Multi-agent-based hierarchical detection and mitigation of cyber attacks in power systems

This paper proposes a multi-agent-based hierarchical detection and mitigation (MHDM) scheme for power systems against physical faults and cyber attacks. The MHDM scheme is designed based on rule-based abnormal detection with a hierarchical structure managed by three-level agents according to the information of each layer. Lower-level and middle-level agents utilize physical measurement, cyber information and the characteristics of physical faults to detect various faults and cyber attacks, and upper-level agent takes advantage of the redundancy of wide-area data to verify the diagnosis results. Through the collaboration of agents at all levels, the MHDM scheme can identify the operation state of the power system, and mitigate the impact of cyber attacks while distinguishing them from physical faults. Simulation studies are undertaken on a platform constructed by interconnecting the New England 39-bus system developed in MATLAB/SIMULINK and a three-level multi-agent system implemented in JADE. Simulation results demonstrate the effectiveness of the MHDM scheme.