Comparison of GRNN and RF algorithms for predicting heat transfer coefficient in heat exchange channels with bulges

In the research of heat transfer, exchanger plays an important role in highly integrated and high-precision thermal management to ensure balance. previous studies, traditional experiments CFD simulations consume lots time computational resources, while transfer correlations have large errors. Hence, this aims establish a reliable method predict Heat Transfer Coefficient (HTC) exchange channels more quickly accurately. paper, General Regression Neural Network (GRNN) Random Forests (RF) models, which are trained by hundreds simulation results, adopted heat-exchange performances with different height bulges. The prediction results show that HTC bulge arrangements accurately predicted, supported R2 > 0.97 both training validation sets. Also, it shows GRNN is applicable than RF. Besides, can be inferred from front-end has significant impact on overall HTC, as does uniformity height. conclusion, machine learning algorithms great potential predicting channels, algorithm may perforce better when calculate other complex problems.