Analysis of new energy vehicle battery temperature prediction by combining BP neural network

Abstract

Abstract With the rapid development of the new energy industry, the safety research of battery technology has become a key topic. This paper focuses on the temperature prediction of new energy vehicle batteries, aiming to improve the safety and efficiency of batteries. Based on the new energy vehicle battery management system, the article constructs a new battery temperature prediction model, SOA-BP neural network, using BP neural network optimized by SOA algorithm. This model can accurately predict the battery temperature, and the effectiveness of its temperature control is verified through experiments. The results show that the SOA-BP neural network model outperforms the traditional BP, CNN, and RNN models in temperature prediction. Regarding evaluation indexes, the model’s root mean square error (RMSE), mean absolute error (MAE), and R2_Score are 0.953, 0.909, and 0.837, respectively. It is worth noting that the model can effectively regulate and control the battery temperatures at different temperatures, ensuring that the maximum temperature difference of each battery is maintained within 5°C. The model can also be used to predict the temperature of the batteries in different temperatures. This battery temperature prediction model not only provides an effective means for predicting and controlling the battery temperature of new energy vehicles, but also provides an essential reference for improving the vehicle’s performance and developing energy management strategies. This study offers a new solution for the safety and efficiency of new energy vehicle batteries.