Intelligent detection method for coal flow foreign objects based on dual attention generative adversarial network
-
摘要:
在煤炭开采过程中混入的异物可能会导致输送带连接处堵塞甚至输送带撕裂等事故，现有的机器学习算法大多采用监督学习的方式自动识别物品类别，而在真实工矿场景下，异常样本稀缺，易导致建模数据集存在严重的样本分布不平衡且显著特征丢失的问题。针对上述问题，提出了一种基于双注意力生成对抗网络(DA−GANomaly)的煤流异物智能检测方法。该方法采用半监督学习的方式，仅需要正常样本完成异物检测模型训练，有效解决了因样本分布不平衡造成的识别精度低、鲁棒性差的问题；在Skip−GANomaly的基础上引入双注意力机制，增强了编码器与解码器之间的信息交流，以抑制无关特征和噪声，同时突出有利于区分异常样本的感兴趣特征，进一步提高模型分类的准确性。实验结果表明：DA−GANomaly模型的分类精确率为79.5%，召回率为83.2%，精确率−召回率曲线下面积（AUPRC）为85.1%；与AnoGAN等5种经典异常检测模型相比，DA−GANomaly模型的综合性能最佳。
Abstract:Foreign objects mixed in during coal mining may cause accidents such as blockage or even tearing of conveyor belt connections. Most existing machine learning algorithms for coal flow foreign objects use supervised learning to automatically recoginze item categories. However, in real industrial and mining scenarios, the scarcity of abnormal samples leads to problems of serious imbalanced sample distribution and significant features lost in the modeling dataset. In order to solve the above problems, a coal flow foreign object intelligent detection method based on dual-attention Skip-GANomaly (DA-GANomaly) is proposed. This method adopts a semi supervised learning approach, which only requires normal samples to complete the training of the foreign object detection model, effectively solving the problems of low recognition accuracy and poor robustness caused by imbalanced sample distribution. On the basis of Skip-GANomaly, a dual attention mechanism is introduced to enhance the information exchange between the encoder and decoder and suppress irrelevant features and noise. It highlights the interesting features that are conducive to distinguishing abnormal samples, and further improves the accuracy of model classification. The experimental results show that the classification accuracy of the DA-GANomaly model is 79.5%, the recall rate is 83.2%, and the area under the precision recall curve (AUPRC) is 85.1%. Compared with 5 classic anomaly detection models such as AnoGAN, the DA-GANomaly model has the best overall performance.
-
图 1 基于DA−GANomaly的煤流异物智能检测模型
Figure 1. Intelligent detection model for coal flow foreign objects based on DA-GANomaly
图 5 DA−GANomaly模型的异常分数分布直方图
Figure 5. Histogram of abnormal fraction distribution of DA-GANomaly model
表 1 生成器网络参数
Table 1. Generator network parameters
网络层级 M1 M2 M3 M4 M5 M6 N1 N2 N3 N4 N5 N6 卷积核尺寸 4×4 4×4 4×4 4×4 4×4 4×4 4×4 4×4 4×4 4×4 4×4 4×4 输出特征维度 64 128 256 512 512 512 512 512 256 128 64 3 输出特征图尺寸 32×32 16×16 8×8 4×4 2×2 1×1 2×2 4×4 8×8 16×16 32×32 64×64 
下载: 导出CSV
表 2 判别器网络参数
Table 2. Discriminator network parameters
网络层级 Q1 Q2 Q3 Q4 Q5 Q6 卷积核尺寸 4×4 4×4 4×4 4×4 4×4 4×4 输出特征维度 64 128 256 512 512 100 输出特征图尺寸 32×32 16×16 8×8 4×4 2×2 1×1
下载: 导出CSV
表 3 数据集划分
Table 3. Dataset partitioning
数据类型 训练集样本数/张 测试集样本数/张 正样本 14 000 600 负样本 0 107 总体样本 14 000 707
下载: 导出CSV
表 5 模型实时性测试结果
Table 5. Real time test results of the model
每秒浮点计算数/109 模型参数量/106个 单帧计算时间/ms 每秒计算帧数 5.30 32.8 7.2 138
下载: 导出CSV
-
[1] 程德强,徐进洋,寇旗旗,等. 融合残差信息轻量级网络的运煤皮带异物分类[J]. 煤炭学报,2022,47(3):1361-1369.CHENG Deqiang,XU Jinyang,KOU Qiqi,et al. Lightweight network based on residual information for foreign body classification on coal conveyor belt[J]. Journal of China Coal Society,2022,47(3):1361-1369. [2] 孟晓娟,张月琴,郝晓丽,等. 多分类深度卷积生成对抗网络的皮带撕裂检测[J]. 计算机工程与应用,2021,57(16):269-275.MENG Xiaojuan,ZHANG Yueqin,HAO Xiaoli,et al. Multi-class deep convolutional generative adversarial networks for belt tear detection[J]. Computer Engineering and Applications,2021,57(16):269-275. [3] RONG Dian,XIE Lijuan,YING Yibin. Computer vision detection of foreign objects in walnuts using deep learning[J]. Computers and Electronics in Agriculture,2019,162:1001-1010. doi: 10.1016/j.compag.2019.05.019 [4] 王卫东,张康辉,吕子奇,等. 基于深度学习的煤中异物机器视觉检测[J]. 矿业科学学报,2021,6(1):115-123.WANG Weidong,ZHANG Kanghui,LYU Ziqi,et al. Machine vision detection of foreign objects in coal using deep learning[J]. Journal of Mining Science and Technology,2021,6(1):115-123. [5] 王超,郝博南,张立亚,等. 基于RetinaNet的煤矿井下输送带异物检测技术[J]. 煤矿机械,2022,43(12):180-183.WANG Chao,HAO Bonan,ZHANG Liya,et al. Conveyor belt foreign object detection technology based on RetinaNet in underground coal mine[J]. Coal Mine Machinery,2022,43(12):180-183. [6] 王燕,郭潇樯,刘新华. 带式输送机大块异物视觉检测系统设计[J]. 机械科学与技术,2021,40(12):1939-1943.WANG Yan,GUO Xiaoqiang,LIU Xinhua. Design of visual detection system for large foreign body in belt conveyor[J]. Mechanical Science and Technology for Aerospace Engineering,2021,40(12):1939-1943. [7] 薛旭升,杨星云,齐广浩,等. 煤矿带式输送机分拣机器人异物识别与定位系统设计[J]. 工矿自动化,2022,48(12):33-41. doi: 10.13272/j.issn.1671-251x.2022100024XUE Xusheng,YANG Xingyun,QI Guanghao,et al. Design of foreign object recognition and positioning system for sorting robot of coal mine belt conveyor[J]. Journal of Mine Automation,2022,48(12):33-41. doi: 10.13272/j.issn.1671-251x.2022100024 [8] 郝帅,张旭,马旭,等. 基于CBAM−YOLOv5的煤矿输送带异物检测[J]. 煤炭学报,2022,47(11):4147-4156.HAO Shuai,ZHANG Xu,MA Xu,et al. Foreign object detection in coal mine conveyor belt based on CBAM-YOLOv5[J]. Journal of China Coal Society,2022,47(11):4147-4156. [9] 武红鑫,韩萌,陈志强,等. 监督和半监督学习下的多标签分类综述[J]. 计算机科学,2022,49(8):12-25.WU Hongxin,HAN Meng,CHEN Zhiqiang,et al. Survey of multi-label classification based on supervised and semi-supervised learning[J]. Computer Science,2022,49(8):12-25. [10] SCHLEGL T,SEEBÖCK P,WALDSTEIN S M,et al. Unsupervised anomaly detection with generative adversarial networks to guide marker discovery[C].25th International Conference on Information Processing in Medical Imaging,Boone,2017:146-157. [11] AKCAY S,ATAPOUR-ABARGHOUEI A,BRECKON T P. GANomaly:semi-supervised anomaly detection via adversarial training[C].14th Asian Conference on Computer Vision,Perth,2018:622-637. [12] AKÇAY S,ATAPOUR-ABARGHOUEI A,BRECKON T P. Skip-GANomaly:skip connected and adversarially trained encoder-decoder anomaly detection[C].International Joint Conference on Neural Networks,Budapest,2019:1-8. [13] RONNEBERGER O,FISCHER P,BROX T. U-net:convolutional networks for biomedical image segmentation[C].18th International Conference on Medical Image Computing and Computer-Assisted Intervention,Munich,2015:234-241. [14] 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):77-83.DU Jingyi,CHEN Rui,HAO Le,et al. Coal mine belt conveyor foreign object detection[J]. Industry and Mine Automation,2021,47(8):77-83. [15] 王涛. 机器视觉技术在煤矿胶带运输系统中的应用[J]. 能源科技,2021,19(2):34-40.WANG Tao. Application of the machine vision technology in the belt conveyor system for coal mines[J]. Energy Science and Technology,2021,19(2):34-40. [16] GUO Menghao,XU Tianxing,LIU Jiangjiang,et al. Attention mechanisms in computer vision:a survey[J]. Computational Visual Media,2022(3):331-368. [17] RAMACHANDRAN P,PARMAR N,VASWANI A,et al. Stand-alone self-attention in vision models[EB/OL]. [2023-03-22].https://arxiv.org/abs/1906.05909. [18] FU Jun,LIU Jing,TIAN Haijie,et al. Dual attention network for scene segmentation[C].IEEE/CVF Conference on Computer Vision and Pattern Recognition,Salt Lake City,2019:3146-3154. [19] ZOU Liang,YU Xinhui,LI Ming,et al. Nondestructive identification of coal and gangue via near-infrared spectroscopy based on improved broad learning[J]. IEEE Transactions on Instrumentation and Measurement,2020,69(10):8043-8052. [20] ZENATI H,FOO C S,LECOUAT B,et al. Efficient GAN-based anomaly detection[EB/OL]. [2023-03-22].https://arxiv.org/abs/1802.06222v1. [21] ZENATI H,ROMAIN M,FOO C S,et al. Adversarially learned anomaly detection[C].IEEE International Conference on Data Mining,Singapore,2018:727-736. -
点击查看大图
计量
- 文章访问数: 138
- HTML全文浏览量: 115
- PDF下载量: 14
- 被引次数: 0
