Disturbance Rejection Control of Morphing Aircraft Based on RBF Neural Networks

To enhance the disturbance rejection of morphing aircraft, an adaptive back-stepping control method based on the RBF neural network is proposed. Firstly, a nonlinear dynamic model of wing-sweep morphing aircraft is established and decomposed into an altitude subsystem and a velocity subsystem. The aerodynamic parameters and model parameters are modeled as functions of sweep angle, and the latter is expressed as pure function of time. Then, an adaptive back-stepping control based on RBFNN algorithm is proposed for the altitude subsystem and the velocity subsystem of the aircraft. The RBF neural network adaptive algorithm is combined with the back-stepping control to estimate the uncertain disturbance during the wing transition process so as to compensate for the control law output by the back-stepping control. By utilizing the fast convergence speed and strong approximation ability of the RBF neural network, uncertain disturbances caused by the deformation are estimated. Compared with the traditional extended state observer, the RBF neural network adaptive algorithm has strong robustness, which can solve the chattering problem caused by wing transition process. A differential auxiliary signal is introduced and a corresponding second-order command filter is designed to replace the continuous differentiation of the virtual control law. This not only avoids ‘explosion of complexity’, but also reduces the large overshoot caused by direct feedback. Thereafter, the weight update law of the RBF adaptive algorithm is obtained by Lyapunov stability theory, and the stability of the control system is proved at the same time. Finally, comparative simulation experiments are used to show that the proposed control strategy is preferable, and the error between actual values and the estimated of the RBF adaptive algorithm can be controlled within 0.05%.