Statistical methods for detection and mitigation of the effect of different types of cyber-attacks and parameter inconsistencies in a real world distribution system

Abstract

Distribution Management systems are effectively used by electric utilities for analyzing and controlling the distribution systems, but inconsistencies in the modelling parameters of the distribution network or cyber-attack conditions may result in failing of the controller to maintain voltage profile in a smart distribution system. In this paper we propose a statistical method based on multiple linear regression (MLR) to develop a robust and reliable capacitor control algorithm to maintain the voltage profile in a distribution system with distributed generators (DG). Two cyber-attacks, namely deception attack and load redistribution attack are modeled in OpenDSS to validate the proposed statistical methods on real American Electric Power (AEP) distribution feeder. A regression based distributed detection algorithm is also proposed for detection of cyber-attack in a distribution system with DG. Regression is based on least squares method making use of data acquired from exact simulations of the AEP System feeder. According to our knowledge this is the first paper which discusses cyber-attack in three phase unbalanced real distribution network.