A Maximum Mean Discrepancy Estimator Enabled Cyber Attack Detection in an Autonomous AC Microgrid

This paper presents a novel scheme to precisely identify cyber attacks, such as controller hijacking and unauthorized data modification, which can compromise an autonomous microgrid's (MG) frequency and voltage stability. The scheme is based on the Maximum Mean Discrepancy (MMD) index, which calculates the unbiased estimates of local voltage/frequency synchronizing tracking errors for each Distributed Energy Resource (DER). Under the compromised situation, MMD accurately and quickly distinguishes the attacked DER from the uncompromised DERs. The efficacy of the proposed cyber-attack detection scheme is tested under various types of cyber attacks on the IEEE-13 bus distribution feeder, modelled in RSCAD and is validated with Real Time Digital Simulator (RTDS) platform integrated with dSPACE 1104 R&D automation controller through Hardware-in-the-Loop (HIL) simulation control environment.