基于点云驱动的道岔尖轨密贴检测方法

林海香; 陆天硕; 曲子贤; 马亚斌

doi:10.19713/j.cnki.43-1423/u.T20251495

您当前的位置：

首页 >

文章列表页 >

基于点云驱动的道岔尖轨密贴检测方法

工程诊所•工匠之家 | 更新时间：2026-07-24

- 基于点云驱动的道岔尖轨密贴检测方法
- Point cloud-driven detection model for close contact of switch point rails
- 铁道科学与工程学报 2026年23卷第7期页码：3512-3524
- 作者机构：
  
  1.兰州交通大学自动化与电气工程学院，甘肃兰州 730070
  2.中国铁路兰州局集团有限公司科技和信息化部，甘肃兰州 730000
- 作者简介：
  
  林海香(1977—)，女，甘肃天水人，副教授，博士，从事交通信息数据挖掘研究；E-mail：linhaixiang@mail.lzjtu.cn
- 基金信息：
  
  国家自然科学基金资助项目(52562055);甘肃省重点研发计划(23YFGA0046;22YF7GA141)
- DOI：10.19713/j.cnki.43-1423/u.T20251495
  中图分类号： U216.3
- 收稿：2025-09-23，
  
  网络首发：2026-07-24，
  
  纸质出版：2026-07-28
- 稿件说明：
移动端阅览
林海香,陆天硕,曲子贤等.基于点云驱动的道岔尖轨密贴检测方法[J].铁道科学与工程学报,2026,23(07):3512-3524.

LIN Haixiang,LU Tianshuo,QU Zixian,et al.Point cloud-driven detection model for close contact of switch point rails[J].Journal of Railway Science and Engineering,2026,23(07):3512-3524.
林海香,陆天硕,曲子贤等.基于点云驱动的道岔尖轨密贴检测方法[J].铁道科学与工程学报,2026,23(07):3512-3524. DOI： 10.19713/j.cnki.43-1423/u.T20251495.

LIN Haixiang,LU Tianshuo,QU Zixian,et al.Point cloud-driven detection model for close contact of switch point rails[J].Journal of Railway Science and Engineering,2026,23(07):3512-3524. DOI： 10.19713/j.cnki.43-1423/u.T20251495.

摘要

铁路道岔尖轨密贴检测是保障铁路运营安全的关键技术环节。针对传统人工检测手段存在巡检盲区、作业风险高等问题，提出基于点云驱动的道岔尖轨密贴非接触检测PMG方法(preprocessing-midline extraction-gap calculation

PMG)。该方法使用红外VCSEL随机时空数字散斑扫描仪获取道岔尖轨表面三维点云数据，通过体素均匀化处理原始数据，消除噪声干扰并保持几何特征完整性。构建融合密度聚类与几何分箱的中线精细化提取算法，利用DBSCAN(density-based spatial clustering of applications with noise)密度聚类算法对二维投影点云进行分割，识别上下轨道簇并处理复杂点云数据，采用

轴等宽分箱结合局部百分位统计方法，自适应地捕获轨道边缘特征，生成离散中点序列。通过多项式拟合对中点序列进行平滑处理，消除离散点噪声干扰，实现中线几何特征的连续化重构，准确表征尖轨缝隙的真实几何趋势。使用KD树(K-Dimensional Tree)高效邻域搜索技术，量化计算中线点与轨道点的密度分布特征，结合指数衰减理论构建密贴值计算公式，利用归一化密度与原始密度的双重影响机制，平衡局部密度变化与全局分布特征，实现尖轨密贴的自动化检测。实验结果表明，该方法能够正确分割道岔尖轨点云，实现密贴状态的判别，重复性精度收敛于0.06 mm，平均测量偏差达到0.038 mm，在关键监测区间内具有良好的线性响应和鲁棒性，响应时间为5.46 s，检测实时性优于人工巡检。该方法具有高精度、非接触、高鲁棒性等检测优势，实现道岔尖轨准确检测的同时保证了作业安全性，为铁路智能维护提供了有效的技术支撑。

Abstract

The inspection of rail point contact density in railway turnouts is a critical technical measure for ensuring railway operational safety. To address issues such as inspection blind spots and high operational risks associated with traditional manual detection methods

this study proposed the Preprocessing-Midline extraction-Gap calculation (PMG) method for turnout switch point non-contact detection driven by point clouds. This method employed an infrared VCSEL random speckle digital scanner to acquire three-dimensional point cloud data of the switch point rail surface. Raw data could undergo voxelization uniform to eliminate noise interference while preserving geometric feature integrity. A refined centerline extraction algorithm integrating density clustering and geometric binning was developed. The DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm could perform robust segmentation on the 2D projected point cloud

effectively identifying upper and lower rail clusters while handling complex data. X-axis equal-width binning combined with local percentile statistics adaptively and robustly captures rail edge features

generating a discrete centerline sequence. Polynomial fitting smooths the center point sequence

effectively eliminating discrete point noise interference. This achieved 3continuous reconstruction of the centerline’s geometric features

accurately representing the true geometric trend of the switch gap. Employing KD-tree (K-Dimensional Tree) efficient neighborhood search technology

the method quantitatively calculated density distribution characteristics between centerline points and track points. Integrating exponential decay theory to construct a close-fitting value calculation formula

it utilized a dual-influence mechanism of normalized density and original density to effectively balance local density variations and global distribution features

enabling automated detection of switch blade close-fitting. Experimental results demonstrate that this method can accurately segment switch point cloud data

enabling the identification of close-fitting conditions. Repeatability accuracy converges to 0.06 mm

with an average measurement deviation of 0.038 mm. It can exhibit excellent linear response and robustness within critical monitoring sections

achieving a response time of 5.46 seconds. The detection real-time capability surpasses that of manual inspections. This method can offer detection advantages such as high precision

non-contact operation

and high robustness. It enables remote switch blade inspection while ensuring operational safety

providing effective technical support for intelligent railway maintenance.

关键词

Keywords

references

杨东升 , 司道林 , 王璞 . 高速铁路道岔的大修周期 [J ] . 铁道建筑 , 2024 , 64 ( 5 ): 1 - 6 .

YANG Dongsheng , SI Daolin , WANG Pu . Overhaul cycle of high-speed railway switches [J ] . Railway Engineering , 2024 , 64 ( 5 ): 1 - 6 .

曲玉福 . 高速铁路道岔扣件系统病害及整治 [J ] . 科技创新与应用 , 2022 , 12 ( 21 ): 127 - 131 .

QU Yufu . Diseases and treatment of turnout fastener system in high-speed railway [J ] . Technology Innovation and Application , 2022 , 12 ( 21 ): 127 - 131 .

杨飞 , 涂文靖 , 魏子龙 , 等 . 铁路工务、电务、供电检测装备发展现状综述 [J ] . 交通运输工程学报 , 2023 , 23 ( 1 ): 47 - 69 .

YANG Fei , TU Wenjing , WEI Zilong , et al . Review on development status of inspection equipment for track maintenance, communication and signaling, and power supply of railway [J ] . Journal of Traffic and Transportation Engineering , 2023 , 23 ( 1 ): 47 - 69 .

左勇 , 任阳 , 杜志华 , 等 . 基于LP-RANSAC算法的路面点云快速移除算法 [J ] . 激光与光电子学进展 , 2023 , 60 ( 14 ): 1428009 .

ZUO Yong , REN Yang , DU Zhihua , et al . Rapid removal algorithm of road surface point cloud based on LP-RANSAC algorithm [J ] . Laser & Optoelectronics Progress , 2023 , 60 ( 14 ): 1428009 .

王东妍 , 于才 , 沈鹍 , 等 . 点云数据在道岔关键节点几何检测应用 [J ] . 测绘通报 , 2024 ( 3 ): 107 - 112 .

WANG Dongyan , YU Cai , SHEN Kun , et al . Application of point cloud data in geometric detection of critical node of turnouts [J ] . Bulletin of Surveying and Mapping , 2024 ( 3 ): 107 - 112 .

SAHEBDIVANI S , AREFI H , MABOUDI M . Rail track detection and projection-based 3D modeling from UAV point cloud [J ] . Sensors , 2020 , 20 ( 18 ): 5220 .

CHEN Xiao , QIN Fei , XIA Chenxi , et al . An innovative detection method of high-speed railway track slab supporting block plane based on point cloud data from 3D scanning technology [J ] . Applied Sciences , 2019 , 9 ( 16 ): 3345 .

GAO Xiang , YANG Ronghao , TAN Junxiang , et al . Floor plan reconstruction from indoor 3D point clouds using iterative RANSAC line segmentation [J ] . Journal of Building Engineering , 2024 , 89 : 109238 .

LI Zhixin , SHAN Jie . RANSAC-based multi primitive building reconstruction from 3D point clouds [J ] . ISPRS Journal of Photogrammetry and Remote Sensing , 2022 , 185 : 247 - 260 .

MIAO Yanlong , LI Shuai , WANG Liuyang , et al . A single plant segmentation method of maize point cloud based on Euclidean clustering and K -means clustering [J ] . Computers and Electronics in Agriculture , 2023 , 210 : 107951 .

ZHOU Jiawen , CHEN Junlin , LI Haibo . An optimized fuzzy K -means clustering method for automated rock disconti nuities extraction from point clouds [J ] . International Journal of Rock Mechanics and Mining Sciences , 2024 , 173 : 105627 .

LI Bin , ZHANG Yonghan , SUN Fuqiang . Deep residual neural network based PointNet for 3D object part segmentation [J ] . Multimedia Tools and Applications , 2022 , 81 ( 9 ): 11933 - 11947 .

CHEN Hui , LIANG Man , LIU Wanquan , et al . An approach to boundary detection for 3D point clouds based on DBSCAN clustering [J ] . Pattern Recognition , 2022 , 124 : 108431 .

XIE Tingting , CHEN Hui , LIU Wanquan , et al . 3D surface segmentation from point clouds via quadric fits based on DBSCAN clustering [J ] . Pattern Recognition , 2024 , 154 : 110589 .

LIU Haibin , TANG Yanglei , WANG Huanjie . Robust and fast point cloud registration for robot localization based on DBSCAN clustering and adaptive segmentation [J ] . Sensors , 2024 , 24 ( 24 ): 7889 .

LEDENTSOV N N , MAKAROV O Y , SHCHUKIN V A , et al . High speed VCSEL technology and applications [J ] . Journal of Lightwave Technology , 2022 , 40 ( 6 ): 1749 - 1763 .

YIN Wei , ZHAO Hang , JI Yifan , et al . High-resolution, wide-field-of-view, and real-time 3D imaging based on spatial-temporal speckle projection profilometry with a VCSEL projector array [J ] . ACS Photonics , 2024 , 11 ( 2 ): 498 - 511 .

涂静敏 , 沈阳 , 李婕 , 等 . 体素与点混合增长的机载点云屋顶平面分割 [J ] . 中国激光 , 2024 , 51 ( 22 ): 205 - 216 .

TU Jingmin , SHEN Yang , LI Jie , et al . Airborne lidar point cloud roof plane segmentation method based on voxel and point hybrid growth [J ] . Chinese Journal of Lasers , 2024 , 51 ( 22 ): 205 - 216 .

ZHAO Baiting , DONG Xiao , GUO Yongcun , et al . PCA dimensionality reduction method for image classification [J ] . Neural Processing Letters , 2022 , 54 ( 1 ): 347 - 368 .

张杰 , 曲洪权 , 柳长安 , 等 . 基于双子空间PCA降维的脑力负荷分类 [J ] . 科学技术与工程 , 2024 , 24 ( 11 ): 4433 - 4438 .

ZHANG Jie , QU Hongquan , LIU Chang’an , et al . Classification of mental workload based on dimension reduction of PCA in two subspaces [J ] . Science Technology and Engineering , 2024 , 24 ( 11 ): 4433 - 4438 .

杨秀建 , 黄甲龙 , 张生斌 , 等 . 基于激光雷达的无人车辆非结构化环境联级点云分割算法 [J ] . 光学学报 , 2025 , 45 ( 8 ): 255 - 268 .

YANG Xiujian , HUANG Jialong , ZHANG Shengbin , et al . Cascaded point cloud segmentation algorithm for unstructured environments based on LiDAR for autonomous vehicles [J ] . Acta Optica Sinica , 2025 , 45 ( 8 ): 255 - 268 .

ZHANG Hua , DUAN Zhenwei , ZHENG Nanshan , et al . An efficient class-constrained DBSCAN approach for large-scale point cloud clustering [J ] . IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 2022 , 15 : 7323 - 7332 .

韩玉川 , 侯贺 , 白云瑞 , 等 . 一种基于边缘系数的闭合点云边缘提取算法 [J ] . 激光与光电子学进展 , 2018 , 55 ( 11 ): 111003 .

HAN Yuchuan , HOU He , BAI Yunrui , et al . A closed point cloud edge extraction algorithm using edge coefficient [J ] . Laser & Optoelectronics Progress , 2018 , 55 ( 11 ): 111003 .

潘兵 , 谢惠民 . 数字图像相关中基于位移场局部最小二乘拟合的全场应变测量 [J ] . 光学学报 , 2007 , 27 ( 11 ): 1980 - 1986 .

PAN Bing , XIE Huimin . Full-field strain measurement based on least-square fitting of local displacement for digital image correlation method [J ] . Acta Optica Sinica , 2007 , 27 ( 11 ): 1980 - 1986 .

GUO Ziye , LIU Hui , SHI Hongyin , et al . KD-tree-based euclidean clustering for tomographic SAR point cloud extraction and segmentation [J ] . IEEE Geoscience and Remote Sensing Letters , 2023 , 20 : 4000205 .

凌晴 , 任峻广 , 孔令杰 , 等 . 基于滞后衰减效应与主控因子筛选策略的库岸滑坡位移预测模型 [J/OL ] . 武汉大学学报(信息科学版) , 1 - 29 [ 2025-12-26 ] . https://doi.org/10.13203/j.whugis20250066 https://doi.org/10.13203/j.whugis20250066 .

LING Qing , REN Junguang , KONG Lingjie , et al . Based on lagged attenuation effect and key factor selection strategy for reservoir bank slope displacement prediction model [J ] . Geomatics and Information Science of Wuhan University , 1 - 29 [ 2025-12-26 ] . https://doi.org/10.13203/j.whugis20250066 https://doi.org/10.13203/j.whugis20250066 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

铁路无人机自主智能巡检平台技术框架与实践

基于能量等效方法的无黏性土细粒流失强度损伤模型

基于交变电场的寒区隧道排水管防结晶试验研究

桥建分离高铁车站噪声声场特性实测与分析

中性盐雾条件下Q370qENH耐候钢腐蚀特征及力学性能研究