Artificial Neural Network for the Classification and Prediction of Air Quality

Abstract

Air pollution is the presence in the atmosphere of substances in an amount that involves discomfort or risk to the health of people and other living beings. They are being produced in different ways, and they can attack different materials, reduce visibility or produce unpleasant odors. Polluted emissions to the atmosphere (which is strongly related to climate change) can aggravate the effects on the health of population, not only indirectly (by the impact on weather phenomena), but immediately, due the direct effects of pollutants for health. In this paper we propose the application of a back propagation neural network to make predictions (one hour later) about the air quality, using estimation methods to complete the missing records of the databases used. This study is based on a four years database obtained from the CICEG atmosphere station, one of the three stations that are located in the city of León, Guanajuato Mexico. Five pollutants are considered as part of the parameters of the model (????????????,????????????????, ????????????, ????????, ????????) in addition to the month, the day and the time when the contaminant were registered. The results show an acceptable accuracy of the model used for the prediction of air quality.