부산대학교 도서관

In this article, reinforcement learning based fault-tolerant control for the nonlinear offshore platform is studied. First, the dynamic model of the nonlinear steel-jacket offshore platform with active mass damper and actuator fault is investigated then an affine nonlinear representation for the offshore platform is obtained. By estimating actuator fault and irregular wave force and inserting their estimation into the cost function, the disturbance control and fault-tolerant control problem (FTC) are converted into optimal control problem. Online policy iteration is used to minimize the performance index. Finally, simulation results show the effectiveness of our method. • The nonlinear model of the offshore platform with actuator fault is considered. • Using a novel observer, various types of the external wave force can be considered. • The proposed method benefits the merits of optimal control and intelligent control. • The proposed method can compensate for fault and disturbance simultaneously. [ABSTRACT FROM AUTHOR]