Structure optimization of air-cooled battery thermal management system based on neural network

Battery thermal management system (BTMS) is essential to the safe operation of electric vehicles. In order to improve the heat dissipation performance of BTMS, the Non-dominated sorting genetic algorithm-2 (NSGA2) combined with neural network is used to optimize the battery pack with multiple objectives. First, the three-dimensional battery pack model is converted into the two-dimensional model to simulate 2000 battery packs, saving much calculation time. Subsequently, five parameters, including the width of the inlet and the outlet, the position of the inlet and the outlet, and the battery spacing, are used as design variables to establish a BP neural network model with a good prediction effect of BTMS. After that, the NSGA2 algorithm is used to optimize the neural network model with multiple objectives. Finally, the final design solution with the lowest maximum temperature in the Pareto solution set is selected and simulated. The results show that the maximum temperature of the optimized battery pack is reduced by 7.5 K, the maximum temperature difference is reduced by 67.4%, and the power consumption is reduced by 26%.