The neural network algorithm and the hybrid model for calculating the carbon dioxide greenhouse gas concentrations

by

Yu.A. Tunakova 1, S.V. Novikova 1, A.R. Shagidullin 2, V.S. Valiev 2

1Kazan National Research Technical University named after A.N. Tupolev – KAI,

  RF, Kazan, K. Marks St., 10

2Research Institute for Problems of Ecology and Mineral Wealth Use of Tatarstan Academy

  of Sciences, RF, Kazan, Daurskaya St., 28

E-mail: juliaprof@mail.ru, artur.shagidullin@tatar.ru

DOI: 10.33075/2220-5861-2023-3-133-140 

UDC 502.15                                                                                                                                                   

Abstract:

As part of the climate agenda, the task of calculating the concentrations of the greenhouse gas carbon dioxide (CO2) is solved. The calculation of CO2 concentrations by standard methods of dispersion calculations is not possible due to the lack of necessary data on the parameters of emission sources. Therefore, when creating a calculation model, traditional and innovative intelligent calculation technologies are combined. The created model has a cascade structure. The use of carbon monoxide concentration CO calculated using a consolidated database of city emission parameters, meteorological conditions, determining the processes of impurity distribution, transformation coefficient (TC) and ozone concentration (O3) as input parameters is justified. The step-by-step method of calculation is developed. At the first stage, a regulated method for calculating CO concentrations at the point under study is used. The second stage is the neural network correction of the results obtained at the first stage with a significant increase in the accuracy of CO concentration calculations. At the third stage, neural networks that directly calculate the CO2 concentration are used. At this stage, TC and O3 concentration are added as predictors to take into account atmospheric transformations. The choice of the neural network that performs the calculation at the third stage is made depending on the assignment of the input data tuple to one of the three blocks identified as a result of clustering using Kohonen’s self-learning neural networks. Approbation of the model is carried out on the example of Nizhnekamsk and Kazan cities. Average model error over the entire data set was less than one percent – 0.87%.

Keywords: emission, neural network, hybrid model, calculation of concentrations, carbon dioxide.

To quote: 

Full text in PDF(RUS)

REFERENCES

  1. Metody raschetov rasseivaniya vybrosov vrednyh (zagryaznyayushchih) veshchestv v atmosfernom vozduhe (Russian standard), Moscow: Ministry of Natural Resources of the Russian Federation, 2017.
  2. Vedenie sistemy raschetnogo monitoringa za sostoyaniem atmosfernogo vozduha dlya vyyavleniya istochnikov zagryazneniya, deyatel’nost’ kotoryh yavlyaetsya prichinoj povyshennoj zagazovannosti atmosfernogo vozduha v gorode Kazani (Maintaining a system of calculated monitoring of the state of atmospheric air to identify sources of pollution, the activity of which is the cause of increased atmospheric air pollution in the city of Kazan), Kazan: IPEN AN RT, 2020, 247 p.
  3. Vedenie sistemy raschetnogo monitoringa za sostoyaniem atmosfernogo vozduha dlya vyyavleniya istochnikov zagryazneniya, deyatel’nost’ kotoryh yavlyaetsya prichinoj povyshennoj zagazovannosti atmosfernogo vozduha v gorode Nizhnekamske (Maintaining a system of calculated monitoring of the state of atmospheric air to identify sources of pollution, the activity of which is the cause of increased atmospheric air pollution in the city of Nizhnekamsk), Kazan: IPEN AN RT, 2020, 177 p.
  4. Novikova S.V., Tunakova Yu.A., SHagidullin A.R., Kuznecova O.N. Ispol’zovanie intellektual’nyh raschetnyh metodov dlya povysheniya tochnosti rezul’tatov raschetnogo monitoringa osnovnyh komponentov vybrosov g.Nizhnekamska (soobshchenie 1) (Use of intelligent computational methods to improve the accuracy of the results of computational monitoring of the main components of emissions from the city of Nizhnekamsk (message 1)). Vestnik Tekhnologicheskogo universiteta, 2020, Vol. 23, No. 9, pp. 85–88.
  5. Novikova S.V., Tunakova YuU.A., SHagidullin A.R., Kuznecova O.N. Ispol’zovanie intellektual’nyh raschetnyh metodov dlya povysheniya tochnosti rezul’tatov raschetnogo monitoringa osnovnyh komponentov vybrosov g.Nizhnekamska (soobshchenie 2) (Use of intelligent computational methods to improve the accuracy of the results of computational monitoring of the main components of emissions from the city of Nizhnekamsk (message 2)). Vestnik Tekhnologicheskogo universiteta, 2020, Vol. 23, No. 9, pp. 89–92.
  6. Grigor’eva I.G., Tunakova Yu.A., Valiev V.S., Aleksandrova A.K., Kuznecova O.N. Ocenka koefficienta transformacii oksidov azota v prizemnom sloe atmosfery Nizhnekam-skogo promyshlennogo uzla (Estimation of the transformation coefficient of nitrogen oxides in the surface layer of the atmosphere of the Nizhnekamsk industrial hub). Vestnik tekhnologicheskogo universiteta, 2015, No. 19, pp. 242–244.

 

Loading