Abstract: The aircraft model shows obvious time-varying characteristic due to the uncertainty of aerodynamic parameters. The fault diagnosis of the flight control systems is a difficult issue. A sensor fault estimation approach based on aerodynamic parameter identification and iterative learning is proposed by taking the longitudinal motion model of some unmanned aerial vehicle as the study subject. The augmented cubature Kalman filter (ACKF) is used for the aerodynamic parameter estimation so that the system model can be identified online. Once a fault comes up, the currently identified aerodynamic parameters are applied to system modeling in the local flight envelope, and a fault estimator is constructed using the iterative learning algorithm. Furthermore,a novel iterative learning algorithm based on the essence of extended state observer (ESO) is designed to improve the fault estimation speed. The fault simulation experiments are conducted to verify the feasibility and effectiveness of the proposed approach.

Key words: control science and technology, sensor fault, flight control system, aerodynamic parameter, augmented cubature Kalman filter, iterative learning, extended state observer