Distributed fault identification and fault-tolerant control for multi-agent systems

This paper studies the consensus-tracking based distributed fault identification (DFI) and distributed fault-tolerant control (DFTC) problems for a multi-agent system modeling by a large-scale power generator network. The objective is to enable all the agents to coordinately detect and identify the faults without losing desirable system performance. First, for the distributed fault diagnosis part, a model-based distributed state observer (DSO) is designed to obtain the residual and then a nonlinear distributed fault estimator is developed to identify the faults. By exploiting the redundancies from the distributed fault diagnosis part, a novel distributed fault-tolerant control law is proposed to enable all the agents to achieve the control goal without losing the system performance while achieving fault identification. The corresponding sufficient conditions for both DFI and DFTC are obtained using graph theory and Lyapunov analysis theory. Practical power system simulation results are provided to illustrate the effectiveness of the proposed methods.