| Popis: |
This paper proposes a novel event-triggered attack detection mechanism for multi-agent DC microgrid system. Under a distributive network framework, each node only collects its neighbours’ relative data to regulate its own output for inner system stabilization. Hardly can an agent identify the source of unexpected outside power flow directly without power line current data, especially under an organized attack composed of voltage variation and corresponding deceptive messages. In order to recognize traitors who broadcast wrong data, aim for system distortion and even splitting, an efficient attack detection and identification strategy is mandatory. After the attack detector is triggered, each relative agent refuses to trust any received data directly before authentication. During two-step verification by comparing theoretical estimated signals with received ones, both self sensors and neighbour nodes would be inspected, and the attacker was difficult to hide himself only except Buck converters were employed globally. Through simulation on MATLAB/SIMULINK, the effectiveness of proposed detection algorithm has been proved. |
