基于CA-PnPNet的焊接接头类型与漏焊检测

打开文本图片集

关键词：计算机视觉；三维点云；焊接接头分类；漏焊检测；PointNet十 + ;PnP3D中图分类号：TP391.4；TG409 文献标识码：A DOI：10.7535/hbkd.2026yx01009

Abstract：Toaddressthe limited3Dstructuralperceptionand insuficientfeaturediscrimination in traditional welding joint clasificationand lack-of-fusion detection methods，thisstudyproposeda3Dpointcloud detectionnetwork that integrated geometric structure modeling withanattention mechanism，termed CA-PnPNet.First，the network was builtupon the PointNet+ + framework，in which a point neighborhood processing in 3D（ PnP3D ）was integrated into multiple feature extractionstages tostrengthenthemodelingoflocal spatial geometricrelationships.Inadition，achannelatentionmodule （CAM）was incorporated toadaptivelyemphasize keyfeaturesbycapturing semantic dependenciesacross channels.Finally， thecollborativeintegrationof thesetwomodulesatdiferentfeaturelayersenabledunifiedenhancementofbothlocalpoint cloud geometric representation and semanticfeatureexpression，resulting in more comprehensive 3D structural characterization.Tovalidatetheefectivenessof themethod，multiplesetsofcomparativeexperiments wereconductedThe results demonstrate that CA-PnPNet achieves an accuracy of 97.7% inthe welding point cloud classification task， outperforming the baseline model by 1.9% ，while improving the inference speed from 33.3 FPS to 36.1 FPS. These results validatethe superioracuracyandreal-time performance of the proposed method.Overall，CA-PnPNet provides aneffective technical reference for intellgent detection and industrial quality monitoring of complex welded structures.

Keywords：computer vision；3Dpoint cloud；welding joint classification；lack-of-fusion detection；PointNet++；PnP3D

智能焊接技术是推动制造业智能化发展的重要组成部分，在航空航天、轨道交通等高端装备制造领域，焊接质量直接决定结构件的服役性能与安全性。（剩余11402字）