Adaptive Neural Network Actuator Failure Compensation Scheme for Quadrotor Trajectory Tracking Control

Aiming at the problems of parameter uncertainty and actuator failure, a trajectory tracking control scheme based on adaptive RBF neural network minimum parameter learning method and dynamic sliding surface is proposed. Firstly, a nonsingular dynamic sliding mode surface is designed, and the terminal sliding mode control law for trajectory tracking of the quadrotor UAV is derived. Secondly, an adaptive RBF neural network is used to compensate the uncertainties and actuator faults of the control system. The single parameter online learning is adopted to replace the whole weight vector online learning, which reduces the complexity of the controller design and avoids the problem of dimension disaster. Thirdly, the convergence and stability of the closed-loop system are proved using Lyapunov theory. Finally, the numerical simulation experiments verify the effectiveness of the control scheme. Compared with the traditional sliding mode control method, the RMSE (Root Mean Square Error) of trajectory tracking is reduced by 8.2%.