Semi-supervised fault diagnosis of rolling bearing based on twin-representation contrastive learning

PDF(2482 KB)

Journal of Vibration and Shock ›› 2025, Vol. 44 ›› Issue (15) : 209-216.

FAULT DIAGNOSIS ANALYSIS

Semi-supervised fault diagnosis of rolling bearing based on twin-representation contrastive learning

CHEN Renxiang1, ZHANG Xu1, YANG Lixia*2, LIANG Dong1, SUN Shizheng1, DONG Shaojiang1

Author information +

History +

Abstract

To address the issue of high annotation costs for bearing fault data in practical engineering, which leads to insufficient labeled samples for supervised model training, a semi-supervised fault diagnosis method for rolling bearings based on contrastive learning with twin representations is proposed. Firstly, Gaussian white noise is added as a data augmentation method to apply different degrees of perturbations to the unlabeled data, generating positive pairs. Simultaneously, a twin self-correcting convolutional neural network with shared weights is constructed to extract high-dimensional features from the positive pairs. Secondly, based on the contrastive learning strategy, a negative cosine similarity loss function is constructed to compare the features of the positive pairs. By maximizing the correlation between features, supervisory information is built for the pre-training stage, promoting the model to learn consistent feature representations of samples from different perspectives in the unlabeled data. Then, a small number of labeled samples are introduced for fine-tuning, establishing the mapping relationship between feature representations and labels. Finally, the test data is input into the fine-tuned encoder model to achieve semi-supervised fault diagnosis of rolling bearings. The proposed method learns the intrinsic structure and feature representations of the data from a large amount of unlabeled data, without relying on an expensive annotation process. Experiments conducted on the collected rolling bearing data and the public HUST bearing dataset verify that the proposed method achieves an accuracy of over 97%, demonstrating its excellent diagnostic performance.

Key words

Rolling bearings / Fault diagnosis / semi-supervision / Comparative learning / self-calibrated convolutions

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

CHEN Renxiang1, ZHANG Xu1, YANG Lixia2, LIANG Dong1, SUN Shizheng1, DONG Shaojiang1. Semi-supervised fault diagnosis of rolling bearing based on twin-representation contrastive learning[J]. Journal of Vibration and Shock, 2025, 44(15): 209-216

References

[1] Li S, Peng Y, Shen Y, et al. Rolling bearing fault diagnosis under data imbalance and variable speed based on adaptive clustering weighted oversampling[J]. Reliability Engineering & System Safety, 2024, 244: 109938.
[2] 王冉, 石如玉, 胡升涵等. 基于声成像与卷积神经网络的轴承故障诊断方法及其可解释性研究[J]. 振动与冲击, 2022, 41(16): 224-231.
Wang R, Shi R Y, Hu S H et al. An acoustic fault diagnosis method of rolling bearings based on acoustic imaging and convolutional neural network[J]. Journal of Vibration and Shock, 2022, 41(16): 224-231.
[3] 雷春丽, 夏奔锋, 薛林林等. 小样本下自校正卷积神经网络的滚动轴承故障识别方法[J]. 仪器仪表学报, 2022, 43(09): 122-130.
Lei C L, Xia B F, Xue L L et al. Fault identification for rolling bearing by self-calibrated convolutional neural network under small samples conditions[J]. Chinese Journal of Scientific Instrument, 2022, 43(09): 122-130.
[4] 王鸽, 吴国新, 刘秀丽. 基于MADCNN的故障诊断方法研究[J]. 电子测量与仪器学报, 2023, 37(3): 187-193.
Wang G, Wu G X, Liu X L. Research on fault diagnosis method based on MADCNN[J]. Journal of Electronic Measurement and Instrumentation, 2023, 37(3): 187-193.
[5] 韩康, 战洪飞, 余军合, 等. 基于空洞卷积和增强型多尺度特征自适应融合的滚动轴承故障诊断[J]. 浙江大学学报(工学版), 2024, 58(06): 1285-1295.
Han K, Zhan H F, Yu J H, et al. Rolling bearing fault diagnosis based on dilated convolution and enhanced multi-scale feature adaptive fusion[J]. Journal of Zhejiang University (Engineering Science Edition), 2024, 58(06): 1285-1295.
[6] Hou Y, Wang J, Chen Z, et al. Diagnosisformer: An efficient rolling bearing fault diagnosis method based on improved Transformer[J]. Engineering Applications of Artificial Intelligence, 2023, 124: 106507.
[7] Yang X, Song Z, King I, et al. A survey on deep semi-supervised learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2022.
[8] 崔建国, 李国庆, 蒋丽英等. 基于深度自编码网络的航空发动机故障诊断[J]. 振动.测试与诊断, 2021, 41(01): 85-89+201-202.
Cui J G, Li G Q, Jiang L Y et al. Aero⁃engine Fault Diagnosis Based on Deep Self⁃Coding Network[J]. Journal of Vibration，Measurement & Diagnosis, 2021, 41(01): 85-89+201-202.
[9] Wang C, Xin C, Xu Z, et al. Mix-VAEs: A novel multisensor information fusion model for intelligent fault diagnosis[J]. Neurocomputing, 2022, 492: 234-244.
[10] 周兴康, 余建波. 基于深度一维残差卷积自编码网络的齿轮箱故障诊断[J]. 机械工程学报, 2020, 56(07): 96-108.
Zhou X K, Yu J B. Gearbox Fault Diagnosis Based on One-dimension Residual Convolutional Auto-encoder[J]. Journal of Mechanical Engineering, 2020, 56(07): 96-108.
[11] Wang L, Rao H, Dong Z, et al. Automatic fault diagnosis of rolling bearings under multiple working conditions based on unsupervised stack denoising autoencoder[J]. Structural Health Monitoring, 2024: 14759217231221214.
[12] Shi M, Ding C, Wang R, et al. Deep hypergraph autoencoder embedding: An efficient intelligent approach for rotating machinery fault diagnosis[J]. Knowledge-Based Systems, 2023, 260: 110172.
[13] 陈仁祥, 张晓, 张旭, 等. 改进掩码自编码器的滚动轴承半监督故障诊断[J]. 仪器仪表学报, 2024, 45(01): 26-33.
CHEN R X, ZHANG X, ZHANG X, et al. Improved semi-supervised fault diagnosis of rolling bearings with mask autoencoder[J]. Chinese Journal of Scientific Instrument, 2019, 45(01): 26-33.
[14] 李巍华, 何琛, 陈祝云等. 基于对称式对比学习的齿轮箱无监督故障诊断方法[J]. 仪器仪表学报, 2022, 43(03): 121- 131.
Li W H, He C, Chen Z Y et al. Unsupervised Fault diagnosis of gearbox based on Symmetric Contrast Learning[J]. Chinese Journal of Scientific Instrument, 2022, 43(03): 121-131.
[15] Ding Y, Zhuang J, Ding P, et al. Self-supervised pretraining via contrast learning for intelligent incipient fault detection of bearings[J]. Reliability Engineering & System Safety, 2022, 218: 108126.
[16] 乌文扬, 陈景龙, 刘莘, 等. 基于无监督特征表示深度Q学习的智能故障诊断方法[J]. 中南大学学报(自然科学版), 2022, 53(05): 1750-1759.
Wu W Y, Chen J L, Liu S, et al. Intelligent fault diagnosis method based on unsupervised featurerepresentation and deep Q-learning[J]. Journal of Central South University(Science and Technology), 2022, 53(5): 1750-1759.
[17] Wang X, Zhang H, Ma K, et al. Internal Contrastive Learning for Generalized Out-of-distribution Fault Diagnosis (GOOFD) Framework[J]. arXiv preprint arXiv:2306.15266, 2023.
[18] 王岩红, 温笑欢, 揭永琴, 等. 基于对比学习的滚动轴承早期故障在线检测方法[J]. 振动与冲击, 2023, 42(14): 229-236.
Wang Y H, Wen X H, Jie Y Q, et al. Online detection method for bearing incipient faults based on contrastive learning[J]. Journal of Vibration and Shock, 2023, 42(14): 229-236.
[19] Liu J J, Hou Q, Cheng M M, et al. Improving convolutional networks with self-calibrated convolutions[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 10096-10105.
[20] Chen X, He K. Exploring simple siamese representation learning[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021: 15750-15758.
[21] 陈仁祥, 谢文举, 徐向阳, 等. 基于数据融合和改进MoCo的工业机器人抖动原因识别[J]. 仪器仪表学报, 2023, 44(07): 112-120.
Chen R X, XIE W J, XU X Y, et al. Recognition of Jitter Causes for Industrial Robots Based on Data Fusion and an Improved MoCo [J]. Chinese Journal of Scientific Instrument, 2023, 44(07): 112-120.
[22] Hong, Hoang Si; Thuan, Nguyen (2023), “HUST bearing”, Mendeley Data,V2