Predicting gaseous and solid air pollution with cyclostationary neural networks

In this chapter, we present a cyclostationary neural network (CNN) architecture to model and estimate hourly the NO2 concentrations, and to obtain a 1–day ahead prediction for the PM10. The cyclostationary nature of the problem guides the construction of the CNN. The strength of this approach particularly lies on its independence from meteorological information, suggesting that the network is able to infer the exogenous data directly from the pollution, therefore being robust with respect to geographical and seasonal changes. Some comparative experimentation (CNNs vs AutoRegressive eXogenous — ARX– models) is reported, based on the data gathered by ARPA (Agenzia Regionale per la Protezione dell’Ambiente — Regional Environmental Protection Agency) of Lombardia (northern Italy).