Detecting Cyberattacks on Electrical Storage Systems through Neural Network Based Anomaly Detection Algorithm
Giovanni Battista Gaggero,
Roberto Caviglia,
Alessandro Armellin,
Mansueto Rossi,
Paola Girdinio,
Mario Marchese
Affiliations
Giovanni Battista Gaggero
Department of Electrical, Electronic and Telecommunications Engineering and Naval Architecture—DITEN, University of Genoa, Via Opera Pia 11A, 16145 Genoa, Italy
Roberto Caviglia
Department of Electrical, Electronic and Telecommunications Engineering and Naval Architecture—DITEN, University of Genoa, Via Opera Pia 11A, 16145 Genoa, Italy
Alessandro Armellin
Department of Electrical, Electronic and Telecommunications Engineering and Naval Architecture—DITEN, University of Genoa, Via Opera Pia 11A, 16145 Genoa, Italy
Mansueto Rossi
Department of Electrical, Electronic and Telecommunications Engineering and Naval Architecture—DITEN, University of Genoa, Via Opera Pia 11A, 16145 Genoa, Italy
Paola Girdinio
Department of Electrical, Electronic and Telecommunications Engineering and Naval Architecture—DITEN, University of Genoa, Via Opera Pia 11A, 16145 Genoa, Italy
Mario Marchese
Department of Electrical, Electronic and Telecommunications Engineering and Naval Architecture—DITEN, University of Genoa, Via Opera Pia 11A, 16145 Genoa, Italy
Distributed Energy Resources (DERs) are growing in importance Power Systems. Battery Electrical Storage Systems (BESS) represent fundamental tools in order to balance the unpredictable power production of some Renewable Energy Sources (RES). Nevertheless, BESS are usually remotely controlled by SCADA systems, so they are prone to cyberattacks. This paper analyzes the vulnerabilities of BESS and proposes an anomaly detection algorithm that, by observing the physical behavior of the system, aims to promptly detect dangerous working conditions by exploiting the capabilities of a particular neural network architecture called the autoencoder. The results show the performance of the proposed approach with respect to the traditional One Class Support Vector Machine algorithm.
