Data-Driven Fault-Tolerant Distributed Formation Control of Multiagent Systems With Switching Topologies

In this paper, we study the problem of data-driven fault-tolerant formation control for linear multiagent systems with actuator faults in a directed communication switching topology. Initially, constructing a distributed observer to generate using the local information as a reference, the convergence of the observer under the switching topology is proved. Subsequently, we leverage reinforcement learning theory to determine the optimal control strategy, even when the parameters of the subsystems within the multi-agent network are not known. In the final step, we apply the acquired control strategy to devise a data-driven fault-tolerant controller, aimed at minimizing fault impacts as they arise. The efficiency of our proposed solution is confirmed through simulation experiments.