谱归一化CycleGAN的轴承故障迁移诊断研究

PDF(2045 KB)

振动与冲击 ›› 2023, Vol. 42 ›› Issue (24) : 282-289.

论文

李洁松1,2，刘韬1,2，伍星2,3

作者信息 +

Spectral normalization CycleGAN for bearing fault transfer diagnosis

LI Jiesong1,2， LIU Tao1,2， WU Xing2,3

Author information +

文章历史 +

摘要

深度学习无需先验特征提取的优点使其受到了工业设备的智能故障诊断领域研究的青睐，但深度学习的低鲁棒性和较高的数据要求阻碍其实际应用。为适应在工业现场复杂多变的工况，本文提出了一种基于谱归一化(spectral normalization，SN)和循环一致对抗网络(cycle-consistent adversarial networks，CycleGAN)的SN_1DCycleGAN网络用于变工况条件下的故障数据迁移生成和诊断。首先，搭建一种适应振动数据生成的1DCycleGAN网络，用于获得同种工况下正常信号与故障信号的映射关系。使用谱归一化对网络进行改进，有效的防止CycleGAN网络训练过程中训练不稳定情况。其次，通过不同工况的正常数据生成自适应的故障数据，实现变工况迁移生成的目的。最后，三种评价指标以及分类器准确率对数据生成质量进行定量评估，并使用仿真与实验信号进行验证。实验结果表明，SN_1DCycleGAN在一维振动信号上具备一定迁移效果，可对变工况数据进行增强，提升分类器的准确率。

Abstract

The advantage of deep learning without prior feature extraction makes it increasingly used in intelligent fault diagnosis of industrial equipment, but the low robustness and high data requirements of deep learning methods hinder practical applications. In order to adapt to the complex and variable working conditions in industrial sites, this paper proposes a SN_1DCycleGAN network based on spectral normalization (SN) and cycle-consistent adversarial networks (CycleGAN) for fault data transfer generation and diagnosis under variable working conditions. Firstly, a 1DCycleGAN network adapted to vibrational data generation is built for obtaining the mapping relationship between the source and target domains. The network is improved using spectral normalization to effectively prevent the training instability situation. Secondly, adaptive target data are obtained by changing the source domain to achieve the purpose of variable work transfer. Finally, the quality of data generation is quantitatively evaluated using three evaluation metrics as well as classifier accuracy, and validated using simulation and experimental signals. The experimental results show that SN_1DCycleGAN has a certain transfer effect on 1D vibration signals, which can enhance the variable working condition data and improve the accuracy of the classifier. The stability and generation quality are better than 1DCycleGAN.

导出引用

李洁松1,2，刘韬1,2，伍星2,3. 谱归一化CycleGAN的轴承故障迁移诊断研究[J]. 振动与冲击, 2023, 42(24): 282-289

LI Jiesong1,2， LIU Tao1,2， WU Xing2,3. Spectral normalization CycleGAN for bearing fault transfer diagnosis[J]. Journal of Vibration and Shock, 2023, 42(24): 282-289

参考文献

[1] ZHANG T, LIU S L, WEI Y, et al. A novel feature adaptive extraction method based on deep learning for bearing fault diagnosis [J]. Measurement, 2021, 185: 110030. [2] LI X H, WAN S K, LIU S J, et al. Bearing fault diagnosis method based on attention mechanism and multilayer fusion network [J]. ISA Transactions, 2022, 128: 550-564 2021. [3] YANG Z-B, ZHANG J-P, ZHAO Z-B, et al. Interpreting network knowledge with attention mechanism for bearing fault diagnosis [J]. Applied Soft Computing, 2020, 97: 106829. [4] 王琦, 邓林峰, 赵荣珍. 基于改进一维卷积神经网络的滚动轴承故障识别[J]. 振动与冲击, 2022, 41(03): 216-223. WANG Q, DENG L F, ZHAO R Z. Fault recognition of rolling bearing based on improved 1D convolutional neural network [J]. Journal of Vibration and Shock, 2022, 41(03): 216-223. [5] 胡若晖, 张敏, 许文鑫. 基于DCGAN和DANN网络的滚动轴承跨域故障诊断[J]. 振动与冲击, 2022, 41(06): 21-29. HU R H, ZHANG Min, XU W X. Cross-domain fault diagnosis of rolling element bearings using DCGAN and DANN [J]. Journal of Vibration and Shock, 2022, 41(06): 21-29. [6] LI W, ZHONG X, SHAO H D, et al. Multi-mode data augmentation and fault diagnosis of rotating machinery using modified ACGAN designed with new framework [J]. Advanced Engineering Informatics, 2022, 52: 101552. [7] GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative Adversarial Networks [J]. arXiv preprint arXiv: 1406.2661, 2014. [8] PAN T Y, CHEN J L, ZHANG T C, et al. Generative adversarial network in mechanical fault diagnosis under small sample: A systematic review on applications and future perspectives [J]. ISA Transactions, 2022, 128: 1-10.2021. [9] MA L, DING Y, WANG Z L, et al. An interpretable data augmentation scheme for machine fault diagnosis based on a sparsity-constrained generative adversarial network [J]. Expert Systems with Applications, 2021, 182: 115234. [10] LIU S, CHEN J L, HE S L, et al. Intelligent fault diagnosis under small sample size conditions via Bidirectional InfoMax GAN with unsupervised representation learning [J]. Knowledge-Based Systems, 2021, 232: 107488. [11] LIU J, ZHANG C H, JIANG X X. Imbalanced fault diagnosis of rolling bearing using improved MsR-GAN and feature enhancement-driven CapsNet [J]. Mechanical Systems and Signal Processing, 2022, 168: 108664. [12] TIAN J H, HAN D Y, LI M D, et al. A multi-source information transfer learning method with subdomain adaptation for cross-domain fault diagnosis [J]. Knowledge-Based Systems, 2022, 243: 108466. [13] ZHU J, PARK T, ISOLA P, et al. Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks; proceedings of the 2017 IEEE International Conference on Computer Vision (ICCV) [C], F 22-29 Oct. 2017, 2017: 2223-2232.2017. [14] 余艳杰, 孙嘉琪, 葛思擘，等. CycleGAN-SN:结合谱归一化和CycleGAN的图像风格化算法 [J]. 西安交通大学学报, 2020, 54(5): 9. YU Y J, SUN J Q, GE S B, et al. CycleGAN-SN:Image Stylization Algorithm Combining Spectral Normalization and CycleGAN [J]. Journal of Xi'an Jiaotong University, 2020, 54(5): 9. [15] GU J, YANG T S, YE J C, et al. CycleGAN denoising of extreme low-dose cardiac CT using wavelet-assisted noise disentanglement [J]. Medical Image Analysis, 2021, 74: 102209. [16] PRAJAPATI G P, SINGH D K, AMIN P P, et al. Voice privacy using CycleGAN and time-scale modification [J]. Computer Speech & Language, 2022, 74: 101353. [17] QIN Y, WANG Z W, XI D J. Tree CycleGAN with maximum diversity loss for image augmentation and its application into gear pitting detection [J]. Applied Soft Computing, 2022, 114: 108130. [18] LI X, ZHANG W, DING Q. Cross-Domain Fault Diagnosis of Rolling Element Bearings Using Deep Generative Neural Networks [J]. IEEE Transactions on Industrial Electronics, 2019, 66(7): 5525-5534. [19] ARJOVSKY M, CHINTALA S, BOTTOU L. Wasserstein GAN [J]. arXiv preprint arXiv: 1701.07875, 2017. [20] MIYATO T, KATAOKA T, KOYAMA M, et al. Spectral Normalization for Generative Adversarial Networks [J]. arXiv preprint arXiv: 1802.05957, 2018. [21] Dong Y J, Li Y Q, Zheng H L, et al. A new dynamic model and transfer learning based intelligent fault diagnosis framework for rolling element bearings race faults: Solving the small sample problem [J]. ISA Transactions 2022; 121:327-348. [22] LIU T, CHEN J, DONG G M, et al. The fault detection and diagnosis in rolling element bearings using frequency band entropy [J]. Proceedings of Institution of Mechanical Engineers Part C Journal of Mechanical Engineering Science. 2013, 227(1): 87-99. [23] Smith W A, Randall R B. Wade A. Smith, Robert B. Randall Rolling element bearing diagnostics using the Case Western Reserve University data: A benchmark study [J]. Mechanical Systems and Signal Processing. 2015, 64–65: 100-131.