ABSTRACTLithium batteries are increasingly favored for energy storage due to their high energy density, long cycle life, and robust charge and discharge rates. However, safety concerns necessitate the implementation of a battery management system (BMS) to monitor battery status, maintain energy balance, and provide failure warnings to ensure safe operation. This paper proposes an efficient BMS for high‐voltage, high‐current lithium battery energy storage. The approach leverages a multihead‐attention‐enhanced long short‐term memory (LSTM) neural network combined with an adaptive unscented Kalman filter to accurately calculate the battery's state of charge (SOC) and state of health (SOH). To improve accuracy, various factors such as temperature and internal resistance were considered. The algorithm was validated through hardware and simulation experiments, with experimental data compared to estimation results to demonstrate its precision. The findings show strong convergence and tracking capabilities, with SOC estimation presenting a maximum error of 1.5% and SOH estimation a maximum error of under 0.4%. We expect that this approach will allow for a more refined evaluation of SOC and SOH in lithium‐ion batteries, potentially improving Li‐ion battery system management.
Read full abstract7-days of FREE Audio papers, translation & more with Prime
7-days of FREE Prime access