Attack Detection and Isolation for Distributed Load Shedding Algorithm in Microgrid Systems

Abstract

Load shedding is usually adopted as an emergency management to cope with large frequency deviation and supply–demand imbalance in a microgrid (MG). As generators and loads are usually highly distributed in the MG, a distributed load shedding strategy is considered in this article. In the proposed strategy, each agent first locally discovers the system's global knowledge. Efficient load shedding decisions are then made with the acquired information. Inspired by the fact that the communication channels are very vulnerable to malicious attacks, we further address the security issue of the considered strategy. It is assumed that an attacker intends to disrupt the information discovery procedure and further deteriorates the system operation by injecting malicious signals. By considering the injected signal as an external input with no prior knowledge, we first establish necessary and sufficient conditions for the misbehaviors to be observed. The design procedure of an unknown input observer is then presented, based on which a detect and isolate mechanism is further developed to distributively detect and isolate the misbehaving agent and mitigate the induced negative effects. The simulation and experimental results finally validate the effectiveness of our schemes.