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。
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.
作者
付光
李泽琳
王成
贺岩
于浩淼
于洋
FU Guang;LI Zelin;WANG Cheng;HE Yan;YU Haomiao;YU Yang(Beijing Shougang Co.,Ltd.,Tangshan 064400,China;Shougang Zhixin Electromagnetic Material Co.,Ltd.,Tangshan 064400,China;Technology Institute of Shougang Group Co.,Ltd.,Beijing 100043,China)
出处
《轧钢》
2021年第6期103-108,共6页
Steel Rolling
关键词
冷轧带钢
压印缺陷
机器视觉
识别率
分类
快速控制
cold rolled strip
imprinting defect
machine vision
recognition rate
classification
fast control