Artificial Intelligence-Based Prediction of Pollution from Power Systems and Transportation Sectors

Due to global warming and climate changes, accurate air quality monitoring and prediction became very essential. The long-term and short-term effects of excess air pollutants in the atmosphere can be detrimental to the health of humans and the environment. This paper focuses on the development of an Artificial Intelligence (AI)-driven tool to predict the pollution from the power generation sector and cars. Two types of dynamic neural networks (NN), time-delay NN and layer recurrent NN, are developed and tested in MATLAB. The proposed NNs are used in the prediction of sulfur dioxide (SO
2
) and particulate matter (PM
2.5
) produced from power plants and cars in Prichard, AL, USA. Each NN consists of seven inputs and two outputs. The inputs included historical power plant emissions, traffic volume data from two highways leading into Prichard, and historical meteorological data from the area. Outputs included the predicted atmospheric concentrations of SO
2
and PM
2.5
in Prichard. The accuracy of each NN is measured using the Root Mean Squared Error (RMSE).