Analysis and development of a prediction model for the temperature-humidity index in broiler houses under regional, environmental, and operational conditions using building energy simulation

In this study, a previously validated Building Energy Simulation (BES) framework was applied to standard broiler house models to simulate the indoor thermal environment under various regional, environmental, and operational conditions.The BES model was constructed based on the 2024 Korean standard design guidelines for broiler houses, and a total of 3,072 simulation scenarios were conducted by combining region (16), weather input data (2), building size (4), building insulation level (4), broiler weight category (3), and cooling-system operation (2).For each scenario, hourly indoor air temperature and relative humidity were generated and subsequently used to calculate the Temperature-Humidity Index (THI).Statistical analyses were then performed by assigning the simulated THI as the dependent variable and the operational factors as independent variables.For factors showing significant differences, Duncan's multiple range test was applied to evaluate interactions and statistical significance among groups.Based on the simulation outcomes, 32 individual regression models were developed for combinations of 16 regions and two cooling-system conditions.For each model, the simulation cases were divided at the case level into approximately 70% training data (67 cases) and 30% test data (29 cases), and predictive performance was evaluated using the independent test dataset.The resulting models achieved a mean test R 2 of 0.9643, with mean RMSE and MAPE values of 0.8101 and 0.8696, respectively.