摘要
针对烧结工艺中,台车箅条在机头位置容易发生脱落的问题,本文采用在机头处搭建箅条脱落检测摄像头,并基于YOLOv3算法检测台车箅条缺失现象。首先根据真实箅条尺寸设计出台车箅条模型,并在不同角度采集不同的箅条缺失图像数据;然后使用LabelImg软件标注箅条缺失图像数据;最后在Pytorch深度学习框架下基于YOLOv3算法训练并检测台车箅条缺失图像数据。实验结果表明:摄像机安装在台车两侧对箅条缺失检测准确度能达到92.31%以上,可为箅条缺失检测摄像机安装位置提供方案,并为烧结台车箅条缺失检测提供技术支持。
Aiming at the problem that the trolley grate bars are easy to fall off at the head position during the sintering process, camera is set up at the sintering machine head position to detect the lack of trolley grate bars.The missing of trolley grate bars is detected based on YOLOv3 algorithm.Firstly, the trolley grate bars model is designed and issued according to the real size of grate bars, and different missing image data of grate bars are collected at different angles.Then, LabelImg is used to label the missing image data of grate bars.Finally, under the Pytoch deep learning framework, missing image data of trolley grate bars is trained and detected based on YOLOv3 algorithm.The results show that the cameras installed on both sides of the trolley can detect the grate bars missing with the accuracy of more than 92.31%,which provides a scheme for the installation position of the grate bars missing detection camera and provides technical support for the trolley grate bars missing detection in the sintering plant.
作者
王月明
李世翔
翟容清
吴永刚
WANG Yueming;LI Shixiang;ZHAI Rongqing;WU Yonggang(School of Information Engineering,Inner Mongolia University of Science and Technology,Baotou 014010,Inner Mongolia,China;Ironmaking plant,Inner Mongolia Baotou Steel Union Co.,Ltd.,Baotou 014010,Inner Mongolia,China)
出处
《烧结球团》
北大核心
2021年第5期30-34,53,共6页
Sintering and Pelletizing
基金
内蒙古自治区关键技术攻关项目(2021GG0045)
内蒙古自然基金资助项目(2019MS06036&2020MS06008)。
关键词
箅条缺失
深度学习
检测
烧结机
YOLOv3
missing grate bars
deep learning
detection
sintering machine
YOLOv3
