Lightweight multi-scale cross channel attention coal flow detection network
-
摘要: 为通过变频调速提高带式输送机运行效率，需要对带式输送机煤流进行检测。现有基于深度学习的带式输送机煤流检测方法难以在模型轻量化和分类准确度之间达到平衡，且很少考虑在特征提取过程中通道权重分布不平衡对检测准确度的影响。针对上述问题，提出了一种轻量化的多尺度跨通道注意力煤流检测网络，该网络由特征提取网络和分类网络组成。将轻量化的残差网络ResNet18作为特征提取网络，并在此基础上引入煤流通道注意力（CFCA）子网络，CFCA子网络采用多个卷积核大小不同的一维卷积，并对一维卷积的输出进行堆叠，以捕获特征图中不同尺度的跨通道交互关系，实现对特征图权重的重新分配，从而提高特征提取网络的语义表达能力。分类网络由3个全连接层构成，其将向量化的特征提取网络的输出作为输入，并对其进行非线性映射，最终得到“煤少”、“煤适中”、“煤多”3类结果的概率分布，通过将煤流检测问题转换为图像分类问题，避免瞬时煤流量波动过大导致带式输送机频繁变频调速的问题，提高带式输送机运行稳定性。实验结果表明，ResNet18+CFCA网络在几乎不增加网络参数量和计算复杂度的情况下，比ResNet18网络在分类准确率上提升了1.6%，可更加有效地区分图像中的前景信息，准确提取煤流特征。Abstract: In order to improve the operating efficiency of belt conveyors through variable frequency speed regulation, it is necessary to detect the coal flow of belt conveyor. The existing deep learning-based coal flow detection methods for belt conveyors are difficult to achieve a balance between model lightweight and classification accuracy. There are few researches on the impact of imbalanced channel weight distribution on detection accuracy in the feature extraction process. In order to solve the above problems, a lightweight multi-scale cross channel attention coal flow detection network is proposed. The network consists of a feature extraction network and a classification network. The lightweight residual network ResNet18 is used as the feature extraction network, and on this basis, the coal flow channel attention (CFCA) subnetwork is introduced. The CFCA subnetwork uses multiple one-dimensional convolutions with different kernel sizes, and stacks the output of one-dimensional convolution to capture cross channel interaction relationships at different scales in the feature map. It achieves the reassignment of feature map weights, thereby improving semantic expression capability of the feature extraction network. The classification network consists of three fully connected layers, which take the output of the vectorized feature extraction network as input and perform nonlinear mapping on it. It ultimately obtains the probability distribution of three types of results: "little coal", "moderate coal", and "much coal". By transforming the coal flow detection problem into an image classification problem, the problem of frequent frequency conversion and speed regulation of belt conveyors caused by excessive fluctuations in instantaneous coal flow is avoided. It improves stability of belt conveyor operation. The experimental results show that the ResNet18+CFCA network improves classification accuracy by 1.6% compared to the ResNet18 network, with almost no increase in network parameters and computational complexity. It can distinguish foreground information in images more effectively and accurately extract coal flow features.
-
图 2 轻量化的多尺度跨通道注意力煤流检测网络结构
Figure 2. Structure of lightweight multi-scale cross channel attention coal flow detection network
图 4 不同网络训练时的损失函数曲线
Figure 4. Loss function curves of different networks during training
表 1 CFCA子网络采用不同大小卷积核的对比结果
Table 1. Comparison results of coal flow channel attention subnetwork using different sizes of convolution kernels
网络 准确率/% ResNet18 93.0 ResNet18+CFCA（K=3） 93.6 ResNet18+CFCA（K=3，5） 93.7 ResNet18+CFCA（K=3，5，7） 94.0 ResNet18+CFCA（K=3，5，7，9） 93.8 
下载: 导出CSV
表 2 CFCA子网络嵌入到不同卷积层的对比结果
Table 2. Comparison results of coal flow channel attention subnetwork embedded in different convolutional layers
CFCA子网络嵌入卷积层的位置 准确率/% Conv1与 Conv2之间 93.5 Conv2与 Conv3之间 93.8 Conv3与 Conv4之间 94.3 Conv4之后 94.0 Conv3与 Conv4之间及Conv4之后 94.6
下载: 导出CSV
表 3 不同网络性能对比结果
Table 3. Performance comparison of different networks
网络 参数量/MiB 运算量/GiB 准确率/% ResNet18 11.178 051 6.282 886 93.0 ResNet18+ECANet 11.219 827 6.283 150 93.8 ResNet18+SENet 11.178 057 6.283 188 93.6 ResNet18+ CFCA 11.178 069 6.283 160 94.6
下载: 导出CSV
-
[1] 王海军,王洪磊. 带式输送机智能化关键技术现状与展望[J]. 煤炭科学技术,2022,50(12):225-239.WANG Haijun,WANG Honglei. Status and prospect of intelligent key technologies of belt conveyor[J]. Coal Science and Technology,2022,50(12):225-239. [2] 张志军. 基于变频技术的煤矿带式输送机节能控制系统研究[J]. 能源与环保,2023,45(7):210-215.ZHANG Zhijun. Research on energy-saving control system of coal mine belt conveyor based on frequency conversion technology[J]. China Energy and Environmental Protection,2023,45(7):210-215. [3] 李阿红. 基于图像处理的矿用带式输送机系统设计及其能效分析[J]. 能源与环保,2023,45(7):228-232,238.LI Ahong. Design and energy efficiency analysis of mining belt conveyor system based on image processing[J]. China Energy and Environmental Protection,2023,45(7):228-232,238. [4] 王志广,王朋飞. 变频控制系统在矿用带式输送机中的应用[J]. 能源与环保,2021,43(9):255-259,265.WANG Zhiguang,WANG Pengfei. Application of frequency conversion control system in belt conveyor used in mine[J]. China Energy and Environmental Protection,2021,43(9):255-259,265. [5] 李亚飞. 变频节能控制系统在带式输送机中的应用研究[J]. 石化技术,2023,30(6):297-298. doi: 10.3969/j.issn.1006-0235.2023.06.103LI Yafei. Research on the application of frequency conversion and energy saving control system in belt conveyor[J]. Petrochemical Industry Technology,2023,30(6):297-298. doi: 10.3969/j.issn.1006-0235.2023.06.103 [6] 于林海. 带式输送机能耗优化控制系统设计与应用[J]. 石化技术,2023,30(5):249-250. doi: 10.3969/j.issn.1006-0235.2023.05.085YU Linhai. Design and application of optimised control system for energy consumption of belt conveyor[J]. Petrochemical Industry Technology,2023,30(5):249-250. doi: 10.3969/j.issn.1006-0235.2023.05.085 [7] 王文龙. 煤矿带式输送机智能化控制系统技术应用[J]. 机械管理开发,2023,38(5):217-218,223.WANG Wenlong. Intelligent control system technology application for coal mine belt conveyor[J]. Mechanical Management and Development,2023,38(5):217-218,223. [8] 王桂忠,叶隆浩. 基于煤流量识别的带式输送机节能控制系统设计与研究[J]. 煤矿机械,2023,44(1):14-17.WANG Guizhong,YE Longhao. Design and research of belt conveyor energy-saving control system based on coal flow recognition[J]. Coal Mine Machinery,2023,44(1):14-17. [9] 张少宾,蒋卫良,芮丰. 矿用带式输送机输送量测量方法现状及发展趋势[J]. 工矿自动化,2019,45(5):100-103. doi: 10.13272/j.issn.1671-251x.2019010055ZHANG Shaobin,JIANG Weiliang,RUI Feng. Current status and development trend of measuring methods of conveying capacity of mine-used belt conveyor[J]. Industry and Mine Automation,2019,45(5):100-103. doi: 10.13272/j.issn.1671-251x.2019010055 [10] 姜玉峰,张立亚,李标,等. 基于单线激光雷达的带式输送机煤流量检测研究[J]. 煤矿机械,2022,43(8):151-153.JIANG Yufeng,ZHANG Liya,LI Biao,et al. Study on coal flow detection of belt conveyor based on single-line LiDAR[J]. Coal Mine Machinery,2022,43(8):151-153. [11] 于海里,孙立超,左胜,等. 基于双激光雷达的带式输送机煤流量检测系统[J]. 工矿自动化,2023,49(7):27-34,59.YU Haili,SUN Lichao,ZUO Sheng,et al. Coal flow detection system for belt conveyor based on dual lidar[J]. Journal of Mine Automation,2023,49(7):27-34,59. [12] 关丙火. 基于激光扫描的带式输送机瞬时煤量检测方法[J]. 工矿自动化,2018,44(4):20-24.GUAN Binghuo. Detection method of instantaneous coal quantity of belt conveyor based on laser scanning[J]. Industry and Mine Automation,2018,44(4):20-24. [13] 崔振国. 基于机器视觉的带式输送机煤量监测系统研究[D]. 徐州: 中国矿业大学, 2021.CUI Zhenguo. Study on coal quantity monitoring system of belt conveyor based on machine vision[D]. Xuzhou: China University of Mining and Technology, 2021. [14] 李学晖. 基于机器视觉和深度学习的带式输送机煤量识别方法研究[D]. 邯郸: 河北工程大学, 2022.LI Xuehui. Research on coal quantity identification method of belt conveyor based on machine vision and deep learning[D]. Handan: Hebei University of Engineering, 2022. [15] 袁敦鹏. 基于三维点云的带式输送机跑偏及堆煤监测技术研究[D]. 徐州: 中国矿业大学, 2022.YUAN Dunpeng. Study on the belt conveyor deviation and coal stacking monitoring based on three-dimensional point cloud[D]. Xuzhou: China University of Mining and Technology, 2022. [16] 王桂梅,李学晖,杨立洁,等. 基于深度学习的永磁直驱带式输送机煤量检测方法研究[J]. 煤炭技术,2022,41(1):188-190.WANG Guimei,LI Xuehui,YANG Lijie,et al. Research on coal quantity detection method of permanent magnet direct drive belt conveyor based on deep learning[J]. Coal Technology,2022,41(1):188-190. [17] 成彦颖. 煤矿井下传送带智能输煤检测的研究[D]. 太原: 太原科技大学, 2021.CHENG Yanying. Research on intelligent coal conveying detection of underground conveyor belt in coal mine[D]. Taiyuan: Taiyuan University of Science and Technology, 2021. [18] XI Tianyu,WANG Jiangning,HAN Yan,et al. Multiple butterfly recognition based on deep residual learning and image analysis[J]. Entomological Research,2022,52(1):44-53. doi: 10.1111/1748-5967.12564 [19] HU Jie, SHEN Li, SUN Gang. Squeeze-and-excitation networks[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 2018. DOI: 10.1109/CVPR.2018.00745. [20] WANG Qilong, WU Banggu, ZHU Pengfei, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, 2020. DOI: 10.1109/CVPR42600.2020.01155. [21] WANG Chengyang,ZHONG Caiming. Adaptive feature pyramid networks for object detection[J]. IEEE Access,2021,9:107024-107032. doi: 10.1109/ACCESS.2021.3100369 -
点击查看大图
图(5) / 表(3)
计量
- 文章访问数: 163
- HTML全文浏览量: 49
- PDF下载量: 21
- 被引次数: 0
