不锈钢薄板激光焊焊缝缺陷检测与识别  

Detection and Identification of Weld Defects in Laser Welding of Stainless Steel Sheet

作　　者：孙露萍 闫志鸿[1] 王俊涛[2] 张方洲 SUN Luping;YAN Zhihong;WANG Juntao;ZHANG Fangzhou(Engineering Research Center for Advanced Manufacturing Technology of Automotive Structural Components,Beijing University of Technology,Beijing 100124,China;China Aviation Comprehensive Technology Institute,Aviation Industry Corporation of China,Beijing 100028,China)

机构地区：[1]北京工业大学机电学院汽车结构部件先进制造技术教育部工程研究中心,北京100124 [2]中国航空工业集团有限公司中国航空综合技术研究所,北京100028

出　　处：《热加工工艺》2021年第19期109-113,共5页Hot Working Technology

基　　金：国家自然科学基金面上项目(51975015)。

摘　　要：不锈钢薄板激光焊焊缝成形尺寸细微,焊接速度快、节拍快,人工进行焊缝缺陷检测已不适合,因此焊缝缺陷检测的自动化是必然趋势。构建了一套激光焊焊缝数字成像(DR)检测系统,并在DR检测图像的基础上研究了焊缝缺陷的自动检测与识别算法。首先设计焊缝和焊接缺陷检测系统,可靠地将焊缝和焊接缺陷从工件背景中分割出来,基于此又设计了焊接缺陷特征提取算法和缺陷类型识别算法,通过二叉树和逻辑回归分类可以将激光焊焊缝中的常见缺陷识别出来,识别正确率达到了工程化水平。The laser welding seam of stainless steel sheet has small forming size, fast welding speed and fast cycle.Manual welding seam defect detection is no longer suitable, so the automation of weld seam defect detection is an inevitable trend. A set of laser welding seam DR inspection system was constructed, and based on the DR inspection image, an automatic detection and recognition algorithm of weld seam defects was studied. The welding seam and welding defect detection system were designed firstly, which could reliably segment the welding seam and welding defect from the workpiece background.Based on this, a welding defect feature extraction algorithm and defect type recognition algorithm were designed. The common defects in laser welding seam were identified by binary tree and logistic regression classification, and the accuracy of identification reached the engineering level.

关 键 词：不锈钢激光焊 数字射线 焊缝检测 缺陷检测与识别 逻辑回归 

分 类 号：TG441.7[金属学及工艺—焊接] TG456.7