Ultrasonic Detection and Classification for Internal Defect of Rail Based on Deep Learning

被引：0
|
作者：
Hu W. [1 ,2 ,3 ]
Qiu S. [1 ,2 ,3 ]
Xu X. [1 ,2 ,3 ]
Wei X. [1 ,2 ,3 ]
Wang W. [1 ,2 ,3 ]
机构：
[1] School of Civil Engineering, Central South University, Changsha
[2] Moe Key Laboratory of Engineering Structures of Heavy-Haul Railway, Changsha
[3] Intelligent Monitoring Research Center of Rail Transit Infrastructure, Central South University, Changsha
来源：
Tiedao Xuebao/Journal of the China Railway Society | 2021年 / 43卷 / 04期
关键词：
Deep learning; Image classification; Internal defect of rail; Ultrasonic detection;
D O I：
10.3969/j.issn.1001-8360.2021.04.014
中图分类号：
学科分类号：
摘要：
The real-time detection and treatment of internal defects of the rail will effectively reduce the risk of accidents. Compared with manual physical detection method that is time-consuming and laborious, ultrasonic detection technology can detect the internal state of the rail in real time, but the identification of defects still relies on manual or image processing technology to process one by one, which is likely to cause missed or wrong detection. This paper proposed a new deep learning-based ultrasonic testing data post-processing method for rail internal defect to realize the automation of defect identification and classification, which is verified and evaluated from three aspects: data, method and interference factors. The results show that Resnet-50 deep residual network has a classification accuracy of 99.3% for the five types of labels, with F1 scores of 99.24% (head kernel defect), 98.5% (web oblique crack), 99% (base crescent crack), 99.75% (joint defect) and 100% (non-defect), respectively. In addition, with good robustness to clutter interference, it is superior to three traditional machine learning methods, ensuring the real-time, accurate and efficient detection and treatment of internal damage to the rail. © 2021, Department of Journal of the China Railway Society. All right reserved.
引用
收藏
页码：108 / 116
页数：8
相关论文
共 21 条
  • [11] SUN Cisuo, LIU Jun, QIN Yong, Et al., Intelligent Detection Method for Rail Flaw Based on Deep Learning, China Railway Science, 39, 5, pp. 51-57, (2018)
  • [12] (2012)
  • [13] (2012)
  • [14] (2010)
  • [15] ABDEL-ZAHER A M, ELDEIB A M., Breast Cancer Classification Using Deep Belief Networks, Expert Systems with Applications, 46, pp. 139-144, (2016)
  • [16] TRAN V T, ALTHOBIANI F, BALL A., An Approach to Fault Diagnosis of Reciprocating Compressor Valves Using Teager-Kaiser Energy Operator and Deep Belief Networks, Expert Systems with Applications, 41, 9, pp. 4113-4122, (2014)
  • [17] HE K M, ZHANG X Y, REN S Q, Et al., Deep Residual Learning for Image Recognition, 2016 IEEE Conference on Computer Vision and Pattern Recognition, pp. 770-778, (2016)
  • [18] DUCHI J C, HAZAN E, SINGER Y., Adaptive Subgradient Methods for Online Learning and Stochastic Optimization, Journal of Machine Learning Research, 12, pp. 2121-2159, (2011)
  • [19] GAVILAN M, BALCONES D, MARCOS O, Et al., Adaptive Road Crack Detection System by Pavement Classification, Sensors, 11, 10, pp. 9628-9657, (2011)
  • [20] HOANG N D, NGUYEN Q L., Automatic Recognition of Asphalt Pavement Cracks Based on Image Processing and Machine Learning Approaches: a Comparative Study on Classifier Performance, Mathematical Problems in Engineering, 7, pp. 1-16, (2018)
123