Application of gradient boosting modeling method in building energy consumption prediction: a case study of a secondary school

In recent years, the energy issue has gained increasing attention from the world, with buildings contributing significantly to global energy consumption and carbon emissions. The issue of energy conservation and carbon reduction in middle school campuses had received little attention, as most schools' energy-saving potential remained untapped. Energy management was often rough, and there was a lack of awareness about energy conservation. This study aims to address the high energy consumption in middle school buildings by proposing an improved bottom-up energy consumption model. Traditional models used in the area expansion method may not accurately reflect actual conditions. The improved method integrated the real energy consumption data and refined predictions using a gradient boosting algorithm. The modeling process included zone classification, data extraction and preprocessing, physical model establishment, and bottom-up analysis. Applied to a secondary school in Beijing, the improved model demonstrated significantly higher accuracy, particularly during high-consumption winter months, with monthly errors kept below 10%. This accuracy is influenced by the sample size and iteration counts, ensuring optimal results. Given the issues of poor energy management and high carbon emissions in middle school campuses, this model excelled in capturing energy use patterns and suggesting potential energy-saving strategies. This innovative modeling technique provided a robust tool for policymakers and stakeholders to evaluate and implement energy conservation measures effectively, which could promote the development of green, sustainable campuses. The method was broadly applicable to all buildings with energy monitoring systems and could contribute to broader energy conservation efforts by enhancing energy efficiency and sustainability.