基于机器视觉技术的汽车制造零件检测系统  

The Detection System of Automobile Manufacturing Parts Based on Machine Vision Technology

作　　者：王金成 刘兴 Wang Jincheng;Liu Xing(FAW Volkswagen Co.,Ltd,Foshan Branch,Foshan 528000)

机构地区：[1]一汽-大众汽车有限公司佛山分公司,佛山528000

出　　处：《汽车工艺与材料》2023年第7期68-72,共5页Automobile Technology & Material

摘　　要：为代替人工完成汽车零件安装状态的检测识别,基于YOLOv5算法并结合PLC、RFID、Python、OpenCV、钉钉即时移动通讯软件设计形成一套完整的视觉检测系统。该系统种所训练得到的样本模型准确率和召回率均在98%以上,从实际运行观察,其识别并准确提醒的概率在99%以上,并且可满足生产节拍需求,优于目前人工检测的实际效果,成本较低具备良好的经济推广性。A complete visual inspection system based on YOLOv5 combining with PLC,RFID,Python,OpenCV and DingTalk,was designed to replace manual inspection and recognition of the installation status of automotive parts.Accuracy and recall rate of the sample models trained in this system are both above 98%.From actual operation observation,the probability of identifying and accurately reminding is above 99%,and it meets the requirements of production takt,superior to the actual results of current manual detection,with low cost and good economic promotion.

关 键 词：整车生产 YOLOv5 PYTHON 检测 

分 类 号：U466[机械工程—车辆工程]