基于3D机器视觉动力电池焊缝质量检测方法  

Inspection Method of Welding Quality about Power Battery Based on 3D Machine Vision

作　　者：莫之剑 范彦斌[2] 彭明仔 MO Zhijian;FAN Yanbin;PENG Mingzai(Guangdong Liyuanheng Intelligent Equipment Co.,Ltd.,Huizhou,Guangdong 516000,China;School of Mechatronics Engineering,Foshan University,Foshan,Guangdong 528000,China;Mechanical and Electrical Engineering College,Guangzhou Vocational College of Technology&Business,Guangzhou 510000,China)

机构地区：[1]广东利元亨智能装备股份有限公司,广东惠州516000 [2]佛山科学技术学院机电工程学院,广东佛山528000 [3]广州科技贸易职业学院机电工程学院,广州510000

出　　处：《机电工程技术》2020年第4期1-3,95,共4页Mechanical & Electrical Engineering Technology

摘　　要：为了解决动力电池在生产过程中人工检测焊缝缺陷容易产生误检、漏检和无法实现自动化检测的问题,设计了一种基于3D机器视觉技术和机器学习的焊缝质量检测方法。采用支持向量机SVM学习焊缝关键特征值,得到最优分类超平面,训练完成后,相机采集图像,通过高斯滤波降噪、图像纠正和缺陷特征提取,然后SVM对缺陷特征值进行分类,判断产品是否存在缺陷。实验证明,该方法实现了100%在线自动化检测焊缝质量,并具有快速、准确的优点,成功地解决了动力电池生产过程中焊缝质量自动化检测难题。In order to solve the problem that the manual inspection of weld defects in the production process of power battery is easy to produce false inspection,missing inspection and unable to realize automatic inspection,a welding quality inspection method based on 3D machine vision technology and machine learning was designed.Support vector machine was used to learn the key characteristic values of weld seam,and the optimal classification hyperplane was obtained.After the training,the images was collected by the camera,denoises through Gaussian filtering,image correction and defect feature extraction,and the defect characteristic values were classified by SVM to determine whether the product has defects.The experiment shows that this method realizes 100%on-line automatic detection of weld quality,and has the advantages of fast and accurate,and successfully solves the problem of automatic detection of weld quality in the production process of power battery.

关 键 词：3D机器视觉 点云 焊缝质量检测 移动窗口 支持向量机SVM 

分 类 号：TG409[金属学及工艺—焊接]