Machine learning-based prediction of heat transport performance in oscillating heat pipe

An oscillating heat pipe (OHP) is a highly efficient cooling system for densely integrated electronic and electric devices operating at high frequencies with generation densities. However, because of the complicated internal flow phase changes, it difficult to predict transport performance OHPs accurately. Such predictions are needed understand fundamental phenomena in optimize OHP design parameters. The objective this study three prediction targets comprising pattern, wall temperature difference between cooled heated sections, rate through machine learning recurrent neural networks. Experiments on ethanol were performed input range 62-125 W obtain time series data pattern images, temperatures, water temperatures. images processed by semantic segmentation subsequently used training models each target. patterns recursively predicted using trained model. then into these two targets. experimental target compared, ability learning-based procedure was demonstrated quantitative agreement statistical values.