Fault Diagonsis And Fault Tolerant Control Based On Adaptive Observer

Actuator fault accommodation for MIMO (multiple-input-multiple-output) system and fault accommodation of a kind of nonlinear system are studied in this dissertation. The proposed method of actuator fault accommodation is simulated in the semi-physical flight control simulation platform.First, the history and the development of FDI (fault detection and isolation) and FTC (fault tolerant control) are surveyed. Also, the adaptive control technique and neural network are introduced.According to adaptive control theory, a new actuator fault accommodation scheme based on adaptive observer for MIMO systems is presented. The method estimates the loss of actuator effectiveness based on the augmented error, which is composed of the output estimation error generated by an adaptive observer and the auxiliary error. The actuator fault estimation can be used to reconfigure the control law to recover the system performance.Furthermore, a fault accommodation method for a class of nonlinear system, which can be changed into output-feedback form, is proposed using a neural-network-based adaptive observer. The system faults can be detected through the residual generated by the proposed observer. The stability of closed-loop error system is guaranteed by Lyapunov stability theory. Then the controller is designed for the faulty system to recover the system performance.Finally, the method of actuator fault accommodation for MIMO system is successfully applied to the simulation of Semi-physical flight control platform, the simulation results show the effectiveness of the proposed methods.