Optimizing energy efficiency: predicting heating load with a machine learning approach and meta-heuristic algorithms
摘要
This study responds to the urgent demand for accurate energy consumption forecasts and the strategic assessment of retrofit methodologies, aiming to achieve energy conservation and emissions reduction simultaneously. Amidst an escalating emphasis on energy-efficient practices, the research pioneers an integration of advanced optimization algorithms within a meticulous heating load (HL) prediction framework. Focused on the intricate domain of HL systems, this study leverages the Radial Basis Function (RBF) model, enhancing predictive precision through the integration of two meta-heuristic algorithms: the Artificial Hummingbird Algorithm (AHA) and the Improved Grey Wolf Optimizer (IGWO). The validation process entails a comprehensive examination of HL data from diverse energy sources, subjected to stringent stability tests. This study introduces three distinctive models—RBF + AHA (RBAH), RBF + IGWO (RBIG), and an independent RBF model—each contributing unique insights for precise HL prediction. Remarkably, the RBAH model emerges as an exemplar, demonstrating an outstanding coefficient of correlation (R2) value of 0.996 and an exceptionally low root mean square error (RMSE) value of 0.605. These findings underscore the unparalleled predictive prowess of the RBAH model in forecasting heating load outcomes, marking it as a noteworthy advancement in the realm of energy prediction models. The study's contributions extend beyond the conventional, providing a robust framework and invaluable insights for advancing energy-efficient practices in building management, thereby addressing contemporary sustainability challenges.