Fault/Noise Reconstruction And Fault-tolerant Control Of Nonlinear Uncertain Systems

The faults will inevitably occur in the actual engineering systems with long-term operation. The system failures are often the main factors influencing the stable operation of the whole system. In order to guarantee the security and reliability of the system itself, the research on the fault detection and diagnosis(FDI) and fault tolerant control(FTC) is developed. Considering that most engineering systems are nonlinear and uncertain, the thesis deeply investigates fault/noise reconstruction and fault-tolerant control of uncertain nonlinear systems.Firstly, the problem for actuator fault reconstruction and fault-tolerant control of uncertain nonlinear systems is investigated. For the uncertain nonlinear systems with actuator faults, an auxiliary augmented state vector consisting of system state and actuator fault is introduced to construct an equivalent augmented system, and based on the augmented system an augmented sliding mode observer is designed for the state estimation and fault reconstruction. Through analyzing theoretically the estimation error, the LMI conditions are given to guarantee the stability of estimation error dynamics. Then based on the state estimation and fault reconstruction, the fault tolerant controller against actuator faults is designed to maintain the stability of system state.Secondly, the problem for sensor fault/measurement noise reconstruction and fault-tolerant control of uncertain nonlinear systems is investigated. For the uncertain nonlinear systems with sensor faults/measurement noises, an auxiliary descriptor state vector consisting of system state and virtual sensor fault(including real sensor fault and measurement noise) is introduced to construct an equivalent augmented descriptor system, and based on the descriptor system a descriptor sliding mode observer is designed for the state estimation and fault reconstruction. Through analyzing theoretically the estimation error, the LMI conditions are given to guarantee the stability of estimation error dynamics. Then based on the state estimation and fault reconstruction, the fault tolerant controller against sensor faults/measurement noises is designed to maintain the stability of system state.Lastly, the problem for multiple fault/noise reconstruction and fault-tolerant control of uncertain nonlinear systems is investigated. For the uncertain nonlinear systems with multiple faults(actuator faults and sensor fault/measurement noise), an auxiliary descriptor state vector consisting of system state, actuator fault and virtual sensor fault is introduced to construct an equivalent augmented descriptor system, and based on the descriptor system a descriptor sliding mode observer is designed for the state estimation and fault reconstruction. Through analyzing theoretically the estimation error, the LMI conditions are given to guarantee the stability of estimation error dynamics. Then based on the state estimation and fault reconstruction, the fault tolerant controller against multiple faults is designed to maintain the stability of system state. The proposed design methods not only effectively solve the difficult problem of continuous reconstruction of multiple fault/noise, but also are suitable for single fault(only actuator fault or sensor fault/ measurement noise) systems discussed previously.