EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Precise energy modeling and green retrofitting optimization of existing buildings based on BIM and deep learning approaches

Xiyang Ge, Sharifah Akmam Syed Zakaria, Chenyu Wang iD, Meng Zhu

PLOS One · 2025

DOI: 10.1371/journal.pone.0337469 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, 2025. Verfasst von Xiyang Ge und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Civil Engineering, Universiti Sains Malaysia, Penang, Malaysia. Hauptthemen laut MEDLINE: Deep Learning, Conservation of Energy Resources, Construction Industry, Models, Theoretical. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 51 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The construction industry has emerged as a major contributor to global energy consumption and greenhouse gas emissions amidst continuously rising worldwide energy demands. Enhancing building energy efficiency represents a critical intervention for achieving energy conservation and emission reduction targets. In the context of smart city development, such optimization efforts provide substantial momentum for sustainable urban growth. This study introduces a novel methodology that integrates Transformer models with Graph Neural Networks (GNNs) to improve the accuracy and operability of building energy efficiency prediction through advanced deep learning techniques. By leveraging Building Information Modeling (BIM) data to model spatial structures and energy consumption patterns, GNNs effectively capture complex relationships between building components, thereby strengthening the characterization of multidimensional interactions within structures. The self-attention mechanism in Transformers enables the model to focus on key factors such as energy consumption hotspots and temporal variations, enhancing learning capabilities across both spatial and temporal dimensions. To further augment optimization performance, we incorporate Generative Adversarial Networks (GANs) to generate diverse green renovation schemes, expanding optimization pathways and enhancing model adaptability and robustness. Experimental validation using BIM data demonstrates that our integrated approach outperforms traditional energy efficiency optimization models, increasing energy savings by nearly 4 These findings establish that BIM data-integrated deep learning optimization methodologies offer significant potential for providing effective energy efficiency prediction and optimization decision support. Such approaches directly contribute to building design and operations in smart cities, advancing the realization of green buildings and sustainable urban development.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningConservation of Energy ResourcesConstruction IndustryModels, TheoreticalNeural Networks, Computer

Autor:innen und Institutionen

  • School of Civil Engineering, Universiti Sains Malaysia, Penang, Malaysia.
  • School of Architectural Engineering, Jiaxing Nanhu University, Jiaxing, China.

ORCID-Profile: Chenyu Wang

Zitieren

APA

Ge, X., Syed Zakaria, S. A., Wang, C., & Zhu, M. (2025). Precise energy modeling and green retrofitting optimization of existing buildings based on BIM and deep learning approaches. PLOS One. https://doi.org/10.1371/journal.pone.0337469

Vancouver

Ge X, Syed Zakaria SA, Wang C, Zhu M. Precise energy modeling and green retrofitting optimization of existing buildings based on BIM and deep learning approaches. PLOS One. 2025. doi:10.1371/journal.pone.0337469

BibTeX
@article{Ge2025,
  author = {Ge, Xiyang and Syed Zakaria, Sharifah Akmam and Wang, Chenyu and Zhu, Meng},
  title = {Precise energy modeling and green retrofitting optimization of existing buildings based on BIM and deep learning approaches},
  journal = {PLOS One},
  year = {2025},
  doi = {10.1371/journal.pone.0337469}
}
Kostenlos starten →