基于点云配准与最近邻搜索的钢轨磨耗测量方法  

A measurement method for rail wear based on point cloudregistration and nearest neighbor search

作　　者：曾杉[1] 王文斌 尹太军 彭建川 刘艳彩 张杰 ZENG Shan;WANG Wenbin;YIN Taijun;PENG Jianchuan;LIU Yancai;ZHANG Jie(Institute of Geographic Sciences and Natural Resources Research,CAS,Beijing 100101,China;China Shenhua Energy Co.Ltd.,Beijing 100011,China;Digitwinology International Co.Ltd.,Qinhuangdao,Hebei 066000,China;Guoneng Baoshen Railway Group Co.Ltd.,Baotou,Inner Mongolia 014000,China)

机构地区：[1]中国科学院地理科学与资源研究所,北京100101 [2]中国神华能源股份有限公司,北京100011 [3]中科吉芯(秦皇岛)信息技术有限公司,河北秦皇岛066000 [4]国能包神铁路集团有限责任公司,内蒙古包头014000

出　　处：《燕山大学学报》2025年第1期55-65,共11页Journal of Yanshan University

基　　金：中国神华能源股份有限公司科技项目(SHGF-21-02);河北省科技计划项目(216Z1704G);中国科学院战略性先导科技专项(XDB0740200)。

摘　　要：提出了基于点云配准和最邻近搜索的方法,以解决钢轨轨腰处钢印噪声导致的轨顶磨耗测量点识别误差较大的问题,并成功实现了钢轨垂直和侧面磨耗点的自动定位。首先,通过坐标系旋转和点云滤波等预处理技术,以钢轨轮廓作为数据单元,获取有效的钢轨配准数据。接着,采用非线性拟合方法拟合轨腰圆弧的圆心,以此作为基准点进行任意状态下的点云初步粗配准。对于在轨腰处出现钢印编号的实际测量情况,采用了轨顶与轨腰点云的ICP加权精配准方案,实现测量轮廓与标准轮廓的精确重合。最后,根据钢轨磨耗计量办法,以标准钢轨轮廓指定位置坐标线为基准线,在配准后的点云数据中,通过最邻近搜索的方法寻找距离基准线最近的坐标,从而精确定位磨耗测量点的位置。实验结果表明,该方法能高效且精确地提取钢轨磨耗测量点。文章最后以三维图的方式展示磨耗测量点与标准轮廓的对比,其特征点提取的标准偏差小于0.1 mm,最大偏差小于0.3 mm。A method based on point cloud registration and nearest neighbor search is proposed to address the significant errors in identifying wear measurement points on the rail head caused by noise from stamped markings on the rail web.Automatic localization of vertical and lateral wear points on the rail is successfully achieved.Initially,preprocessing techniques such as coordinate system rotation and point cloud filtering are employed to use the rail profile as a data unit and obtain effective rail registration data.Subsequently,a nonlinear fitting method is used to fit the center of the arc of the rail web,serving as the reference point for preliminary coarse registration of the point cloud in any state.For the actual measurement scenario where stamped numbers appear on the rail web,an ICP(Iterative Closest Point)weighted fine registration scheme for the rail head and rail web point clouds is adopted to achieve precise alignment of the measured profile with the standard profile.Finally,according to the rail wear measurement method,the specified position coordinate line of the standard rail profile is used as the reference line.In the registered point cloud data,the nearest coordinate to the reference line is identified through the nearest neighbor search method,thereby accurately locating the wear measurement points.Experimental results demonstrate that this method can efficiently and accurately extract rail wear measurement points.A three-dimensional graph comparing the wear measurement points with the standard profile is shown as the article′s conclusion.The standard deviation of the extracted feature points is less than 0.1 mm,and the maximum deviation is less than 0.3 mm.

关 键 词：钢轨磨耗 点云预处理 加权点云配准 最近邻搜索 

分 类 号：U216.3[交通运输工程—道路与铁道工程]