Optimal Consensus Model-Free Control for Multi-agent Systems Subject to Switching Topologies: Using Action Reinforcement Learning Method

This article investigates the optimal tracking problem of the multi-agent systems under switching topology by using action reinforcement learning method. Firstly, based on the action enhancement signal function, the performance index function of each agent is designed to be independent of neighbor nodes. Then, the action reinforcement learning algorithm is developed to improve the agent’s ability to use long-term information. Based on Bellman’s optimality principle, it is proved that the value function can converge to the optimum. Finally, the correctness of the theoretical is verified by a numerical simulation.