AdaBoost-Based Cyberattack Detection Algorithm for Battery Systems Providing Frequency Regulation

The intermittent nature of renewable energy sources may require the integration of an increasing amount of battery energy storage systems (BESSs) in the electrical grid. A BESS can provide multiple services such as frequency regulation and backup. Simultaneously, internet-of-things (IoT) enabled technologies are increasingly being applied in the BESS domain. IoT integration is not supported by any specific regulations with respect to BESS design and operation, which can increase the vulnerability of utility-scale batteries to cyberattacks. Cyberattacks can compromise the ability of BESSs to provide an adequate and reliable response to the system's demand. This paper presents a machine-learning-based algorithm for cyberattack detection that can decrease the vulnerability of a BESS to cyber threats. It utilizes Adaptive Boosting to forecast the state of charge and detect potentially corrupted data. Moreover, this paper presents mathematical models of cyberattacks than can be applied against a BESS, providing frequency regulation. The benefits of applying the novel cyberattack detection algorithm are demonstrated in the example of the simulated dataset. The dataset is generated based on parameters of the largest grid-connected BESS in Denmark as well as real frequency data of the Nordic Region.