Optimization of air-cooling technology for LiFePO4 battery pack based on deep learning

Abstract

The forced air-cooling system is applied extensively in the battery thermal management system (BTMS) to ensure temperature uniformity because of the simple structure and low cost. In this paper, a BTMS of LiFePO4 cuboid battery module with adding different additional airflow outputs in typical U-type cooling system is designed to optimize temperature uniformity. Because the features of these additional airflow outputs, including positions, numbers and areas of them, influence the temperature of battery pack simultaneously, a nine-layer fully connected deep network AI model is built to find the relationship between temperature of pack and these features of additional airflow outputs in BTMS. The temperatures of batteries with different additional airflow outputs are simulated by a three-dimensional computational fluid dynamics (CFD) model. Comparing with the CFD simulation results, the AI model's mean absolute errors of maximum temperature and temperature differences are 0.046% and 0.99%, respectively. An optimal BTMS air-cooling design is found as a U-type structure adding three 4 mm additional airflow outputs with gaps of 40 mm, 120 mm, 10 mm, and 48 mm after comparison of 765,846 BTMS structures. The maximum temperature is decreased by 6.22 K, while the temperature difference is reduced by 40.36% compared to original design.