RESEARCH ON ARTIFICIAL INTELLIGENCE ALGORITHMS FOR BUILDING ENERGY CONSUMPTION PREDICTION

Abstract

The application of machine learning technologies is increasingly gaining attention for the optimization of Energy Management Systems (EMS). This research focuses on the application and performance comparison of three regression algorithms: Linear Regression (LR), Support Vector Regression (SVR), and Random Forest (RF), to address the problem of forecasting energy consumption (grid power and solar power) at Building C, Can Tho University of Technology. The evaluation results, using the metrics of MSE, RMSE, R², and MAE, indicate that the Linear Regression model exhibits the best performance in forecasting both grid power and solar power consumption, particularly in terms of error. Random Forest also demonstrates good forecasting capabilities, effectively capturing the trend of solar power data (R² = 1.00), while SVR shows higher error rates. Predicting energy consumption in buildings provides valuable information to support efficient investment decisions for building repair and renovation.