Contrastive learning (CL), as a commonly used method for mechanical fault diagnosis, has achieved good application results, but it is prone to introducing false negatives and redundant easy negatives, especially in strong noise environments. To address these limitations, a novel framework, named filtered CL with kernel attention (FCL-KA), is proposed in this paper. Specifically, a new spatially adaptive encoding module is constructed, which can enable dynamic receptive field calibration for precise feature capture under high-noise interference, thus overcoming the limitations of fixed receptive fields and capturing discriminative features from noisy signals. Furthermore, a dual-threshold filtered contrastive mechanism is designed to explicitly eliminate high-similarity false negatives, screen out low-information easy negatives, and build more reliable prediction models. Finally, experimental verification was conducted on two rolling bearing fault datasets, and the results showed that FCL-KA exhibited superior classification and diagnostic performance in different evaluation metrics.
ChenXYanRXueY, et al. Deep transfer learning for bearing fault diagnosis: a systematic review since 2016. IEEE Trans Instrum Meas2023; 72: 1–21. https://doi.org/10.1109/TIM.2023.3244237
2.
YanRShenFSunC, et al. Knowledge-driven deep learning for mechanical fault diagnosis: a survey. IEEE Trans Ind Inform2022; 18(12): 9152–9164.
3.
LiWHuangRLiJ, et al. A perspective survey on deep transfer learning for fault diagnosis in industrial scenarios: theories, applications and challenges. Mech Syst Signal Process2022; 167: 108487. https://doi.org/10.1016/j.ymssp.2021.108487
4.
HanTLiuCYangL, et al. A novel adversarial learning framework in deep convolutional neural networks for intelligent diagnosis of mechanical faults. Knowl Based Syst2021; 218: 106098. https://doi.org/10.1016/j.knosys.2018.12.019
5.
ZhangZShaoHZuoMJ, et al. Deep transfer learning with multi-scale feature fusion for cross-domain fault diagnosis. IEEE Trans Ind Electron2023; 70(1): 865–875.
6.
LiJWuXLiS. Rolling bearing fault diagnosis based on VMD-MPE and improved SVM. Shock Vib2021; 2021: 1–13.
7.
LiuYLiuGZhuY. Rolling bearing fault diagnosis based on EEMD and optimized random forest. IEEE Access2022; 10: 56269–56280.
8.
NiQTanJYangS. Matrix-based classifier with structural regularization for bearing fault diagnosis. IEEE Trans Instrum Meas2021; 70: 1–11.
9.
LiXZhangWDingQ. Deep learning-based remaining useful life estimation of bearings using multi-scale feature extraction. Reliab Eng Syst Saf2021; 182: 106073. https://doi.org/10.1016/j.ress.2018.11.011
RuanDWangJYanJ, et al. CNN parameter design based on fault signal analysis and its application in bearing fault diagnosis. Adv Eng Inform2023; 55: 101877. https://doi.org/10.1016/j.aei.2023.101877
12.
SunCHongMHanH. Domain adaptation network with double adversarial mechanism for intelligent fault diagnosis. IEEE Trans Instrum Meas2021; 70: 1–12. https://doi.org/10.3390/app11177983
13.
ZhangWPengGLiC. Bearings fault diagnosis based on convolutional neural networks with attention mechanism. IEEE Trans Ind Inform2022; 18(5): 3089–3099.
14.
DongSLiuQYangB. Multi-scale dynamic graph convolutional network for fault diagnosis of rotating machinery under time-varying speeds. Mech Syst Signal Process2023; 188: 110012.
15.
ZhangCWannCDYanR. Physics-informed deep learning for machinery fault diagnosis. IEEE Trans Ind Electron2023; 70(6): 6215–6225.
16.
CaiLYinHLinJ, et al. A multiattribute learning model for zero-sample mechanical fault diagnosis. IEEE Trans Ind Inform2024; 20(7): 9633–9643. https://doi.org/10.1109/TII.2024.3383459
17.
WangYYanJHuGQ. Class-aware semi-supervised contrastive learning with pseudo-label guidance for bearing fault diagnosis. IEEE Trans Ind Electron2024; 71(4): 4056–4066. https://doi.org/10.1109/INDIN58382.2024.10774422
18.
DingYLiWWangJL. Time-frequency transformer with self-supervised contrastive learning for machinery fault diagnosis. IEEE Trans Instrum Meas2023; 72: 1–12.
19.
PengXYangBYanR. Regime-invariant contrastive learning for cross-domain fault diagnosis. IEEE Trans Syst Man Cybern Syst2024; 54(2): 1120–1132.
20.
ChenTKornblithSNorouziM, et al. A simple framework for contrastive learning of visual representations. Int J Comput Vis2021; 129(11): 3065–3086.
21.
HeKFanHWuY, et al. Momentum contrast for unsupervised visual representation learning. IEEE Trans Patt Anal Mach Intell2021; 43(3): 982–998.
22.
DingAQinYWangB, et al. Evolvable graph neural network for system-level incremental fault diagnosis of train transmission systems. Mech Syst Signal Process2024; 210: 111175. https://doi.org/10.1016/j.ymssp.2024.111175
23.
LiuHPanHZhengJ, et al. A new robust projection distributed broad learning under redundant samples and noisy environment. IEEE Trans Instrum Meas2024; 73: 1–9. https://doi.org/10.1109/TIM.2024.3460883
24.
ZhangJZouJSuZ. A class-aware supervised contrastive learning framework for imbalanced fault diagnosis. Knowl Based Syst2022; 252: 109437. https://doi.org/10.1016/j.knosys.2022.109437
CaiRGaoWPengl, et al. Debiased contrastive learning with supervision guidance for industrial fault detection. IEEE Trans Ind Inform2024; 20(11): 12814–12825. https://doi.org/10.1109/TII.2024.3424561
27.
ChengTChengHFangY, et al. CPCL: conceptual prototypical contrastive learning for few-shot text classification. J Intell Fuzzy Syst2023; 45(6): 11963–11975. https://doi.org/10.3233/JIFS-231570
28.
MengZLuoCLiJ, et al. Research on fault diagnosis of rolling bearing based on lightweight model with multiscale features. IEEE Sens J2023; 23(12): 13236–13247. https://doi.org/10.1109/JSEN.2023.3270880
