Applicability of smell agent optimization and Tasmanian devil optimization hybridized with ANFIS and SVR as reliable solutions in estimation of cooling load in buildings

Research conducted on the prediction of building energy consumption plays a growingly crucial role in the identification of optimal control strategies to combat excessive energy usage. The energy-efficient architectural design relies on the modeling of cooling and heating loads (CLs and HLs). These models define the prerequisites for cooling and heating systems, which are essential for maintaining a comfortable indoor air environment. Analytical models for energy-efficient buildings offer a precise assessment of the influence of various architectural designs. Nonetheless, the deployment of these tools may entail a substantial manual effort and an extended time commitment and is contingent on user expertise. Considering these factors, this paper employs machine learning techniques to introduce two distinctive approaches to estimate the cooling load of residential buildings, presented in the form of intricate mathematical formulations. These approaches encompass Smell Agent Optimization (SAO), Tasmanian Devil Optimization (TDO), and hybridization of a Support Vector Regression (SVR) and an adaptive neuro-fuzzy interface system (ANFIS), specifically denoted as ANTD, ANSA, SVTD, and SVSA. The results generated from each of the proposed models undergo assessment using a range of performance metrics. Furthermore, it is feasible to ascertain the most efficient model through a comparative analysis of their coefficient of determination (R²) and root mean square error (RMSE). Moreover, the results indicate that ANTD surpassed the other alternative models, emerging as the superior forecasting model, characterized by the highest R² value of 0.992 and the lowest RMSE value of 0.85 KW, while the SVR model was identified as the weakest model which exhibited RMSE 2.532 KW and R² values of 0.944, respectively, during the training phase. Accurate prediction of cooling loads enables better management of energy consumption in residential, commercial, and industrial buildings. By understanding the expected cooling demand, building operators can optimize HVAC (Heating, Ventilation, and Air Conditioning) systems for energy efficiency, thereby reducing utility costs and environmental impact.