A New Back-propagation Algorithm for Modelling Air Quality Time Series

Authors:	Giuseppe Nunnari, University of Catania, Italy Cannavò Flavio, University of Catania, Italy
Topic:	8.3 Modelling & Control of Environmental Systems
Session:	Modeling and Control of Environmental Systems
Keywords:	Stochastic modelling, Air pollution, Forecasts, Neural Networks, Backpropagation

Abstract

In this paper a new Back-propagation algorithm appropriately studied for modelling air pollution time series is proposed. The underlying idea is that of modifying the error definition in order to improve the capability of the model to forecast episodes of poor air quality. In the paper five different expressions of error definition are proposed and their performances are rigorously evaluated by using an appropriate set of indices, in the framework of a real case study which refer to the modelling of 1 hour average maximum daily Ozone concentration recorded in the industrial area of Melilli (Siracusa, Italy).Results indicate that despite the traditional and the proposed version of Back-propagation performs quite similarly in terms of success index, this latter algorithm performs better in terms of the percentage of exceedences correctly forecast.