基于点云处理的机车车辆车底中心鞘螺栓故障检测方法  

Failure Detection Method for Center Sheath Bolts of Locomotive and Rolling Stock Based on Point Cloud Processing

作　　者：卢海林 冯其波[1,2] 徐昌源 LU Hailin;FENG Qibo;XU Changyuan(Key Laboratory of Luminescence and Optical Information,Ministry of Education,Beijing Jiaotong University,Beijing 100044,China;Dongguan Nannar Electronics Technology Co.,Ltd.,Dongguan 523050 Guangdong,China)

机构地区：[1]北京交通大学发光与光信息教育部重点实验室,北京100044 [2]东莞市诺丽电子科技有限公司,广东东莞523050

出　　处：《铁道机车车辆》2022年第5期67-73,共7页Railway Locomotive & Car

基　　金：东莞市引进创新科研团队项目(201536000600028)。

摘　　要：针对机车车辆车底中心鞘螺栓松动和缺失这一常见故障,提出了一种基于三维点云处理的检测算法。首先对目标点云进行下采样和离群点的移除,然后根据欧式距离对点云进行聚类分割,最后通过计算分割后每个点云包围盒形状系数和平均Z坐标实现螺栓点云定位,进而通过采样一致性算法拟合点云平面计算出螺栓松动程度,识别螺栓故障。将文中提出的算法与传统的基于2D图像的识别算法进行了对比,最后在地铁车辆智能巡检机器人中进行了应用,结果表明,该算法检测精度约为0.1 mm,且检测速度快,可满足实时检测的要求。Aiming at the common failure of loose and missing bolts in the center sheath of rolling stock,a detection algorithm based on 3D point cloud processing is proposed.First,the target point cloud is down-sampled and outliers are removed,then the point cloud is clustered and segmented according to euclidean distance,and finally the bolt point cloud positioning is achieved by calculating the bounding box shape coefficient and average Z coordinate of each point cloud after segmentation.Then use the sampling consistency algorithm to fit the point cloud plane to calculate the bolt looseness and identify the bolt failure.The algorithm proposed in this paper is compared with the traditional 2D image-based recognition algorithm.Finally,it is applied to the intelligent inspection robot of subway vehicles.The results show that the detection accuracy of the algorithm is about 0.1 mm,and the detection speed is fast which can meet requirements for real-time detection.

关 键 词：螺栓故障检测 支持向量机(SVM) 点云聚类 点云处理 

分 类 号：U269.3[机械工程—车辆工程]