Bulanık Modelleme ve CNN-BILSTM Ağı Kullanılarak Ray Yüzeyi Kusurlarının Tespiti ve Sınıflandırılması için Kararlı İki Aşamalı Yaklaşım
Öz
Ray yüzeyindeki kusurlar, demiryolu taşımacılığını olumsuz etkileyen önemli bir sorun teşkil etmektedir. Bu tür kusurların zamanında tespit edilmesi, hat üzerindeki tehlikelerin belirlenmesi ve erken bakımın sağlanması sayesinde ciddi kazaların önlenmesine olanak tanır. Ancak, yüzeydeki pas ve yağ gibi safsızlıklar nedeniyle kusurların doğru şekilde tespit edilmesi oldukça zordur. Yüzey kusurlarının tespitinde temel amaç, yüksek doğruluk elde ederek gerçek kusurlar ile yalancı alarmların karıştırılmasını engellemektir. Bu çalışmada, söz konusu zorlukların üstesinden gelmek amacıyla iki aşamalı, hızlı bir ray yüzeyi kusur tespit yöntemi önerilmektedir. İlk olarak, demiryolu görüntüsünde rayları içeren piksellerin farklılıkları analiz edilerek ray çıkarımı gerçekleştirilir. Daha sonra, ön işleme adımında oluşabilecek gürültüler giderildikten sonra, sağlam bir ray görüntüsünün histogramı bulanık üyelik fonksiyonu kullanılarak modellenir. Kusurlu ray, bu bulanık üyelik değerlerine göre bölütlenir ve kusur olup olmadığı belirlenir. Kusur türünün sınıflandırılmasında ise Evrişimsel Sinir Ağı- Çift Yönlü Uzun Kısa Süreli Bellek (CNN-BILSTM) yöntemi kullanılır. Önerilen yüzey kusur tespit yöntemi, iki farklı kıyaslama veri kümesinde (Rail Surface Defect Datasets - RSDD-I/II) mevcut yöntemlerden daha yüksek performans göstermiştir. Ayrıca, kusur türünü belirlemede %98,00 oranında tanıma başarısı elde edilmiştir.
Anahtar Kelimeler
Demiryolu Görüntü Segmentasyonu Bulanık Ölçüm Evrişimsel Sinir Ağları Yüzey Kusurları Kusur Tespiti
Kaynakça
- [1] L. Zhuang, H. Qi, & Z. Zhang, “The automatic rail surface multi-flaw identification based on a deep learning powered framework,” IEEE Trans. Intell. Transp. Syst., 2021.
- [2] H. Yang, Y. Wang, J. Hu, J. He, Z. Yao, & Q. Bi, “Segmentation of track surface defects based on machine vision and neural networks,” IEEE Sensors J., vol. 22, no. 2, pp. 1571–1582, 2021.
- [3] M. Sevi, İ. Aydın, E. Akın, “Detection of rail surface defects based on ensemble learning of YOLOv5,” Demiryolu Mühendisliği, no. 17, pp. 115-132, Jan. 2023. doi: 10.47072/demiryolu.1205483.
- [4] M. Guerrieri, G. Parla, & C. Celauro, “Digital image analysis technique for measuring railway track defects and ballast gradation,” Measurement, vol. 113, pp. 137–147, 2018.
- [5] Y. Jiang et al., “Non-contact ultrasonic detection of rail surface defects in different depths,” in 2018 IEEE Far East NDT New Technology & Application Forum (FENDT), pp. 46–49, IEEE, 2018.
- [6] G. Piao, J. Li, L. Udpa, J. Qian, & Y. Deng, “Finite-element study of motion-induced eddy current array method for high-speed rail defects detection,” IEEE Trans. Magn., vol. 57, no. 12, pp. 1–10, 2021.
- [7] Y. Santur, M. Karaköse, & E. Akin, “A new rail inspection method based on deep learning using laser cameras,” in 2017 Int. Artificial Intelligence and Data Processing Symposium (IDAP), pp. 1–6, IEEE, 2017.
- [8] H. Zhong, L. Liu, J. Wang, Q. Fu, & B. Yi, “A real-time railway fastener inspection method using the lightweight depth estimation network,” Measurement, vol. 189, p. 110613, 2022.
- [9] H. Wang, M. Li, & Z. Wan, “Rail surface defect detection based on improved Mask R-CNN,” Comput. Electr. Eng., vol. 102, p. 108269, 2022.
- [10] J. Gan, J. Wang, H. Yu, Q. Li, & Z. Shi, “Online rail surface inspection utilizing spatial consistency and continuity,” IEEE Trans. Syst., Man, Cybern.: Syst., vol. 50, no. 7, pp. 2741–2751, 2018.
- [11] H. Zhang et al., “Automatic visual detection system of railway surface defects with curvature filter and improved Gaussian mixture model,” IEEE Trans. Instrum. Meas., vol. 67, no. 7, pp. 1593–1608, 2018.
- [12] D. Zhang et al., “Two deep learning networks for rail surface defect inspection of limited samples with line-level label,” IEEE Trans. Ind. Informat., vol. 17, no. 10, pp. 6731–6741, 2020.
- [13] Z. Zhang, M. Liang, and Z. Wang, “A deep extractor for visual rail surface inspection,” IEEE Access, vol. 9, pp. 21798–21809, 2021.
- [14] Y. Wu et al., “Hybrid deep learning architecture for rail surface segmentation and surface defect detection,” Comput.-Aided Civ. Infrastruct. Eng., vol. 37, no. 2, pp. 227–244, 2022.
- [15] H. Yang et al., “Deep learning and machine vision-based inspection of rail surface defects,” IEEE Trans. Instrum. Meas., vol. 71, pp. 1–14, 2021.
- [16] H. Li et al., “Ensemble model for rail surface defects detection,” PLoS One, vol. 17, no. 5, p. e0268518, 2022.
- [17] İ. Aydın & E. Akın, “Two-stage rail defect classification based on fuzzy measure and convolutional neural networks,” in Intelligent and Fuzzy Systems: Digital Acceleration and The New Normal-Proceedings of the INFUS 2022 Conf., vol. 1, pp. 769–776, Springer, Cham, 2022.
- [18] Z. Tu, S. Wu, G. Kang, & J. Lin, “Real-time defect detection of track components: Considering class imbalance and subtle difference between classes,” IEEE Trans. Instrum. Meas., vol. 70, pp. 1–12, 2021.
- [19] İ. Aydin, E. Akin, & M. Karakose, “Defect classification based on deep features for railway tracks in sustainable transportation,” Appl. Soft Comput., vol. 111, p. 107706, 2021.
- [20] A. K. Jain & F. Farrokhnia, “Unsupervised texture segmentation using Gabor filters,” Pattern Recognit., vol. 24, no. 12, pp. 1167–1186, 1991.
- [21] J. C. Lagarias, J. A. Reeds, M. H. Wright, & P. E. Wright, “Convergence properties of the Nelder-Mead simplex method in low dimensions,” SIAM J. Optim., vol. 9, no. 1, pp. 112–147, 1998.
- [22] H. Zhai & Z. Ma, “Detection algorithm of rail surface defects based on multifeature saliency fusion method,” Sensor Rev., ahead-of-print, 2022.
- [23] H. Yu et al., “A coarse-to-fine model for rail surface defect detection,” IEEE Trans. Instrum. Meas., vol. 68, no. 3, pp. 656–666, 2018.
- [24] V. Badrinarayanan, A. Handa, and R. Cipolla, “Segnet: A deep convolutional encoder-decoder architecture for robust semantic pixel-wise labelling,” arXiv preprint arXiv:1505.07293, 2015.
- [25] Z. Zhou, M. M. Rahman Siddiquee, N. Tajbakhsh, & J. Liang, “Unet++: A nested u-net architecture for medical image segmentation,” in Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support, pp. 3–11, Springer, Cham, 2018.
- [26] H. Zhao, J. Shi, X. Qi, X. Wang, & J. Jia, “Pyramid scene parsing network,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., pp. 2881–2890, 2017.
- [27] Y. Liu, H. Xiao, J. Xu, & J. Zhao, “A rail surface defect detection method based on pyramid feature and lightweight convolutional neural network,” IEEE Trans. Instrum. Meas., vol. 71, pp. 1–10, 2022.
- [28] X. Ni, H. Liu, Z. Ma, C. Wang, & J. Liu, “Detection for rail surface defects via partitioned edge feature,” IEEE Trans. Intell. Transp. Syst., 2021.
- [29] L. Yang, S. Xu, J. Fan, E. Li, & Y. Liu, “A pixel-level deep segmentation network for automatic defect detection,” Expert Syst. Appl., vol. 215, p. 119388, 2023.
- [30] S. Faghih-Roohi, S. Hajizadeh, A. Núñez, R. Babuska, & B. De Schutter, “Deep convolutional neural networks for detection of rail surface defects,” in 2016 Int. Joint Conf. Neural Netw. (IJCNN), pp. 2584–2589, IEEE, 2016.
A Stable Two-Stage Approach for Detection and Classification of Rail Surface Defects Using Fuzzy Modeling and CNN-BILSTM Network
Öz
Surface defects on railway tracks pose a significant challenge that adversely affects railway transportation. Timely detection of such defects enables the identification of potential hazards on the track and facilitates early maintenance, thereby preventing serious accidents. However, accurate detection of these defects is quite difficult due to surface impurities such as rust and oil. The primary objective in detecting surface defects is to achieve high accuracy to prevent confusion between actual defects and false alarms. In this study, a fast, two-stage method for detecting rail surface defects is proposed to address these challenges. First, rail extraction is performed by analyzing the differences in pixels containing rails in railway images. Then, after eliminating noise that may occur during the preprocessing step, the histogram of a sound rail image is modeled using a fuzzy membership function. The defective rail is segmented based on these fuzzy membership values to determine the presence of a defect. For the classification of defect types, a Convolutional Neural Network–Bidirectional Long Short-Term Memory (CNN-BILSTM) method is employed. The proposed surface defect detection method has demonstrated superior performance compared to existing methods on two benchmark datasets (Rail Surface Defect Datasets – RSDD-I/II). Furthermore, a classification accuracy of 98.00% was achieved in identifying defect types.
Anahtar Kelimeler
Railway Image Segmentation Fuzzy Measure Convolutional Neural Networks Surface Defects Defect Detection
Kaynakça
- [1] L. Zhuang, H. Qi, & Z. Zhang, “The automatic rail surface multi-flaw identification based on a deep learning powered framework,” IEEE Trans. Intell. Transp. Syst., 2021.
- [2] H. Yang, Y. Wang, J. Hu, J. He, Z. Yao, & Q. Bi, “Segmentation of track surface defects based on machine vision and neural networks,” IEEE Sensors J., vol. 22, no. 2, pp. 1571–1582, 2021.
- [3] M. Sevi, İ. Aydın, E. Akın, “Detection of rail surface defects based on ensemble learning of YOLOv5,” Demiryolu Mühendisliği, no. 17, pp. 115-132, Jan. 2023. doi: 10.47072/demiryolu.1205483.
- [4] M. Guerrieri, G. Parla, & C. Celauro, “Digital image analysis technique for measuring railway track defects and ballast gradation,” Measurement, vol. 113, pp. 137–147, 2018.
- [5] Y. Jiang et al., “Non-contact ultrasonic detection of rail surface defects in different depths,” in 2018 IEEE Far East NDT New Technology & Application Forum (FENDT), pp. 46–49, IEEE, 2018.
- [6] G. Piao, J. Li, L. Udpa, J. Qian, & Y. Deng, “Finite-element study of motion-induced eddy current array method for high-speed rail defects detection,” IEEE Trans. Magn., vol. 57, no. 12, pp. 1–10, 2021.
- [7] Y. Santur, M. Karaköse, & E. Akin, “A new rail inspection method based on deep learning using laser cameras,” in 2017 Int. Artificial Intelligence and Data Processing Symposium (IDAP), pp. 1–6, IEEE, 2017.
- [8] H. Zhong, L. Liu, J. Wang, Q. Fu, & B. Yi, “A real-time railway fastener inspection method using the lightweight depth estimation network,” Measurement, vol. 189, p. 110613, 2022.
- [9] H. Wang, M. Li, & Z. Wan, “Rail surface defect detection based on improved Mask R-CNN,” Comput. Electr. Eng., vol. 102, p. 108269, 2022.
- [10] J. Gan, J. Wang, H. Yu, Q. Li, & Z. Shi, “Online rail surface inspection utilizing spatial consistency and continuity,” IEEE Trans. Syst., Man, Cybern.: Syst., vol. 50, no. 7, pp. 2741–2751, 2018.
- [11] H. Zhang et al., “Automatic visual detection system of railway surface defects with curvature filter and improved Gaussian mixture model,” IEEE Trans. Instrum. Meas., vol. 67, no. 7, pp. 1593–1608, 2018.
- [12] D. Zhang et al., “Two deep learning networks for rail surface defect inspection of limited samples with line-level label,” IEEE Trans. Ind. Informat., vol. 17, no. 10, pp. 6731–6741, 2020.
- [13] Z. Zhang, M. Liang, and Z. Wang, “A deep extractor for visual rail surface inspection,” IEEE Access, vol. 9, pp. 21798–21809, 2021.
- [14] Y. Wu et al., “Hybrid deep learning architecture for rail surface segmentation and surface defect detection,” Comput.-Aided Civ. Infrastruct. Eng., vol. 37, no. 2, pp. 227–244, 2022.
- [15] H. Yang et al., “Deep learning and machine vision-based inspection of rail surface defects,” IEEE Trans. Instrum. Meas., vol. 71, pp. 1–14, 2021.
- [16] H. Li et al., “Ensemble model for rail surface defects detection,” PLoS One, vol. 17, no. 5, p. e0268518, 2022.
- [17] İ. Aydın & E. Akın, “Two-stage rail defect classification based on fuzzy measure and convolutional neural networks,” in Intelligent and Fuzzy Systems: Digital Acceleration and The New Normal-Proceedings of the INFUS 2022 Conf., vol. 1, pp. 769–776, Springer, Cham, 2022.
- [18] Z. Tu, S. Wu, G. Kang, & J. Lin, “Real-time defect detection of track components: Considering class imbalance and subtle difference between classes,” IEEE Trans. Instrum. Meas., vol. 70, pp. 1–12, 2021.
- [19] İ. Aydin, E. Akin, & M. Karakose, “Defect classification based on deep features for railway tracks in sustainable transportation,” Appl. Soft Comput., vol. 111, p. 107706, 2021.
- [20] A. K. Jain & F. Farrokhnia, “Unsupervised texture segmentation using Gabor filters,” Pattern Recognit., vol. 24, no. 12, pp. 1167–1186, 1991.
- [21] J. C. Lagarias, J. A. Reeds, M. H. Wright, & P. E. Wright, “Convergence properties of the Nelder-Mead simplex method in low dimensions,” SIAM J. Optim., vol. 9, no. 1, pp. 112–147, 1998.
- [22] H. Zhai & Z. Ma, “Detection algorithm of rail surface defects based on multifeature saliency fusion method,” Sensor Rev., ahead-of-print, 2022.
- [23] H. Yu et al., “A coarse-to-fine model for rail surface defect detection,” IEEE Trans. Instrum. Meas., vol. 68, no. 3, pp. 656–666, 2018.
- [24] V. Badrinarayanan, A. Handa, and R. Cipolla, “Segnet: A deep convolutional encoder-decoder architecture for robust semantic pixel-wise labelling,” arXiv preprint arXiv:1505.07293, 2015.
- [25] Z. Zhou, M. M. Rahman Siddiquee, N. Tajbakhsh, & J. Liang, “Unet++: A nested u-net architecture for medical image segmentation,” in Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support, pp. 3–11, Springer, Cham, 2018.
- [26] H. Zhao, J. Shi, X. Qi, X. Wang, & J. Jia, “Pyramid scene parsing network,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., pp. 2881–2890, 2017.
- [27] Y. Liu, H. Xiao, J. Xu, & J. Zhao, “A rail surface defect detection method based on pyramid feature and lightweight convolutional neural network,” IEEE Trans. Instrum. Meas., vol. 71, pp. 1–10, 2022.
- [28] X. Ni, H. Liu, Z. Ma, C. Wang, & J. Liu, “Detection for rail surface defects via partitioned edge feature,” IEEE Trans. Intell. Transp. Syst., 2021.
- [29] L. Yang, S. Xu, J. Fan, E. Li, & Y. Liu, “A pixel-level deep segmentation network for automatic defect detection,” Expert Syst. Appl., vol. 215, p. 119388, 2023.
- [30] S. Faghih-Roohi, S. Hajizadeh, A. Núñez, R. Babuska, & B. De Schutter, “Deep convolutional neural networks for detection of rail surface defects,” in 2016 Int. Joint Conf. Neural Netw. (IJCNN), pp. 2584–2589, IEEE, 2016.
Ayrıntılar
| Birincil Dil | Türkçe |
|---|---|
| Konular | Veri İletişimleri |
| Bölüm | Bilimsel Yayınlar (Hakemli Araştırma ve Derleme Makaleler) |
| Yazarlar | |
| Yayımlanma Tarihi | 31 Temmuz 2025 |
| Gönderilme Tarihi | 14 Haziran 2025 |
| Kabul Tarihi | 23 Temmuz 2025 |
| Yayımlandığı Sayı | Yıl 2025 Sayı: 22 |