Set-Induced Anomaly Detectors for Networked Power Systems under Bias Injection Cyber-Attacks

This paper addresses the concept of a set-induced anomaly detector of bias injection cyber-attacks affecting the load frequency control loop of a networked power system. An adversary corrupts the frequency sensor measurements causing abnormal system behavior. A set-theoretic methodology is used for the extraction of a convex and compact polyhedral robust invariant set under the overall discretized network dynamics. An attack is considered disclosed when the state vector exits the invariant set. Simulation studies demonstrate the impact of an intermittent attack on a two-area power plant and provide an assessment of the proposed detector, when the attack happens simultaneously with changes in the power load demand.