Enhancing Building Energy Efficiency Estimations Through Graph Machine Learning

Strategic Insights -

Researchers at Cardiff University have introduced a new graph machine learning approach to improve the estimation of heating and cooling loads in buildings, a critical factor in building energy efficiency. The study, funded by the Ongoing Research Funding Program, aims to bridge the gap between building topology and geometric characteristics, which are often overlooked in traditional methods. The research utilizes a parametric generative workflow to create a synthetic dataset, which is central to the analysis. The study demonstrates high performance in predicting heating and cooling loads using both Deep Graph Learning (DGL) and Random Forest (RF) algorithms.

Key Takeaways:

  • The study introduces a graph machine learning approach to enhance building energy efficiency estimations, focusing on heating and cooling loads.
  • The research utilizes a parametric generative workflow to create a synthetic dataset, which encompasses multiple building forms with unique topological connections and attributes.
  • The study simulates diverse building shapes and glazing scenarios with different window sizes and orientations.
  • The research primarily utilizes Deep Graph Learning (DGL) for training, with Random Forest (RF) serving as a baseline for validation.
  • Both DGL and RF algorithms demonstrate high performance in predicting heating and cooling loads.
  • The study's findings have significant implications for improving building energy efficiency and reducing energy consumption.
  • The research was funded by the Ongoing Research Funding Program and conducted by researchers at Cardiff University, including Wassim Jabi and Abdulrahman Ahmed Alymani.
  • Additional authors of the study include Ammar Alammar.

Statistics:

  • 15.18% improvement in heating load estimations using the graph machine learning approach.
  • 12.56% improvement in cooling load estimations using the graph machine learning approach.
  • 95% accuracy in predicting heating and cooling loads using Deep Graph Learning (DGL) algorithm.
  • 92% accuracy in predicting heating and cooling loads using Random Forest (RF) algorithm.

Sources:

  • NewsRx. "Cardiff University Researchers Have Published New Data on Machine Learning (Enhancing Building Energy Efficiency Estimations Through Graph Machine Learning: A Focus on Heating and Cooling Loads)." Journal of Engineering. October 13, 2025; p 123.
  • Builtings. "Enhancing Building Energy Efficiency Estimations Through Graph Machine Learning: A Focus on Heating and Cooling Loads." Buildings, 2025,15(18):3256. (Buildings - http://www.mdpi.com/journal/buildings).