State of Charge Estimation of Lithium-ion Battery Pack based on CNN-LSTM

The state of charge (SOC) of a battery is a critical parameter in the management of lithium-ion batteries, and high-precision SOC estimation is beneficial for BMS and energy managers to control battery usage patterns. Data related to the charging and discharging cycles of the battery are not only time-series, but also have certain spatial relationships between feature variables. Therefore, this study introduces a method for SOC of lithium-ion battery packs based on a joint CNN-LSTM network structure, which firstly obtains the extraction of feature relationships across different data dimensions of lithium-ion batteries using convolutional neural networks (CNNs), and then extracts the time-series relationships among them through LSTM network structure, to comprehensively capture the spatiotemporal features of the battery pack dataset. The experimental results demonstrate that the average error in battery SOC prediction is reduced when using the combined CNN-LSTM network model compared with that of the separate LSTM network, and the estimation method has high accuracy and versatility, which has a good application prospect.