A machine learning and deep learning based approach to predict the thermal performance of phase change material integrated building envelope

Abstract

This study aims to develop a machine learning and deep learning-based model for thermal performance prediction of PCM integrated roof building. Performance prediction is carried out using the newly proposed MKR index. Five machine learning and one deep learning technique are explored in order to predict the thermal performance of PCM integrated roof considering variations in thermophysical properties of PCM. Total 500 data points are generated using numerical simulations considering variations in thermophysical properties of PCM. The five machine learning models used in this study are Random forest regression, Extra trees regression, Gradient boosting regression, Extreme Gradient boosting regression, and Catboost regression. The results indicate that Gradient boosting regression is the best-performing model compared to other machine learning models. An artificial neural network is used as a deep learning approach for predicting the MKR index. The ANN-based model performed best among all five machine learning models and proved its efficacy in training, testing, and sensitivity analysis with the independent dataset.