Rolling bearing fault diagnosis method based on SConvNeXt-ECMS and DBO-RELM model
-
摘要:
针对现有基于深度学习的滚动轴承故障诊断方法存在准确度不高、泛化性较差的缺点，提出了一种基于SConvNeXt-ECMS（the ConvNeXt network based on shuffled convolution-efficient channel and multi-scale spatial attention module）与DBO-RELM（dung beetleoptimizer regularized extreme learning machine）的滚动轴承故障诊断模型。将ECMS注意力机制与分流卷积模块融入ConvNeXt网络，提升ConvNeXt网络的特征提取能力；使用蜣螂优化算法完成参数寻优后的RELM替换网络原有分类层，提升网络对相近特征的分辨能力；利用哈尔滨工业大学航空轴承故障数据集仿真实验，验证所提分流卷积对ConvNeXt网络的提升效果；使用帕德博恩大学数据集进行滚动轴承混合故障诊断实验，验证所提SConvNeXt-ECMS与DBO-RELM模型的分类效果。仿真实验结果表明：所提SConvNeXt网络在航空轴承故障分类任务中，准确率可达100%，优于其他现有网络；帕德博恩大学滚动轴承混合故障诊断实验表明，所提ECMS注意力机制以及DBO-RELM方法均对原网络的性能有进一步的提升，新模型对滚动轴承混合故障的诊断准确率最高可达99.94%，相较于其他现有的滚动轴承故障诊断模型，均具有更高的故障诊断准确率和更强的泛化能力。
-
关键词:
- 故障诊断 /
- 滚动轴承 /
- 分流卷积 /
- 注意力机制 /
- 正则化极限学习机（RELM） /
- 蜣螂优化算法（DBO）
Abstract:In view of the existing shortcomings of deep learning-based rolling bearing fault diagnosis methods such as low accuracy and poor generalization, a rolling bearing fault diagnosis model based on SConvNeXt-ECMS （the ConvNeXt network based on shuffled convolution-efficient channel and multi-scale spatial attention module） and DBO-RELM （dung beetle optimizer regularized extreme learning machine） was proposed. The ECMS attention mechanism and shunt convolution module were integrated into the ConvNeXt network to improve the feature extraction capability of the ConvNeXt network. The RELM after parameter optimization by using the dung beetle optimization algorithm was used to replace the original classification layer of the network to improve the network’s resolution of similar features. The aviation bearing fault data set simulation experiment from Harbin Institute of Technology was used to verify the improvement effect of the proposed shunt convolution on the ConvNeXt network. The data set from the University of Paderborn was used to conduct a rolling bearing hybrid fault diagnosis experiment to verify the classification effect of the SConvNeXt-ECMS and DBO-RELM models proposed by the author. Simulation experiment results showed that the proposed SConvNeXt network had an accuracy of up to 100% in the aviation bearing fault classification task, which was better than other existing networks; the rolling bearing hybrid fault diagnosis test showed that the ECMS attention mechanism and the DBO-RELM method had further improved the performance of the original network. The new model can diagnose rolling bearing hybrid faults with a maximum accuracy of 99.94%. Compared with other existing rolling bearing fault diagnosis models, it has higher fault diagnosis accuracy and stronger generalization ability.
-
图 1 Morlet小波变换生成时频图
Figure 1. Morlet wavelet transform to generate time-frequency diagram
图 8 不同网络对航空轴承故障数据集识别准确率
Figure 8. Accuracy of identifying aviation bearing fault datasets using different networks
图 9 不同网络对帕德博恩数据集识别准确率
Figure 9. Recognition accuracy of different networks on the Paderborn dataset
图 10 不同注意力模型对帕德博恩数据集识别准确率
Figure 10. Recognition accuracy of different attention models on the Paderborn dataset
图 11 对测试集的预测混淆矩阵和分类效果
Figure 11. Prediction confusion matrix and classification effect on the test set
图 12 不同分类算法对帕德博恩数据集识别准确率
Figure 12. Recognition accuracy of different classification algorithms for Paderborn dataset
图 13 各分类算法对测试集的分类效果和预测混淆矩阵
Figure 13. Classification effect of each classification algorithm on test set and the prediction of confusion matrix
表 1 混合域可分离卷积与深度可分离卷积参数表
Table 1. Mixed-domain separable convolution and depth-separable convolution parameter table
参数 混合域可分离卷积 深度可分离卷积 输入大小 $ {W_1} \times {H_1} $ $ {W_1} \times {H_1} $ 核大小 $ （k\times 1,1\times k） $ $ k \times k $ 卷积次数 2 1 输出大小 $ {W_2} \times {H_2} $ $ {W_2} \times {H_2} $ 计算量 $ （2{H_1} - 2） （{W_1} - 2） \times k $ $ （{H_1} - 2） （{W_1} - 2） \times {k^2} $ 注：表中W、H分别表示特征的宽和高；k代表卷积核大小。 
下载: 导出CSV
表 2 普通卷积与膨胀卷积参数对照
Table 2. Comparison of ordinary convolution and dilated convolution parameters
参数 普通卷积 膨胀卷积 输入大小 $ {W_1} \times {H_1} \times {C_1} $ $ {W_1} \times {H_1} \times {C_1} $ 卷积核大小 $ k \times k $ $ k \times k $ 膨胀因子 0 2 感受野 $ {k^2} $ $ { （k + 2） ^2} $ 参数量 $ {k^2}{C_1}{C_2} $ $ {k^2}{C_1}{C_2} $ 注：表中C1为卷积过程中输入特征的通道数，C2为卷积过程中输出特征的通道数。
下载: 导出CSV
表 3 航空轴承数据集样本分布
Table 3. Sample distribution of aviation bearing dataset
故障类型 损伤等级 标签 样本数 健康 0 1200 内圈 1级 1 1200 外圈 1级 2 1200 外圈 2级 3 1200
下载: 导出CSV
表 4 滚动轴承故障样本分布
Table 4. Sample distribution of rolling bearing faults
实验工况 故障类型 代号 损伤等级 标签 样本数 训练集样本数 测试集样本数 N09-M07-F10 健康 K001 0 1200 960 240 外圈 KA04 1级 1 1200 960 240 KA16 2级 2 1200 960 240 混合 KB23 1级 3 1200 960 240 KB27 2级 4 1200 960 240 内圈 KI04 1级 5 1200 960 240 KI18 2级 6 1200 960 240
下载: 导出CSV
表 5 分类算法参数设置
Table 5. Classification algorithm parameter settings
分类器 参数设置 KNN 近邻参数设置为5 PSO-SVM 局部搜索能力为2，全局搜索能力为2，
粒子群数为10，迭代次数为100GWO-KELM 灰狼种群规模为2，终止迭代为100 JS-KELM 水母种群规模为10，终止迭代为100 DBO-RELM 蜣螂种群规模为30，所有代理比例为
6∶6∶7∶11终止迭代为100
下载: 导出CSV
-
[1] 曹蔚,苟臻元,韩昭,等. 基于GWO-TVF-EMD方法的行星齿轮箱齿面剥落故障诊断[J]. 航空动力学报,2021,36(11): 2418-2429. CAO Wei,GOU Zhenyuan,HAN Zhao,et al. Fault diagnosis of tooth surface spalling of planetary gearbox based on GWO-TVF-EMD method[J]. Journal of Aerospace Power,2021,36(11): 2418-2429. (in ChineseCAO Wei, GOU Zhenyuan, HAN Zhao, et al. Fault diagnosis of tooth surface spalling of planetary gearbox based on GWO-TVF-EMD method[J]. Journal of Aerospace Power, 2021, 36(11): 2418-2429. (in Chinese) [2] 丁闯,张兵志,冯辅周,等. 局部均值分解和排列熵在行星齿轮箱故障诊断中的应用[J]. 振动与冲击,2017,36(17): 55-60. DING Chuang,ZHANG Bingzhi,FENG Fuzhou,et al. Application of local mean decomposition and permutation entropy in fault diagnosis of planetary gearboxes[J]. Journal of Vibration and Shock,2017,36(17): 55-60. (in ChineseDING Chuang, ZHANG Bingzhi, FENG Fuzhou, et al. Application of local mean decomposition and permutation entropy in fault diagnosis of planetary gearboxes[J]. Journal of Vibration and Shock, 2017, 36(17): 55-60. (in Chinese) [3] LAZAKIS I,DIKIS K,MICHALA A L,et al. Advanced ship systems condition monitoring for enhanced inspection,maintenance and decision making in ship operations[J]. Transportation Research Procedia,2016,14: 1679-1688. doi: 10.1016/j.trpro.2016.05.133 [4] VERSTRAETE D,FERRADA A,DROGUETT E L,et al. Deep learning enabled fault diagnosis using time-frequency image analysis of rolling element bearings[J]. Shock and Vibration,2017,2017: 5067651. [5] HE Zhiyi,SHAO Haidong,DING Ziyang,et al. Modified deep autoencoder driven by multisource parameters for fault transfer prognosis of aeroengine[J]. IEEE Transactions on Industrial Electronics,2022,69(1): 845-855. doi: 10.1109/TIE.2021.3050382 [6] YANG Huixin,LI Xiang,ZHANG Wei. Interpretability of deep convolutional neural networks on rolling bearing fault diagnosis[J]. Measurement Science and Technology,2022,33(5): 055005. doi: 10.1088/1361-6501/ac41a5 [7] 张西宁,刘书语,余迪,等. 改进深度卷积神经网络及其在变工况滚动轴承故障诊断中的应用[J]. 西安交通大学学报,2021,55(6): 1-8. ZHANG Xining,LIU Shuyu,YU Di,et al. Improved deep convolutional neural network with applications to bearing fault diagnosis under variable conditions[J]. Journal of Xi’an Jiaotong University,2021,55(6): 1-8. (in ChineseZHANG Xining, LIU Shuyu, YU Di, et al. Improved deep convolutional neural network with applications to bearing fault diagnosis under variable conditions[J]. Journal of Xi’an Jiaotong University, 2021, 55(6): 1-8. (in Chinese) [8] 宫文峰,陈辉,WANG Danwei,等. 基于改进CNN-GAP-SVM的船舶电力变换器快速故障诊断方法[J]. 计算机集成制造系统,2022,28(5): 1370-1384. GONG Wenfeng,CHEN Hui,WANG Danwei,et al. Fast fault diagnosis method of marine electrical converter based on improved CNN-GAP-SVM algorithm[J]. Computer Integrated Manufacturing Systems,2022,28(5): 1370-1384. (in ChineseGONG Wenfeng, CHEN Hui, WANG Danwei, et al. Fast fault diagnosis method of marine electrical converter based on improved CNN-GAP-SVM algorithm[J]. Computer Integrated Manufacturing Systems, 2022, 28(5): 1370-1384. (in Chinese) [9] 张思源,纪洪泉,刘洋. 基于ISCNN-LightGBM的轴承故障诊断[J]. 控制理论与应用,2023,40(4): 753-760. ZHANG Siyuan,JI Hongquan,LIU Yang. Bearing fault diagnosis based on ISCNN-LightGBM[J]. Control Theory and Applications,2023,40(4): 753-760. (in ChineseZHANG Siyuan, JI Hongquan, LIU Yang. Bearing fault diagnosis based on ISCNN-LightGBM[J]. Control Theory and Applications, 2023, 40(4): 753-760. (in Chinese) [10] 王太勇,宫立明,王鹏,等. 基于KD-DenseNet的旋转机械故障诊断模型[J]. 振动与冲击,2020,39(16): 39-45,80. WANG Taiyong,GONG Liming,WANG Peng,et al. A fault diagnosis model of rotating machinery based on KD-DenseNet[J]. Journal of Vibration and Shock,2020,39(16): 39-45,80. (in ChineseWANG Taiyong, GONG Liming, WANG Peng, et al. A fault diagnosis model of rotating machinery based on KD-DenseNet[J]. Journal of Vibration and Shock, 2020, 39(16): 39-45, 80. (in Chinese) [11] 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. (in ChineseDENG 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. (in Chinese) [12] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, NV, US: IEEE, 2016: 770-778.HE K M,ZHANG X Y,REN S Q,et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, NV, US:IEEE,2016: 770-778. [13] LIU Zhuang,MAO Hanzi,WU Chaoyuan,et al. A ConvNet for the 2020s[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway,US: IEEE,2022: 11966-11976. [14] 尹文哲,夏虹,彭彬森,等. 基于CNN-SVM的核电厂轴承故障诊断方法[J]. 哈尔滨工程大学学报,2023,44(3): 410-417. YIN Wenzhe,XIA Hong,PENG Binsen,et al. Bearing fault diagnosis method in nuclear power plants based on CNN-SVM[J]. Journal of Harbin Engineering University,2023,44(3): 410-417. (in ChineseYIN Wenzhe, XIA Hong, PENG Binsen, et al. Bearing fault diagnosis method in nuclear power plants based on CNN-SVM[J]. Journal of Harbin Engineering University, 2023, 44(3): 410-417. (in Chinese) [15] 程军圣,于德介,邓乾旺,等. 连续小波变换在滚动轴承故障诊断中的应用[J]. 中国机械工程,2003,14(23): 2037-2040. CHENG Junsheng,YU Dejie,DENG Qianwang,et al. Application of continuous wavelet transform to fault diagnosis of rolling bearings[J]. China Mechanical Engineering,2003,14(23): 2037-2040. (in ChineseCHENG Junsheng, YU Dejie, DENG Qianwang, et al. Application of continuous wavelet transform to fault diagnosis of rolling bearings[J]. China Mechanical Engineering, 2003, 14(23): 2037-2040. (in Chinese) [16] GU Xiaohui,YANG Shaopu,LIU Yongqiang,et al. Compound faults detection of the rolling element bearing based on the optimal complex Morlet wavelet filter[J]. Proceedings of the Institution of Mechanical Engineers,Part C: Journal of Mechanical Engineering Science,2018,232(10): 1786-1801. doi: 10.1177/0954406217710673 [17] 张永宏,邵凡,赵晓平,等. 基于多标签零样本学习的滚动轴承故障诊断[J]. 振动与冲击,2022,41(11): 55-64,89. ZHANG Yonghong,SHAO Fan,ZHAO Xiaoping,et al. Rolling bearing fault diagnosis based on multi-label zero-shot learning[J]. Journal of Vibration and Shock,2022,41(11): 55-64,89. (in ChineseZHANG Yonghong, SHAO Fan, ZHAO Xiaoping, et al. Rolling bearing fault diagnosis based on multi-label zero-shot learning[J]. Journal of Vibration and Shock, 2022, 41(11): 55-64, 89. (in Chinese) [18] CHOLLET F. Xception: deep learning with depthwise separable convolutions[C]// IEEE Conference on Computer Vision and Pattern Recognition. Piscataway,US: IEEE,2017: 1800-1807. [19] ROY S K,DUBEY S R,CHATTERJEE S,et al. FuSENet: fused squeeze-and-excitation network for spectral-spatial hyperspectral image classification[J]. IET Image Processing,2020,14(8): 1653-1661. doi: 10.1049/iet-ipr.2019.1462 [20] 朱佩荣,刘勇智,刘棕成,等. 基于CEEMD与改进的ELM旋转整流器故障诊断[J]. 北京航空航天大学学报,2023,49(5): 1166-1175. ZHU Peirong,LIU Yongzhi,LIU Zongcheng,et al. Fault diagnosis of synchronous generator rotating rectifier based on CEEMD and improved ELM[J]. Journal of Beijing University of Aeronautics and Astronautics,2023,49(5): 1166-1175. (in ChineseZHU Peirong, LIU Yongzhi, LIU Zongcheng, et al. Fault diagnosis of synchronous generator rotating rectifier based on CEEMD and improved ELM[J]. Journal of Beijing University of Aeronautics and Astronautics, 2023, 49(5): 1166-1175. (in Chinese) [21] 高斐,李洪儒,许葆华. 基于ICPSO优化的极限学习机在故障诊断中的应用[J]. 中国机械工程,2013,24(20): 2753-2757. GAO Fei,LI Hongru,XU Baohua. Applications of extreme learning machine optimized by ICPSO in fault diagnosis[J]. China Mechanical Engineering,2013,24(20): 2753-2757. (in ChineseGAO Fei, LI Hongru, XU Baohua. Applications of extreme learning machine optimized by ICPSO in fault diagnosis[J]. China Mechanical Engineering, 2013, 24(20): 2753-2757. (in Chinese) [22] CHEN Xu,QI Xiaoli,WANG Zhenya,et al. Fault diagnosis of rolling bearing using marine predators algorithm-based support vector machine and topology learning and out-of-sample embedding[J]. Measurement,2021,176: 109116. doi: 10.1016/j.measurement.2021.109116 [23] XUE Jiankai,SHEN Bo. Dung beetle optimizer: a new meta-heuristic algorithm for global optimization[J]. The Journal of Supercomputing,2023,79(7): 7305-7336. doi: 10.1007/s11227-022-04959-6 [24] HOU Lei,YI Haiming,JIN Yuhong,et al. Inter-shaft bearing fault diagnosis based on aero-engine system: a benchmarking dataset study[J]. Journal of Dynamics,Monitoring and Diagnostics,2023,2(4): 228-242. -
点击查看大图
计量
- 文章访问数: 556
- HTML浏览量: 410
- PDF量: 54
- 被引次数: 0