Motion Control of AUV Vertical Plane Based on RBF Neural Network Backstepping Sliding Mode

Abstract

Aiming at the vertical plane motion issue of autonomous underwater vehicle (AUV) with modeling uncertainty and external perturbation, a backstepping sliding mode control algorithm is proposed on the base of multiple radical basis function (RBF) neural networks. Firstly, the kinematics and dynamics equations of AUV are established. Secondly, according to the second method of Lyapunov, a robust sliding mode controller on the base of backstepping control algorithm is designed to eliminate the errors of states and improve the response speed. At the same time, multiple RBF networks are adopted to adaptively compensate for the uncertainty or nonlinear term in the AUV motion equation and external perturbation. Finally, the stability analysis for the AUV control system is given based upon the second method of Lyapunov. The simulation results demonstrate that: The presented controller can enable the AUV to reach the desired depth at a quick speed and high accuracy. In comparison to the traditional sliding mode controller, the given method possesses higher adaptability and better dynamics performance.