Fault Tolerant Control Design For Hybrid Systems

The hybrid systems (HS) considered in this thesis consist of a series of continuous modes and a logic switching scheme. These modes are switched among each other according to a given switching law. The study of HS is motivated by the fundamentally hybrid nature of many modern systems. Over the last decade, significant progress has taken place in the control theories of HS. Faults in automated processes will often cause undesired reactions of a controlled plant. Fault Tolerant Control (FTC) aims at guaranteeing the system goal to be achieved in spite of faults. Until now, only a few FTC results have been reported for HS. Since many complex systems have to be modeled by HS, the FTC theory of HS needs to be developed for its theoretical values as well as practical ones. Faulty behaviors and FTC techniques of HS are investigated systematically in this thesis. Two main kinds of faults are considered: Continuous fault that affects each continuous system mode; Discrete fault that affects the switching condition. The FTC design has two main objectives: To maintain the continuous performances including various stabilities of the origin and the output tracking/regulation behaviors along the trajectories of HS; To maintain the discrete specifications that have to be followed by HS, e.g. a desired switching sequence.Firstly, for HS with various switching, e.g., time dependent switching, state dependent switching, impulsive switching and stochastic switching, a set of FTC methods based on continuous system theories are proposed to maintain the systems' continuous performance. Two natural ideas are considered: One way is first to design FTC law to stabilize each faulty mode, and then apply the stability results of HS. Another way is to research directly the stability of HS without reconfiguring the controller in each unstable faulty mode. The FTC goal can be achieved if the negative effects resulting from unstable faulty modes are compensated for by that of stable modes. Secondly, for HS where discrete specifications are imposed, a set of schemes are derived from discrete event system (DES) point of view to keep these discrete specifications. The key idea is to reconfigure the discrete part and combine the reachability of the continuous dynamics, such that the specification is maintained. Two major DES models, i.e. finite state machine and Petri net are discussed.Finally, based on HS approaches, several novel supervisory FTC schemes are developed to improve non-hybrid system's performance. The proposed FTC schemes do not need a series of models or filters to isolate the fault, but only rely on a simple controller switching scheme. The stability of the system during the fault diagnosis and FTC delay can be guaranteed. The materials in the monograph have explicit and broad practical backgrounds. Many examples are taken to illustrate the potential of the obtained theoretical results, e.g. Circuit systems; DC motors; CPU process; Manufacturing system; Intelligent transportation systems and electric automated vehicles, etc.