Fault diagnosis and adaptive reconfiguration control for actuator loop in aeronautic propulsion system

A method of fault diagnosis and adaptive reconfiguration control based on dynamical recurrent neural network was presented for actuator loop in aeronautic propulsion system. Network weighted values were modified on-line to compensate for zero-deflection and uncertainty of actuator loop. The threshold-value differentiate method was applied for real-timely watching and diagnosing the actuator loop fault. After that, adaptive reconfiguration control was realized with the output of dynamical recurrent neural network model substituting for fault sensor output to closed loop. Through digital simulation research for the actuator fault mode, the results show that the method put forward can exactly diagnose the fault of actuator loop in time and effectively realize adaptive reconfiguration control.