Optimal group consensus control for the second‐order agents in the coopetition networks via adaptive dynamic programming and event‐triggered methods

Abstract

AbstractThis article investigates the optimal group consensus control problem for the second‐order multi‐agent systems by using event‐triggered and adaptive dynamic programming methods. The proposed method only uses the interaction information among the agents. Comprehensively considering the coopetition coupling interactions among the agents, a novel tracking error function is addressed for meeting the requirement of group consensus. On this basis, we use the Bellman optimality principle to formulate the optimal group consensus problem. To implement the proposed method, the actor–critic neural networks are used to approximate the iterative performance index functions and the control policies in real time. Meanwhile, the event‐triggered condition with filter function is installed for each agent to reduce the control cost. Therefore, the weights of actor–critic networks are updated only at the event‐triggered instants. The weight estimation errors and the control policy estimation errors are proved uniformly ultimately bounded and the Zeno behavior in the system can be avoided. Finally, the comparative simulation results show that the proposed method can greatly reduce the update times of the control policy of the system.