Research on Data Processing System Model for the Effect of Low Temperature on Electric Vehicle Battery Performance

A model for predicting battery performance based on GNN is presented. Through the construction of the topology structure of battery cells and the strong representation capability of the FNN, the model can be used to capture the complex relationships among the battery cells, so as to better predict the performance of the battery in the low-temperature environment. Compared with the traditional SVM and random forest, the FNN model is more effective in capturing the relationships among the battery cells. The experiment results show that the GNN model can greatly increase the precision of the key parameters such as battery capacity, current and inner resistance. The prediction error of GNN model is less than 3.5% in low temperature environment, and that of traditional method is usually above 5%. Moreover, the GNN model is more robust and stable to deal with the variation of the battery's performance in different low-temperature conditions. This study not only offers a new way to optimize the low-temperature performance of EV batteries, but also provides a valuable reference for the study on the prediction of battery performance.