A convolutional neural network analysis of a heat pipe with Hybrid Nanofluids
K. Kumararaja, C. Srinivasa Kumar, B. Sıvaraman
Abstract
Heat pipes are one of the most efficient devices for transporting heat from one region to another. It was utilised in a wide range of cooling applications. Heat pipe’s thermal behaviours are deeply interlinked with the outlet temperature. To predict the behaviours of the heat pipe, a model was essential to predict the outlet temperature. In this work, Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Deep Neural Network (DNN) and Deep Neural Network with Dropouts (DNN-DO) have been used to predict the outlet temperature of the heat pipe. The result shows that, out of four, the most precise model was CNN, followed by DNN-DO, DNN and ANN. Compared to the ANN model, CNN’s mean absolute error (MAE) is reduced by 66.3545% and the R2 score is increased by 15.8346%. In the case of percentage error (PE), 95.0649% data predicted between ±1.5% by the CNN model.