Cyber Resilience of Battery Energy Storage System Scheduling under Adversarial Attack on Load Forecasting

Battery Energy Storage System (BESS) scheduling plays an important role in the modernization of power grid, yet its growing reliance on load forecasting introduces critical cyber risks. Misled load forecasts can lead to operational costs increase and compromise system reliability and stability. However, the resilience of BESS scheduling to such attacks is largely overlooked. To address this gap, this paper proposes an attack- schedulingvalidation framework to evaluate how adversarial attacks on load forecasting practically impacts BESS scheduling especially in peak shaving. Specifically, we present how to integrate attacked load forecasts into BESS scheduling model. We further propose an optimization-based validation model to assess the operational feasibility of such attacked BESS scheduling under actual load conditions. Experiments show that injecting adversarial attacks on load forecasting can lead to either increased investment costs or power system anomaly during peak hours. These findings underscore the importance to consider cyber resilience in future BESS scheduling.