瓦斯抽采作业中钻孔深度的精确测量对于确保煤矿安全生产至关重要。然而，由于煤矿井下瓦斯抽采钻场场景的独特性，使得现有的面向通用场景的图像处理方法在处理井下抽采钻场视频时漏检和误检现象频发，从而无法保证钻杆计数的精度。因此，本文提出一种面向煤矿井下复杂场景改进的YOLOv8和ByteTrack算法以及钻机工作状态判别模型。通过分析钻机卡盘与夹持器的相对运动轨迹，将钻进与退钻轨迹分类，对钻进轨迹的峰值计数获取钻进的钻杆数量，从而实现智能化钻杆计数。

针对井下瓦斯抽采钻场场景特性，本文构建了一个全新的煤矿井下瓦斯抽采复杂场景钻场数据集，作为本文钻杆计数方法研究的数据基础。该数据集覆盖了小目标、目标与背景颜色相似、目标背景运动三种复杂场景下的钻孔施工过程图像。本文根据不同场景的特点，研究基于图像处理的钻杆计数方法。

针对钻机的卡盘、夹持器在小目标、目标与背景颜色相似场景下难检测的问题，提出一种面向钻场复杂场景改进的YOLOv8_CBiPW目标检测算法。首先，引入CBAM注意力机制，结合PConv局部卷积计算，使模型更加关注卡盘与夹持器的区域位置信息，降低背景干扰。此外，融合BiFPN特征金字塔网络，增强模型对小目标的特征提取。最后，通过改进损失函数，提高模型对不同场景下目标的定位精度。通过实验验证，本文提出的算法的各项指标均有不同程度的提升，其中目标识别精度和mAP50达到了93.6%和79.1%，验证了在煤矿井下复杂场景下YOLOv8_CBiPW目标检测模型的优越性。

针对ByteTrack目标追踪算法处理目标背景运动场景的不足，提出了一种基于背景运动补偿的ByteTrack_SRAN算法。首先，将YOLOv8_CBiPW作为检测器与ByteTrack追踪算法融合，使算法更好地适应钻场复杂场景下卡盘与夹持器的跟踪任务。其次，引入SURF与RANSAC算法进行目标背景特征提取和最优仿射矩阵计算。结合ByteTrack中的卡尔曼滤波重新投影目标位置与运动方向，对当前帧预测框的位置信息矫正，实现跟踪过程中的数据关联与轨迹预测。经过实验验证，对比原模型，本文提出基于背景运动补偿的ByteTrack_SRAN算法降低了目标ID的切换次数，目标追踪的跟踪精度和轨迹关联的准确性分别提升了3.6%和2.6%，证明了该模型在处理背景运动场景时的追踪稳定性。

最后，提出一种基于钻机运动特性的状态判别模型与钻杆计数方法。基于ByteTrack_SRAN获取卡盘的运动轨迹并进行轨迹拟合。通过进退钻轨迹周期变化率的差异判别轨迹状态，将钻进与退钻轨迹分类。通过对钻进状态下的轨迹峰值计数，获得打入钻杆数量。经过实验证明，本文提出的钻杆计数方法取得了显著的效果。在钻场复杂场景下计数精度达到98.4%，常规场景下达到99.5%，满足煤矿井下瓦斯抽采复杂场景下智能化钻杆计数需求。

The accurate measurement of borehole depth in gas extraction operation is very important to ensure the safety of coal mine production. However, due to the uniqueness of the scene of underground gas extraction drilling field, the existing image processing methods for general scenes often miss and misdetect when processing the video of underground gas extraction drilling field, so that the precision of drill pipe counting can not be guaranteed. Therefore, this paper proposes an improved YOLOv8 and ByteTrack algorithms and a working state identification model of drilling rig for complex underground scenarios. By analyzing the relative motion track of the drill chuck and the gripper, the drilling and undrilling tracks are classified, and the number of drill pipe in the drilling stage is obtained, so as to realize the intelligent drill pipe count.

According to the scene characteristics of underground gas extraction drilling field, this paper constructs a new drilling field data set of complex scene of underground gas extraction in coal mine, which is the data basis for the research of drill pipe counting method in this paper. The data set covers the drilling process images in three complex scenarios: small target, target and background color similar, target background movement. According to the characteristics of different scenes, this paper studies the drill pipe counting method based on image processing.

Aiming at the problem that it is difficult to detect the chuck and the holder of the drilling rig in the scene of small target and similar color between the target and the background, an improved YOLOv8_CBiPW target detection algorithm based on the complex scene of the drilling field is proposed. By introducing CBAM attention mechanism and combining with PConv local convolution calculation, the model pays more attention to the regional position information of chuck and gripper and reduces background interference. In addition, the BiFPN feature pyramid network is integrated to enhance the feature extraction of small targets. Finally, by improving the loss function, the localization accuracy of the model in different scenarios is improved. Through experimental verification, all indexes of the proposed algorithm have been improved to varying degrees, among which the target recognition accuracy and mAP50 reach 93.6% and 79.1%, which verifies the superiority of the YOLOv8_CBiPW target detection model in complex scenarios in coal mines.

A ByteTrack_SRAN algorithm based on background motion compensation is proposed to solve the shortcomings of ByteTrack algorithm in processing target background motion scenes. First, YOLOv8_CBiPW is used as a detector and ByteTrack tracking algorithm is integrated, so that the algorithm can better adapt to the tracking task of chuck and gripper in the complex scene of drilling field. Secondly, SURF and RANSAC algorithms are introduced to extract the target background features and calculate the optimal affine matrix. Combined with Kalman filter in ByteTrack, the target position and movement direction are re-projected, and the position information of the current frame prediction box is corrected, so as to realize data association and track prediction in the tracking process. After experimental verification, compared with the original model, the proposed ByteTrack_SRAN algorithm based on background motion compensation reduces the switching times of target ID, and the tracking accuracy and trajectory association accuracy of target tracking are increased by 3.6% and 2.6% respectively, which proves the tracking stability of this model when processing background motion scenes.

Finally, a state discrimination model and drill pipe counting method based on drill motion characteristics are proposed. Based on ByteTrack_SRAN, the motion trajectory of chuck is obtained and the trajectory fitting is carried out. According to the difference of the cyclic change rate of the drilling trajectory, the trajectory state is judged, and the drilling trajectory and de-drilling trajectory are classified. The number of drill pipe driven is obtained by counting the peak value of the track under drilling condition. The experiment proves that the drill pipe counting method proposed in this paper has achieved remarkable results. In the complex scene of drilling site, the counting accuracy can reach 98.4%, and in the conventional scene, it can reach 99.5%, which meets the intelligent drill pipe counting requirements in the complex scene of underground gas extraction in coal mine.

[1]刘峰,郭林峰,张建明,等.煤炭工业数字智能绿色三化协同模式与新质生产力建设路径[J].煤炭学报,2024,49(01):1-15.

[2]中国矿产资源报告2022[J].自然资源情报,2023,(10):2.

[3]王恩元,张国锐,张超林,等.我国煤与瓦斯突出防治理论技术研究进展与展望[J].煤炭学报,2022,47(01):297-322.

[4]秦波涛,马东.采空区煤自燃与瓦斯复合灾害防控研究进展及挑战[J/OL].煤炭学报,1-18[2024-04-20].

[5]郭世斌,胡国忠,朱家锌等.顶板瓦斯抽采巷布置位置智能预测方法[J/OL].煤炭科学技术:1-13[2024-04-20].

[6]陈本良,袁亮,薛生等.淮南矿区煤层顶板分段压裂水平井抽采技术及效果研究[J/OL].煤炭科学技术:1-11[2024-04-20].

[7]王国法,刘峰,孟祥军等.煤矿智能化(初级阶段)研究与实践[J].煤炭科学技术,2019,47(08):1-36.

[8]王国法.煤矿智能化最新技术进展与问题探讨[J].煤炭科学技术,2022,50(01):1-27.

[9]王国法,庞义辉,任怀伟,等.智慧矿山系统工程及关键技术研究与实践[J].煤炭学报,2024,49(01):181-202.

[10]范京道,徐建军,张玉良,等.不同煤层地质条件下智能化无人综采技术[J].煤炭科学技术,2019,47(03):43-52.

[11]程德强,钱建生,郭星歌,等.煤矿安全生产视频AI识别关键技术研究综述[J].煤炭科学技术,2023,51(02):349-365.

[12]孙志飞,吴银成,胡云.钻杆长度测量方法[J].工矿自动化,2015,41(03):51-53.

[13]王小龙,张军.钻孔数据处理与三维显示软件设计及应用[J].煤矿安全,2021,52(10):130-135.

[14]王岚,陈龙,张鹏,等.随钻测井钻孔深度测量装置的设计[J].煤田地质与勘探,2017,45(03):144-146.

[15]朱祖扬,刘江涛,杨明清等.井筒微芯片示踪器的深度定位研究[J].钻采工艺,2021,44(01):7-12.

[16]李柯泉,陈燕,刘佳晨,等.基于深度学习的目标检测算法综述[J].计算机工程,2022,48(07):1-12.

[17]彭业勋.煤矿井下钻杆计数方法研究[D].西安科技大学,2019.

[18]董立红,王杰,厍向阳.基于改进Camshift算法的钻杆计数方法[J].工矿自动化,2015,41(01):71-76.

[19]方杰,李振璧,夏亮.基于ECO-HC的钻杆计数方法[J].煤炭技术,2021,40(11):186-189.

[20]Zhou M, Yuan J, Gao Z, et al. Drill pipe counting method based on local dense optical flow estimation[C]//Image and Graphics: 11th International Conference, ICIG 2021, Haikou, China, August 6–8, 2021, Proceedings, Part I 11. Springer International Publishing, 2021: 443-454.

[21]Tingjiang T, Changfang G, Guohua Z, et al. Research and application of downhole drilling depth based on computer vision technique[J]. Process Safety and Environmental Protection, 2023, 174: 531-547.

[22]高瑞,郝乐,刘宝,等.基于改进ResNet网络的井下钻杆计数方法[J].工矿自动化,2020,46(10):32-37.

[23]杜京义,党梦珂,乔磊,等.基于改进时空图卷积神经网络的钻杆计数方法[J].工矿自动化,2023,49(01):90-98.

[24]姚超修,胡亚磊.基于视频识别的煤矿井下钻杆计数算法[J].煤炭技术,2023,42(08):203-206.

[25]党伟超,姚远,白尚旺,等.煤矿探水卸杆动作识别研究[J].工矿自动化,2020,46(07):107-112.

[26]Yu Y, Zhao J, Yi C, et al. Drill-Rep: Repetition counting for automatic shot hole depth recognition based on combined deep learning-based model[J]. Engineering Applications of Artificial Intelligence, 2023, 123: 106302.

[27]张栋,姜媛媛.基于改进MobileNetV2的钻杆计数方法[J].工矿自动化,2022,48(10):69-75.

[28]张栋, 姜媛媛. 融合注意力机制与逆残差结构的轻量级钻机目标检测方法[J]. 电子测量与仪器学报, 2022,36(11):201-210.

[29]Hussain M. YOLO-v1 to YOLO-v8, the rise of YOLO and its complementary nature toward digital manufacturing and industrial defect detection[J]. Machines, 2023, 11(7): 677.

[30]Fu H, Song G, Wang Y. Improved YOLOv4 marine target detection combined with CBAM[J]. Symmetry, 2021, 13(4): 623.

[31]Du F J, Jiao S J. Improvement of lightweight convolutional neural network model based on YOLO algorithm and its research in pavement defect detection[J]. Sensors, 2022, 22(9): 3537.

[32]Zhang J, Wei X, Zhang L, et al. YOLO v7-ECA-PConv-NWD Detects Defective Insulators on Transmission Lines[J]. Electronics, 2023, 12(18): 3969.

[33]Wu T, Dong Y. YOLO-SE: Improved YOLOv8 for remote sensing object detection and recognition[J]. Applied Sciences, 2023, 13(24): 12977.

[34]Zhang Y, Sun P, Jiang Y, et al. Bytetrack: Multi-object tracking by associating every detection box[C]//European conference on computer vision. Cham: Springer Nature Switzerland, 2022: 1-21.

[35]Han L, Wang J, Zhang Y, et al. Research on adaptive ORB-SURF image matching algorithm based on fusion of edge features[J]. IEEE Access, 2022, 10: 109488-109497.

[36]Zhang F, Li X, Wang T, et al. High-Precision Satellite Video Stabilization Method Based on ED-RANSAC Operator[J]. Remote Sensing, 2023, 15(12): 3036.

[37]Zhang Q, He X, Yang S, et al. Solving dynamic multi-objective problems using polynomial fitting-based prediction algorithm[J]. Information Sciences, 2022, 610: 868-886.

[38]蒋玉英,陈心雨,李广明,等.图神经网络及其在图像处理领域的研究进展[J].计算机工程与应用,2023,59(07):15-30.

[39]Du T, Wang D, Qian X. Study on 3-dimensional stereoscopic image model in intelligent coal mine[J]. Energy Reports, 2022, 8: 291-299.

[40]Cheng G, Chen J, Wei Y, et al. A Coal Gangue Identification Method Based on HOG Combined with LBP Features and Improved Support Vector Machine[J]. Symmetry, 2023, 15(1): 202.

[41]Liang Y, Mao S, Zheng M, et al. Study on the Prediction of Low-Index Coal and Gas Outburst Based on PSO-SVM[J]. Energies, 2023, 16(16): 5990.

[42]Huang X, Li Z, Jin Y, et al. Fair-AdaBoost: Extending AdaBoost method to achieve fair classification[J]. Expert Systems with Applications, 2022, 202: 117240.

[43]许德刚,王露,李凡.深度学习的典型目标检测算法研究综述[J].计算机工程与应用,2021,57(08):10-25.

[44]Lv Z, Wang W, Xu Z, et al. Fine-grained object detection method using attention mechanism and its application in coal–gangue detection[J]. Applied Soft Computing, 2021, 113: 107891.

[45]Yan P, Sun Q, Yin N, et al. Detection of coal and gangue based on improved YOLOv5. 1 which embedded scSE module[J]. Measurement, 2022, 188: 110530.

[46]Song P, Li P, Dai L, et al. Boosting R-CNN: Reweighting R-CNN samples by RPN’s error for underwater object detection[J]. Neurocomputing, 2023, 530: 150-164.

[47]Kumar L S, Hariharasitaraman S, Narayanasamy K, et al. AlexNet approach for early stage Alzheimer’s disease detection from MRI brain images[J]. Materials Today: Proceedings, 2022, 51: 58-65.

[48]Luo M, Tian Y, Zhang S, et al. Individual tree detection in coal mine afforestation area based on improved Faster RCNN in UAV RGB images[J]. Remote Sensing, 2022, 14(21): 5545.

[49]Yin G, Yu M, Wang M, et al. Research on highway vehicle detection based on faster R-CNN and domain adaptation[J]. Applied Intelligence, 2022, 52(4): 3483-3498.

[50]Xu Q, Kuang W, Zhang Z, et al. Sppnet: A single-point prompt network for nuclei image segmentation[C]//International Workshop on Machine Learning in Medical Imaging. Cham: Springer Nature Switzerland, 2023: 227-236.

[51]Zhang Y, Wang J, Yu Z, et al. Research on intelligent detection of coal gangue based on deep learning[J]. Measurement, 2022, 198: 111415.

[52]Tulbure A A, Tulbure A A, Dulf E H. A review on modern defect detection models using DCNNs–Deep convolutional neural networks[J]. Journal of Advanced Research, 2022, 35: 33-48.

[53]Redmon J, Divvala S, Girshick R, et al. You only look once: Unified, real-time object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 779-788.

[54]Bochkovskiy A, Wang C Y, Liao H Y M. Yolov4: Optimal speed and accuracy of object detection[J]. arXiv preprint arXiv:2004.10934, 2020.

[55]Yan P, Sun Q, Yin N, et al. Detection of coal and gangue based on improved YOLOv5. 1 which embedded scSE module[J]. Measurement, 2022, 188: 110530.

[56]Terven J, Cordova-Esparza D. A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NAS[J]. arXiv preprint arXiv:2304.00501, 2023.

[57]姚兰,赵永恒,施雨晴,等.一种基于视频分析的高速公路交通异常事件检测算法[J].计算机科学,2020,47(08):208-212.

[58]Revach G, Shlezinger N, Ni X, et al. KalmanNet: Neural network aided Kalman filtering for partially known dynamics[J]. IEEE Transactions on Signal Processing, 2022, 70: 1532-1547.

[59]Alam M Z, Jamalipour A. Multi-agent DRL-based Hungarian algorithm (MADRLHA) for task offloading in multi-access edge computing Internet of Vehicles (IoVs)[J]. IEEE Transactions on Wireless Communications, 2022, 21(9): 7641-7652.

[60]Xu X, Qiao H, Ma X, et al. An automatic wheat ear counting model based on the minimum area intersection ratio algorithm and transfer learning[J]. Measurement, 2023, 216: 112849.

[61]Yang J, Delpha C. An incipient fault diagnosis methodology using local Mahalanobis distance: Detection process based on empirical probability density estimation[J]. Signal Processing, 2022, 190: 108308.

[62]Liu J, Wang L, Shi Z. Dynamic modelling of the defect extension and appearance in a cylindrical roller bearing[J]. Mechanical Systems and Signal Processing, 2022, 173: 109040.

[63]Bewley A, Ge Z, Ott L, et al. Simple online and realtime tracking[C]//2016 IEEE international conference on image processing (ICIP). IEEE, 2016: 3464-3468.

[64]Yang D, Miao C, Liu Y, et al. Improved foreign object tracking algorithm in coal for belt conveyor gangue selection robot with YOLOv7 and DeepSORT[J]. Measurement, 2024: 114180.

[65]He L, Wu F, Du X, et al. Cascade-SORT: A robust fruit counting approach using multiple features cascade matching[J]. Computers and Electronics in Agriculture, 2022, 200: 107223.

[66]Zhang Y, Sun P, Jiang Y, et al. Bytetrack: Multi-object tracking by associating every detection box[C]//European conference on computer vision. Cham: Springer Nature Switzerland, 2022: 1-21.

[67]Zhang Y, Sun P, Jiang Y, et al. Bytetrack: Multi-object tracking by associating every detection box[C]//European conference on computer vision. Cham: Springer Nature Switzerland, 2022: 1-21.

[68]闫保永,曹柳,张家贵.煤层顶板裂隙带瓦斯抽采技术与装备探索[J].煤炭科学技术,2020,48(10):60-66.

[69]Xu P, Zhou Z, Geng Z. Safety monitoring method of moving target in underground coal mine based on computer vision processing[J]. Scientific reports, 2022, 12(1): 17899.

[70]Abdou M A. Literature review: Efficient deep neural networks techniques for medical image analysis[J]. Neural Computing and Applications, 2022, 34(8): 5791-5812.

[71]Zhang Y, Guo Z, Wu J, et al. Real-time vehicle detection based on improved yolo v5[J]. Sustainability, 2022, 14(19): 12274.

[72]Wang N, Wang Y, Wei Y, et al. Marine vessel detection dataset and benchmark for unmanned surface vehicles[J]. Applied Ocean Research, 2024, 142: 103835.

[73]Meng F, Qin Y, Wu Y, et al. A subtle defect recognition method for catenary fastener in high-speed railroad using destruction and reconstruction learning[J]. Advanced Engineering Informatics, 2024, 60: 102393.

[74]Lou H, Duan X, Guo J, et al. DC-YOLOv8: small-size object detection algorithm based on camera sensor[J]. Electronics, 2023, 12(10): 2323.

[75]Jin X, Xie Y, Wei X S, et al. Delving deep into spatial pooling for squeeze-and-excitation networks[J]. Pattern Recognition, 2022, 121: 108159.

[76]Karthik R, Vaichole T S, Kulkarni S K, et al. Eff2Net: An efficient channel attention-based convolutional neural network for skin disease classification[J]. Biomedical Signal Processing and Control, 2022, 73: 103406.

[77]Zeng Q, Zhou G, Wan L, et al. Detection of Coal and Gangue Based on Improved YOLOv8[J]. Sensors, 2024, 24(4): 1246.

[78]Dombi J, Jónás T. The generalized sigmoid function and its connection with logical operators[J]. International Journal of Approximate Reasoning, 2022, 143: 121-138.

[79]Park S, Yeo Y J, Shin Y G. PConv: simple yet effective convolutional layer for generative adversarial network[J]. Neural Computing and Applications, 2022, 34(9): 7113-7124.

[80]Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 2117-2125.

[81]Liu S, Qi L, Qin H, et al. Path aggregation network for instance segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 8759-8768.

[82]Tan M, Pang R, Le Q V. Efficientdet: Scalable and efficient object detection[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 10781-10790.

[83]Kang J S, Kang J K, Kim J J, et al. Neural architecture search survey: A computer vision perspective[J]. Sensors, 2023, 23(3): 1713.

[84]Tian Y, Su D, Lauria S, et al. Recent advances on loss functions in deep learning for computer vision[J]. Neurocomputing, 2022, 497: 129-158.

[85]Tong Z, Chen Y, Xu Z, et al. Wise-IoU: bounding box regression loss with dynamic focusing mechanism[J]. arXiv preprint arXiv:2301.10051, 2023.

[86]Zhang Y, Zhang W, Yu J, et al. Complete and accurate holly fruits counting using YOLOX object detection[J]. Computers and Electronics in Agriculture, 2022, 198: 107062.

[87]Usmani K, O’Connor T, Wani P, et al. 3D object detection through fog and occlusion: passive integral imaging vs active (LiDAR) sensing[J]. Optics Express, 2023, 31(1): 479-491.

[88]Maharjan P, Vanfossan L, Li Z, et al. Fast LoG SIFT Keypoint Detector[C]//2023 IEEE 25th International Workshop on Multimedia Signal Processing (MMSP). IEEE, 2023: 1-5.

[89]Barath D, Cavalli L, Pollefeys M. Learning to find good models in RANSAC[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022: 15744-15753.

[90]Kunis S, Nagel D, Strotmann A. Multivariate Vandermonde matrices with separated nodes on the unit circle are stable[J]. Applied and Computational Harmonic Analysis, 2022, 58: 50-59.