Looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networks
To deal with the looseness of fasteners on the conductor rails of the medium and low speed maglev railways, a detection algorithm based on deep convolutional networks was proposed for loose detection of two kinds of fasteners, i.e. the base mounting bolt and the connecting plate screw. The first ste...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Department of Electric Drive for Locomotives
2022-07-01
|
| Series: | 机车电传动 |
| Subjects: | |
| Online Access: | http://edl.csrzic.com/thesisDetails#10.13890/j.issn.1000-128X.2022.04.025 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850186491089649664 |
|---|---|
| author | LI Cheng CHEN Jianxiong LIN Jun KANG Gaoqiang |
| author_facet | LI Cheng CHEN Jianxiong LIN Jun KANG Gaoqiang |
| author_sort | LI Cheng |
| collection | DOAJ |
| description | To deal with the looseness of fasteners on the conductor rails of the medium and low speed maglev railways, a detection algorithm based on deep convolutional networks was proposed for loose detection of two kinds of fasteners, i.e. the base mounting bolt and the connecting plate screw. The first step was to locate the fastener area on the conductor rail for the sake of eliminating background interference; the second was to analyze any change of the fastener position for looseness detection. This algorithm was executed and verified through deep convolutional networks. Firstly, the areas of these two kinds of fasteners were located through YOLO V2 network; secondly, segmentation was conducted at the edge of the connecting plate, the insulator, the bolt and the screw, and the head of the connecting plate screw through the Mask R-CNN network simultaneously; finally, looseness of fasteners was detected by judging whether the position of any segmented part was changed. The proposed defect detection algorithm was tested using the data of the medium and low speed maglev system in Changsha, which revealed a looseness detection accuracy of over 90% for both the base mounting bolts and the connecting plate screws, and a recall rate of over 94%. The test results show that the method presented in this paper can accurately identify the loosened fasteners on the conductor rails of the medium and low speed maglev railways. |
| format | Article |
| id | doaj-art-81b83380a6d448d2a3496c834e3e12e5 |
| institution | OA Journals |
| issn | 1000-128X |
| language | zho |
| publishDate | 2022-07-01 |
| publisher | Editorial Department of Electric Drive for Locomotives |
| record_format | Article |
| series | 机车电传动 |
| spelling | doaj-art-81b83380a6d448d2a3496c834e3e12e52025-08-20T02:16:19ZzhoEditorial Department of Electric Drive for Locomotives机车电传动1000-128X2022-07-0117217930835406Looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networksLI ChengCHEN JianxiongLIN JunKANG GaoqiangTo deal with the looseness of fasteners on the conductor rails of the medium and low speed maglev railways, a detection algorithm based on deep convolutional networks was proposed for loose detection of two kinds of fasteners, i.e. the base mounting bolt and the connecting plate screw. The first step was to locate the fastener area on the conductor rail for the sake of eliminating background interference; the second was to analyze any change of the fastener position for looseness detection. This algorithm was executed and verified through deep convolutional networks. Firstly, the areas of these two kinds of fasteners were located through YOLO V2 network; secondly, segmentation was conducted at the edge of the connecting plate, the insulator, the bolt and the screw, and the head of the connecting plate screw through the Mask R-CNN network simultaneously; finally, looseness of fasteners was detected by judging whether the position of any segmented part was changed. The proposed defect detection algorithm was tested using the data of the medium and low speed maglev system in Changsha, which revealed a looseness detection accuracy of over 90% for both the base mounting bolts and the connecting plate screws, and a recall rate of over 94%. The test results show that the method presented in this paper can accurately identify the loosened fasteners on the conductor rails of the medium and low speed maglev railways.http://edl.csrzic.com/thesisDetails#10.13890/j.issn.1000-128X.2022.04.025conductor rail of medium and low speed maglev railwaysfastener loosenessYOLO V2 networkMask R-CNN networkRadon transform |
| spellingShingle | LI Cheng CHEN Jianxiong LIN Jun KANG Gaoqiang Looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networks 机车电传动 conductor rail of medium and low speed maglev railways fastener looseness YOLO V2 network Mask R-CNN network Radon transform |
| title | Looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networks |
| title_full | Looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networks |
| title_fullStr | Looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networks |
| title_full_unstemmed | Looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networks |
| title_short | Looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networks |
| title_sort | looseness detection of fasteners on conductor rails of medium and low speed maglev railways based on deep convolutional networks |
| topic | conductor rail of medium and low speed maglev railways fastener looseness YOLO V2 network Mask R-CNN network Radon transform |
| url | http://edl.csrzic.com/thesisDetails#10.13890/j.issn.1000-128X.2022.04.025 |
| work_keys_str_mv | AT licheng loosenessdetectionoffastenersonconductorrailsofmediumandlowspeedmaglevrailwaysbasedondeepconvolutionalnetworks AT chenjianxiong loosenessdetectionoffastenersonconductorrailsofmediumandlowspeedmaglevrailwaysbasedondeepconvolutionalnetworks AT linjun loosenessdetectionoffastenersonconductorrailsofmediumandlowspeedmaglevrailwaysbasedondeepconvolutionalnetworks AT kanggaoqiang loosenessdetectionoffastenersonconductorrailsofmediumandlowspeedmaglevrailwaysbasedondeepconvolutionalnetworks |