부산대학교 도서관

This paper investigates the surge velocity control of an autonomous surface vehicle (ASV) in the context of a specific performance index subject to partially unknown model nonlinearities. An extended state observer (ESO)-based actor-critic structure is proposed such that an optimal surge velocity control performance is achieved in terms of a predefined performance cost function. First, a second-order ESO is developed to recover the unknown model parameter and time-varying environmental disturbances. Next, a Hamilton-Jacobi-Bellman equation is established based on the estimated model by using the ESO. Then, two neural networks, namely, the critic neural network and the actor neural network, are developed and the optimal surge force is presented based on the actor neural network. It is proven that the closed-loop system is input-to-state practically stable. Simulation results validate the proposed ESO-based actor-critic framework's effectiveness in achieving optimal surge velocity control for the ASV.