Development of SO2 Prediction Model with Artificial Neural Network for Pune City

Abstract In this study, an artificial neural network is proposed to predict SO2 concentrations for Pune city, the major industrial metropolis of Maharashtra, India. The developed artificial neural network models involve meteorological parameters viz., temperature, relative humidity and wind speed and historical data on observed SO2 concentrations for 6 yrs (2004-2010) as input. The subsequent SO2 concentration for one day ahead being the output parameter of this study is estimated. The developed model is based on three-layer neural network trained by a back-propagation algorithm with number of epoch. The models accurately match the trend of SO2 concentrations for one day ahead upto 86.9%. Keywords : Artificial neural network, SO2, back-propagation, transform, Generalized Feed Forward Network