RBFNN Variable Structure Controller for MIMO System and Application to Ship Rudder/Fin Joint Control

Abstract

Aiming at a class of multiple-input multiple-output (MIMO) system with uncertainty, a sliding mode control algorithm based on neural network disturbance observer is designed and applied to ship yaw and roll joint stabilization. The nonlinear disturbance observer is finished by radial basis function neural network and with that a terminal sliding mode control algorithm is proposed. The asymptotic stability of closed-loop system is proved based on Lyapunov theorem. The proposed control law is applied to anti-roll control under simulative wave disturbances. Simulation results verified robustness and effectiveness of the suggested algorithm. A good anti-rolling effect is achieved and yaw angle is also reduced greatly with less mechanical loss.