煤炭是我国重要的能源供给。在煤炭的生产运输的过程中，带式输送机扮演着至关重要的角色。但开采的原煤常常伴随着矸石、铁器等异物，其存在严重威胁着带式运输机的正常运行。因此本文主要研究基于深度学习的非煤异物检测方法，旨在对带式运输机工作过程中出现的非煤异物及时的发现，避免带式运输机的损害，保证煤矿生产的安全高效。本文进行的主要研究如下：

       针对煤矿视频监控系统采集到的图像质量较低的问题，本文提出了一种基于联合增强算法的低质量带式运输机图像预处理模型，通过降噪、增强和分割的方法，对矿下低质量带式运输机图像进行预处理。该预处理方法可以降低噪声，提高整体亮度和对比度，增强图像细节信息，为后续异物检测提供了更好的依据。

       本文采集非煤异物样本数据，并设计制作非煤异物数据集。针对异物样本的特殊性，采用数据增强的方法，扩充非煤异物数据集数量；针对模型受数据限制，可能出现泛化性能差的问题，本文在数据增强的基础上，提出一种针对中间隐藏层的混类增强方法，改善数据集质量，为后续构建异物检测模型打下了基础。

      本文通过对主流目标检测模型的对比研究，选择YOLO检测模型为带式运输机异物检测模型的基准模型，针对异物检测任务进行了改进激活函数和解耦候选框等方面的改进优化。并通过引入视觉注意力机制和改进加强特征提取网络，进一步提升改进YOLO异物检测模型的检测性能。改进后模型对矸石类异物识别精度达到88.14%，铁器类异物识别精度达到87.10%，并能保证每秒27 FPS的检测速度。

       经过实验分析验证，本文构建的带式运输机非煤异物检测系统，可以实时 有效地检测出带式运输机工作工程中出现的异物，在带式运输机防护上具有一定的实用价值。

Coal is an important energy supply in our country. In the process of coal production, belt conveyor plays a vital role. But the raw coal is often accompanied by gangue, iron and other foreign bodies, its existence seriously threatens the normal operation of the belt conveyor. Therefore, this paper mainly studies the method of non-coal foreign body detection based on deep learning, aiming at finding the non-coal foreign body in time, avoiding the damage of belt conveyor, and ensuring the safety and high efficiency of coal mine production. The main research carried out in this paper is as follows:

In view of the complex working environment of belt conveyor and the low image quality captured by the video monitoring system of coal mine, this paper presents a low quality belt conveyor image preprocessing model based on the joint enhancement algorithm, which adopts the methods of noise reduction, enhancement and segmentation to preprocess the image of low quality belt conveyor under the mine. It can reduce noise, improve the overall brightness and contrast, enhance image details, and provide a better basis for the subsequent recognition of non-coal foreign bodies.

This paper has collected the sample data of non-coal foreign bodies and designed and made the data set of non-coal foreign bodies. In view of the particularity of the research object, this paper expands the number of non-coal foreign body dataset by means of data enhancement, and in view of the problem that the model is limited by data and may lead to poor generalization performance, this paper proposes a hybrid enhancement method for the middle hidden layer based on data enhancement to improve the quality of the dataset and lay a foundation for the subsequent construction of non-coal foreign body detection model.

Through the comparative study of mainstream target detection models, this paper selects Yolo detection model as the benchmark model of foreign object detection model of belt conveyor, and improves and optimizes the foreign object detection task in terms of improved activation function and decoupling candidate box. By introducing visual attention mechanism and improving feature extraction network, the detection performance of Yolo foreign object detection model is further improved. The recognition accuracy of the improved model for gangue foreign matters is 88.14%, and the recognition accuracy of iron foreign matters is 87.10%, and the detection speed of 27fps per second can be guaranteed.

Through experimental analysis, the non-coal foreign body detection system of belt conveyor constructed in this paper can effectively detect non-coal foreign body on belt conveyor, and has certain practical value in tearing protection of belt conveyor.

[1]能源新闻[OL]. 2022-02-28.

[2] 元松明,寇子明,秦国栋,等. 长距离大坡度下运带式输送机在孟家窑煤业的应用[J]. 煤炭工程,2021,53(3):98-102.

[3] 武林海.带式输送机常见故障检测及防治系统研究[J]. 煤矿机械，2019,40（2）：145-147.

[4]陆小翠,苗长云,王巍,等. X 光强力输送带无损检测系统及其网络传输的设计[J].工矿自动化,2008,(1): 62-64

[5]Dobrot, Dan. Vulcanization of Rubber Conveyor Belts with Metallic Insertion Using Ultrasounds[J]. Procedia Engineering, 2015, 100:1160-1166.

[6]徐善永,黄友锐,冯涛,等. 基于SMO算法的皮带撕裂红外图像检测方法[J]. 现代电子技术,2020,43(11):37-40,46. DOI:10.16652/j.issn.1004-373x.2020.11.009.

[7]Hou C , Qiao T , Qiao M , et al. Research on Audio-Visual Detection Method for Conveyor Belt Longitudinal Tear[J]. IEEE Access, 2019, PP(99):1-1.

[8]Yang Y , Hou C , Qiao T , et al. Longitudinal tear early-warning method for conveyor belt based on infrared vision[J]. Measurement, 2019, 147:106817.

[9]乔铁柱.输送带纵向撕裂可见光与红外双目视觉在线检测系统研究[D].太原:太原理工大学, 2015.

[10]Tan C , Yang J . Research on extraction of image gray information and texture features of coal and gangue image[J]. Industry and Mine Automation, 2017.

[11]何晓明. 基于X射线的煤与矸石自动识别方法研究[D]. 东北大学, 2013.

[12]Zhao Y D , He X . Recognition of Coal and Gangue Based on X-Ray[J]. Applied Mechanics & Materials, 2013, 275-277:2350-2353.

[13]Zou L , X Yu, Li M , et al. Nondestructive Identification of Coal and Gangue via Near-Infrared Spectroscopy Based on Improved Broad Learning[J]. IEEE Transactions on Instrumentation and Measurement, 2020, PP(99):1-1.

[14]宋亮, 刘善军, 虞茉莉,等. 基于可见-近红外和热红外光谱联合分析的煤和矸石分类方法研究[J]. 光谱学与光谱分析, 2017, 37(2):7.

[15]Wang X , Wang S , Y Guo, et al. Dielectric and geometric feature extraction and recognition method of coal and gangue based on VMD-SVM[J]. Powder Technology, 2021(7).

[16]吴守鹏. 基于机器视觉的运煤皮带异物识别方法研究[D].北京:中国矿业大学,2019.

[17]吕志强. 复杂环境下煤矿皮带运输异物图像识别研究[D].北京:中国矿业大学,2020.

[18]Wang Y , Wang Y , Dang L . Video detection of foreign objects on the surface of belt conveyor underground coal mine based on improved SSD[J]. Journal of Ambient Intelligence and Humanized Computing, 2020:1-10.

[19]杜京义, 陈瑞, 郝乐,等. 煤矿带式输送机异物检测[J]. 工矿自动化, 2021, 47(8):7.

[20]杜慧, 邓潇潇, 李建, 等. 智能的图像滤波去噪算法[J]. 计算机工程与设计, 2015(8): 2180-2184.

[21]He K, Sun J, Tang X. Guided image filtering[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2013, 35(6): 1397-1409.

[22]Ham B , Cho M , Ponce J . Robust Guided Image Filtering Using Nonconvex Potentials[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 40(1):192-207.

[23]Li C Y, Guo J C, Cong R M, et al. Underwater Image Enhancement by Dehazing With Minimum Information Loss and Histogram Distribution Prior[J]. IEEE Transactions on Image Processing, 2016, 25(12): 5664-5677.

[24]Yao Z, Lai Z, Wang C. Image Enhancement Based on Equal Area Dualistic Sub-image and Non-parametric Modified Histogram Equalization Method[C]//2016 9th International Symposium on Computational Intelligence and Design (ISCID). IEEE, 2016, 1: 447-450.

[25]Maki S, Konno T, Maeda H. Image enhancement in computerized tomography for sensitive diagnosis of liver cancer and semiquantitation of tumor selective drug targeting with oily contrast medium[J]. Cancer, 2015, 56(4): 751-757.

[26]姚拓中, 左文辉, 安鹏, 等. 渐进式前景更新和区域关联的图像协同分割[J]. 中国图象图形学报, 2019, 24(03): 46–55.

[27]Bai M, Urtasun R. Deep watershed transform for instance segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017: 5221-5229.

[28]王静. 图像分割算法的研究与应用[D]. 西安：延安大学, 2015.

[29]张海涛, 李雅男. 阈值标记的分水岭彩色图像分割[J]. 中国图象图形学报, 2015, 20(12): 1602-1611.

[30]杨丹, 赵海滨, 龙哲等. MATLAB图像处理实例详解[M]. 清华大学出版社. 2-5.

[31]NX Thảo. Similarity measures of picture fuzzy sets based on entropy and their application in MCDM[J]. Pattern Analysis and Applications, 2019.

[32]韩幸. 分析矸石山环境污染治理的对策建议[J]. 当代化工研究, 2019, (3):16-17.

[33]Msa A , Ams A , Hsa B . High-dimensional dynamics of generalization error in neural networks[J]. Neural Networks, 2020, 132:428-446.

[34]Yun S , Han D , Chun S , et al. CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features[C]// International Conference on Computer Vision.

[35]Schmidhuber J. Deep learning in neural networks: An overview[J]. Neural Networks, 2015, 61: 85-117.

[36]Ren S , He K , Girshick R , et al. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(6):1137-1149.

[37]Dai J , Li Y , He K , et al. R-FCN: Object Detection via Region-based Fully Convolutional Networks[J]. Curran Associates Inc. 2016.

[38]Szegedy C , Reed S , D Erhan, et al. Scalable, High-Quality Object Detection[J]. Computer Science, 2014.

[39]Liu W , Anguelov D , Erhan D , et al. SSD: Single Shot MultiBox Detector[J]. Springer, Cham, 2016.

[40]Redmon J , Divvala S , Girshick R , et al. You Only Look Once: Unified, Real-Time Object Detection[J]. IEEE, 2016.

[41]Wang C Y , Liao H , Wu Y H , et al. CSPNet: A New Backbone that can Enhance Learning Capability of CNN[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). IEEE, 2020.

[42]Liu S , Qi L , Qin H , et al. Path Aggregation Network for Instance Segmentation[J]. IEEE, 2018.

[43]Redmon J , Farhadi A . YOLOv3: An Incremental Improvement[J]. arXiv e-prints, 2018.

[44]Wang C Y , Liao H , Wu Y H , et al. CSPNet: A New Backbone that can Enhance Learning Capability of CNN[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). IEEE, 2020.

[45]He K , Zhang X , Ren S , et al. Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2014, 37(9):1904-16.

[46]Bochkovskiy A , Wang C Y , Liao H . YOLOv4: Optimal Speed and Accuracy of Object Detection[J]. 2020.

[47]Yu J , Jiang Y , Wang Z , et al. UnitBox: An Advanced Object Detection Network[J]. ACM, 2016.

[48]Misra D . Mish: A Self Regularized Non-Monotonic Neural Activation Function[J]. 2019.

[49]Chen K M , Cofer E M , Zhou J , et al. Selene: a PyTorch-based deep learning library for sequence data[J]. Nature Methods, 2019, 16(4):315.

[50]李鹤喜,李记花,李威龙. 一种基于注意机制和卷积神经网络的视觉模型[J]. 计量学报,2021,42(7):840-845.

[51]张超. P-R曲线与模型评估问题研究[J]. 现代信息科技, 2020, 4(4):3.

[52]Everingham M , Eslami S , Gool L V , et al. The Pascal Visual Object Classes Challenge: A Retrospective[J]. International Journal of Computer Vision, 2015, 111(1):98-136.

[53]胡学敏, 童秀迟, 郭琳,等. 基于深度视觉注意神经网络的端到端自动驾驶模型[J]. 计算机应用, 2020.

[54]Tang Z , Liu X , Li Y , et al. Multi-Atlas Brain Parcellation Using Squeeze-and-Excitation Fully Convolutional Networks[J]. IEEE Transactions on Image Processing, 2020, PP(99):1-1.

[55]Ding X , Zhang X , Ma N , et al. RepVGG: Making VGG-style ConvNets Great Again[J]. 2021.

[56]Woo S , Park J , Lee J Y , et al. CBAM: Convolutional Block Attention Module[J]. Springer, Cham, 2018.

[57]Tan M.,Pang R.,and Le Q.EfficientDet:Scalable and Efficient Object Detection[C]// 2020 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA,2020:10778-10787.

[58]Asghar MZ, Albogamy FR, Al-Rakhami MS, et al Facial Mask Detection Using Depthwise Separable Convolutional Neural Network Model During COVID-19 Pandemic. Front Public Health. 2022 Mar 7;10:855254.