Prediction of Air Pollutants Concentration Emitted from Kirkuk Cement Plant Based on Deep Learning and Gaussian Equation Outputs

Researchers are interested in developing techniques to monitor, manage and predict the risks of gases and particles emitted from cement factories, which have a direct and negative impact on human health. Deep learning (DL) is a critical component of data mining, which further involves statistics and prediction. In this study, we developed a deep learning prediction model called the Deep Pollutant Prediction Model (DPPM). The data used for DPPM are separated into two types: observed data from a pollution monitoring station of the Institute of Mental Health in Ahmedabad City, India coded as (GJ001), to validate the model and simulated data generated using the Gaussian Plume Model for the hypothetical receptor (Laylan District, Kirkuk, Iraq) to predict the pollution that emitted from Kirkuk Cement Plant 5 km apart from the study area. The findings indicated that the DPPM has high efficiency in both Allahabad and Laylan stations, with more closed results for the data in the Laylan station, which is based on the Gaussian equation simulated data. Since the highest loss function value in the Laylan is 0.0221 of the CaO parameter, while it is 4.466 of the AQI parameter for the Allahabad Station, and the smallest loss function value in the Laylan is equal to 0.0041of both Fe2O3 and MgO parameters, it corresponds to 0.038 of Xylene for the Allahabad station. The results of the study proved that data continuity and non-volatility produce excellent outcomes for DPPM.