Structure design and performance optimization of liquid cooling plate of power battery based on machine learning and genetic algorithm

The structural parameters of the liquid cooling plate (LCP) have a significant impact on the temperature control of the power battery. In this paper, the rib-column composite structure parameters are optimized by combining the CAWOA-BP neural network and NSGA-II optimization algorithm, and compared with the original structure, its Nu and thermal enhancement coefficient (TEC) are effectively improved. Firstly, the rib-column composite structure was coupled to the serpentine flow channel LCP, and the results showed that the elliptical column and staggered semicircular-rib composite structure EC-SSR had the best performance compared to other composite structures, and its TEC increased by 34.47%. Then, five structural parameters were extracted from it as variables, and Nu and the friction coefficient f were used as target functions to construct the dataset. Than the CAWOA-BP neural network was trained by a training set consisting of 155 sets of sample points, and then the optimal structural parameters were determined by an optimization algorithm and numerical simulation. The Nu of the optimum structure is increased from 2.89 to 6.32 and its TEC is increased by 51.33% compared to the initial structure, although its friction coefficient f is increased from 0.41 to 1.24. Some insights into cooling structures were provided.