Accessibility navigation


A machine-learning-based approach to predict residential annual space heating and cooling loads considering occupant behaviour

Li, X. and Yao, R. ORCID: https://orcid.org/0000-0003-4269-7224 (2020) A machine-learning-based approach to predict residential annual space heating and cooling loads considering occupant behaviour. Energy, 212. 118676. ISSN 0360-5442

Full text not archived in this repository.

It is advisable to refer to the publisher's version if you intend to cite from this work. See Guidance on citing.

To link to this item DOI: 10.1016/j.energy.2020.118676

Abstract/Summary

Energy consumption for space heating and cooling typically accounts for more than 40% of residential household energy consumption. An accurate and fast prediction of space heating and cooling loads aids energy conservation and carbon emission reduction by relieving the simulation burden for optimisation design, which consider various building characteristics combinations. This study aims to develop machine learning based load prediction model for residential building, five machine-learning models have been utilised for the prediction of residential building space heating and cooling load intensities, with occupant behaviour innovatively accounted as predictor variable. Their prediction performances are compared with each other. The five machine-learning models used in this study are linear kernel support vector regression, polynomial kernel support vector regression, Gaussian radial basis function kernel support vector regression, linear regression, and artificial neural networks. The results indicate that the Gaussian radial basis function kernel support vector regression is the best-performing model, with training time of less than 35s as well as less than 4% normalised mean absolute error and normalised root-mean-square error for both cooling and heating load prediction. The sample size of training and validation set for Gaussian radial basis function kernel support vector regression model is suggested as 200 samples. A data-driven machine-learning-based prediction model is an alternative to complex simulation tools in aiding the decision making of both building design and retrofit processes.

Item Type:Article
Refereed:Yes
Divisions:Science > School of the Built Environment > Energy and Environmental Engineering group
ID Code:95603
Publisher:Elsevier

University Staff: Request a correction | Centaur Editors: Update this record

Page navigation