NEURAL NETWORK CALCULATION OF CARBON DIOXIDE CONCENTRATIONS

Yu.A. Tunakova, S.V. Novikova, A.R. Shagidullin, V.S. Valiev
  Download PDF
Abstract: Reducing the carbon footprint is currently one of the priorities for the world economy. To do this, it is necessary to reduce greenhouse gas emissions, as well as to develop methods for monitoring greenhouse gases in the atmospheric air to ensure control over the effectiveness of decisions taken. Considering the complexity of the processes of dispersion of gases in the atmospheric air, neural network modeling methods have significant advantages in determining the concentrations of atmospheric impurities. This article presents a method for calculating the concentration of carbon dioxide in the atmospheric air using a designed and trained cascade neural network model, which makes it possible to take into account the complex influence of meteorological factors and local dispersion conditions when calculating concentrations. The first level of the model is the calculation of the concentration of carbon monoxide according to the known parameters of the emission sources of this substance using the regulated method for calculating the dispersion of impurities in the atmosphere in the Unified program for calculating dispersion "Ecolog-City". The second level is a neural network, which corrects the concentration calculated at the first step according to the specified meteorological parameters to increase the modeling accuracy. The third level is a neural network that allows calculating the concentration of carbon dioxide based on the concentration of carbon monoxide obtained at the previous step, as well as the measured values of the coefficient of chemical transformation and concentration of atmospheric ozone. The resulting cascade model was tested on the territory of Nizhnekamsk. The achieved accuracy of calculating the concentration of carbon dioxide was more than 95%. Thus, the presented technology makes it possible to expand the capabilities of the local monitoring system in conditions of an insufficient number of measurements of carbon dioxide.
Index terms: emissions into the atmosphere, greenhouse gas, neural network, calculation of concentrations, carbon dioxide

Contacts

Russia, 659305, Altai region, Biysk,
Trofimova Street, 27, room 404B
Tel. + 7-923-162-93-27
(executive secretary -
Golykh Roman Nikolayevich)
e-mail: info@s-sibsb.ru

The certificate