基于机器视觉的钢水自动化取样

Molten steel automatic sampling based on machine vision

钢水分析是钢铁生产重要的过程控制环节。针对现有人工钢水取样危险系数高、工作环境恶劣,基于机器视觉的钢水自动化取样系统设计,采用机械臂完成钢水自动化取样。基于CCD获取的钢水样件图像,设计了分区特征提取算法,实现了图像坐标系下抓取位置及抓取方向的识别。通过摄像机标定、手眼标定、末端转轴映射函数,完成了图像坐标系下抓取位置、方向到机械臂基座坐标系下抓取位置、姿态的转化。实验验证了设计系统的有效性,针对直径6 mm、长度为50 mm的钢水样件把手,机械臂定位抓取位置偏差不高于2 mm,方向偏差不高于2°,精度满足应用需求。

Molten steel analysis is an important step of process control in iron and steel production.In view of the high⁃risk coefficient and abominable working environment of the existing manual molten steel sampling,a molten steel automatic sampling system based on machine vision is proposed,in which the mechanical arm is used to complete automatic sampling.The molten steel samples are captured by CCD,and a partition feature extraction algorithm is designed to recognize grasping position and grasping direction in the image coordinates.On the basis of the camera calibration,hand⁃eye calibration and end effector hinge mapping function,the grasping position and grasping direction in the image coordinates are mapped to the grasping position and grasping posture in coordinate system of mechanical arm base.The experiments have verified the effectiveness of the designed system.For the molten steel sample handle with a diameter of 6 mm and a length of 50 mm,the positioning and grasping position deviation of the mechanical arm is less than 2 mm,and the direction deviation is less than 2°,which meets the application requirements.