基于VAE-GAN和FLCNN的不均衡样本轴承故障诊断方法

摘要
图/表
参考文献(26)
相关文章 (15)

全文: PDF (3059 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要针对滚动轴承故障诊断中样本分布不均衡引起的模型泛化能力差、诊断精度低的问题，从两个方面展开研究：(1)故障样本增广：提出结合变分自编码器(VAE)和生成对抗网络(GAN)的VAE-GAN样本增广模型；(2)改进分类算法：提出基于焦点损失(FL)和卷积神经网络(CNN)的FLCNN样本分类模型。在此基础上，将VAE-GAN和FLCNN融合，构建VAE-GAN+FLCNN轴承故障诊断模型。首先，将样本量少的故障类输入VAE-GAN模型，通过交替训练编码网络、生成网络和判别网络，学习出真实故障样本的数据分布，从而实现故障样本的增广；然后用增广后的数据样本训练FLCNN分类模型，完成轴承故障识别。实验对比结果表明，所提方法能够有效提升样本不均衡条件下的轴承故障诊断效果，拥有更高的Recall值和F1-score值。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	张永宏1
	张中洋1
	赵晓平2
	3
	王丽华1
	邵凡1
	吕凯扬2

关键词 ：滚动轴承, 变分自编码器, 生成对抗网络, 焦点损失, 故障诊断

Abstract：Aiming at the problem of poor generalization ability and low diagnosis accuracy caused by unbalanced sample in rolling bearing fault diagnosis, this paper studied it from two aspects:(1)fault sample augmentation: a sample augmentation model based on VAE-GAN which combining variational auto-encoder (VAE) and enerative adversarial network (GAN) was proposed;(2)improvement of classification algorithm: a FLCNN classification model based on focus loss(FL) and convolutional neural network(CNN) was proposed. On this basis, the VAE-GAN+FLCNN bearing fault diagnosis model was constructed by fusing VAE-GAN and FLCNN. Firstly, the fault categories with few samples were input to the VAE-GAN model, the data distribution of real fault samples was learned by training encoding network, generating network and discriminating network alternately, and more fault samples were generated; then the augmented data were input into FLCNN for bearing fault identification. Experimental results show that the proposed method can effectively improve the bearing fault diagnosis effect under the condition of unbalanced samples, and has higher Recall value and F1-score value.

Key words： rolling bearing variational auto-encoder generative adversarial network focal loss fault diagnosis

收稿日期: 2020-11-23 出版日期: 2022-05-15

引用本文:

张永宏1，张中洋1，赵晓平2,3，王丽华1，邵凡1，吕凯扬2. 基于VAE-GAN和FLCNN的不均衡样本轴承故障诊断方法[J]. 振动与冲击, 2022, 41(9): 199-209.
ZHANG Yonghong1, ZHANG Zhongyang1, ZHAO Xiaoping2,3, WANG Lihua1, SHAO Fan1, L Kaiyang2. Bearing fault diagnosis method based on VAE-GAN and FLCNN unbalanced samples. JOURNAL OF VIBRATION AND SHOCK, 2022, 41(9): 199-209.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2022/V41/I9/199

[1] 刘永强，李翠省，廖英英. 基于EEMD和自相关函数峰态系数的轴承故障诊断方法[J]. 振动与冲击，2017，36 (2): 111-116.
LIU Yongqiang，LI Cuixing，LIAO Yingying. Fault diagnosis method for rolling bearings based on EEMD and autocorrelation function kurtosis [J]. Journal of Vibration and Shock，2017，36(2): 111-116.
[2] LI Y，XU M，HUANG W，et al. An improved EMD method for fault diagnosis of rolling bearing [C] // Prognostics and System Health Management Conference. Chengdu，China:
IEEE，2017: 1-5.
[3] 祝小彦，王永杰，张钰淇，等. 基于自适应最优Morlet小波的滚动轴承故障诊断[J]. 振动、测试与诊断，2018，
38(5): 1021-1029.
ZHU Xiaoyan，WANG Yongjie，ZHANG Yuqi，et al. Method of incipient fault diagnosis of bearing based on adaptive optimal Morlet wavelet [J]. Journal of Vibration，Measurement & Diagnosis，2018，38(5): 1021-1029.
[4] 王新，闫文源. 基于变分模态分解和SVM的滚动轴承故障诊断[J]. 振动与冲击，2017，36(18): 252-256.
WANG Xin，YAN Wenyuan. Fault diagnosis of roller bearings based on the variational model decomposition and SVM [J]. Journal of Vibration and Shock，2017，36 (18): 252-256.
[5] SHAO H，JIANG H，WANG F，et al. Rolling bearing fault diagnosis using adaptive deep belief network with dual-tree complex wavelet packet [J]. ISA Transactions，2017，69:
187-201.
[6] 吴春志，江鹏程，冯辅周，等. 基于一维卷积神经网络的齿轮箱故障诊断[J]. 振动与冲击，2018，37(22): 51-56.
WU Chunzhi，JIANG Pengcheng，FENG Fuzhou，et al. Faults diagnosis method for gearboxes based on a 1-D convolutional neural network [J]. Journal of Vibration and Shock，2018，37(22): 51-56.
[7] CHAWLA N V，BOWYER K W，HALL L O，et al. SMOTE: Synthetic minority over-sampling technique [J]. Journal of Artificial Intelligence Research，2002，16(1): 321-357.
[8] DOUZAS G，BACAO F，LAST F. Improving imbalanced learning through a heuristic oversampling method based on k-means and SMOTE [J]. Information Sciences，2018，465: 1-20.
[9] 姚培，王仲生，姜洪开，等. 不均衡数据下基于CS-Boosting的故障诊断算法[J]. 振动，测试与诊断，2013(01): 119-123+177.
YAO Pei，WANG Zhongsheng，JIANG Hongkai，et al. Fault diagnosis method based on CS-Boosting for unbalanced training data [J]. Journal of Vibration，Measurement & Diagnosis，2013(01): 119-123+177.
[10] HE H，GARCIA E A. Learning from imbalanced Data [J]. IEEE Transactions on Knowledge and Data Engineering，2009，21(9): 1263-1284.
[11] ZHAO X P，ZHANG Y H，SHAO F. A Multifault diagnosis method of gear box running on edge equipment [J]. Security and Communication Networks，2020，2020(3): 1-13.
[12] GOODFELLOW I J，POUGET-ABADIE J，MIRZA M，et al. Generative adversarial networks [J]. Advances in Neural Information Processing Systems，2014，3:2672-2680.
[13] KINGMA D P，WELLING M. Auto-encoding variational Bayes [C] // International Conference on Learning Representations，2014.
[14] DENTON E，GROSS S，FERGUS R. Semi-supervised learning with context-conditional generative adversarial networks [J]. 2016.
[15] 邹秀芳，朱定局. 生成对抗网络研究综述[J]. 计算机系统应用，2019，028(11): 1-9.
ZOU Xiufang，ZHU Dingju. Review on generative adversarial network [J]. Computer System & Applications，2019，028(11): 1-9.
[16] MIYATO T，KOYAMA M. cGANs with projection discriminator [J]. arXiv preprint arXiv:1802.05637，2018.
[17] 王威，孙长杰，王丽，等. 应用深度学习的生成对抗网络行星齿轮箱故障诊断技术研究[J]. 机械科学与技术，2020，39(1): 117-123.
WANG Wei，SUN Changjie，WANG Li，et al. Research on fault diagnosis technology of planetary gear box using deep learning generative adversarial network [J]. Mechanical Science and Technology for Aerospace Engineering. 2020，39(1): 117-123.
[18] CABRERA D，SANCHO F，LONG J，et al. Generative adversarial networks selection approach for extremely imbalanced fault diagnosis of reciprocating machinery [J]. IEEE Access，2019，PP(99): 1-1.
[19] 张周磊，李垣江，李梦含，等. 基于深度学习的永磁同步电机故障诊断方法[J]. 计算机应用与软件，2019，36 (10): 123-129.
ZHANG Zhoulei，LI Yuanjiang，LI Menghan，et al. Fault diagnosis method of permanent magnet synchronous motor based on depth learning [J]. Computer Applications and Software，2019，36(10): 123-129.
[20] QI Z F，Liu Q Q，WANG J，et al. Battle damage assessment based on an improved Kullback-Leibler divergence sparse autoencoder [J]. Frontiers of Information Technology & Electronic Engineering，2017，18(12): 1991-2000.
[21] KINGMA D P，SALIMANS T，WELLING M. Variational dropout and the local reparameterization trick [J]. Computer ence，2015.
[22] LINT Y，GOYAL P，GIRSHICK R，et al. Focal loss for dense object detection [C]//2017 IEEE International Conference on Computer Vision.Venice:IEEE，2017.
[23] RADFORD A，METZ L，CHINTALA S. Unsupervised representation learning with deep convolutional generative adversarial networks [J]. arXiv preprint arXiv:1511.06434，2015.
[24] 赵晓平，吴家新，钱承山，等. 基于多任务深度学习的齿轮箱多故障诊断方法[J]. 振动与冲击，2019，38(23) :
271-278.
ZHAO Xiaoping，WU Jiaxin，QIAN Chengshan，et al. Multi-fault diagnosis for gearboxes based on multi-task deep learning [J]. Journal of Vibration and Shock，2019，38(23): 271-278.
[25] 郑华林，王超，潘盛湖，等. 基于EEMD和分层阈值的磁记忆信号降噪方法研究[J]. 工程设计学报，2020，27 (04) : 433-440.
ZHENG Hualin，WANG Chao，PAN Shenghu et al.
Research on noise reduction method of magnetic memory signal based on EEMD and layered threshold [J]. Chinese Journal of Engineering Design，2020，27(04): 433-440.
[26] 才子昕，王馨月，徐剑，等. 样本自适应的不平衡分类器 [J]. 计算机科学，2019，46(01): 94-99.
CAI Zixin，WANG Xinyue，XU Jian，et al. Sample adaptive classifier for imbalanced Data [J]. Computer Science，2019，46(01): 94-99.