Defect detection method of high-speed rail catenary based on few-shot learning

Siliang Guo; Chaochao Zheng; Xian Wu; Zhongli Wang

doi:10.1117/12.2646931

7 September 2022 Defect detection method of high-speed rail catenary based on few-shot learning

Siliang Guo, Chaochao Zheng, Xian Wu, Zhongli Wang

Proceedings Volume 12329, Third International Conference on Artificial Intelligence and Electromechanical Automation (AIEA 2022); 123291F (2022) https://doi.org/10.1117/12.2646931
Event: Third International Conference on Artificial Intelligence and Electromechanical Automation (AIEA 2022), 2022, Changsha, China

Abstract

Catenary is the core component of high-speed railway traction power supply system, its operation and maintenance safety are directly related to the safe operation of high-speed railway. In order to solve the problem of various parts and defects but rare actual defect samples on catenary support device, a few-shot defect detection method based on manifold smoothing and graph neural network is proposed. Firstly, the catenary parts are extracted from coarse to fine in two stages through the Yolo v5 target detection network. Then, the wide residual network is used to capture the feature representation of the input image, and the graph neural network is used to construct the undirected graph to transmit the label information from the labeled support samples to the unlabeled query samples. Thus, the prediction of the input image is obtained. Moreover, embedding propagation is used to smooth the low dimensional manifold constructed by graph neural network. It updates the nodes in the graph according to the surrounding node states, so as to alleviate the distribution deviation generated in the process of few-shot learning, improve the classification accuracy of defective samples and enhance the generalization performance of the model. Through the defect detection of three parts of the catenary, the average comprehensive defect detection accuracy is 93.9%. The results show that in the case of a variety of catenary parts and a small number of samples, the proposed few-shot defect detection method can accurately locate the parts in the catenary and effectively detect the defect samples.

Citation Download Citation

Siliang Guo, Chaochao Zheng, Xian Wu, and Zhongli Wang "Defect detection method of high-speed rail catenary based on few-shot learning", Proc. SPIE 12329, Third International Conference on Artificial Intelligence and Electromechanical Automation (AIEA 2022), 123291F (7 September 2022); https://doi.org/10.1117/12.2646931

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
9 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Defect detection

Dielectrics

Neural networks

Image processing

Data modeling

Target detection

Statistical modeling

Show All Keywords

Keywords/Phrases

Search In:

Publication Years