DBPapers
DOI: 10.5593/sgem2017H/43/S19.070

THE MODEL OF NEURAL NETWORKS FOR FORECASTING OF DISTRIBUTION OF THE AIR POLLUTION IN THE URBAN AREA

N. Plugotarenko, O. Korotkova
Thursday 23 November 2017 by Libadmin2017

References: 17th International Multidisciplinary Scientific GeoConference SGEM 2017, www.sgemviennagreen.org, SGEM2017 Vienna GREEN Conference Proceedings, ISBN 978-619-7408-28-7 / ISSN 1314-2704, 27 - 29 November, 2017, Vol. 17, Issue 43, 553-560 pp; DOI: 10.5593/sgem2017H/43/S19.070

ABSTRACT

The monitoring of quality of the urban air environment takes a big part in reaching good environmental conditions. The model of neural networks for forecasting of distribution of nitrogen dioxide in the surface layer of the air environment is shown in this research. The model is based on meteorological data and the information of sources of emissions of polluting substances. The optimal network configuration was determined on the basis of standard deviation and the quality of education. This is the cascade network with direct propagation of signal and back-propagation of error. The network consists of two layers with 11 neurons. The training of network conducted the method of Levenberg–Marquardt based on environmental monitoring data. It’s shown in retrainingthat standard deviation decreases more even more than for 2 years. The results of the model are visualized in ArcGis program. The results of the modeling give the information about the distribution of air nitrogen dioxide with a regard of the different meteorological conditions and the contribution of industrial pollution in the air environment of the city, enabling a timely response and management decisions.

Keywords: monitoring, neural network , modeling, cascade network, pollutants