首页 | 官方网站   微博 | 高级检索  
     

冷轧带钢连退机组周期性压印缺陷快速控制技术
引用本文:付光,李泽琳,王成,贺岩,于浩淼,于洋.冷轧带钢连退机组周期性压印缺陷快速控制技术[J].轧钢,2021,38(6):103-108.
作者姓名:付光  李泽琳  王成  贺岩  于浩淼  于洋
作者单位:1.北京首钢股份有限公司,河北 唐山 064400; 2.首钢智新电磁材料有限公司,河北 唐山 064400; 3.首钢集团有限公司技术研究院,北京 100043
摘    要:针对冷轧带钢连退机组周期性压印缺陷问题,介绍了一种快速控制技术。该技术以机器视觉为基础,通过有效构建数据库、并手动增加限制规则,对初期压印缺、中期压印缺陷、严重压印缺陷的识别率达到97.5%,分类准确率达95.8%。同时,还介绍了一种辊径排查控制方法,其可快速确定发生压印缺陷的炉辊并进行快速处理,周期性压印缺陷的排查及解决周期由1~1.5卷缩短到1/3~2/3卷左右。

关 键 词:冷轧带钢  压印缺陷  机器视觉  识别率  分类  快速控制  
收稿时间:2021-01-12

Fast control technology for periodic imprinting defect of cold rolled strip in continuous annealing unit
FU Guang,LI Zelin,WANG Cheng,HE Yan,YU Haomiao,YU Yang.Fast control technology for periodic imprinting defect of cold rolled strip in continuous annealing unit[J].Steel Rolling,2021,38(6):103-108.
Authors:FU Guang  LI Zelin  WANG Cheng  HE Yan  YU Haomiao  YU Yang
Affiliation:1. Beijing Shougang Co., Ltd., Tangshan 064400, China; 2. Shougang Zhixin Electromagnetic Material Co., Ltd.,Tangshan 064400, China; 3. Technology Institute of Shougang Group Co.,Ltd.,Beijing 100043,China
Abstract:Aiming at the periodic imprinting defect of cold rolled strip in continuous annealing unit, a fast control technology was introduced. The technology was based on machine vision, by effectively constructing a database and manually adding restriction rules, the recognition rate of initial imprinting defects, mid-term imprinting defects, and serious imprinting defects reached 97.5%, and the classification accuracy rate reached 95.8%. At the same time, a roll diameter inspection and control method was introduced, which can quickly determine the hearth roller with imprinting defects and perform rapid processing. The inspection and resolution cycle of periodic imprinting defects could be shortened from 1 to 1.5 coils to 1/ 3~2/3 coils.
Keywords:cold rolled strip  imprinting defect  machine vision  recognition rate  classification  fast control  
点击此处可从《轧钢》浏览原始摘要信息
点击此处可从《轧钢》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号