«Automation and Informatization of the fuel and energy complex»

Experience of application of a neural network approach with additional network training on a limited volume of samples to solve the problem of classification of the objects, surrounding a gas pipeline

UDC: 004.032.26+528.7
DOI: 10.33285/0132-2222-2021-12(581)-14-20

Authors:

¹ JSC "Gazprom diagnostics", St. Petersburg, Russian Federation
² National University of Oil and Gas "Gubkin University", Moscow, Russian Federation

Keywords: neural network, artificial intelligence, gas transportation network, risk-based model

Annotation:

The article considers the actual problem of classifying objects which surround a main gas pipeline on the basis of a limited set of aerial photography data. The applicability of the use of neural network methods of already pre-trained artificial intelligence for solving this problem is shown. The accuracy and losses during the retraining of the neural network are estimated. In the course of the work, frameworks and software tools were selected, an overview and selection criteria were presented. The results of detecting constructions on a test aerial photograph, which were not included in the training kit, are presented. The prospects and directions of further research of neural network methods applicability for solving problems of managing the reliability and integrity of a unified gas supply system are indicated.

Bibliography:

1. Sistema upravleniya tekhnicheskim sostoyaniem i tselostnost'yu ob"ektov GTS PAO "Gazprom" / S.V. Nefedov, V.M. Silkin, G.A. Mil'ko-Butovskiy [i dr.] // Gazovaya prom-st'. – 2017. – № 3(749). – S. 14–20. 2. Zhuchkov K.N., Zav'yalov A.P. Sovershenstvovanie nauchno-metodicheskikh podkhodov k informatsionno-analiticheskomu obespecheniyu sistemy diagnosticheskogo obsluzhivaniya oborudovaniya i truboprovodov kompressornykh stantsiy // Oborudovanie i tekhnologii dlya neftegazovogo kompleksa. – 2021. – № 1(121). – S. 104–108. – DOI: 10.33285/1999-6934-2021-1(121)-104-108 3. Otraslevaya integrirovannaya informatsionno-upravlyayushchaya sistema. Informatsionno-upravlyayushchaya sistema transportirovkoy gaza i gazovogo kondensata (IUS T) PAO "GAZPROM". – URL: https://www.primegroup.ru/projects/pao-gazprom-4/ 4. Vasilchenko M., Zavyalov A., Zhuchkov K. Increasing the stability of a spatially distributed information system using a robust algorithm for filtering anomalous measurements // Information Technology in Industry. – 2020. – Vol. 8, No 3. – P. 1–7. 5. Plesnyaev V.A., Zhuchkov K.N. Razvitie informatsionnoy sistemy otsenki tekhnicheskogo sostoyaniya "Infotekh" // Oborudovanie i tekhnologii dlya neftegazovogo kompleksa. – 2019. – № 2(110). – S. 34–37. – DOI: 10.33285/1999-6934-2019-2(110)-34-37 6. Kratkosrochnoe stokhasticheskoe prognozirovanie dinamiki potokov i zapasov gaza v edinoy sisteme gazosnabzheniya Rossii na osnove rekurrentnykh neyronnykh setey / N.A. Kislenko, A.V. Belinskiy, A.S. Kazak, O.I. Belinskaya // Avtomatizatsiya, telemekhanizatsiya i svyaz' v neftyanoy promyshlennosti. – 2021. – № 9(578). – S. 27–35. – DOI: 10.33285/0132-2222-2021-9(578)-27-35 7. Podol'skiy A.K. Primenenie metodov iskusstvennogo intellekta v neftegazovoy promyshlennosti // Sovremennaya nauka. – 2016. – № 3. – S. 33–36. 8. Mandrik I.E., Shakhverdiev A.Kh., Suleymanov I.V. Otsenka i prognoz nefteotdachi na osnove modelirovaniya neyronnymi setyami // Neft. khoz-vo. – 2005. – № 10. – S. 36–39. 9. Machine learning powered by principal component descriptors as the key for sorted structural fit of XANES / A. Martini, A.A. Guda, S.A. Guda [et al.] // Physical Chemistry Chemical Physics. – 2021. – Vol. 33. – P. 17873–17887. – DOI: 10.1039/D1CP01794B 10. Mak-Kallok U.S., Pitts V. Logicheskoe ischislenie idey, otnosyashchikhsya k nervnoy aktivnosti = A logical calculus of the ideas immanent in nervous activity // Avtomaty: sb. / pod red. K.E. Shennona i Dzh. Makkarti. – M.: Izd-vo inostr. lit., 1956. – S. 363–384. 11. Hebb D.O. The organization of behavior a neuropsychological theory. – New York: Wiley, 1949. – 368 p. 12. Rozenblatt F. Printsipy neyrodinamiki: pertseptrony i teoriya mekhanizmov mozga = Principles of Neurodynamic: Perceptrons and the Theory of Brain Mechanisms. – M.: Mir, 1965. – 480 s. 13. Dechter R. Learning while searching in constraint-satisfaction problems // Proc. of the 5th National Conf. on Artificial Intelligence, Philadelphia, PA, USA, Aug. 11–15. – 1986. – Vol. 1. – P. 178–185. 14. Ivakhnenko A.G., Lapa V.G. Kiberneticheskie predskazyvayushchie ustroystva. – Kiev: Naukova dumka, 1965. – 214 s. 15. Backpropagation Applied to Handwritten Zip Code Recognition / Y. LeCun, B. Boser, J.S. Denker [et al.] // Neural Computation. – 1989. – Vol. 1, Issue 4. – P. 541–581. – DOI: 10.1162/neco.1989.1.4.541 16. Sholle F. Glubokoe obuchenie na Python. – SPb.: Piter, 2018. – 400 s. 17. Poynter Ya. Programmirovanie na PyTorch: Sozdanie prilozheniy glubokogo obucheniya. – SPb.: Piter, 2020. – 256 s. 18. Sirota A.A., Mitrofanova E.Yu., Milovanova A.I. Analiz algoritmov poiska ob"ektov na izobrazheniyakh s ispol'zovaniem razlichnykh modifikatsiy svertochnykh neyronnykh setey // Vestn. Voronezhskogo gos. un-ta. Ser.: Sistemnyy analiz i informatsionnye tekhnologii. – 2019. – № 3. – S. 123–137.