Hybrid machine learning application with integration of meta-heuristic algorithm for prediction of cooling load

Cooling Load, the energy needed for temperature control in a space, is vital for energy conservation, efficient management, and planning. Accurate energy consumption predictions are crucial for resource optimization and sustainability. Improving predictive models is essential for energy system efficacy as technology advances. This research introduces hybrid machine learning models integrated with advanced optimization techniques designed to estimate Cooling Load in buildings precisely. By merging ML and optimization, it aims to innovate in predicting and managing cooling energy needs, contributing to overall sustainability in the built environment. The research utilizes the Support Vector Regression model combined with the Fire Hawk Optimizer and the Leader Harris Hawks optimization method to achieve this goal. A thorough set of experiments is carried out, covering various architectural variables like orientation, glazing area, relative compactness, building height, surface area, roof area, and wall area. A comparative assessment is then performed to assess how well the suggested models can predict outcomes. As indicated by the results, the SVR+LHHO model, which fuses the Support Vector Regression model with the Leader Harris Hawks optimization, achieved the highest correlation coefficient at an impressive 99.4%. Furthermore, it showcased the smallest statistical Root Mean Square Error value, recording a minimal 0.745. These findings provide strong evidence of the outstanding predictive capabilities as well as the effectiveness of the SVR+LHHO model in predicting Cooling Load.