V.V. Belousov1, O.V. Druzhinina2, E.R. Korepanov3, I.V. Makarenkova4, V.V. Maksimova5

1-5 FRС «Computer Science and Control» of RAS (Moscow, Russia)

The development of instrumental and methodological support for the construction and analysis of neural network models in the problems of diagnosing the state of technical systems is an urgent direction that is associated with the introduction of digital technologies. The use of neural networks for data processing in the problems of fault detection and assessment of the technical condition of elements and nodes of transport systems makes it possible to expand the standard capabilities of information and control systems. The objectives of the paper include: the application of neural network modeling methods to solving problems aimed at identifying malfunctions of elements and nodes of transport systems; analysis of models of functioning of axle boxes of railway cars using data mining; determination of neural network parameters and creation of a machine learning algorithm for technical diagnostics of axle box failures. An approach to the modeling of technical systems has been developed, aimed at developing methods for detecting malfunctions of axle boxes of wagons using data mining. The analysis of methods based on mathematical statistics and neural network methods for detecting malfunctions of axle boxes is carried out. The formulation of the classification problem is considered in relation to the analysis of malfunctions of axle boxes by temperature characteristics. The construction of a neural network of a suitable architecture is proposed, taking into account the features used in practice when determining warming boxes. A machine learning algorithm for neural networks has been developed to solve the classification problem. The results can be used in the problems of creating methodological and instrumental support for solving problems of technical diagnostics using artificial intelligence methods. The considered approach to the modeling of technical systems can be used in the development of intelligent transport systems and the improvement of digital twin technologies.

Belousov V.V., Druzhinina O.V., Korepanov E.R., Makarenkova I.V., Maksimova V.V. The use of neural networks for solving classi-fication problems in diagnosing malfunctions of transport systems. Neurocomputers. 2022. V. 24. № 4. Р. 18-27.
DOI: https://doi.org/10.18127/j19998554-202204-02 (in Russian)

1. Mironov A.A., Obrazcov V.L., Pavljukov A.Je. Teorija i praktika beskontaktnogo teplovogo kontrolja buksovyh uzlov v poezdah. Ekaterinburg: RPF «Assorti». 2012 (In Russian).
2. Mironov A.A., Pavljukov A.Je., Saltykov D.N. Kompleks vychislitel'nyh modelej dlja issledovanija processov kontrolja uzlov podvizhnogo sostava po infrakrasnomu izlucheniju. Mir izmerenij. 2014. № 6. S. 21–27 (In Russian).
3. Prohorov A., Lysachev M. Cifrovoj dvojnik. Analiz, trendy, mirovoj opyt. Pod red. A. Borovkova. M.: OOO «Al'jansPrint». 2020 (In Russian).
4. Dzjuba Ju. V., Pavlovskij A. A., Umanskij V. I. Cifrovaja zheleznaja doroga. Tehnologicheskij uroven'. Perspektivy nauki i obrazovanija. 2018. № 1(31). S. 208–213 (In Russian).
5. Tjagunov AA. Cifrovaja transformacija v sfere transporta. Nauka i tehnologii zheleznyh dorog. 2021. № 2. S. 13–21 (In Russian).
6. Belousov V.V., Druzhinina O.V., Korepanov Je.R., Makarenkova I.V., Maksimova V.V. Podhod k ocenke tehnicheskogo sostojanija jelementov i uzlov transportnyh sistem s primeneniem metodov nejrosetevogo modelirovanija i tehnologii cifrovyh dvojnikov. Nejrokomp'jutery: razrabotka, primenenie. 2021. T. 23. № 5. S. 5–20 (In Russian).
7. Belousov V. V., Druzhinina O. V., Korepanov Je. R., Makarenkova I. V, Maksimova V. V. O podhodah k sozdaniju i ispol'zovaniju cifrovyh dvojnikov dlja jelementov i uzlov transportnyh sistem. Materialy VII Mezhdunar. nauch.-praktich. konf. «Sistemy upravlenija, slozhnye sistemy: modelirovanie, ustojchivost', stabilizacija, intellektual'nye tehnologii». Elec: EGU im. I.A. Bunina. 2021. S. 79-85. (In Russian).
8. Brink H., Richards Dzh., Feverolf M. Mashinnoe obuchenie. SPb: Piter. 2017.
9. Hastie T., Tibshirani R., Friedman J. The elements of statistical learning. Springer. 2017.
10. Pugachev V.S., Sinicyn I.N. Teorija stohasticheskih sistem. M.: Izd-vo «Logos». 2000; 2003 (In Russian).
11. Mironov A.A., Tagirov A.F., Obrazcov V.L. Avtomatizirovannaja sistema kontrolja podvizhnogo sostava (ASK PS). Bezopasnost' dvizhenija, sovershenstvovanie konstrukcij vagonov i resursosberegajushhie tehnologii v vagonnom hozjajstve. Ekaterinburg: Izd-vo Ural'skogo gosud. un-ta putej soobshhenija. 2003. S. 65–70 (In Russian).
12. Hajkin S. Nejronnye seti: polnyj kurs. Izd. 2-e, ispr. M.-SPb: Dialektika. 2019 (In Russian).
13. Galushkin A.I. Nejronnye seti: osnovy teorii. M.: Gorjachaja linija – Telekom. 2012 (In Russian).
14. Modeli nejronnyh setej (s uchitelem). Mnogoslojnyj perseptron [Jelektronnyj resurs]. Rezhim dostupa https://scikit-learn.ru/1-17-neural-network-models-supervised/#classification. svobodnyj. Zagl. s jekrana. Jaz. rus., angl. (In Russian).
15. Gorban' A. N. Obobshhennaja approksimacionnaja teorema i vychislitel'nye vozmozhnosti nejronnyh setej. Sibirskij zhurnal vychislitel'noj matematiki. 1998. T. 1. № 1. S. 11–24. (In Russian).
16. Sinicyn I.N., Druzhinina O.V., Belousov V.V., Masina O.N., Petrov A.A. Opyt razrabotki instrumental'no-metodi-cheskogo obespechenija dlja reshenija zadach modelirovanija upravljaemyh dinamicheskih sistem s primeneniem tehnologij mashinnogo obuchenija i otechestvennyh programmno-apparatnyh sredstv. Nelinejnyj mir. 2019. T. 17. № 4. S. 5–19 (In Russian).
17. Druzhinina O.V., Korepanov Je.R., Belousov V.V., Masina O.N., Petrov A.A. Opyt razrabotki metodov i sredstv nejro-setevogo modelirovanija nelinejnyh sistem na baze otechestvennoj vychislitel'noj platformy «Jel'brus 801-PC». Nelinejnyj mir. 2020. T. 18. № 2. S. 5–17 (In Russian).