Component-based machine learning for performance prediction in building design

Research output: Contribution to journalArticleResearchpeer review

Authors

  • Philipp Florian Geyer
  • Sundaravelpandian Singaravel

External Research Organisations

  • KU Leuven
View graph of relations

Details

Original languageEnglish
Pages (from-to)1439-1453
Number of pages15
JournalApplied energy
Volume228
Early online date13 Jul 2018
Publication statusPublished - 15 Oct 2018
Externally publishedYes

Abstract

Machine learning is increasingly being used to predict building performance. It replaces building performance simulation, and is used for data analytics. Major benefits include the simplification of prediction models and a dramatic reduction in computation times. However, the monolithic whole-building models suffer from a limited transfer of models and their data to other contexts. This imposes a vital limitation on the application of machine learning in building design. In this paper, we present a component-based approach that develops machine learning models not only for a parameterized whole building design, but for parameterized components of the design as well. Two decomposition levels, namely construction level components (wall, windows, floors, roof, etc.), and zone-level components, are examined. Results in test cases show that, depending on how far the cases deviate from the training case and its data, high prediction quality may be achieved with errors as low as 3.7% for cooling and 3.9% for heating.

Keywords

    Building performance prediction, Building simulation, Component-based machine learning, Parametric systems modeling, Systems engineering

ASJC Scopus subject areas

Cite this

Component-based machine learning for performance prediction in building design. / Geyer, Philipp Florian; Singaravel, Sundaravelpandian.
In: Applied energy, Vol. 228, 15.10.2018, p. 1439-1453.

Research output: Contribution to journalArticleResearchpeer review

Geyer PF, Singaravel S. Component-based machine learning for performance prediction in building design. Applied energy. 2018 Oct 15;228:1439-1453. Epub 2018 Jul 13. doi: 10.1016/j.apenergy.2018.07.011
Geyer, Philipp Florian ; Singaravel, Sundaravelpandian. / Component-based machine learning for performance prediction in building design. In: Applied energy. 2018 ; Vol. 228. pp. 1439-1453.
Download
@article{d63ed2e1e80447779e5d0f2b7077c5ad,
title = "Component-based machine learning for performance prediction in building design",
abstract = "Machine learning is increasingly being used to predict building performance. It replaces building performance simulation, and is used for data analytics. Major benefits include the simplification of prediction models and a dramatic reduction in computation times. However, the monolithic whole-building models suffer from a limited transfer of models and their data to other contexts. This imposes a vital limitation on the application of machine learning in building design. In this paper, we present a component-based approach that develops machine learning models not only for a parameterized whole building design, but for parameterized components of the design as well. Two decomposition levels, namely construction level components (wall, windows, floors, roof, etc.), and zone-level components, are examined. Results in test cases show that, depending on how far the cases deviate from the training case and its data, high prediction quality may be achieved with errors as low as 3.7% for cooling and 3.9% for heating.",
keywords = "Building performance prediction, Building simulation, Component-based machine learning, Parametric systems modeling, Systems engineering",
author = "Geyer, {Philipp Florian} and Sundaravelpandian Singaravel",
note = "Funding Information: The research was funded by Starting Grant STG-14-00346 of the KU Leuven and by the Deutsche Forschungsgemeinschaft (DFG) in Researcher Unit 2363 “Evaluation of building design variants in early phases using adaptive levels of development”, in Subproject 4 “System-based Simulation of Energy Flows”. ",
year = "2018",
month = oct,
day = "15",
doi = "10.1016/j.apenergy.2018.07.011",
language = "English",
volume = "228",
pages = "1439--1453",
journal = "Applied energy",
issn = "0306-2619",
publisher = "Elsevier BV",

}

Download

TY - JOUR

T1 - Component-based machine learning for performance prediction in building design

AU - Geyer, Philipp Florian

AU - Singaravel, Sundaravelpandian

N1 - Funding Information: The research was funded by Starting Grant STG-14-00346 of the KU Leuven and by the Deutsche Forschungsgemeinschaft (DFG) in Researcher Unit 2363 “Evaluation of building design variants in early phases using adaptive levels of development”, in Subproject 4 “System-based Simulation of Energy Flows”.

PY - 2018/10/15

Y1 - 2018/10/15

N2 - Machine learning is increasingly being used to predict building performance. It replaces building performance simulation, and is used for data analytics. Major benefits include the simplification of prediction models and a dramatic reduction in computation times. However, the monolithic whole-building models suffer from a limited transfer of models and their data to other contexts. This imposes a vital limitation on the application of machine learning in building design. In this paper, we present a component-based approach that develops machine learning models not only for a parameterized whole building design, but for parameterized components of the design as well. Two decomposition levels, namely construction level components (wall, windows, floors, roof, etc.), and zone-level components, are examined. Results in test cases show that, depending on how far the cases deviate from the training case and its data, high prediction quality may be achieved with errors as low as 3.7% for cooling and 3.9% for heating.

AB - Machine learning is increasingly being used to predict building performance. It replaces building performance simulation, and is used for data analytics. Major benefits include the simplification of prediction models and a dramatic reduction in computation times. However, the monolithic whole-building models suffer from a limited transfer of models and their data to other contexts. This imposes a vital limitation on the application of machine learning in building design. In this paper, we present a component-based approach that develops machine learning models not only for a parameterized whole building design, but for parameterized components of the design as well. Two decomposition levels, namely construction level components (wall, windows, floors, roof, etc.), and zone-level components, are examined. Results in test cases show that, depending on how far the cases deviate from the training case and its data, high prediction quality may be achieved with errors as low as 3.7% for cooling and 3.9% for heating.

KW - Building performance prediction

KW - Building simulation

KW - Component-based machine learning

KW - Parametric systems modeling

KW - Systems engineering

UR - http://www.scopus.com/inward/record.url?scp=85049658932&partnerID=8YFLogxK

U2 - 10.1016/j.apenergy.2018.07.011

DO - 10.1016/j.apenergy.2018.07.011

M3 - Article

AN - SCOPUS:85049658932

VL - 228

SP - 1439

EP - 1453

JO - Applied energy

JF - Applied energy

SN - 0306-2619

ER -