Active Fault Tolerant Control Of Dynamic Systems With Actuator Faults

Fault tolerant control (FTC) is receiving more and more attention due to the increasing demand for higher safety and reliability of dynamic systems. The basic function of an FTC system is that it can guarantee satisfactory performance not only when all the components in the system are healthy but also faulty. Up to now, many results have been achieved on FTC for linear systems. However, for nonlinear systems and switched systems, the results are limited. In this dissertation, the problem of FTC against actuator faults for a class of nonlinear systems and switched linear systems is considered.Inspired by the limited results in FTC for nonlinear systems with actuator faults, we firstly study the the FTC problem for a class of nonlinear systems with matched actuator faults. A generalized fault estimation observer is designed to collect the fault information. Based on the information, a fault tolerant controller is designed and its effectiveness is proved. Then, we consider the FTC problem of feedback linearizable systems with unmatched actuator faults. Based on the former mentioned fault estimation observer and the cascade design method, we design a fault tolerant controller and proved its effectiveness.During the research, we find that the time instant at which the controller switches from the original one to the fault tolerant one, i.e., reconfiguration instant, can greatly influence the transient performance of an active FTC system. However, there are few results on how to choose an appropriate reconfiguration instant. We study the optimal reconfiguration instant problem for a typical active FTC scheme. A two-stage optimization method is proposed based on differential games, which provides not only theoretical basis but also algorithm for the problem.Finally, the FTC problem of switched linear systems with switching faults is considered, which has almost no results in the literature. For a class of time-driven switched systems, periodically switched linear systems, we design a fault detection and diagnosis (FDD) unit by optimization method, which overcomes the difficulty of fault diagnosis. To overcome the difficulty of recovering the normal performance, we reconfigure the post-fault system by an idea of 'mode adapter', i.e., making the subsystems that can be healthily switched to recover the behaviors of the ones that can not be switched to. For a class of state-driven switched linear systems, we introduce the concept of 'cushion' areas to design the normal controller and the FDD unit simultaneously, which overcomes the difficulty of fault isolation. Based on the fault isolation result, we then design an FTC law for the post-fault system and prove its effectiveness theoretically.