Prescribed-time fuzzy resilient learning formation control for nonlinear multiagent systems under DoS attacks

This paper is concerned with the prescribed-time optimal formation control problem for nonlinear multi-agent systems (MASs) under denial-of-service (DoS) attacks. The core challenge lies in estimating the leader’s information when communication between agents is intermittently disrupted by attacks. To overcome this problem, a distributed prescribed-time resilient observer is established for each agent to estimate the leader’s state. In view of the estimated results, a novel scale function and transformation relationship are designed, ensuring that the formation tracking error reaches a predefined accuracy within a prescribed time. Meanwhile, a shifting function is developed in the coordinate transformation to eliminate the constraint on the initial formation tracking error. Subsequently, the actor-critic architecture-based reinforcement learning (RL) algorithm together with the backstepping technique is utilized to optimize system performance. By employing Lyapunov stability theory, the boundedness of all closed-loop system signals is proven. Finally, the superiority of the designed method is validated through a simulation example.