Decentralized Zero-sum Games for Nonlinear Systems Based on Off-policy Learning Scheme

In this paper, the decentralized zero-sum differential games for nonlinear interconnected systems are investigated based on off-policy integral reinforcement learning algorithm. First, the decentralized zero-sum differential games of the overall system can be transformed into seeking the Nash equilibrium solution of each isolated subsystem. In order to obtain the Nash equilibrium solution, the off-policy integral reinforcement learning algorithm is introduced to solve the related HamiltonJacobi-Isaacs equations. The system dynamics can be completely unknown when implementing this proposed algorithm. In addition, three neural networks are constructed to approximate the optimal cost function, the optimal control policy and the worst disturbance policy, respectively. Finally, the validity of the proposed algorithm is verified through simulation results.