Research on fault detection of belt conveyor drum based on improved YOLOv5s
-
摘要: 针对目前带式输送机滚筒故障检测方法检测效率低、识别准确率不高、特征提取能力较差等问题,提出了一种基于改进YOLOv5s的带式输送机滚筒故障检测方法。在YOLOv5s网络模型中增加了小尺寸检测层,使尺寸较小的滚筒故障更易被检测到;在Backbone和Neck间引入卷积注意力机制(CBAM),提高目标检测的准确率;在Neck中引入高效通道注意力机制(ECA),增强对滚筒故障的特征提取能力。实验结果表明:① 在满足实时检测要求的前提下,改进YOLOv5s网络模型识别平均准确率均值达94.46%,较改进前提升了1.65%。② 改进YOLOv5s网络模型对滚筒开焊、包胶磨损和包胶脱落检测的平均准确率分别为95.29%,96.43%,91.65%,较改进前分别提高了1.56%,0.89%和2.50%。设计了基于改进YOLOv5s的带式输送机滚筒故障检测系统,并对该系统进行验证:①实验平台测试结果表明:基于改进YOLOv5s的带式输送机滚筒故障检测系统对滚筒开焊、包胶磨损和包胶脱落检测的平均准确率分别达95.29%,96.43%,91.65%,3种故障检测的平均准确率均值达94.46%,检测速度约为14帧/s。 ② 现场测试结果表明:包胶磨损和包胶脱落的置信度分别为0.92,0.97,且能准确地识别出滚筒的故障类型和位置,说明基于改进YOLOv5s的带式输送机滚筒故障检测系统具有可行性。Abstract: At present, the detection efficiency of belt conveyor drum fault detection methods is low, the recognition accuracy is not high, and the feature extraction capability is poor. In order to solve the above problems, a belt conveyor drum fault detection method based on improved YOLOv5s is proposed. A small-sized detection layer has been added to the YOLOv5s network model, making it easier to detect smaller drum faults. The method introduces the convolutional block attention module (CBAM) between the Backbone and Neck to improve the accuracy of target detection. The method introduces efficient channel attention mechanism (ECA) in Neck to enhance feature extraction capabilities for drum faults. The experimental results show the following points. ① On the premise of meeting the real-time detection requirements, the average recognition accuracy of the improved YOLOv5s network model reaches 94.46%, which is 1.65% higher than before the improvement. ② The average accuracy of the improved YOLOv5s network model for detecting drum opening, rubber coating wear, and rubber coating detachment are 95.29%, 96.43%, and 91.65%, respectively, which are 1.56%, 0.89%, and 2.50% higher than before the improvement. A belt conveyor drum fault detection system based on improved YOLOv5s is designed and validated. ① The experimental platform test results show that the average accuracy of the belt conveyor drum fault detection system based on improved YOLOv5s for drum welding, rubber coating wear, and rubber coating detachment detection reach 95.29%, 96.43%, and 91.65%, respectively. The average accuracy of the three types of faults reaches 94.46%, and the detection speed is about 14 frames/s. ② The on-site test results show that the confidence levels for rubber coating wear and rubber coating detachment are 0.92 and 0.97, respectively. The fault type and location of the drum can be accurately identified. This indicates that the improved YOLOv5s-based belt conveyor drum fault detection system is feasible.
-
表 1 训练超参数
Table 1. Training hyperparameterss
训练超参数 值 初始学习率 0.01 学习率衰减 0.000 1 动量 0.973 批处理大小 32 训练批次 300 图像输入尺寸 416×416 表 2 消融实验结果
Table 2. Ablation experimental results
改进策略 平均准确率均值/% 帧速率/(帧·s−1) 小尺寸检测层 CBAM ECA × × × 92.81 32.969 √ × × 93.76 26.366 × √ × 93.88 32.375 × × √ 93.87 31.238 √ √ √ 94.46 25.227 表 3 YOLOv5s网络模型改进前后性能对比
Table 3. Performance comparison before and after YOLOv5s network model improvement
模型 类别 平均
准确率/%平均
准确率均值/%参数量 YOLOv5s 滚筒开焊 93.73 92.81 7 066 239 包胶磨损 95.54 包胶脱落 89.15 改进YOLOv5s 滚筒开焊 95.29 94.46 7 210 354 包胶磨损 96.43 包胶脱落 91.65 表 4 带式输送机滚筒故障检测系统检测性能
Table 4. Detection performance of the belt conveyor drum falut detection system
% 类别 精确率 召回率 平均准确率 平均准确率均值 滚筒开焊 93.10 90.98 95.29
94.46包胶磨损 97.12 95.33 96.43 包胶脱落 98.21 90.85 91.65 -
[1] ANDREJIOVA M,GRINCOVA A,MARASOVA D. Measurement and simulation of impact wear damage to industrial conveyor belts[J]. Wear,2016,368:400-407. [2] 刘洋. 机器视觉的输送带纵向撕裂故障检测系统信号采集器的研究[D]. 天津: 天津工业大学, 2016.LIU Yang. Study on signal collector of conveyor belt longitudinal tear fault detection system for machine vision[D]. Tianjin: Tianjin Polytechnic University, 2016. [3] 韩越. 带式输送机驱动滚筒轴承故障特征提取分析研究[J]. 煤矿机械,2021,42(10):162-165.HAN Yue. Analysis and study on fault feature extraction of driving roller bearing of belt conveyor[J]. Coal Mine Machinery,2021,42(10):162-165. [4] 李丹宁,郑闯. 一种模糊神经网络的采煤机滚筒温度实时故障预警方法[J]. 煤炭科学技术,2021,49(增刊1):161-166.LI Danning,ZHENG Chuang. A real-time fault early warning method of shearer drum temperature based on fuzzy neural network[J]. Coal Science and Technology,2021,49(S1):161-166. [5] 张强. 基于新型检测方法的带式输送机滚筒故障诊断[J]. 机械管理开发,2022,37(6):144-145,151.ZHANG Qiang. Fault diagnosis of belt conveyor roller based on new detection method[J]. Mechanical Management and Development,2022,37(6):144-145,151. [6] 丁秀荣,薛正福,王芝兰. 矿用带式输送机滚筒故障检测系统应用研究[J]. 能源与环保,2022,44(4):205-210.DING Xiurong,XUE Zhengfu,WANG Zhilan. Application research on fault detection system of mine belt conveyor roller running[J]. China Energy and Environmental Protection,2022,44(4):205-210. [7] 李现国,李斌,刘宗鹏,等. 井下视频行人检测方法[J]. 工矿自动化,2020,46(2):54-58.LI Xianguo,LI Bin,LIU Zongpeng,et al. Underground video pedestrian detection method[J]. Industry and Mine Automation,2020,46(2):54-58. [8] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]. The 29th IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 2016: 779-788. [9] BOCHKOVSKIY A, WANG C Y, LIAO H Y, et al. YOLOv4: optimal speed and accuracy of object detection[EB/OL]. [2021-06-04]. https://arxiv.org/abs/2004.10934. [10] REDMON J, FARHADI A. YOLOv3: an incremental improvement[EB/OL]. [2021-06-04]. https://arxiv.org/abs/1804.02767. [11] SINGH S K. Multiple fault detection of rolling bearing through ensemble empirical mode decomposition of vibration signal[J]. International Journal of Engineering and Advanced Technology,2019,9(2):2724-2726. doi: 10.35940/ijeat.B3562.129219 [12] 潘杨,张守京,杨文彬. 基于改进YOLOv5的棉花异纤检测方法[J]. 棉纺织技术,2022,50(10):37-43.PAN Yang,ZHANG Shoujing,YANG Wenbin. Detection method of foreign fiber in cotton based on improved YOLOv5[J]. Cotton Textile Technology,2022,50(10):37-43. [13] 孙耀泽,高军伟. 基于改进YOLOv5的轮对踏面缺陷检测[J]. 激光与光电子学进展,2022,59(22):228-234.SUN Yaoze,GAO Junwei. Defect detection of wheel set tread based on improved YOLOv5[J]. Laser & Optoelectronics Progress,2022,59(22):228-234. [14] LIN T S, DOLLAR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]. IEEE Conference on Computer Vision and Pattern Recognition(CVPR), Honolulu, 2017: 936-944. [15] LIU Shu, QI Lu, QIN Haifeng, et al. Path aggregation network for instance segmentation[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 2018: 8759 -8768. [16] 章程军,胡晓兵,牛洪超. 基于改进YOLOv5的车辆目标检测研究[J]. 四川大学学报(自然科学版),2022,59(5):79-87.ZHANG Chengjun,HU Xiaobing,NIU Hongchao. Vehicle object detection based on improved YOLOv5 method[J]. Journal of Sichuan University(Natural Science Edition),2022,59(5):79-87. [17] 柏罗,张宏立,王聪. 基于高效注意力和上下文感知的目标跟踪算法[J]. 北京航空航天大学学报,2022,48(7):1222-1232.BAI Luo,ZHANG Hongli,WANG Cong. Target tracking algorithm based on efficient attention and context awareness[J]. Journal of Beijing University of Aeronautics and Astronautics,2022,48(7):1222-1232. [18] 袁祎铭,韩婷婷,丁佳骏,等. 基于高效通道注意力机制的龙格库塔去雨网络[J]. 计算机应用,2022,42(增刊1):305-309.YUAN Yiming,HAN Tingting,DING Jiajun,et al. Runge kutta network based on efficient channel attention mechanism for image deraining[J]. Journal of Computer Applications,2022,42(S1):305-309. [19] 韩兴,张红英,张媛媛. 基于高效通道注意力网络的人脸表情识别[J]. 传感器与微系统,2021,40(1):118-121. doi: 10.13873/J.1000-9787(2021)01-0118-04HAN Xing,ZHANG Hongying,ZHANG Yuanyuan. Facial expression recognition based on high efficient channel attention network[J]. Transducer and Microsystem Technologies,2021,40(1):118-121. doi: 10.13873/J.1000-9787(2021)01-0118-04 [20] 应宇航,任泰安,李伟,等. 一种基于Jetson Nano深度学习的生活垃圾智能分类桶[J]. 计算技术与自动化,2023,42(2):151-157.YING Yuhang,REN Tai'an,LI Wei,et al. A Kind of intelligent classified garbage bin based on Jetson Nano deep learning[J]. Computing Technology and Automation,2023,42(2):151-157. [21] 苏羽康,林鹏程,郭佳. 基于Jetson Nano的智能快递柜设计与实现[J]. 物联网技术,2022,12(7):53-54,58.SU Yukang,LIN Pengcheng,GUO Jia. Design and implementation of intelligent express cabinet based on Jetson Nano[J]. Internet of Things Technologies,2022,12(7):53-54,58. -