Analysis of the applicability of various machine learning methods for forecasting hourly gas consumption in the context of gas transmission companies
UDC: 681.5:622.279
DOI: 10.33285/2782-604X-2023-6(599)-5-14
Authors:
PANKRATOV SERGEY N.
1,
KAZAK ALEXANDER S.2,
LOBANOV ANDREY N.1,
GORLOV DMITRY V.2
1 Gazprom, St. Petersburg, Russia
2 NIIgazekonomika, Moscow, Russia
Keywords: machine learning, neural networks, linear regression, short-term forecasting, gas consumption, dispatch control
Annotation:
The development of reasonable management decisions in the process of operational dispatch control of the modes of the Unified Gas Supply System (UGSS) requires predicting changes of external factors affecting the UGSS functioning. The most important external factor is the volume of gas consumption. On the basis of forecasts for the gas supply to consumers, dispatching tasks are formed for gas transmission, gas producing companies, for underground gas storage facilities. At the same time, due to the large length of the gas transmission system (GTS) and the technological features of production facilities, the response time of the entire GTS to a change in the operational situation can take several days. For this reason, the accuracy of forecasting internal gas consumption for the UGSS GTS as the main expense item of the balance sheet is a key condition for making the right management decisions in the operational dispatch control of the UGSS GTS.
Bibliography:
1. Šebalj D., Mesarić J., Dujak D. Predicting Natural Gas Consumption – A Literature Review // Proc. of the 28th Central European Conf. on Information and Intelligent Systems, Sept. 27–29, 2017, Varaždin, Croatia. – P. 293–300. – URL: https://bib.irb.hr/datoteka/895588.clanak.pdf
2. Shyh-Jier Huang, Kuang-Rong Shih. Short-term load forecasting via ARMA model identification including non-Gaussian process considerations // IEEE Transactions on Power Systems. – 2003. – Vol. 18, Issue 2. – P. 673–679. – DOI: 10.1109/TPWRS.2003.811010
3. Al-Hamadi H.M., Soliman S.A. Long-term/mid-term electric load forecasting based on short-term correlation and annual growth // Electric Power Systems Research. – 2005. – Vol. 74, Issue 3. – P. 353–361. – DOI: 10.1016/j.epsr.2004.10.015
4. Razvitie sistemy podderzhki prinyatiya dispetcherskikh resheniy na baze sovremennykh podkhodov k operativnomu kratkosrochnomu prognozirovaniyu otbora gaza potrebitelyami v zone ESG / S.N. Pankratov, V.V. Dedkova, A.N. Lobanov [i dr.] // Gazovaya prom-st'. – 2021. – № S4(825). – S. 58–64.
5. Myuller A., Gvido S. Vvedenie v mashinnoe obuchenie s pomoshch'yu Python. Rukovodstvo dlya spetsialistov po rabote s dannymi: per. s angl. – Perm': Gevissta, 2017. – 393 s.
6. Vapnik V.N., Chervonenkis A.Ya. Teoriya raspoznavaniya obrazov. – M.: Nauka, 1974. – 415 s.
7. Vorontsov K.V. Lektsii po lineynym algoritmam klassifikatsii. – URL: http://www.machinelearning.ru/wiki/images/6/68/voron-ML-Lin.pdf
8. Kuhn M. Building Predictive Models in R Using the caret Package // J. of Statistical Software. – 2008. – Vol. 28, Issue 5. – P. 1–26. – DOI: 10.18637/jss.v028.i05
9. Muggeo V.M.R. Estimating regression models with unknown break-points // Statistics in Medicine. – 2003. – Vol. 22, Issue 19. – P. 3055–3071. – DOI: 10.1002/sim.1545
10. R Gazprom 173-2019. Kratkosrochnoe operativnoe planirovanie potokov gaza po Edinoy Sisteme Gazosnabzheniya. – SPb.: Gazprom ekspo, 2020.
Back to issue