Resilience-Oriented Restoration Strategy of Networked Microgrids Considering Grid Topology Against Data Intrusion Attacks

This paper presents a novel method for detecting and restoring networked microgrid (MG) systems compromised by topology attacks. A data intrusion attack detection (DIAD) system, utilizing machine learning techniques, is employed to identify tampered or malfunctioning smart meters. Simultaneously, an enhanced topology identification (TI) based graph learning algorithm is proposed to determine the exact fault locations and identify restoration MG zones for pre-event and post-event. A mixed integer linear programming (MILP) oriented approach is then utilized to optimize the load restoration process in networked MG areas. The goal is to rapidly restore critical loads with minimal losses, taking advantage of flexible fault support resources such as grid support storage systems (GSSs), photovoltaic systems (PV s), electric vehicle stations (EVSs), and mobile generators. The propounded model is evaluated, and the results show its effectiveness in handling various topology attack cases for load recovery.