摘要
激光焊接过程中,强光导致焊接节点的视觉图像质量较低,降低了定位准确性;为了优化特殊材质的激光焊接节点定位精准度;以AH36钢为例,设计一种激光焊接节点视觉定位方法;在获取AH36钢的激光焊接节点区域图像后,通过EGDNet算法获取焊接节点边缘图像;利用Shi-Tomasi算法检测焊接边缘图像的角点,计算节点在AH36钢激光焊接图像中的位置坐标,完成AH36钢激光焊接节点定位;实验结果表明,该方法的焊接节点采集质量高;对x,y,z三个方向上的定位误差低于0.2 mm;最大定位时间为0.75 s;对于不同方案的平均定位误差始终较小,低于0.5 mm;在基准实验环境下的定位误差平均值为0.2 mm,标准差为0.05 mm,光照变化时的平均值为0.35 mm,标准差为0.08 mm,杂质引入时的平均值为0.4 mm,标准差为0.1 mm;证明了所提技术具有较高的定位精准度和定位速度。
In the process of laser welding,strong light causes the low quality of visual images at welding nodes,which reduces positioning accuracy.To optimize the positioning accuracy of laser welding nodes with special materials.Taking AH36 steel as an example,a visual positioning method for laser welding nodes is designed.After obtaining the laser welding node area image of AH36 steel,the edge image of welding nodes is obtained through EGDNet algorithm.The Shi-Tomasi algorithm is used to detect the corners of welding edge image,calculate the position coordinates of the nodes in the laser welding image of AH36 steel,and complete the node positioning of AH36 steel laser welding.The experimental results show that the welding node acquisition quality of this method is high;The positioning errors of x,y,and z directions are less than 0.2 mm;The maximum positioning time is 0.75 s;The average positioning error for different schemes is always small,below 0.5 mm;The average positioning error in the benchmark experimental environment is 0.2 mm,with a standard deviation of 0.05 mm.The average value is 0.35 mm when the light changes,with a standard deviation of 0.08 mm.The average value are introduced is 0.4 mm when impurities,with a standard deviation of 0.1 mm.It is proven that the proposed technology has high positioning accuracy and speed.
作者
洪祥
张海越
宋骐
HONG Xiang;ZHANG Haiyue;SONG Qi(Nanjing Daqo Electric Research Institute Co.,Ltd.,Nanjing 211000,China;Urbana Champaign Branch School University of Illinois,Champaign,IL,USA 61820;Department of Information and Intelligence,University of Science and Technology of China,Hefei 230000,China)
出处
《计算机测量与控制》
2023年第11期299-305,314,共8页
Computer Measurement &Control
基金
国家工信部智能船舶项目(CBG4N21-4-2)。
