H∞ optimal tracking control for remotely operated vehicle

Abstract

To deal with this problem for tracking the depth-varying trajectory of remotely operated vehicle (ROV), state variables is introduced to system transformation for converting trajectory tracking problem into an optimal control problem. For this system, the H∞ optimal control is added basing on the adaptive dynamic programming algorithm (ADP), and the problem is regarded as the process of a two-player zero-sum differential game. Then we use the critic network to estimate the value function, and propose a online policy iteration algorithm to solve the HJI equation basing on the actor network and the disturbance network. Considering the limited output of the controller, we introduce a non-quadratic functional into the performance index function to solve the saturation problem. By using the Lyapunov stability theorem, we prove that the state of the closed-loop system and the weight estimation error of the neural network are uniformly bounded. Finally, an example is used to prove the feasibility and effectiveness of the method.