<![CDATA[Closed-type poultry houses support stable production performance by maintaining a controlled microenvironment that promotes optimal poultry growth. However, many farms still rely on manual monitoring of environmental parameters such as temperature, humidity, and ammonia concentration, resulting in delayed responses, reduced productivity, and increased environmental stress on poultry. These limitations highlight the need for predictive and automated systems that can monitor and forecast environmental conditions in real time. Previous studies have shown that LSTM networks are effective for nonlinear time-series forecasting. However, when applied independently, LSTM models often face difficulties in capturing linear seasonal patterns and long-term trends inherent in poultry house environmental data. Therefore, this study proposes a hybrid forecasting framework that integrates LSTM and SARIMA models to simultaneously capture nonlinear temporal dependencies and linear seasonal components. Environmental parameters, including temperature, litter moisture, and ammonia concentration, were collected using SHT31, Soil Moisture, and MQ137 sensors. The collected data were processed using a Python-Flask backend system, stored in MongoDB, and visualized through a cross-platform web interface developed using Flutter. Experimental results demonstrate that the proposed LSTM–SARIMA model achieves strong predictive performance, with MAE = 0.62, MSE = 0.55, RMSE = 0.58, MAPE = 7.89%, and R² = 0.86. These findings indicate that the proposed method effectively supports early warning systems and real-time microclimate monitoring, enabling faster environmental control responses and reducing production losses caused by unstable poultry house conditions.]]>
Copyrights © 2026
