. An observer-based RFDF is presented as a residual generator, and the residual dynam-ical system is modeled as a novel stochastic nonlinear NCS with state-delay, model uncertainty, external disturbance. A performance index is proposed to ensure robust-ness to data packet dropout, disturbance, model uncertainty, and sensitivity to fault. Sufficient condition for asymptotically mean-square stable of this residual dynamical system is derived. Then, the desired RFDF is obtained, which is constructed in terms of certain LMI.3. The design of fault tolerant control (FTC) for NCS with actuator fault and random data packet dropout is investigated. The data packet dropout in both S-C link and C-A link is described by two switches, which can be modeled as a discrete event system with known rate. After introducing the matrix of actuator failure, the closed-loop NCS is developed. Applying the theory of asynchronous dynamical system, the sufficiency of exponential stability for the NCS is obtained.4. The output-feedback FTC problem for NCS with random access is studied. The network-induced delay and data packet dropout are assumed to be in both S-C link and C-A link. A novel switched output feedback controller is proposed. Then, the model of closed-loop NCS with actuator (complete/partial) failure is established, which is also a discrete-time Markovian jump uncertain system. The sufficient and necessary conditions for exponential mean square stability of such NCS is derived. Moreover, the sufficient condition for designing the controllers are presented.5. A robust fault-tolerant controller design problem for NCS with random data packet dropout in both S-C link and C-A link is addressed. The random packet dropout, sensor failures and actuator failures are characterized by a binary random variable. By designing a dynamic fault-tolerant controller, a novel stochastic NCS model with state-delay, model uncertainty, disturbance, probabilistic sensor failure and actuator failure is proposed. The sufficient condition for asymptotically mean-square stable of the NCS is derived and the controlled output satisfies a new H∞performance constraint. The parameters of the fault-tolerant controller is obtained by solving the LMI.Finally, some open issues and the future work in RFD and FTC of NCSs are dis-cussed. |