Forecasting air quality by using ANNs

The artificial neural networks (ANNs) have been extensively used in air pollution prediction because of their flexibility to deal with processes involv- ing non linear and complex data and/or to solve articulate problems in which a priori knowledge is incomplete or noisy. In this study, we trained different ANNs in assessing the capability of models for the prediction of air quality. The air pollution data from two monitoring stations in Pescara (Central Italy), along with some meteorological parameters, were used in forecasting Nitrogen Dioxide (NO2) levels, one day in advance, in the area of interest. The evaluation of obtained results shows that the degree of success in forecasting NO2 is promising.