AI-enabled energy baselines for verified building decarbonization.

Area of Science:

• Building Science
• Artificial Intelligence
• Energy Systems

Background:

• Quantifying energy savings in operational buildings is complex due to dynamic factors like occupancy and weather.
• Existing methods struggle to accurately assess real-world energy performance post-retrofit.

Purpose of the Study:

• To develop and validate an AI-enabled framework for dynamic energy baselining and savings verification.
• To align energy savings metrics with policy requirements, such as Singapore's Green Mark certification.

Main Methods:

• Integration of high-resolution operational data with a hybrid LSTM-XGBoost model for energy use prediction.
• Development of Energy Conservation Calculation (ECC) logic for auditable savings metrics.
• Deployment and validation across diverse building types over three years.

Main Results:

• The AI framework demonstrated strong predictive performance (RMSE < 5%) across various building types.
• Verified cumulative emissions reductions of 3221 tCO₂e and energy use intensity improvements over 60% in retrofitted buildings.
• The framework proved effective for operational optimization and generating audit-ready sustainability reports.

Conclusions:

• Combining dynamic AI-based baselining with standardized energy accounting enables reliable verification of decarbonization in operational buildings.
• The proposed framework supports building performance optimization, certification, and sustainability-linked finance.
• This approach addresses the challenge of accurately quantifying energy savings in dynamic building environments.