An ensemble model for the energy consumption prediction of residential buildings

The HVAC unit helps reduce overall energy consumption. ML models can enhance HVAC performance by accurately predicting a building's energy consumption and load utilization. Therefore, this study presents a stacked ensemble model that incorporates extreme gradient boosting (XGB), decision tree (DT), and Random Forest (RF) algorithms to predict the energy consumption of heating and cooling loads (HL and CL) in buildings. The performance of the proposed stacked ensemble is compared to other machine learning predictive models such as Ridge, Lasso, K Nearest Neighbor (KNN), Support Vector Regression (SVR), and Artificial Neural Network (ANN). Bayesian optimization is used to determine the hyperparameter values of the ML algorithms. The results show that the proposed predictive model has the lowest root mean square value (RMSE) of 0.484 and 0.948 for HL and CL, respectively, compared to other machine learning models. Additionally, the efficacy of the stack model is evaluated using a time series dataset about HVAC energy consumption in residential buildings. The simulation results indicate that the stack model outperformed the other prediction models, achieving a root mean square error (RMSE) of 0.1810. In conclusion, the proposed predictive model is more efficient than traditional models in forecasting energy consumption by HL, CL, and HVAC systems.