PhD Defense - Nina Kharlamova - Cyber-Secure Operation of Battery Energy Storage Systems Providing Grid Services

#battery #energystorage #PhDdefense #distributed #systems #grids #smart

Battery energy storage systems have vulnerabilities that make them susceptible to cyberattacks, a challenge that needs to be addressed as the energy sector relies more heavily on the storage. The thesis investigates the potential security concerns that can compromise the work of a BESS providing grid services. To tackle this issue, the methods of cybersecure BESS design were explored and a novel method for detecting and safeguarding against cyberattacks on BESSs as part of a Virtual Power Plant (VPP) was developed. This novel detection algorithm can identify and mitigate the BESS-disabling attack, all while ensuring the system continues to function as required. The approach involves two key strategies: forecasting the future behavior of a BESS and leveraging the resources of the VPP, which acts as an aggregator for multiple BESSs.

To produce the forecast of a BESS behavior, the literature related to time series forecast in the smart grid domain. The thesis compares six machine-learning algorithms that were shortlisted based on the literature review. The algorithms were compared on a realistic dataset of a BESS providing frequency regulations and validated on a real dataset that was obtained using the BESS in PowerLabDK providing frequency regulations in Denmark. To predict BESS behavior and identify any abnormal activities that might indicate an attack, a powerful technique called Adaptive Boosting is employed. This enables the algorithm to forecast the BESS's state of charge (SOC) with precision. Additionally, the Moving Target Defense (MTD) algorithm is implemented, which enhances the sensitivity of the detection mechanism without increasing false positives. MTD modifies VPP performance based on the cyberattack trigger and evaluated the response of the system to the predefined modification.

Furthermore, to counter even more intelligent attackers, the defense mechanism with hidden MTD is enhanced, adding an extra layer of protection. The modification of the performance of BESSs that are part of a VPP is done with regard to the knowledge regarding the measuring and estimation error. In this way, the attacker cannot detect the modifications done by the cyber defense algorithm since they remain within the error range. 

To prove the effectiveness of the algorithm, it was tested using realistic data, simulating various BESS-disabling cyberattacks. The study proposes a BESS-disabling cyberattack that can shut down a BESS completely using the limited knowledge about the system. The results shows that the algorithm successfully detects and neutralizes all applied cyberattacks, ensuring the reliable operation of the BESS.