煤矿井下承人装置违规检测研究  

Research on Violation Detection of Bearing Device in Coal Mine

作　　者：张艳花 白尚旺[1] ZHANG Yanhua;BAI Shangwang(Department of Computer Science and Technology,Taiyuan University of Science and Technology,Taiyuan 030024)

机构地区：[1]太原科技大学计算机科学与技术学院,太原030024

出　　处：《计算机与数字工程》2023年第3期700-705,共6页Computer & Digital Engineering

基　　金：山西省中科院科技合作项目(编号:20141101001);山西省重点研发计划(一般)工业项目(编号:201703D121042-1);山西省社会发展科技项目(编号:20140313020-1);太原科技大学校博士科研启动基金(编号:20162036)资助。

摘　　要：针对井下承人装置违规检测大多是监工人员现场监督效率低下、矿井光照不均匀的问题,提出利用自适应伽马变换图像增强与改进的YOLOv3网络相结合的实时检测方法。首先构建了矿井承人装置数据集,采用自适应伽马变换对井下图像进行亮度校正,然后改进了YOLOv3的分类器,为了进一步提高模型的性能,引入SENet结构来增强网络的全局接受范围。在自制数据集上的实验结果表明,嵌入SENet结构的YOLOv3网络可以达到94.6%的准确率和91.8%的召回率,可以显著提高井下承人装置违规的检测精度。同时,该网络检测速度每秒达到42.54帧,满足实时性要求。Aiming at the problem that the illegal detection of underground bearer devices is mostly caused by the low efficien⁃cy of on-site supervision by supervisors and uneven lighting in the coal mine,a real-time detection method combining adaptive gam⁃ma transform image enhancement and improved YOLOv3 network is proposed.A data set of coal mine bearer device is constructed,the adaptive gamma transformation is used to correct the brightness of mine images,and then the YOLOv3 classifier is improved.In order to further improve the performance of the model,the SENet structure was introduced to enhance the global acceptance range of the network.The experimental results on the self-made data set show that the YOLOv3 network embedded with the SEnet structure can achieve an accuracy rate of 94.6%and a recall rate of 91.8%,which can significantly improve the detection accuracy of under⁃ground bearer device violations.At the same time,the network detection speed reaches 42.54 frames per second,which meets re⁃al-time requirements.

关 键 词：承人装置违规检测 图像增强 伽马变换 YOLOV3 SENet 

分 类 号：TD563[矿业工程—矿山机电] TP39[自动化与计算机技术—计算机应用技术]