Predicting energy consumption for residential buildings using ANN through parametric modeling. (November 2021)

• Record Type: Journal Article
• Authors: Elbeltagi, Emad; Wefki, Hossam
• Is Part Of: Energy reports. Volume 7(2021)
• Page Start: 2534
• Page End: 2545
• Publication Date: 2021-11
• Subjects: Energy simulation -- Parametric analysis -- Artificial Neural Network -- Residential buildings -- Conceptual design phase; Power resources -- Periodicals; Energy industries -- Periodicals; Power resources; Periodicals; Electronic journals; 621.04205
• DOI: 10.1016/j.egyr.2021.04.053
• Languages: English
• ISSNs: 2352-4847

Abstract

Abstract: Controlling buildings energy consumption is a great practical significance. During early design stage, accurate and rapid prediction of energy consumption could provide a quantitative basis for energy-saving designs. Currently, the key problem that are still facing designers is the interoperability between building modeling and energy simulation tools. In addition, design challenges gained recognition due to the complexity of the prevalence of large numbers of independent interrelated variables. Artificial Neural Networks (ANNs) are the most broadly applied artificial intelligence method in buildings' performance field due to its competence to handle nonlinear variables' relationships accurately and promptly. This paper presents a methodology based on the ANNs to improve the prediction of energy usage for residential buildings in early design stages. The model is created using a dataset resulted from the calculation of energy consumption by simulating multiple design options with randomly input variables. The proposed methodology can mitigate technical barriers while integrating and automating available commercial tools into a workflow from a parametric model to the simulation of building energy. The developed ANN model is evaluated and validated and used to predict the energy consumption with acceptable accuracy. Finally, a user-friendly interface is designed to facilitate energy consumption prediction without any experience in modeling and simulation tools acting as a decision support tool, which is simple, reliable and easy to use.