An efficient model‐free adaptive optimal control of continuous‐time nonlinear non‐zero‐sum games based on integral reinforcement learning with exploration

• DOI: 10.1049/cth2.12610
• Publication Date: 2023-12-25T04:45:07Z
• Authors: Lei Guo; Wenbo Xiong; Yuan Song; Dongming Gan
• Keywords: Zero-sum game

Abstract To reduce the learning time and space occupation, this study presents a novel model‐free algorithm for obtaining Nash equilibrium solution of continuous‐time nonlinear non‐zero‐sum games. Based on integral reinforcement method, new HJ equation that can quickly cooperatively determine strategies all players is proposed. By leveraging neural network approximation gradient descent simultaneous adaptive tuning laws are provided both critic actor weights. These facilitate estimation optimal value function policy without requiring knowledge or identification system's dynamics. The closed‐loop system stability convergence weights guaranteed through Lyapunov analysis. Additionally, enhanced to number auxiliary NNs used in critic. simulation results two‐player game validate effectiveness proposed algorithm.