Optimal Anomaly Threshold Selection for Monitoring Grid-Connected Lithium-Ion Battery Storage Systems

Safety of grid-connected Lithium-ion battery systems is a challenging issue. With advancements in Lithium-ion battery monitoring, early signs of battery failure can be effectively detected. However, the determination of warning thresholds remains a great challenge for different systems in commercial operation. To fill this gap, this paper presents an optimal anomaly threshold selection method for Lithium-ion Battery storage monitoring from an insurance perspective. The proposed method incorporates Kullback-Leibler Divergence (KLD) measurement and false alarm rate evaluation in the threshold selection. As a result, a threshold can be selected through the proposed method without increasing extra burden.