AI-Driven Energy Efficiency in Building Design: New Study Outlines Potential for Significant Reductions in Energy Consumption

Strategic Insights -

Research from Kingston University in London, United Kingdom, has published a study on the integration of artificial intelligence (AI) in automating the prediction and optimization of energy performance in residential buildings. The study aims to explore the potential of AI in enhancing energy efficiency, a key concern in architectural design, by precisely measuring and optimizing building energy saving. The research developed a hybrid stacked model combining a deep Feedforward Neural Network (FNN) with Extreme Gradient Boosting (XGB), achieving an impressive Coefficient of Determination (R²) of 0.99 and Mean Absolute Percentage Error (MAPE) of 0.02 across all targets.

Key Takeaways:

  • The research emphasized the importance of energy efficiency in architectural design, with Governments increasingly committed to prioritizing energy efficiency initiatives and implementing energy measures to address climate change and its impacts.
  • The integration of AI in automating the prediction and optimization of energy performance in residential buildings shows promise in reducing energy consumption while maintaining thermal comfort.
  • The developed hybrid stacked model, combining FNN with XGB, emerged as the top performer, achieving a Coefficient of Determination (R²) of 0.99 and Mean Absolute Percentage Error (MAPE) of 0.02 across all targets.
  • Feature importance analysis revealed that occupant behavior and infiltration play the most significant role in energy performance, surpassing structural and building envelope characteristics.
  • The research concluded that the developed workflow, AI-driven optimization framework, and robust hybrid modeling approach offer novel tools for energy-efficient building design and retrofitting.
  • The findings are particularly valuable for architects, urban planners, and policymakers seeking scalable, data-driven solutions to reduce energy consumption while maintaining thermal comfort.

Statistics:

  • The developed hybrid stacked model achieved a Coefficient of Determination (R²) of 0.99 on Energy Use Intensity (EUI), Predicted Percentage Dissatisfied (PPD), and Heating Load targets.
  • The model achieved a Mean Absolute Percentage Error (MAPE) of 0.02 across all targets.
  • The research utilized seven machine learning (ML) models, including Linear Regression (LR), Decision Trees (DT), Random Forest Regressor (RFR), Gradient Boosting Machines (GBM), Support Vector Regressor (SVR), K-Nearest Neighbors (KNN), and Extreme Gradient Boosting (XGB).

Sources:

  • AI-enhanced automation of building energy optimization using a hybrid stacked model and genetic algorithms: Experiments with seven machine learning techniques and a deep neural network. Results in Engineering, 2025,26():104994.
  • DOI: https://doi-org.sdpl.idm.oclc.org/10.1016/j.rineng.2025.104994 (Free version available at https://doi.org/10.1016/j.rineng.2025.104994)
  • Publisher: Elsevier
  • Authors: Mohammad H. Mehraban, Samad ME Sepasgozar, Alireza Ghomimoghadam, Behrouz Zafari
  • Keywords: Kingston University, London, United Kingdom, Europe, Cyborgs, Mathematics, Neural Networks, Machine Learning, Emerging Technologies.