Deep learning for building energy consumption prediction

Kadir Amasyali, Nora El-Gohary

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In recent years, building energy consumption prediction gained a lot of research attention due to its importance in energy efficiency-related decision making. With the advancements in data analytics and machine learning, there has been numerous studies on developing data-driven building energy consumption prediction models based on support vector machines (SVM), artificial neural networks (ANN), and other statistical regression algorithms. These studies showed that each algorithm has its own advantages and disadvantages for different cases and that, therefore, the algorithms should be selected based on the specific application. However, none of the existing research efforts tested the effectiveness of deep learning - which is shown to outperform other machine learning algorithms in many other fields -in building energy consumption prediction. To address this gap, this paper (1) presents a deep learning-based model to predict cooling energy consumption of a building based on outdoor weather conditions (e.g., outdoor temperature), and (2) compares the prediction performance and computational efficiency of the deep learning-based model against other machine learning and statistical regression-based benchmark models. In order to generate a labelled dataset for training the models, a building was modelled and simulated by EnergyPlus in five locations. The models - the deep learning model as well as the other benchmark models - were trained using the simulation-generated data and the performance was evaluated in terms of accuracy and computational efficiency. The testing results showed that deep learning can be successfully applied to the field of building energy consumption prediction.

Original languageEnglish (US)
Title of host publication6th CSCE-CRC International Construction Specialty Conference 2017 - Held as Part of the Canadian Society for Civil Engineering Annual Conference and General Meeting 2017
PublisherCanadian Society for Civil Engineering
Pages466-474
Number of pages9
ISBN (Print)9781510878419
StatePublished - Jan 1 2017
Event6th CSCE-CRC International Construction Specialty Conference 2017 - Held as Part of the Canadian Society for Civil Engineering Annual Conference and General Meeting 2017 - Vancouver, Canada
Duration: May 31 2017Jun 3 2017

Publication series

Name6th CSCE-CRC International Construction Specialty Conference 2017 - Held as Part of the Canadian Society for Civil Engineering Annual Conference and General Meeting 2017
Volume1

Conference

Conference6th CSCE-CRC International Construction Specialty Conference 2017 - Held as Part of the Canadian Society for Civil Engineering Annual Conference and General Meeting 2017
CountryCanada
CityVancouver
Period5/31/176/3/17

    Fingerprint

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Building and Construction

Cite this

Amasyali, K., & El-Gohary, N. (2017). Deep learning for building energy consumption prediction. In 6th CSCE-CRC International Construction Specialty Conference 2017 - Held as Part of the Canadian Society for Civil Engineering Annual Conference and General Meeting 2017 (pp. 466-474). (6th CSCE-CRC International Construction Specialty Conference 2017 - Held as Part of the Canadian Society for Civil Engineering Annual Conference and General Meeting 2017; Vol. 1). Canadian Society for Civil Engineering.