Multi-objective parameter optimization of the Z-type air-cooling system based on artificial neural network

Abstract

The performance of electric vehicles is significantly influenced by the battery thermal management system (BTMS). In this article, the Z-type air cooling BTMS is studied, and a method combining an artificial neural network (ANN) model as a surrogate model with non-dominated sorting genetic algorithm II (NSGA-II) for multi-objective optimization is proposed to enhance the heat dissipation performance. The ANN model is trained by using 1529 sets of numerical simulation data of Z-type air cooling system, and the fast and high-precision prediction of heat dissipation performance is realized, which effectively avoids the repeated calculation of CFD. The design variables studied include geometric structure parameters and inlet wind speed (vinlet), and the target variables are maximum temperature (Tmax), temperature difference (ΔTmax) and pressure loss (ΔP). From the new perspective of genetic evolution process of NSGA-II, it is proved that the design of tapered inlet and outlet is better than that of flat and expansion outlet, the wide channel near the inlet and narrow channel near the outlet are more conducive to the improvement of heat dissipation performance. Further analysis shows that there is a linear relationship among vinlet, Tmax and ΔP in the Pareto solution set of Z-type air-cooling system, which simplifies the Pareto model effectively, and engineers can predict the optimal performance of BTMS by this relationship. In addition, through sensitivity analysis, it is found that in the Z-type BTMS, the operating parameter vinlet has the greatest influence on the heat dissipation performance and pressure loss, while the geometric parameters have less influence. Compared with the benchmark case, the optimized design scheme reduces Tmax, ΔTmax and ΔP by 1.97 K, 6.32 K and 2.82 Pa, respectively, which significantly improves the heat dissipation performance.