管屏拼焊焊缝表面缺陷的激光视觉检测方法

Laser vision detection method of surface defects in weld of tailor-welded tube panel

  • 摘要: 针对管屏多道焊缝表面缺陷存在焊缝定位困难、计算量过大的问题,文中提出一种管屏拼焊焊缝表面缺陷激光视觉检测方法,建立基于激光视觉的管屏焊缝检测系统,并基于激光视觉相机拍摄管屏图像。分析管屏特征并结合形态学操作定位焊缝,使用高维多项式拟合解决焊缝缺陷快速检测问题,实现了对管屏拼焊表面缺陷的快速检测。结果表明,该检测方案能识别的最小缺陷为1 mm2,检测速度为0.438 m/s,提升了管屏焊缝表面缺陷的检测效率,解决了曲面焊缝定位难、数据量大的难题,实现快速对大量焊缝轮廓进行检测,并符合应用要求。创新点: (1)提出了一种管屏拼焊焊缝表面缺陷激光视觉检测方案,提升管屏焊缝表面缺陷检测效率。(2)提出了基于高阶曲面的焊缝缺陷检测方法,实现快速对大量焊缝轮廓进行检测。(3)管屏拼焊焊缝表面缺陷激光视觉检测方法经过数据集验证,效率和精度均满足应用要求。

     

    Abstract: In view of surface defects of multi-pass welds of pipe panels, it was difficult to locate welds and calculation amount was too large. A laser vision detection method of surface defects in welds of tailor-welded tube panel was proposed. The weld detection system of tube panel based on laser vision was established, and image of tube panel was captured based on laser vision camera. Characteristics of tube panel were analyzed and welds were positioned by combining morphological operations, and high-dimensional polynomial fitting was used to solve the problem of rapid detection of weld defects, which realized rapid detection of surface defects of tailor-welded tube panels. The results showed that the detection method could identify a minimum defect with 1 mm2 and a detection speed was 0.438 m/s, which improved the detection efficiency of surface defects of tube panel welds, solved problems of difficult positioning of curved welds and a large amount of data, and realized rapid detection of a large number of weld profiles and met application requirements.Highlights: (1) A laser vision detection method of surface defects of welds of tube panel was proposed to improve detection efficiency of surface defects of welds of tube panel.(2) A weld defect detection method based on high-order surface was proposed to quickly detect a large number of weld profiles.(3) The laser visual inspection method of surface defects of welds of tube panel was verified by dataset, and efficiency and accuracy met application requirements.

     

/

返回文章
返回