基于YOLOv5的烧结机台车篦条缺失检测研究  

Research on missing grate detection of sintering machine trolley based on YOLOv5

作　　者：翟容清 陈波[1] 王月明[1] ZHAI Rongqing;CHEN Bo;WANG Yueming(Information Engineering School,Inner Mongolia University of Science and Technology,Baotou 014010,China)

机构地区：[1]内蒙古科技大学信息工程学院,内蒙古包头014010

出　　处：《内蒙古科技大学学报》2022年第3期227-231,共5页Journal of Inner Mongolia University of Science and Technology

基　　金：内蒙古自治区自然基金资助项目(2020MS06008,2019MS06036);内蒙古自治区关键技术攻关项目(2021GG0045).

摘　　要：在烧结工艺中,篦条的缺失会使台车底部出现缺口,烧结矿掉落造成生产事故,严重将导致烧结系统停产.针对以上问题,提出了在烧结机机头处基于目标检测算法对篦条缺失进行检测的方案.通过采集篦条缺失图像,使用图片标注工具Labellmg对篦条缺失样本进行标注,构建数据集.在Pytorch深度学习框架下,采用目标检测YOLOv5算法对样本进行训练,使用训练权重对测试集中的图像进行检测及准确率分析,结果表明:使用YOLOv5检测篦条缺失mAP值可达99.5%,其训练后的权重可以检测篦条缺失图像.最后,基于PyQt构建烧结机台车篦条缺失检测系统,对篦条缺失进行实时检测,避免生产事故的发生,为烧结机台车篦条缺失检测提供解决方案.In the sintering process,the lack of grate will make a gap at the bottom of the trolley,and the agglomerate will fall,resulting in production accidents,which will seriously lead to the shutdown of the sintering system.In view of the above problems,a scheme based on a target detection algorithm was proposed to detect the lack of grate at the head of the sintering machine.By collecting the missing grate images,Labellmg was used to label the missing grate samples and build the dataset.Under the deep learning framework called Pytorch,the target detection YOLOv5 algorithm was used to train the samples,and the training weights were used to detect and analyze the accuracy of the images in the test dataset.The results show that the mAP value of detecting the missing grate bar with YOLOv5 can reach 99.5%,and the trained weights can detect the image of the missing grate bar.Finally,based on PyQt,a detection system for the lack of grate bar of sintering machine trolley was constructed to detect the lack of grate bar in real-time,which may avoid the occurrence of production accidents and provide a solution for the missing detection of grate bar of sintering machine trolley.

关 键 词：篦条缺失 深度学习 目标检测 YOLOv5 PyQt 

分 类 号：TF046.4[冶金工程—冶金物理化学]