DEVELOPMENT OF A NEURAL NETWORK STRUCTURE FOR OSCILLATION CLASSIFICATION
DOI:
https://doi.org/10.26906/SUNZ.2024.4.064Keywords:
artificial neural networks, Growing Neural Gas, GNG, Growing When Required, GWR, railway transport, analysis of train oscillationsAbstract
The paper examines today's urgent problem of increasing the maximum speed of railway transport, limitations and obstacles on the way to its solution and the possibility of overcoming these obstacles. The possibility of increasing the maximum speed of the train on those sections of the railway track, where it is permissible, was separately considered. To realize this possibility, the task of detecting, analyzing and classifying the vibrations that occur during train movement is relevant. A review of research devoted to the use of neural networks in related subject areas has been conducted. According to the results of the research analysis, it is proposed to use the Growing When Required (GWR) neural network – Growing Neural Gas modification to perform the task, which is optimal for performing the task of analyzing and classifying train oscillations and has the ability to learn more without damaging previously learned information. The structure of the GWR neural network has been developed. The algorithm of the GWR neural network is presented.Downloads
References
Заковоротный А. Ю. Синтез автоматизированной системы управления подвижным составом на основе геометрической теории управления и нейронных сетей [Электронный ресурс] : дис. ... д-ра техн. наук: спец. 05.13.07 / Александр Юрьевич Заковоротный ; науч. консультант Дмитриенко В. Д. ; Нац. техн. ун-т “Харьков. политехн. ин-т”. – Харьков, 2017. – 433 с. – Библиогр.: с. 326 – 358. – рус.
Deep convolutional neural networks for detection of rail surface defects / S. Faghih-Roohi, S. Hajizadeh, A. Núñez, R. Babuska, B. De Schutter // 2016 International joint conference on neural networks (IJCNN). IEEE. 2016. С. 2584–2589.
Carboni, M., Crivelli, D., 2020. An acoustic emission based structural health monitoring approach to damage development in solid railway axles. Int. J. Fatigue 139, 105753.
Lee, J.S., Hwang, S.H., Choi, I.Y., Kim, I.K., 2018. Prediction of track deterioration using maintenance data and machine learning schemes. J. Transp. Eng. A Syst. 144 (9), 04018045.
Khouzani, A.H.E., Golroo, A., Bagheri, M., 2017. Railway maintenance management using a stochastic geometrical degradation model. J. Transp. Eng. A Syst. 143 (1), 04016002.
Заковоротный А. Ю. Моделирование и оптимизация процессов управления движением дизель-поездов / В.Д. Дмитриенко, А.Ю. Заковоротный / [монография]. – Харьков: Изд. центр «HTMT», 2013. – 248 с.
Fritzke, B. (1995). A growing neural gas network learns topologies. In G. Tesauro, D. S. Touretzky, & T. K. Leen (Eds.), (pp. 625–632). Advances in Neural Information Processing Systems 7 (NIPS’94), Cambridge: MIT Press.
Marsland, S., Shapiro, J., & Nehmzow, U. (2002). A self-organising network that grows when required. Neural Networks, 15, 1041–1058.
