Sequential air pollution emission estimation using a hybrid deep learning model and health-related ventilation control in a pig building

Ammonia (NH3), carbon dioxide (CO2), and hydrogen sulfide (H2S) are predominant gases that are responsible for indoor air quality and air pollution emitted from pig buildings. They are critical for the health of pigs, farm workers, and people living nearby. To achieve an accurate estimation of gas emissions, firstly, hybrid deep learning driven sequential Concentration Transport Emission Model (DL-CTEM) was proposed to estimate the emissions of NH3, CO2, and H2S from a pig building. Then, optimal ventilation control strategies were put forward to improve health-related gas concentrations and air pollution from the pig building. Fifty-three days of hourly measurements data were divided into training data and test data for the DL-CTEM. It was shown that the mean errors between the measurements and the predictions of the proposed model for NH3, CO2, and H2S concentrations were 0.1 ppm, 79.2 ppm, and 106.3 ppb, respectively. The proposed model outperformed when it was built with an optimal structure in the long short-term memory (LSTM) layer. The mean emission rates of NH3, CO2, and H2S based on DL-CTEM were 4.2 mg min(-1), 2887.5 mg min(-1), and 2.1 mu g min(-1). It could provide a feasible way for air pollution emission estimation and health-related ventilation control in a pig building.

Automated summary

This study proposes a deep learning-driven model to estimate the emissions of ammonia, carbon dioxide, and hydrogen sulfide from pig buildings. It also suggests optimal ventilation control strategies to improve indoor air quality and health. Experimental results show that the model has high prediction accuracy and provides a feasible method for estimating air pollution emissions and ventilation control in pig buildings.