Battery temperature Prediction in electric vehicles using bayesian regularization

This study is focused on developing a new temperature prediction model to be used in battery thermal management systems. Electric vehicle (EV) application is considered as a case study however, the proposed model is applicable in other applications too. The final goal is to improve batteries’ performance, durability, and safety. By specifically examining two types of batteries, which are Lithium Iron Phosphate (LFP) and Nickel Cobalt Aluminum (NCA), the proposed model utilizes Bayesian Regularization to precisely predict variations in the battery’s surface temperature in an EV application. The present study experimentally evaluates the accuracy of the proposed model for prediction of the batteries’ surface temperature under various conditions. According to the results, average errors of less than 0.1°C and 0.3°C are achieved when predicting the batteries’ surface temperature in 30 and 90 seconds ahead. This study is expected to have an impact on the advancement of EVs’ battery technologies by improving the battery’s performance and safety.