浏览全部资源
扫码关注微信
基于轻量级卷积神经网络的地铁轨道线路状态检测系统
Track line status detection system for subway based on lightweight convolutional network
- 机车电传动 2022年第2期 页码：97-104
- 作者机构：
中车株洲电力机车研究所有限公司，湖南 株洲 412001
- 作者简介：
林 军（1986—），男，博士，高级工程师，主要从事图像信号处理、深度学习等研究； E-mail: linjun1@csrzic.com
- 基金信息：湖湘英才项目(2020RC3095);国家重大技术装备攻关工程项目（系列化中国标准地铁列车研制及试验）
DOI：10.13890/j.issn.1000-128X.2022.02.014
中图分类号： TN249;U231.4纸质出版日期：2022-03-10，
收稿日期：2022-02-18，
- 稿件说明：
扫 描 看 全 文
林军, 袁浩, 刘悦, 等. 基于轻量级卷积神经网络的地铁轨道线路状态检测系统[J]. 机车电传动, 2022,(2):97-104.
LIN Jun, YUAN Hao, LIU Yue, et al. Track line status detection system for subway based on lightweight convolutional network[J]. Electric drive for locomotives, 2022,(2):97-104.
林军, 袁浩, 刘悦, 等. 基于轻量级卷积神经网络的地铁轨道线路状态检测系统[J]. 机车电传动, 2022,(2):97-104. DOI： 10.13890/j.issn.1000-128X.2022.02.014.
LIN Jun, YUAN Hao, LIU Yue, et al. Track line status detection system for subway based on lightweight convolutional network[J]. Electric drive for locomotives, 2022,(2):97-104. DOI： 10.13890/j.issn.1000-128X.2022.02.014.
摘要
轨道作为承载车辆运行的重要部件，其工作状态对地铁运营的安全性有重要影响。传统人工巡检或者采用轨检车的检测模式只能在正线停运后进行作业，工作效率低。针对该问题，提出一种基于地铁运营列车的轨道线路状态检测系统，该系统采用高速线扫相机对轨道线路进行实时图像采集，并将采集数据送入轻量级RegNet骨干神经网络提取图像深层特征。在此基础上，加入双向特征金字塔网络进行多层次特征融合。最后将融合特征输入目标检测头实现轨道线路状态的实时检测。结合基于云边协同的困难样本挖掘以及模型部署加速技术，算法可实现高准确率、高实时性的检测性能。试验表明，该系统针对11类钢轨伤损及扣件状态的检测平均准确率（mAP）达到0.951，推理速度大于20 f/s
满足地铁在载客运营同时对轨道线路状态进行实时检测的需求。
Abstract
As an important component to carry running train
the working state of the track has an important impact on the safety of the subway operation. Traditional manual detection or rail inspection vehicles can not operate until subway operation ends
leading to low work efficiency. Aiming at this problem
an track line status detection system based on subway passenger trains was proposed. A high frequency line-scanning camera was used to collect real-time images and a lightweight backbone RegNet was used to extract the deep features of track line images. Then
a BiFPN was added to fuse multi-layer features. Finally
the features were send to a object detection head for track line status real-time detection. Combined with the hard sample mining based on cloud-edge collaboration and model acceleration technologies
the algorithm can meet the detection performance of high accuracy and real-time. The experiments show that the mean average precision of detection for 11 types of rail failure and fastener status reaches 0.951
and the inference speed is greater than 20 f/s
which proves that the system can realize real-time track line status detection during the subway passenger operation.
关键词
地铁车辆钢轨损伤扣件检测深度学习云边协同钢轨
Keywords
metro vehiclesrail failurefastener detectiondeep learningcloud-edge collaborationrail
references
TIAN G, GAO B, GAO Y, et al. Review of railway rail defect non-destructive testing and monitoring[J]. Chinese Journal of Scientific Instrument, 2016, 37(8): 1763-1780.
朱崇巧. 钢轨波浪磨耗检测算法研究[D]. 北京: 北京交通大学, 2018.
ZHU Chongqiao. Research on detection algorithm for rail corrugation[D]. Beijing: Beijing Jiaotong University, 2018.
魏秀琨, 所达, 魏德华, 等. 机器视觉在轨道交通系统状态检测中的应用综述[J]. 控制与决策, 2021, 36(2): 257-282.
WEI Xiukun, SUO Da, WEI Dehua, et al. A survey of the application of machine vision in rail transit system inspection[J]. Control and Decision, 2021, 36(2): 257-282.
DUBEY A K, JAFFERY Z A, et al. Maximally stable extremal region marking-based railway track surface defect sensing[J]. IEEE Sensors Journal, 2016, 16(24): 9047-9052.
YUAN Xiaocui, WU Lushen, CHEN Huawei. Rail image segmentation based on Otsu threshold method[J]. Optics and Precision Engineering, 2016, 24(7): 1772-1781.
LI Qingyong, REN Shengwei. A real-time visual inspection system for discrete surface defects of rail heads[J]. IEEE Transactions on Instrumentation and Measurement, 2012, 61(8): 2189-2199.
HE Zhendong, WANG Yaonan, MAO Jianxu, et al. Research on inverse P-M diffusion-based rail surface defect detection[J]. Acta Automatica Sinica, 2014, 40(8): 1667-1679.
MANDRIOTA C, NITTI M, ANCONA N, et al. Filter-based feature selection for rail defect detection[J]. Machine Vision and Applications, 2004, 15(4): 179-185.
RESENDIZ E, HART J M, AHUJA N. Automated visual inspection of railroad tracks[J]. IEEE Transactions on Intelligent Transportation Systems, 2013, 14(2): 751-760.
XIA Yiqi, XIE Fengying, JIANG Zhiguo. Broken railway fastener detection based on Adaboost algorithm[C]//IEEE. 2010 International Conference on Optoelectronics and Image Processing. Haikou, China: IEEE, 2010: 313-316.
FENG Hao, JIANG Zhiguo, XIE Fengying, et al. Automatic fastener classification and defect detection in vision-based railway inspection systems[J]. IEEE Transactions on Instrumentation and Measurement, 2014, 63(4): 877-888.
SOUKUP D, HUBER-MÖRK R. Convolutional neural networks for steel surface defect detection from photometric stereo images[M]//BEBIS G, BOYLE R, PARVIN B, et al. Advances in visual computing. Cham: Springer, 2014: 668-677.
FAGHIH-ROOHI S, HAJIZADEH S, NÚÑEZ A, et al. Deep convolutional neural networks for detection of rail surface defects[C]//IEEE. 2016 International Joint Conference on Neural Networks (IJCNN). Vancouver, BC: IEEE, 2016: 2584-2589.
WEI Xiukun, WEI Dehua, SUO Da, et al. Multi-target defect identification for railway track line based on image processing and improved YOLOv3 model[J]. IEEE Access, 2020, 8: 61973-61988.
WEI Xiukun, YANG Ziming, LIU Yuxin, et al. Railway track fastener defect detection based on image processing and deep learning techniques: a comparative study[J]. Engineering Applications of Artificial Intelligence, 2019, 80: 66-81.
FENG Jianghua, YUAN Hao, HU Yunqing, et al. Research on deep learning method for rail surface defect detection[J]. IET Electrical Systems in Transportation, 2020, 10(4): 436-442.
李忠海, 白秋阳, 王富明, 等. 基于语义分割的钢轨表面缺陷实时检测系统[J]. 计算机工程与应用, 2021, 57(12): 248-256.
LI Zhonghai, BAI Qiuyang, WANG Fuming, et al. Real-time detection system of rail surface defects based on semantic segmentation[J]. Computer Engineering and Applications, 2021, 57(12): 248-256.
GIBERT X, PATEL V M, CHELLAPPA R. Deep multitask learning for railway track inspection[J]. IEEE Transactions on Intelligent Transportation Systems, 2017, 18(1): 153-164.
TU Zhenwei, WU Songrong, KANG Gaoqiang, et al. Real-time defect detection of track components: considering class imbalance and subtle difference between classes[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70: 1-12.
林军, 潘文波, 游俊, 等. 基于多传感器融合的视觉感知困难样本挖掘方法[J]. 机车电传动, 2021(6): 93-99.
LIN Jun, PAN Wenbo, YOU Jun, et al. Hard example mining method for visual perception based on multi-sensor fusion[J]. Electric Drive for Locomotives, 2021(6): 93-99.
RADOSAVOVIC I, KOSARAJU R P, GIRSHICK R, et al. Designing network design spaces[C]//IEEE. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle, WA: IEEE, 2020: 10425-10433.
TAN Mingxing, PANG Ruoming, LE Q V. EfficientDet: scalable and efficient object detection[C]//IEEE. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle, WA: IEEE, 2020: 10778-10787.
LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]//IEEE. 2017 IEEE International Conference on Computer Vision (ICCV). Venice: IEEE, 2017: 2999-3007.
HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]//IEEE. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE, 2016: 770-778.
LIU Ze, LIN Yutong, CAO Yue, et al. Swin transformer: hierarchical vision transformer using shifted windows[C]//IEEE. 2021 IEEE/CVF International Conference on Computer Vision (ICCV). Montreal: IEEE, 2021.
LIU Shu, QI Lu, QIN Haifang, et al. Path aggregation network for instance segmentation[C]//IEEE. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018: 8759-8768.
REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.
REDMON J, FARHADI A. YOLOv3: an incremental improvement[EB/OL]. (2018-04-08) [2022-02-05]. https://arxiv.org/abs/1804.02767https://arxiv.org/abs/1804.02767.
JOCHER G, STOKEN A, BOROVEC J, et al. Ultralytics/yolov5: v4.0 - nn.silu() activations, weights & biases logging, pytorch hub integration[EB/OL]. (2021-01-05) [2022-02-05]. https://zenodo.org/record/4418161#.Yl-9o8jHH4Ehttps://zenodo.org/record/4418161#.Yl-9o8jHH4E.
0
浏览量
38
下载量
0
CSCD
3
CNKI被引量
- 网络PDF下载
- WORD下载
- Meta-XML下载
- BibTeX下载
- Endnote下载
- RefMan 下载
- NoteExpress下载
- NoteFirst下载
关联资源
相关文章
相关作者
相关机构
- 技术支持由北京北大方正电子有限公司提供 湘ICP备20010475号-1
京公网安备11010802024621 - 本系统建议在Chrome、 IE9+ 以上版本浏览器阅读本站内容，360浏览器请切换至极速模式
- Cookies帮助我们提供服务并提供个性化体验。使用本网站，即表示您同意我们使用Cookies