Research On Adaptive Observer-based Actuator Failure Compensation Control

Some adaptive observer-based actuator failure compenstation control problems are studied inthis dissertation. The proposed methods are verified on the explicit model of helicopter and themodel of Boeing747respectively.First, the history and development of FTC (fault tolerance control) are surveyed. Also thedevelopment of adaptive control technique is introduced. Based on the adaptive control theory,new actuator failure compensation schemes based on adaptive observer are presented for systemwith unknown paramteres and immeasurable state variables. In Chapter2, an adaptive observer isestablished, for the time invariant system which is in particular form, to obtain the estimated statevariables and the observation errors are shown stable and asymptotic convergence by Lyapunovstability theory. Then those estimated state variables can be used to construct the adaptivecontroller, which is capable of compensating the stuck failure. The designed control schemeensures the closed-loop signals are bounded and asymptotic tracking performance. Finally, theeffectiveness of proposed method will be verified on the explicit model of helicopter.Furthermore, in Chapter3, a new synthetical adaptive observer scheme is proposed to relaxsome assumptions used in the former design. The stability of the observer and asymptoticconvergence of the observation error are also guaranteed by the Lyapunov stability theory. Thenan adaptive controller, which is able to compensate the parametrizable time-variant failures, isconstructed by using the estimated state variables. The boundedness of closed-loop signals andasymptotic tracking performance are ensured by the proposed control scheme. The simulationresults verify the desired system performance in the presence of actuator.In Chapter4, an adaptive observer is formed to estimate the states of system with relativedegree. It is proved that the observation error is bounded and the adaptive observer is stable. Thenan adaptive controller is formed to compensate the parametrizable time-variant failures. Theclosed-loop signals and tracking error are proved to be bounded and the simulation of Boeing747shown the effectiveness of the proposed method.