Bias-policy iteration based adaptive dynamic programming for unknown continuous-time linear systems

In this paper, a bias-policy iteration method for solving the data-driven optimal control problem of unknown continuous-time linear systems is proposed. Firstly, a model-based bias-policy iteration method is given and its convergence is rigorously proved. Then the data-driven implementation for the proposed method is then introduced without using the information of the system matrices. The relationship between the proposed method and the existing policy iteration method and value iteration method is also analyzed. Compared with the existing policy iteration method, the most significant advantage of the proposed method is that, by adding a bias parameter, the condition of the initial admissible controllers can be further relaxed. Simulation examples verify the effectiveness of the proposed bias-policy iteration method.