基于目标检测的输送带损伤检测方法  

Conveyor Belt Damage Detection Method Based on Object Detection

作　　者：陈云 吴敬兵[1] 叶涛[1] CHEN Yun;WU Jing-bing;YE Tao(School of Mechanical and Electrical Engineering,Wuhan University of Technology,Wuhan 430070,China)

机构地区：[1]武汉理工大学机电工程学院,武汉430070

出　　处：《自动化与仪表》2022年第11期64-69,共6页Automation & Instrumentation

摘　　要：目前,对矿用输送带损伤检测的研究主要是针对输送带的撕裂,为了更精确地对输送带进行检测,将损伤类型分为击穿、破损、撕裂、表面划伤4种,并提出了一种基于目标性检测的矿用输送带损伤检测方法。通过Yolov3(you only look once)目标检测网络对输送带损伤类型进行识别分类。实验结果表明,Yolov3目标检测网络与其他目标性检测网络相比,检测精度高且检测速度快,在矿用输送带损伤数据集上,对定义的4种损伤类型检测的平均精度值达到92.3%,帧率达101帧/s。At present,the research on damage detection of mining conveyor belts is mainly aimed at the tearing of conveyor belts.In order to detect the conveyor belts more accurately,the damage types are divided into four types:breakdown,damage,tearing,and surface scratches.A damage detection method for mining conveyor belts based on object detection is proposed.The type of conveyor belt damage is identified and classified by the Yolov3(you only look once)object detection network.The experimental results show that the Yolov3 object detection network has high detection accuracy and fast detection speed compared with other object detection networks.On the mining conveyor belt damage data set,the average accuracy of the four defined damage types is 92.3%;The frame rate reaches 101 frames/s.

关 键 词：深度学习 目标性检测 矿用输送带 损伤类型 Yolov3 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]