A game-based distributed fault-tolerant control method for multi-agent systems

This paper studies the problem of game based distributed optimization control for first-order multi-agent sys-tems with actuator faults. In non-cooperative games, the action information of all agents usually needs to be known using a centralized optimization solution method. Conversely, the agent cannot obtain the action information of non-neighbor agents in a distributed design. To obtain action information of non-neighbor agents, a distributed estimation algorithm based on leader-follower consensus is designed. In addition, this work also considers actuator faults that occur in any agent, due to which the MAS may fail in reaching Nash equilibrium. To solve this problem, A fault-tolerant optimization control method is proposed guaranteeing that the MAS can still reach the Nash equilibrium while the actuator faults occur. Two simulation examples demonstrate that the proposed fault-tolerant control algorithm for multi-agent system is feasible.