Ensemble Learning Method for Forecasting HVAC System Demand

The efficiency of buildings’ energy use is one of the energy transition’s three fundamental pillars. It is a current issue that many countries are actively promoting as a means of lowering energy dependence by eliminating wasteful energy losses and environmental effect. Controlling energy flows within buildings is actually crucial for maximizing occupant comfort and timely use of consumer goods. In this brief study, ensemble machine learning techniques with different model structure, training process, interpretability, and performance, are developed to create a trustworthy tool for HL (Heat loading) and CL (Cool loading) estimation for future intelligent urban ecosystems. The results demonstrate that XGBoost and GBM with exhaustive feature selection provide accurate and more robust predictions. Moreover, the finding reveals that the proposed method handle better mixed Data.