Forecasts using neural network versus Box-Jenkins methodology for ambient air quality monitoring data

This study explores ambient air quality forecasts using the conventional ti me-series approach and a neural network. Sulfur dioxide and ozone monitorin g data collected from two background stations and an industrial station are used. Various learning methods and varied numbers of hidden layer processi ng units of the neural network model are tested. Results obtained from the time-series and neural network models are discussed and compared on the bas is of their performance for 1-step-ahead and 24-step-ahead forecasts. Altho ugh both models perform well for 1-step-ahead prediction, some neural netwo rk results reveal a slightly better forecast without manually adjusting mod el parameters, according to the results. For a 24-step-ahead forecast, most neural network results are as good as or superior to those of the time-ser ies model. With the advantages of self-learning, self-adaptation, and paral lel processing, the neural network approach is a promising technique for de veloping an automated short-term ambient air quality forecast system.