Fault diagnosis of mining rolling bearings based on Superlet Transform and OD-ConvNeXt-ELA
-
摘要:
针对现有矿用滚动轴承故障诊断方法存在特征提取能力有限、泛化性欠佳的问题，提出了一种基于超小波变换（SLT）与OD−ConvNeXt−ELA的矿用滚动轴承故障诊断方法。以ConvNeXt−T为基础，引入批归一化（BN）技术以提高网络的泛化性，使用全维动态卷积（ODConv）替换原有的深度可分离卷积，以提高网络的适应性，引入高效局部注意力（ELA）以使网络聚焦关键位置特征，构建了矿用滚动轴承故障诊断OD−ConvNeXt−ELA网络模型；为充分利用OD−ConvNeXt−ELA网络模型的图像特征提取能力，选用SLT将采集的滚动轴承一维振动信号转换为二维时频图像后输入OD−ConvNeXt−ELA进行模型训练。选用凯斯西储大学（CWRU）和帕德博恩大学（PU）轴承数据集进行故障诊断实验，结果表明：对于单一工况下的CWRU轴承数据集，OD−ConvNeXt−ELA平均故障诊断准确率为99.65%，较ConvNeXt−T提高了1.61%；对于跨工况下的CWRU轴承数据集，OD−ConvNeXt−ELA平均故障诊断准确率为87.50%，较ConvNeXt−T提高了3.30%；对于跨工况下的PU轴承数据集，OD−ConvNeXt−ELA平均故障诊断准确率为89.33%，较ConvNeXt−T提高了3.46%；基于SLT与OD−ConvNeXt−ELA的矿用滚动轴承故障诊断方法在跨轴承、跨工况及噪声干扰下具有准确率高、泛化能力强的优势。
Abstract:In response to the limitations of current fault diagnosis methods for mining rolling bearings, which suffer from limited feature extraction capabilities and poor generalization, a fault diagnosis method based on Superlet Transform (SLT) and OD-ConvNeXt-ELA was proposed. Built upon ConvNeXt-T, Batch Normalization (BN) technology was introduced to improve the network's generalization ability. Omni-dimensional Dynamic Convolution (ODConv) replaced the original depthwise separable convolution to enhance the adaptability of the network. Efficient Local Attention (ELA) was incorporated to focus the network on key feature locations. This formed the OD-ConvNeXt-ELA network model for fault diagnosis of mining rolling bearings. To fully leverage the image feature extraction ability of the OD-ConvNeXt-ELA model, SLT was used to convert the collected one-dimensional vibration signal of the rolling bearing into a two-dimensional time-frequency image, which was then input into the OD-ConvNeXt-ELA for model training. Fault diagnosis experiments were conducted using the bearing datasets from Case Western Reserve University (CWRU) and Paderborn University (PU). The results showed that for the CWRU bearing dataset under a single operating condition, the average fault diagnosis accuracy of OD-ConvNeXt-ELA was 99.65%, which was an improvement of 1.61% over ConvNeXt-T. For the CWRU bearing dataset under cross-operating conditions, the average fault diagnosis accuracy of OD-ConvNeXt-ELA was 87.50%, which was an improvement of 3.30% over ConvNeXt-T. For the PU bearing dataset under cross-operating conditions, the average fault diagnosis accuracy of OD-ConvNeXt-ELA was 89.33%, an improvement of 3.46% over ConvNeXt-T. The fault diagnosis method based on SLT and OD-ConvNeXt-ELA shows high accuracy and strong generalization ability under cross-bearing, cross-operating conditions, and noise interference.
-
-
![]()
图 3 不同时频变换方法下的时频图像
Figure 3. Time-frequency images with different time-frequency transform methods
![]()
图 4 单一工况下不同网络在CWRU轴承数据集的故障诊断准确率
Figure 4. Fault diagnosis accuracy of different networks on CWRU bearing dataset under a single operating condition
![]()
图 5 单一工况下不同网络的故障诊断准确率和损失值曲线
Figure 5. Fault diagnosis accuracy and loss curves of different networks under a single operating condition
![]()
图 6 跨工况下不同网络在CWRU轴承数据集上的故障诊断准确率
Figure 6. Fault diagnosis accuracy of different networks on CWRU bearing dataset across operating conditions
![]()
图 7 跨工况下不同网络在CWRU轴承数据集上的混淆矩阵
Figure 7. Confusion matrix of different networks on CWRU bearing dataset across operating conditions
![]()
图 8 跨工况下不同网络分类结果可视化
Figure 8. Visualization of classification results for different networks across operating conditions
![]()
图 9 跨工况下不同网络在PU轴承数据集上的故障诊断准确率
Figure 9. Fault diagnosis accuracy of different networks on PU bearing dataset across operating conditions
![]()
图 10 跨工况下不同网络在PU轴承数据集上的混淆矩阵
Figure 10. Confusion matrix of different networks on PU bearing dataset across operating conditions
![]()
图 11 不同信噪比下不同网络故障诊断准确率
Figure 11. Fault diagnosis accuracy of different networks at different signal-to-noise ratios
表 1 CWRU轴承数据集工况划分
Table 1 Division of operating conditions in CWRU bearing dataset
工况 负载功率/W 转速/（r·min−1） A 0 1 797 B 746 1 772 C 1 491 1 750 D 2 237 1 730 
下载: 导出CSV
表 2 CWRU轴承数据集故障类型划分
Table 2 Classification of fault types in CWRU bearing dataset
轴承状态 故障直径/mm 样本个数 标签 健康 0 200 1 内圈故障 0.177 8 200 2 滚动体故障 0.177 8 200 3 外圈故障 0.177 8 200 4 内圈故障 0.355 6 200 5 滚动体故障 0.355 6 200 6 外圈故障 0.355 6 200 7 内圈故障 0.533 4 200 8 滚动体故障 0.533 4 200 9 外圈故障 0.533 4 200 10
下载: 导出CSV
表 3 不同时频变换方法下轴承故障诊断准确率
Table 3 Bearing fault diagnosis accuracy using different time-frequency transform methods
时频变换方法 准确率/% A工况 B工况 C工况 D工况 STFT 95.57 94.69 96.15 98.18 CWT 96.10 93.93 97.46 97.44 SLT 97.29 97.60 99.74 99.12
下载: 导出CSV
表 4 PU轴承数据集工况划分
Table 4 Division of operating conditions in PU bearing dataset
工况 转矩/（N·m） 径向力/N E 0.7 1 000 F 0.1 1 000 G 0.7 400
下载: 导出CSV
表 5 PU轴承数据集故障类型划分
Table 5 Classification of fault types in PU bearing dataset
轴承状态 故障等级 样本个数 标签 健康 — 400 I 外圈故障 1级 400 II 内圈故障 1级 400 III 复合故障 2级 400 IV
下载: 导出CSV
-
[1] 杨春才,李向磊,吕晓伟. 煤机设备轴承故障诊断方法[J]. 工矿自动化,2023,49(12):147-151. YANG Chuncai,LI Xianglei,LYU Xiaowei. Diagnosis method for bearing faults in coal mining equipment[J]. Journal of Mine Automation,2023,49(12):147-151.
[2] 韩争杰,牛荣军,马子魁,等. 基于注意力机制改进残差神经网络的轴承故障诊断方法[J]. 振动与冲击,2023,42(16):82-91. HAN Zhengjie,NIU Rongjun,MA Zikui,et al. Bearing fault diagnosis methods based on an attentional-mechanism-improved residual neural network[J]. Journal of Vibration and Shock,2023,42(16):82-91.
[3] 郭俊锋,谭宝宏,王智明. 基于MDAM−GhostCNN的滚动轴承故障诊断方法[J/OL]. 北京航空航天大学学报:1-15[2024-05-13]. https://doi.org/10.13700/j.bh.1001-5965.2023.0224. GUO Junfeng,TAN Baohong,WANG Zhiming. Fault diagnosis method of rolling bearing based on MDAM-GhostCNN[J/OL]. Journal of Beijing University of Aeronautics and Astronautics:1-15[2024-05-13]. https://doi.org/10.13700/j.bh.1001-5965.2023.0224.
[4] 雷春丽,夏奔锋,薛林林,等. 基于MTF−CNN的滚动轴承故障诊断方法[J]. 振动与冲击,2022,41(9):151-158. LEI Chunli,XIA Benfeng,XUE Linlin,et al. Rolling bearing fault diagnosis method based on MTF-CNN[J]. Journal of Vibration and Shock,2022,41(9):151-158.
[5] 常淼,沈艳霞. 基于改进卷积神经网络的风电轴承故障诊断策略[J]. 电力系统保护与控制,2021,49(6):131-137. CHANG Miao,SHEN Yanxia. Fault diagnosis strategy of a wind power bearing based on an improved convolutional neural network[J]. Power System Protection and Control,2021,49(6):131-137.
[6] 邓飞跃,吕浩洋,顾晓辉,等. 基于轻量化神经网络Shuffle−SENet的高速动车组轴箱轴承故障诊断方法[J]. 吉林大学学报(工学版),2022,52(2):474-482. DENG Feiyue,LYU Haoyang,GU Xiaohui,et al. Fault diagnosis of high-speed train axle bearing based on a lightweight neural network Shuffle-SENet[J]. Journal of Jilin University (Engineering and Technology Edition),2022,52(2):474-482.
[7] LIU Zhuang,MAO Hanzi,WU Chaoyuan,et al. A ConvNet for the 2020s[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition,New Orleans,2022:11966-11976.
[8] 刘立波,郗思宇,邓箴. 结合改进ConvNeXt网络与知识蒸馏的天气识别[J]. 光学精密工程,2023,31(14):2123-2134. DOI: 10.37188/OPE.20233114.2123 LIU Libo,XI Siyu,DENG Zhen. Weather recognition combining improved ConvNeXt models with knowledge distillation[J]. Optics and Precision Engineering,2023,31(14):2123-2134. DOI: 10.37188/OPE.20233114.2123
[9] 杨环宇,王军,吴祥,等. 一种坐标通道注意力深度学习网络的军用飞机识别方法[J]. 兵工学报,2024,45(7):2128-2143. YANG Huanyu,WANG Jun,WU Xiang,et al. A method for military aircraft recognition using a coordinate attention-based deep learning network[J]. Acta Armamentarii,2024,45(7):2128-2143.
[10] 郭盼盼,张文斌,崔奔,等. 基于增强深度卷积神经网络的滚动轴承多工况故障诊断方法[J/OL]. 振动工程学报:1-14[2024-06-25]. https://kns.cnki.net/kcms/detail/32.1349.TB.20230830.1129.002.html. GUO Panpan,ZHANG Wenbin,CUI Ben,et al. Fault diagnosis method of rolling bearing under multiple working conditions based on enhanced deep convolutional neural network[J/OL]. Journal of Vibration Engineering:1-14[2024-06-25]. https://kns.cnki.net/kcms/detail/32.1349.TB.20230830.1129.002.html.
[11] SEGU M,TONIONI A,TOMBARI F. Batch normalization embeddings for deep domain generalization[J]. Pattern Recognition,2023,135. DOI: 10.1016/j.patcog.2022.109115.
[12] IOFFE S,SZEGEDY C,IOFFE S,et al. Batch normalization:accelerating deep network training by reducing internal covariate shift[C]. 32nd International Conference on Machine Learning,Lille,2015:448-456.
[13] LI Yufei,XIN Yufei,LI Xinni,et al. Omni-dimensional dynamic convolution feature coordinate attention network for pneumonia classification[J]. Visual Computing for Industry,Biomedicine,and Art,2024,7(1):17-29. DOI: 10.1186/s42492-024-00168-5
[14] XU Wei,WAN Yi. ELA:efficient local attention for deep convolutional neural networks[EB/OL]. [2024-06-25]. https://arxiv.org/abs/2403.01123v1.
[15] MOCA V V,BÂRZAN H,NAGY-DĂBÂCAN A,et al. Time-frequency super-resolution with superlets[J]. Nature Communications,2021,12(1). DOI: 10.1038/s41467-020-20539-9.
[16] 曾志超,徐玥,王景玉,等. 基于SOE−YOLO轻量化的水面目标检测算法[J]. 图学学报,2024,45(4):736-744. ZENG Zhichao,XU Yue,WANG Jingyu,et al. A water surface target detection algorithm based on SOE-YOLO lightweight network[J]. Journal of Graphics,2024,45(4):736-744.
[17] 段洁利,于世伟,解明坤,等. 基于一维轻量化CNN的山地索道轴承故障诊断[J]. 农业工程学报,2023,39(14):70-79. DOI: 10.11975/j.issn.1002-6819.202304173 DUAN Jieli,YU Shiwei,XIE Mingkun,et al. Fault diagnosis of mountain ropeway bearings based on one-dimensional lightweight CNN[J]. Transactions of the Chinese Society of Agricultural Engineering,2023,39(14):70-79. DOI: 10.11975/j.issn.1002-6819.202304173
[18] SMITH W A,RANDALL R B. Rolling element bearing diagnostics using the Case Western Reserve University data:a benchmark study[J]. Mechanical Systems and Signal Processing,2015,64:100-131.
[19] LESSMEIER C,KIMOTHO J K,ZIMMER D,et al. Condition monitoring of bearing damage in electromechanical drive systems by using motor current signals of electric motors:a benchmark data set for data-driven classification[C]. PHM Society European Conference,Bilbao,2016. DOI: 10.36001/phme.2016.v3i1.1577.
[20] 窦桂东,白艺硕,王均利,等. 基于深度学习的矿井滚动轴承故障诊断方法[J]. 工矿自动化,2024,50(1):96-103,154. DOU Guidong,BAI Yishuo,WANG Junli,et al. A fault diagnosis method for mine rolling bearings based on deep learning[J]. Journal of Mine Automation,2024,50(1):96-103,154.
[21] LIU Chunyang,BAN Yuxuan,LI Hongyu,et al. Accurate recognition method for rolling bearing failure of mine hoist in strong noise environment[J]. Machines,2023,11(6). DOI: 10.3390/machines11060632.
计量
- 文章访问数: 55
- HTML全文浏览量: 22
- PDF下载量: 14
