基于三维点云的结构件搭接焊缝识别方法研究.docxVIP

机械工程师MECHANICAL

机械工程师

MECHANICALENGINEER

机械工程师ME

机械工程师

MECHANICALENGINEER

PAGE35网址：

PAGE35

网址：电邮：hrbengineer@163.com圆园25年第5期

PAGE36圆园

PAGE36

圆园25年第5期网址：电邮：hrbengineer@163.com

基于三维点云的结构件搭接焊缝识别方法研究

晏祖根，曾宏涛，付斌，齐明，闫明，翟一帆

（哈尔滨商业大学轻工学院，哈尔滨150028）

摘要：为提高大型结构件焊接过程的自动化程度，建立自动化焊接机器人系统，设计一种基于三维点云技术的焊缝识别提取方法。针对结构光传感器获取试验件点云数据，采用统计滤波和半径滤波法进行点云预处理，利用八叉树的点云下采样方法，实现对点云数据的简化；通过RANSAC算法拟合平面，求取拟合平面交线，确定点云数据中符合距离阈值的点，最后通过整体最小二乘法拟合焊缝直线。试验表明，相比于点云曲率的采样方法，八叉树的点云下采样方法在效率上提高了

65%，整体最小二乘法拟合焊缝直线具有计算简单、偏差小的优点，精度满足大型结构件焊接工艺要求。

关键词：结构件焊接；三维点云；焊缝识别提取；平面拟合

中图分类号:TP278 文献标志码:粤 文章编号：员园园圆原圆猿猿猿（圆园25）05原园035原园5

ResearchonLapWeldIdentificationMethodBasedonThree-dimensionalPointCloud

YANZugen,ZENGHongtao,FUBin,QIMing,YANMing,ZHAIYifan

(SchoolofLightIndustry,HarbinUniversityofCommerce,Harbin150028,China)

Abstract:Toenhancetheautomationofweldingprocessesforlargestructuralcomponents,anautomatedweldingrobotsystemisdeveloped,alongwithaweldidentificationandextractionmethodbasedonthree-dimensionalpointcloudtechnology.Thepointclouddataoftestpiecesisobtainedusingastructuredlightsensor,andpreprocessedusingstatisticalfilteringandradiusfiltering.Additionally,anoctreepointclouddownsamplingmethodisemployedtosimplifythedata.AplaneisfittedusingtheRANSACalgorithmtoobtaintheintersectionline,followedbydeterminingpointsinthepointclouddatathatmeetdistancethresholdcriteria.Finally,theentireleastsquaresmethodisusedtofittheweldline.Experimentalresultsdemonstratea65%improvementinefficiencycomparedtosamplingmethodsbasedonpointcloudcurvature.Thewholeleastsquaresmethodoffersadvantagessuchassimplecalculationandminimaldeviationwhilemeetingaccuracyrequirementsforlarge-sizedstructuralpartsweldingprocesses.

Keywords:structuralpartswelding;three-dimensionalpointcloud;weldidentificationandextraction;planefitt