| 25 | 0 | 39 |
| 下载次数 | 被引频次 | 阅读次数 |
针对轨道交通轨旁设备隐蔽螺栓松动、脱落及侵限隐患的检测难题，提出了一种融合图像识别与激光扫描技术的智能巡检系统。该系统以双轨式自走行机器人为载体，集成高清相机与激光雷达，能够在轨道交通隧道桥梁实现自主巡航，通过改进的YOLO-Y算法实现小目标螺栓缺陷的精准识别，并创新性引入动态非单调聚焦机制(Wise-IoU)与Slide Loss分段加权策略，显著提升了低质量图像下的检测鲁棒性，实现了基于图像的轨旁设备固定螺栓松动、脱落缺陷的精确检测；同时设计了轨旁设备侵限检测方案，通过将激光点云三维建模与ICP拼接算法结合，构建了轨旁设备毫米级精度空间模型，提出了基于位移偏差与向量拓扑关系的侵限预警方法。现场测试表明，该系统在杭海城际铁路中缺陷识别准确率达98.08%,可有效实现轨旁设备状态的“事前控制”,为轨道交通智慧运维提供了创新解决方案。
Abstract：Aiming at the detection challenges of hidden bolt loosening, falling-off, and clearance intrusion risks in rail transit trackside equipment, this paper proposes an intelligent inspection system integrating image recognition and laser scanning technologies. Based on a dual-rail self-propelled robot, the system integrates high-definition cameras and lidar, enabling autonomous cruising in rail transit tunnels and bridges. It achieves accurate identification of small-target bolt defects through the improved YOLO-Y algorithm, and innovatively introduces a dynamic non-monotonic focusing mechanism(Wise-IoU) and a Slide Loss segmented weighting strategy, which significantly enhances detection robustness under low-quality images and realizes precise image-based detection of loosening and falling-off defects in trackside equipment fixing bolts. Meanwhile, the system incorporates a trackside equipment clearance intrusion detection scheme and related devices. By combining 3D modeling of laser point clouds with the ICP registration algorithm, it constructs a millimeter-level precision spatial model of trackside equipment and proposes a clearance intrusion warning method based on displacement deviation and vector topological relations. Field tests show that the system achieves a defect recognition accuracy of 98.08% in the Hangzhou-Haining Intercity Railway, effectively enabling “pre-event control” of trackside equipment status and providing an innovative solution for smart operation and maintenance of rail transit.
[1] 曾笑雨，刘苏，张奇.基于事故统计分析的城市轨道交通运营安全和可靠性研究[J].安全与环境工程，2012,19(1):90-94.ZENG Xiaoyu,LIU Su,ZHANG Qi.Research on the safety and reliability of urban mass transit operation based on the statistical analysis of accidents [J].Safety and Environmental Engineering,2012,19 (1):90-94.
[2] 蔡静娴，毕江海，王继军，等.基于改进Faster R-CNN的轨旁设备识别算法研究[J].铁道科学与工程学报，2022,19(10):3107-3116.CAI Jingxian,BI Jianghai,WANG Jijun,et al.Research on trackside equipment recognition algorithm based on improved Faster R-CNN [J].Journal of Railway Science and Engineering,2022,19 (10):3107-3116.
[3] 杜馨瑜.电务轨旁设备外观巡检图像增强算法[J].中国铁道科学，2016,37(6):97-105.DU Xinyu.Algorithm for appearance inspection image enhancement of trackside communication and signal equipment [J].China Railway Science,2016,37 (6):97-105.
[4] 黄丰.杭州至海宁城际铁路轨道医院智能运维平台的探索与实践[J].城市轨道交通研究，2024,27(1):234-238.HUANG Feng.Exploration and practice of intelligent operation-maintenance platform for Hang-Hai intercity railway track hospital [J].Urban Mass Transit,2024,27 (1):234-238.
[5] 李梓豪，唐超.基于高清工业相机的盾构隧道裂缝智能识别算法分析[J].测绘通报，2021(8):83-87.LI Zihao,TANG Chao.Analysis of intelligent identification algorithm for shield tunnel cracks based on high-definition industrial camera [J].Surveying and Mapping Bulletin,2021(8):83-87.
[6] WANG L W,LIU Z S,SIU W C,et al.Deep lightening network for low-light image enhancement[C]//2020 IEEE International Symposium on Circuits and Systems (ISCAS).IEEE,2020:1-5.
[7] LYU F,LU F,WU J,et al.MBLLEN:Low-light image/video enhancement using CNNs[C]//BMVC.2018:220.
[8] REDMON J,DIVVALA S,GIRSHICK R,et al.You Only Look Once:Unified,real-time object detection[C]//Computer Vision & Pattern Recognition.IEEE,2016:779-788.
[9] 朱久牛，李立明，柴晓冬，等.基于改进Yolov5算法的地铁车辆轴箱盖螺栓定位方法研究[J].智能计算机与应用，2023,13(10):29-34.ZHU Jiuniu,LI Liming,CHAI Xiaodong,et al.Research on bolt positioning method for axle box cover of metro vehicles based on improved Yolov5 algorithm [J].Intelligent Computer and Application,2023,13 (10):29-34.
[10] TONG Z,CHEN Y,XU Z,et al.Wise-IoU:Bounding box regression loss with dynamic focusing mechanism[J].arXiv preprint,2023 arXiv:2301.10051.
[11] YU Z,HUANG H,CHEN W,et al.YOLO-Facev2:A scale and occlusion aware face detector[J].Pattern Recognition,2024,155:110714.
[12] 杨敏，燕樟林，高利业，等.基于点云数据的地铁隧道限界智能检测[J].建筑技术，2022,53(11):1571-1575.YANG Min,YAN Zhanglin,GAO Liye,et al.Intelligent detection of subway tunnel gauge based on point cloud data [J].Building Technology,2022,53 (11):1571-1575.
[13]安哲立,马伟斌,袁振宇,等.基于激光扫描技术的铁路隧道限界检测与分析[J].地下空间与工程学报,2022, 18(增刊2):874-883.AN Zheli, MA Weibin, YUAN Zhenyu, et al. Railway tunnel gauge detection and analysis based on laser scanning technology[J]. Journal of Underground Space and Engineering, 2022,18(Sup2):874-883.
[14]张亚勇.基于移动激光扫描三维点云的限界检测技术研究[J].铁道标准设计, 2021, 65(5):44-48.ZHANG Yayong. Research on gauge detection technology based on mobile laser scanning 3D point cloud[J].Railway Standard Design, 2021, 65(5):44-48.
[15] RUSINKIEWICZ S, LEVOY M. Efficient variants of the ICP algorithm[C]//Proceedings Third International Conference on 3-D Digital Imaging and Modeling. IEEE,2001:145-152.
[16]安哲立,周鸣亮,马伟斌,等.基于移动激光扫描技术的隧道限界检测[J].工程勘察, 2023, 51(11):42-46.AN Zheli, ZHOU Mingliang, MA Weibin, et al. Tunnel clearance detection based on mobile laser scanning technology[J]. Engineering Survey, 2023, 51(11):42-46.
[17]刘德儿,王永君,闾国年.基于向量代数的点与多边形拓扑关系的推理[J].大地测量与地球动力学, 2011, 31(2):89-93.LIU Deer, WANG Yongjun, LYU Guonian. Reasoning of topological relationship between points and polygons based on vector algebra[J]. Geodesy and Geodynamics,2011, 31(2):89-93.
[18]赵正阳,张梓鸿,黄慧昌,等.基于激光点云数据的城市轨道交通限界侵限判别算法[J].城市轨道交通研究,2023, 26(10):36-42.ZHAO Zhengyang, ZHANG Zihong, HUANG Huichang,et al. Identification algorithm for urban rail transit gauge violation based on laser point cloud data[J]. Urban Mass Transit, 2023, 26(10):36-42.
基本信息:
DOI：10.20213/j.cnki.tdcl.2026.01.003
中图分类号:U239.5
引用信息:
[1]王儒,申路.轨道交通轨旁设备螺栓检测及限界检测技术研究[J].铁道车辆,2026,64(01):21-27+137.DOI:10.20213/j.cnki.tdcl.2026.01.003.
基金信息:
浙江省交通运输厅2024年科技计划项目(2024039)
2026-02-20
2026-02-20