ВИЗУАЛЬНЫЙ ОСМОТР ЖЕЛЕЗНОДОРОЖНОЙ ИНФРАСТРУКТУРЫ НА ОСНОВЕ УГЛУБЛЕННОГО ОБУЧЕНИЯ: СИСТЕМАТИЧЕСКОЕ РАССМОТРЕНИЕ МЕТОДОВ, МОДЕЛЕЙ И ТРУДНОСТЕЙ В РЕАЛИЗАЦИИ
DOI:
https://doi.org/10.52167/1609-1817-2025-139-4-412-426Ключевые слова:
обследование железнодорожной инфраструктуры, углубленное обучение, компьютерное зрение, обнаружение неисправностей железных дорог, YOLOv5, машинное обучениеАннотация
В этой статье мы предлагаем обзор новейших методов глубокого обучения, используемых при проверке железнодорожной инфраструктуры с помощью систем визуализации. Традиционные методы проверки железных дорог не всегда эффективны, масштабируемы или точны, поэтому для обеспечения безопасности железных дорог необходимо использовать компьютерное зрение и искусственный интеллект. Исследование разбивает более 200 статей на важные темы, связанные с обнаружением дефектов на поверхности железной дороги, проблемами с крепежом и шпалами, включением синтетических данных, использованием различных типов железнодорожных участков и мультитехнологической интеграцией. Для каждого домена мы рассматриваем современные прогрессивные модели, такие как cnns, YOLOv5, U-Net, Faster R-CNN и Vision Transformers, обсуждая их использование, преимущества и недостатки. Проблема применения модели на практике анализируется путем изучения обобщений, оперативных результатов, понимания того, как она изучается, и ограничений аппаратного обеспечения. Мы также находим области, в которых необходим прогресс, и предлагаем будущие решения, чтобы уменьшить разницу между тем, какие академические открытия и какие отрасли могут использовать.
Библиографические ссылки
[1] Li, Y.; Chen, C.; Wang, K. Automatic Detection of Rail Surface Defects Based on Vision and Deep Learning Techniques//IEEE Access. – 2020. Т. 8. – С. 134676–134688. – DOI: https://doi.org/10.1109/ACCESS.2020.3010993
[2] Wang, H.; Song, K.; Li, Y. A Review of Railway Track Inspection Using Machine Vision and Deep Learning//Measurement. – 2021. Т. 176. – С. 109145. – DOI: https://doi.org/10.1016/j.measurement.2021.109145
[3] Pan, Z.; Wang, C.; He, Y. Real-Time Detection of Railway Fasteners Using Improved YOLOv5 and Edge Deployment//Sensors. – 2022. Т. 22. – С. 1235. – DOI: https://doi.org/10.3390/s22031235
[4] Lin, Y.; Li, Q.; Zhang, Z. Deep Learning-Based Crack Detection on Railway Sleepers Using U-Net Architecture//Automation in Construction. – 2020. Т. 118. – С. 103291. – DOI: https://doi.org/10.1016/j.autcon.2020.103291
[5] Kang, J.; Sun, G.; Tang, T. Railway Surface Defect Detection Using Residual Attention Network//Computers in Industry. – 2021. Т. 127. – С. 103400. – DOI: https://doi.org/10.1016/j.compind.2021.103400
[6] He, M.; Zhang, L.; Luo, S. Semantic Segmentation for Railway Track Using DeepLabv3+ with Multi-scale Features // IEEE Transactions on Intelligent Transportation Systems. – 2021. – DOI: https://doi.org/10.1109/TITS.2021.3055562
[7] Sun, Y.; Shao, C.; Meng, D. Challenges and Advances in Intelligent Railway Infrastructure Inspection // Transportation Research Part C: Emerging Technologies. – 2020. Т. 118. – С. 102732. – DOI: https://doi.org/10.1016/j.trc.2020.102732
[8] Zhao, W.; Liu, J.; Li, T. Rail Surface Defect Detection Based on GANs and Transfer Learning // IEEE Transactions on Industrial Informatics. – 2021. – DOI: https://doi.org/10.1109/TII.2021.3057515
[9] Zhang, Y.; Zhang, Z.; Zhou, X. A Survey of AI-Based Methods for Railway Inspection // Journal of Rail Transport Planning & Management. – 2023. Т. 25. – С. 100348. – DOI: https://doi.org/10.1016/j.jrtpm.2023.100348
[10] Sadeghi, J.; Askarinejad, H. A review of rail defects and their detection techniques // Journal of Transportation Technologies. – 2014. – URL: https://www.scirp.org/journal/paperinformation.aspx?paperid=49902 (дата обращения: 23.05.2025)
[11] Weston, P. et al. A review of ultrasonic rail inspection technology//Insight-Non-Destructive Testing and Condition Monitoring. – 2005. – URL: https://www.bindt.org/Journal-of-NDT/ (дата обращения: 23.05.2025)
[12] Li, Y. et al. Deep learning for vision-based railway inspection: A review//IEEE Access. – 2020. – DOI: https://doi.org/10.1109/ACCESS.2020.3002607
[13] Kang, J. et al. Railway surface defect detection using convolutional neural networks // Computers in Industry. – 2021. – DOI: https://doi.org/10.1016/j.compind.2021.103460
[14] Pan, Z. et al. Automatic fastener detection using improved YOLOv5 model // Sensors. – 2022. – DOI: https://doi.org/10.3390/s22082930
[15] Lin, Y. et al. Semantic segmentation of railway components using DeepLabv3+// Automation in Construction. – 2021. – DOI: https://doi.org/10.1016/j.autcon.2020.103291
[16] Wang, K.; Song, K. Challenges in dataset construction for railway defect detection//Measurement. – 2022. – DOI: https://doi.org/10.1016/j.measurement.2022.110096
[17] He, M.; Luo, S. Domain adaptation techniques for railway vision systems//Pattern Recognition Letters. – 2022. – DOI: https://doi.org/10.1016/j.patrec.2022.02.012
[18] Zhao, W. et al. Trustworthy AI in railway maintenance// Journal of Rail Transport Planning & Management. – 2023. – DOI: https://doi.org/10.1016/j.jrtpm.2023.100369
[19] Zakeri, A.; Nguyen, T.; Zhang, Y. Towards interpretable deep learning models for railway safety // Transportation Research Part C. – 2023. – DOI: https://doi.org/10.1016/j.trc.2023.104225
[20] Wang, H.; Song, K.; Li, Y. A Review of Railway Track Inspection Using Machine Vision and Deep Learning//Measurement. – 2021. Т. 176. – С. 109145. – DOI: https://doi.org/10.1016/j.measurement.2021.109145
[21] Ahmed, S.; Khan, R.; Lee, S. Comparative Analysis of Deep Learning Models for Crack Detection in Buildings//ResearchGate. – 2023. – URL: https://www.researchgate.net/publication/371348774 (дата обращения: 23.05.2025)
[22] Kim, H.; Lee, D. SAC-ConvLSTM: A Novel Spatio-Temporal Deep Learning-Based Algorithm//Expert Systems with Applications. – 2024. – DOI: https://doi.org/10.1016/j.eswa.2024.122648
[23] Chen, Y.; Xu, L.; Zhang, Q. A Review of Computer Vision-Based Crack Detection Methods in Civil Engineering//Remote Sensing. – 2022. Т. 16(16). – С. 2910. – DOI: https://www.mdpi.com/2072-4292/16/16/2910
[24] Huang, M.; Lin, Z.; Cao, J. Visual Analysis of Spatiotemporal Data Predictions with Deep Learning//Applied Sciences. – 2021. – Т. 11(13). – С. 5853. – DOI: https://doi.org/10.3390/app11135853
[25] Wang, X.; Zhou, J.; Chen, K. Sleeper Crack Detection Using CNN versus SVM on Real-world Data//Journal of Infrastructure Systems. – 2020. – Т. 26(1). – DOI: https://doi.org/10.1061/(ASCE)IS.1943-555X.0000520
[26] Li, Y.; Chen, C.; Wang, K. Automatic Detection of Rail Surface Defects Based on Vision and Deep Learning Techniques // IEEE Access. – 2020. – Т. 8. – С. 134676–134688. – DOI: https://doi.org/10.1109/ACCESS.2020.3011273
[27] Chen, Y.R.; Chao, K.; Kim, M.S. Machine vision technology for agricultural applications // Computers and Electronics in Agriculture. – 2002. – Т. 36. – С. 173–191. – DOI: https://doi.org/10.1016/S0168-1699(02)00100-X
[28] Min, Y.Z.; Yue, B.; Ma, H.F. Rail surface defects detection based on gray scale gradient characteristics of image // Chinese Journal of Scientific Instrument. – 2018. – Т. 9. – С. 220–229. – URL: https://www.instrument.com.cn (дата обращения: 23.05.2025)
[29] Lohmeier, S.P.; Rajaraman, R.; Ramasami, V.C. Development of an ultra-wideband radar system for vehicle detection at railway crossings // IEEE International Geoscience and Remote Sensing Symposium. – 2002. – Т. 6. – С. 3692–3694. – DOI: https://doi.org/10.1109/IGARSS.2002.1026870
[30] Li, Q.; Ren, S. A visual detection system for rail surface defects // IEEE Transactions on Systems, Man, and Cybernetics – Part C: Applications and Reviews. – 2012. – Т. 42. – С. 1531–1542. – DOI: https://doi.org/10.1109/TSMCC.2012.2209182
[31] Banik, P.P.; Saha, R.; Kim, K.D. An Automatic Nucleus Segmentation and CNN model based classification method of white blood cell // Expert Systems with Applications. – 2020. – Т. 149. – С. 113211. – DOI: https://doi.org/10.1016/j.eswa.2020.113211
[32] Liu, J.; Li, B.; Xiong, Y.; He, B.; Li, L. Integrating the symmetry image and improved sparse representation for railway fastener classification and defect recognition // Mathematical Problems in Engineering. – 2015. – Т. 2015. – С. 462528. – DOI: https://doi.org/10.1155/2015/462528
[33] Sun, X.; Gu, J.; Huang, R.; Zou, R.; Palomares, B.G. Surface defects recognition of wheel hub based on improved Faster R-CNN//Electronics. – 2019. – Т. 8. – С. 481. – DOI: https://doi.org/10.3390/electronics8050481
[34] Yuan, H.; Chen, H.; Liu, S.; Lin, J.; Luo, X. A deep convolutional neural network for detection of rail surface defect // Proceedings of the 2019 IEEE Vehicle Power and Propulsion Conference. – Hanoi, Vietnam, 14–17 October 2019. – С. 1–4. – DOI: https://doi.org/10.1109/VPPC46532.2019.8952355
[35] Song, Z.; Yamada, T.; Shitara, H.; Takemura, Y. Detection of damage and crack in railhead by using eddy current testing // Journal of Electromagnetic Analysis and Applications. – 2011. – Т. 3. – С. 546–550. – DOI: https://doi.org/10.4236/jemaa.2011.313084
[36] Kang, J.; Sun, G.; Tang, T. Railway Surface Defect Detection Using Residual Attention Network // Computers in Industry. – 2021. – Т. 127. – С. 103400. – DOI: https://doi.org/10.1016/j.compind.2021.103400
[37] Zhang, Y.; Zhang, Z.; Zhou, X. A Survey of AI-Based Methods for Railway Inspection // Journal of Rail Transport Planning & Management. – 2023. – Т. 25. – С. 100348. – DOI: https://doi.org/10.1016/j.jrtpm.2023.100348
Загрузки
Опубликован
Как цитировать
Выпуск
Раздел
Категории
Лицензия
Copyright (c) 2025 Гульсипат Абишева, Бибигуль Разахова, Толеген Айдынов, Nikolaj Goranin
Это произведение доступно по лицензии Creative Commons «Attribution-NonCommercial-NoDerivatives» («Атрибуция — Некоммерческое использование — Без производных произведений») 4.0 Всемирная.
