Adaptive Optimal Control for Multi-player Systems with Completely Unknown Dynamic

In this paper, we propose model-free adaptive dynamic programming based on identifier-critic neural networks for a multi-player system where the dynamics are completely unknown. First, we develop an identifier network that identifies a multi-player system with completely unknown dynamics. Then, based on a critical neural network, each player is approximated to minimize the cost function for each player resulting in the best controller for each player. Subsequently, the Lyapunov method is used to prove that the NN estimation errors and the system states are uniformly ultimate boundedness (UUB). Finally, the effectiveness of the proposed method is verified by a numerical simulation.