Fault detection and diagnosis for actuator based on neural network observer

An algorithm of actuator fault detection and diagnosis was presented for a class of nonlinear systems with unknown function based on neural network observer. It is not necessary that the matching condition and available state be assumed in the system. A neural network was used to approximate the nonlinear item of the monitored system to improve the accuracy of state estimation, and the state estimation error is proved to be zero asymptotically. The residual of the observer can be used for fault detection, and an adaptive law was adopted to estimate the fault on-line. The simulation shows the effectiveness of the proposed methodology.