Data-Driven Fault-Tolerant Bipartite Consensus for Multiagent Systems With Directed Topology

This article investigates the model-free fault-tolerant bipartite consensus of multiagent systems under directed topology. The radial basis function neural network (RBFNN)-based fault estimation technique is constructed for acquiring unknown actuator faults information directly, in which the topology structure and the information interaction among agents are adequately considered. Compared with the existing method, updating weights using RBFNN estimation is avoided. By utilizing the obtained fault estimation, a distributed model-free adaptive fault-tolerant control (FTC) strategy is developed to achieve bipartite consensus. Unlike other bipartite consensus control techniques, the constructed FTC mechanism does not require accurate system model and structure information, and uses solely the agents' input/output data. Finally, a simulation is performed to verify the proposed mechanism's efficacy.