Research report: A proposed hybrid machine learning and physics-based optimization framework for energy-efficient smart buildings

Improving energy efficiency in smart buildings without sacrificing occupant comfort has long been a difficult problem, because building energy systems are complex and highly dynamic. A new study proposes a hybrid framework that combines machine learning, a physics-based thermal model of the building, and optimization for intelligent building energy management. Rather than relying solely on data-driven or physics-based approaches, the framework integrates the complementary benefits of both to enhance energy prediction and control performance.

The framework was evaluated on the ASHRAE Great Energy Predictor III benchmark dataset. After data preprocessing, the data were randomly split into three sets: 70 percent for training, 15 percent for validation, and 15 percent for testing. Model performance was assessed using root mean square error (RMSE), mean absolute error (MAE), and energy savings, and the results were compared against baseline models including Linear Regression, Random Forest, and Artificial Neural Network.

On the complete benchmark dataset, the proposed system achieved the best results among the comparison methods, with an RMSE of 0.24 kW, an MAE of 0.19 kW, and energy savings of up to 26.4 percent. According to the authors, these results show the effectiveness of incorporating machine learning with physics-based thermal modelling and optimization in enhancing both prediction accuracy and operational energy efficiency.

Overall, the study reports that the hybrid framework significantly improves HVAC energy prediction accuracy and energy efficiency compared with conventional methods, demonstrating its potential as a practical intelligent energy management solution for smart buildings. The work is published as open access research in Energy Informatics, with the article shared early to provide faster access to peer-reviewed, accepted research; this version is subject to further edits and will be replaced by the final Version of Record.

Subject of Research: A hybrid machine learning and physics-based optimization framework for energy-efficient smart building energy management

Article Title: A proposed hybrid machine learning and physics-based optimization framework for energy-efficient smart buildings

Article References: Ka’bi, A. A., AbuJbara, K., Al-Abdullah, K., Aljarajreh, H., & Zaki, M. (2026). A proposed hybrid machine learning and physics-based optimization framework for energy-efficient smart buildings. Energy Informatics. https://doi.org/10.1186/s42162-026-00687-w

Image Credits: AI Generated

DOI: 10.1186/s42162-026-00687-w

Keywords: smart buildings, machine learning, physics-based modelling, optimization, energy efficiency, HVAC, energy prediction, ASHRAE Great Energy Predictor III, building energy management, energy savings, proposed, hybrid