Multi-objective optimization of liquid-cooled battery thermal management system with biomimetic fractal channels using artificial neural networks and response surface methodology

In order to reduce the operating temperature of batteries for energy storage and automotive power, and ensure their safety during operation, a cooling plate with biomimetic fractal channels is proposed for the battery thermal management system (BTMS). The temperature standard deviation (Tstd) of the batteries and the performance evaluation criterion (PEC) of the cooling plate are used as evaluation indicators. The influences of fractal channels structural parameters on the flow and heat transfer performance of the cooling plate and the thermal characteristics of the batteries are studied, and the results show that both the Tstd and the PEC present the same trend with the change of five structures of the fractal channels. In order to obtain smaller Tstd and larger PEC, as to find the optimal structural parameters to improve the overall performance of BTMS, two multi-objective optimization methods were used in this work to optimize the structural parameters of the cooling plate: (1) a combination of artificial neural network and genetic algorithm (ANN-GA) and (2) a combination of response surface methodology and genetic algorithm (RSM-GA). For the two optimization methods compared with the initial structure of the single-factor analysis, PEC increased by 3.93 % and 0.72 %, respectively, while Tstd decreased by 7.79% and 25.45%, respectively. It can be seen that the optimal structure obtained through ANNGA has a greater improvement in the performance of the cooling plate, while the optimal structure obtained through RSM-GA has a greater enhancement in the battery temperature uniformity.