Article

Received 25.01.2026

Revised 18.05.2026

Accepted 25.06.2026

Published 04.07.2026

Retrieved from Vol. 29, No. 1, 2026

Pages 71 -84

Katkalo, I., & Holub, H. (2026). Predictive maintenance of railway point machines using digital twin and deep learning. Transport Systems and Technologies, 29(1), 71-84. https://doi.org/10.32703/2617-9040-2026-47-6

Predictive maintenance of railway point machines using digital twin and deep learning

Ihor Katkalo*, Halyna Holub

ihor.katkalo@icloud.com

Point machine drives are critical components of railway signalling infrastructure, with their failures accounting for 15-20% of all signal system malfunctions across European networks, causing significant operational and economic losses. This study aimed to develop a predictive maintenance approach for 1,520 mm-gauge point machine drives using a digital twin, enabling early fault detection from power electrical signatures recorded during each switch cycle. The methodology combined a four-layer digital twin architecture (physical layer, data layer, analytics layer, and decision-making layer) with a comparison of three deep learning architectures: one-dimensional convolutional neural network (1D-CNN), long short-term memory network (LSTM), and a hybrid CNN-LSTM model. These were benchmarked against logistic regression, support vector machines (SVM), and Random Forest on a synthetic, physics-based dataset comprising 5,000 power curves for SP-6M and VSP-220 point machines. Models were evaluated on five-class state classification and 50-cycle degradation forecasting. The hybrid CNN-LSTM achieved the highest classification performance (macro F1 = 0.938), while LSTM yielded the best degradation forecast (R² = 0.937); both substantially outperformed the strongest baseline model (Random Forest: F1 = 0.871, R² = 0.845). Cross-generator validation and artefact ablation experiments confirmed that deep learning advantages arise from capturing transferable signal shape features rather than memorising simulation artefacts. A recommended deployment configuration includes 1D-CNN at the edge for real-time classification, LSTM at the server level for degradation forecasting, and Random Forest as an intermediate solution for operators with limited labelled data. The practical significance of these results lies in providing evidence-based guidelines for implementing the digital twin analytics layer on 1520 mm railway networks, supporting a transition from scheduled to condition-based maintenance and reducing unplanned downtime

control; diagnostics; power signature analysis; monitoring; architecture; modelling
  1. Givnan, S., Chalmers, C., Fergus, P., Ortega-Martorell, S., Whitaker, R., & Al-Jumeily, D. (2022). Anomaly detection using autoencoder reconstruction upon industrial motors. Sensors, 22(9), article number 3166. doi: 10.3390/s22093166.
  2. Hamadache, M., Dutta, S., Olaby, O., Kestelyn, X., & Raison, B. (2019). On the fault detection and diagnosis of railway switch and crossing systems: An overview. Applied Sciences, 9(23), article number 5129. doi: 10.3390/app9235129.
  3. Hector, I., & Panjanathan, R. (2024). Predictive maintenance in Industry 4.0: A survey of planning models and machine learning techniques. PeerJ Computer Science, 10, article number e2016. doi: 10.7717/peerj-cs.2016.
  4. Holub, H., Voronko, I., Bachynskyi, V., & Korchovyi, V. (2025). Integrated optimization of machine learning algorithms and cloud resources for big data processing in transportation systems. Science and Technology Today, 11(52), 1990-2000. doi: 10.52058/2786-6025-2025-11(52)-1990-2000.
  5. Hu, Z., Zhang, Y., Lv, R., Yan, J., & Wang, M. (2022). Data-driven technology of fault diagnosis in railway point machines: review and challenges. Transportation Safety and Environment, 4(4), article number tdac036. doi: 10.1093/tse/tdac036.
  6. Kaewunruen, S., Sresakoolchai, J., & Lin, Y.-H. (2023). Digital twins for managing railway bridge maintenance, resilience, and climate change adaptation. Sensors, 23(1), article number 252. doi: 10.3390/s23010252.
  7. Kostenko, K.L., Serdiuk, T.M., & Skalko, V.V. (2023). Robotization of diagnostics of railway automation and communication systems. Science and Transport Progress, 3(103), 5-12. doi: 10.15802/stp2023/292720.
  8. Ou, D., Xue, R., & Cui, K. (2019). A data-driven fault diagnosis method for railway turnouts. Transportation Research Record, 2673(4), 448-457. doi: 10.1177/0361198119837222.
  9. Reetz, S., Neumann, T., Schrijver, G., van den Berg, A., & Buursma, D. (2024). Expert system based fault diagnosis for railway point machines. Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit, 238(3), 322-334. doi: 10.1177/09544097231195656.
  10. Sherstinsky, A. (2020). Fundamentals of Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) network. Physica D: Nonlinear Phenomena, 404, article number 132306. doi: 10.1016/j.physd.2019.132306.
  11. Shi, Z., Du, Y., & Yao, X. (2023). Fault diagnosis of ZDJ7 railway point machine based on improved DCNN and SVDD classification. IET Intelligent Transport Systems, 17(8), 1649-1674. doi: 10.1049/itr2.12357.
  12. Siroklyn, I.M., & Kladko, A.S. (2018). Modern approaches to building intelligent systems for monitoring the technical condition of rolling stock. Information and Control Systems on Railway Transport, 4, 49-49.
  13. Sresakoolchai, J., & Kaewunruen, S. (2023). Railway infrastructure maintenance efficiency improvement using deep reinforcement learning integrated with digital twin based on track geometry and component defects. Scientific Reports, 13(1), article number 2439. doi: 10.1038/s41598-023-29526-8.
  14. Sun, Y., Cao, Y., Xie, G., & Wen, T. (2020). Condition monitoring for railway point machines based on sound analysis and support vector machine. Chinese Journal of Electronics, 29(4), 786-792. doi: 10.1049/cje.2020.06.007.
  15. Sysyn, M., Gruen, D., Gerber, U., Nabochenko, O., & Kovalchuk, V. (2019). Turnout monitoring with vehicle based inertial measurements of operational trains: A machine learning approach. Communications – Scientific Letters of the University of Zilina, 21(1), 42-48. doi: 10.26552/com.C.2019.1.42-48.
  16. Tao, F., Xiao, B., Qi, Q., Cheng, J., & Ji, P. (2022). Digital twin modeling. Journal of Manufacturing Systems, 64, 372-389. doi: 10.1016/j.jmsy.2022.06.015.
  17. Thompson, E.A., Lu, P., Alimo, P.K., Atuobi, H.B., Akoto, E.T., & Abbew, C.K. (2025). Revolutionizing railway systems: A systematic review of digital twin technologies. High-speed Railway, 3(3), 238-250. doi: 10.1016/j.hspr.2025.05.005.
  18. van Dinter, R., Tekinerdogan, B., & Catal, C. (2022). Predictive maintenance using digital twins: A systematic literature review. Information and Software Technology, 151, article number 107008. doi: 10.1016/j.infsof.2022.107008.
  19. Zhang, C., Zhang, X., Peng, X., & Li, Y. (2019). Fault diagnosis of railway point machines using the locally connected autoencoder. Applied Sciences, 9(23), article number 5139. doi: 10.3390/app9235139.
  20. Zonta, T., da Costa, C.A., da Rosa Righi, R., de Lima, M.J., da Trindade, E.S., & Li, G.P. (2020). Predictive maintenance in the Industry 4.0: A systematic literature review. Computers & Industrial Engineering, 150, article number 106889. doi: 10.1016/j.cie.2020.106889.

https://doi.org/10.32703/2617-9040-2026-47-6