基于机器视觉的焊缝焊前定位算法研究

在汽车动力锂离子电池的生产中, 中间侧板和端板的焊接是一个重要环节, 焊缝质量很大程度影响着整体模组质量。为保证焊接的定位精度, 文章提出一种基于机器视觉的焊缝焊前定位算法。首先, 对采集到的图像进行灰度化、滤波去噪声、感兴趣区域设置、二值化的预处理; 然后, 通过改进的 Boudaoud 骨架细化算法提取待焊区域骨架, 用灰度投影法结合亚像素双插值法分离无关点, 筛选出所需亚像素点集, 并利用随机抽样一致性算法拟合 2 条直线; 最后, 利用长度和角度信息确定焊接位置坐标。实验结果证明, 文章算法可以精确提取出待焊区域焊缝中心线骨架, 对于焊缝的焊前定位精度高、定位速度快, 可以很好地满足实际工业生产需求。

The quality of the weld between the septum and end plates is a crucial factor in the production of automotive traction lithium-ion batteries. This weld significantly impacts the overall module quality. To ensure welding accuracy, this paper proposes a pre-weld positioning algorithm based on machine vision. The captured image undergoes pre-processing, including grayscaling, filtering, denoising, setting the region of interest, and binarization. The improved Boudaoud skeleton refinement algorithm is then used to extract the skeleton of the area to be welded. The irrelevant points are separated using the grayscale projection method combined with the sub-pixel bi-interpolation method, screening out the required sub-pixel point set, and two straight lines are fitted using the random sample consensus algorithm. Finally, the welding position coordinates are determined based on length and angle information. The experiments have demonstrated that the algorithm can accurately extract the weld centerline in the target welding area, ensuring high accuracy and speed for pre-weld positioning, which satisfies the requirements of industrial production.