Fault Detection And Diagnosis Based On Analytical Model

Aiming to solve some existing problems and follow trends in fault detection and diagnosis (FDD), combined with new results obtained in relative disciplines, the research work about FDD was carried out. The main work in this thesis is summarized as follows.Firstly, an algorithm of actuator fault diagnosis was developed for a class of nonlinear control systems with plant modeling uncertainties and sensor modeling uncertainties based on the observer. The assumptions that plant modeling uncertainties and sensor modeling uncertainties are less than two constants, and two certain functions are considered. The adaptive compensation term was adopted in the observer to compensate the fault and modeling uncertainties after the actuator had a bias. The dead-zone operator which used in the adaptive algorithm improves the algorithm's robustness. The adaptive law was introduced to estimate the fault. The simulation results show the effectiveness of the proposed methodology.Secondly, the problem of fault detection and diagnosis for a class of nonlinear systems with unknown uncertainty was studied in this thesis. A neural network was constructed to approximate the fault on-line. The dead-zone operator adopted in the adaptive algorithm improves the algorithm's robustness. The adaptive compensation term in the observer improves the performance of the observer.Lastly, an algorithm of actuator fault diagnosis was developed for a class of nonlinear-time-delay systems with plant modeling uncertainties based on the adaptive observer. The compensation term in the observer reduces the influence caused by the system modeling uncertainties, so that the observed error converges to zero. An adaptive law was adopted to estimate the fault. By theoretical analysis, the designed observer is proved to be stable with the observer error converging to zero.Through the research work in this paper, the problems of FDD for the several classes of uncertain nonlinear systems have been properly solved. By making use of some effective methods including adaptive compensation, neural network observer, sliding-mode observer, several class of problems of FDD for nonlinear systems are...