Adaptive fault-tolerant controller for morphing aircraft based on the L2 gain and a neural network

Abstract

This paper presents an adaptive neural network fault-tolerant controller approach for the parameters of morphing aircraft that change violently during changing sweep angle processes and with morphing mechanisms that are prone to failure. We establish a nonlinear dynamic model of a morphing aircraft and analyze the variation laws of the aircraft parameters at different sweep angles. Considering the dynamic changes of the aircraft in morphing and morphing mechanism failure, by using recursive damped least squares (RDLS) to estimate the aircraft system model and offset the adverse effects of parameter mutation on the system, a radial basis function neural network (RBFNN) is used to further compensate for unmodeled parts such as actuator delay. The L2 gain is used to design a robust controller to constrain the errors generated by the RDLS and RBFNN in the learning process. The closed-loop stability is guaranteed by Lyapunov stability theory and the technique of compact set. Lastly, the comparison results of simulation indicate that this method satisfies the design criteria in the morphing process and fault situations, confirming the robustness of the controller.