Model-based Fault Diagnosis Of Networked Control Systems

With the rapid development of networks, more and more close-loop control systems communicate via networks. Under this background, Networked Control Systems (NCS) become a very popular research area in recent years. NCS can be defined as a control system with at least one close-loop through the network. NCS is the a kind of distributed system which contributes to its various advantages, such as information sharing, resources conserving, etc. comparing to traditional control systems. However, with the affects from networks, a lot of new topics including network-induced delay, packets loss, asynchronism, etc. which never occurs in traditional control systems, need further study. In addition, in order to improve the reliability of NCS, fault diagnosis of NCS is also worth to research.This thesis introduces the basic definition and classification of fault diagnosis, and makes a detailed introduction about the history and basic idea and techniques of model-based fault diagnosis. The thesis also introduces the NCS'background, advantages and network-induced issues comparing to traditional control systems. In addition, the thesis discusses the information-scheduling model of NCS, and also presents its augmentation model.In this dissertation, traditional model-based fault diagnosis techniques are applied into the periodic time-varying information-scheduling model, including Luenberger Observer approach, Unknown Input Observer approach, Eigenstructure Assignment approach. Moreover, the thesis discusses the network-induced delay issue in the information-scheduling model, analyses the affects from time delay less than one sampling period. In this thesis, the random time delay in information-scheduling model is treated as model uncertainty, and a model-based diagnosis system is designed via optimal parity relation approach.Simulations of diagnosis systems in this thesis are based on Matlab, and all achieved the expected results.