SOC estimation of Li-ion battery with a focus on force features
Accurately estimating the state of charge(SOC) is crucial for ensuring the reliable operation of Li-ion battery.A data-driven SOC estimation method based on multidimensional features,with a specific focus on integrating force signals is introduced.The stress characteristics of Li-ion battery undergo Savitzky-Golay(S-G) filtering,forming in an optimized and reconstructed stress signal. A back propagation(BP) neural network,incorporating the sparrow search algorithm(SSA),is proposed,elevating the global optimization capability of the neural network.The method is evaluated under constant current(CC) and federal urban driving schedule (FUDS) conditions.Within the BP neural network,SOC estimation considering stress features reduces the root mean square error (RMSE) by 89.1％ and the mean absolute error(MAE) by 88.8％ compared to solely based on electrical signals.The SSA-BP neural network,considering stress features in SOC estimation,maintains error within 0.3％,showcasing higher robustness and precision.
state of charge(SOC)Li-ion batteryforceneural networksparrow search algorithm(SSA)
国家科技期刊平台
NSTL
万方数据
