Prediction of Lithium-ion Battery SOC Based on the Fusion of MHA and ConvolGRU

Abstract

Abstract Overcharge and overdischarge of the battery can be avoided if the state of charge of lithium-ion battery is predicted accurately. A prediction method combined with convolution layer, gated cycle unit and multi-heads attention mechanism is proposed in this paper in order to improve the prediction accuracy of SOC. The data set uses the data of battery charging and discharging under FUDS conditions and DST conditions from the University of Maryland. The window sliding technology is used in the data preprocessing part. Finally, the prediction effect of the fusion model proposed in this paper is verified by Pycharm simulation. The average absolute error, root mean square error and maximum prediction error of the model are 0.53%, 0.67% and 0.4% respectively, which proves that the SOC can be predicted accurately by this model.