Along with the continuous progress and development of electronic technology andnetwork technology, the application of network based control systems become a new trend inthe development of automation control filed. The research of networked control systemshave also rapidly developed and become a new hot spot in the study of networked controlsystems. In this paper,based on using network to transport the information, Bonulli model,Linear Matrix Inequlity (LMI) method, Markov theory and robust theory are applied to studynetworked control systems which contain time delay and packet dropout. Some kinds of stateobservers are designed and the effectiveness of results are also validated by simulationresearch. The main contents are summarize as follows:(1) The problem of fault detection of networked control systems that contains uncertainlong time delay is studied. According to network transport time delay, which is more than onesample period and is uncertain too, the system is modeled into Markov jump system. Base onMarkov jump system, state observer was adopted to designed and Lyapunov function isselected to analyze the stability of the observer error system and detect faults in this kind ofnetworked control systems.(2) Fault detection of a kind of networked control system with stochastic random packetdropout is discussed. Networked control systems which contain this kind of problem aremodeled as a special system that has Bernoulli parameters. State observer is designed topropose sufficient conditions for asymptotical stability of state error system and detect faults.(3) According to situations that the networked control systems contain long delay andpacket dropout. Two methods are discussed to study this kind of networked control systems.The first method divide packet dropout into two different situations, dropout or not. Then atime delay compensation based fault observer is designed to study fault detection. The othermethod use deal with uncertain long delay specially, the delay is modeled as uncertain parts ofthe system. Then packet dropout and uncertain external disturbance are considered to designstate observer as a reference model of optimal filter. On the basis of that, a optimal filter isadopted to detect faults. |