Due to lengthy data collection and manual annotation challenges, fault datasets for high-speed train axle box bearings often contain mislabeled samples. This study proposes a semi-supervised method that integrates multi-indicator noise filtering with robust training to counteract performance degradation caused by label noise. A classifier-distinguished adversarial autoencoder is employed, which utilizes reconstruction error, latent features, and discriminator scores in a voting mechanism to separate clean and noisy labels. A two-stage training strategy is then applied: pretraining a residual network on clean label samples, followed by semi-supervised joint training on all data. Evaluated on two public datasets with 50% symmetric noise, the method achieves an average accuracy of over 90%, outperforming comparable approaches.
LuCCaiC. Overview on safety management and maintenance of high-speed railway in China. Transp Geotech2020; 25: 100397.
4.
ZhaoHLiangJLiuC. High-speed EMUs: characteristics of technological development and trends. Engineering2020; 6: 234–244.
5.
WangZZhangWYinZ, et al. Effect of vehicle vibration environment of high-speed train on dynamic performance of axle box bearing. Veh Syst Dyn2019; 57: 543–563.
6.
LiuJDuS. Dynamic analysis of a high-speed railway train with the defective axle bearing. Int J Acoust Vib2020; 25: 525–531.
7.
XuGHouDQiH, et al. High-speed train wheel set bearing fault diagnosis and prognostics: a new prognostic model based on extendable useful life. Mech Syst Signal Process2021; 146: 107050.
8.
HeMHeD. A new hybrid deep signal processing approach for bearing fault diagnosis using vibration signals. Neurocomputing2020; 396: 542–555.
9.
AttouiIOudjaniBBoutassetaN, et al. Novel predictive features using a wrapper model for rolling bearing fault diagnosis based on vibration signal analysis. Int J Adv Manuf Technol2020; 106: 3409–3435.
10.
RuanDWangJYanJ, et al. CNN parameter design based on fault signal analysis and its application in bearing fault diagnosis. Adv Eng Inform2023; 55: 101877.
11.
HuQSiX-SZhangQ-H, et al. A rotating machinery fault diagnosis method based on multi-scale dimensionless indicators and random forests. Mech Syst Signal Process2020; 139: 106609.
12.
SunBLiuX. Significance support vector machine for high-speed train bearing fault diagnosis. IEEE Sens J2023; 23: 4638–4646.
13.
WangHZhengJXiangJ. Online bearing fault diagnosis using numerical simulation models and machine learning classifications. Reliab Eng Syst Saf2023; 234: 109142.
14.
NiQJiJCHalkonB, et al. Physics-Informed Residual Network (PIResNet) for rolling element bearing fault diagnostics. Mech Syst Signal Process2023; 200: 110544.
15.
SongBLiuYFangJ, et al. An optimized CNN-BiLSTM network for bearing fault diagnosis under multiple working conditions with limited training samples. Neurocomputing2024; 574: 127284.
16.
HeDWuJJinZ, et al. AGFCN:A bearing fault diagnosis method for high-speed train bogie under complex working conditions. Reliab Eng Syst Saf2025; 258: 110907.
17.
SongBZhaoSDangL, et al. A survey on learning from data with label noise via deep neural networks. Syst Sci Control Eng2025; 13: 2488120.
18.
SongHKimMParkD, et al. Learning from noisy labels with deep neural networks: a survey. IEEE Trans Neural Netw Learn Syst2023; 34: 8135–8153.
19.
LiangXLiuXYaoL. Review–a survey of learning from noisy labels. ECS Sens Plus2022; 1: 021401.
20.
ZhangXWuBZhangX, et al. A novel assessable data augmentation method for mechanical fault diagnosis under noisy labels. Measurement2022; 198: 111114.
21.
DunkinFLiXLiH, et al. MgCNL: a sample separation approach via multi-granularity balls for fault diagnosis with the interference of noisy labels. IEEE Trans Autom Sci Eng2025; 22: 7748–7761.
22.
ZhongJYangYMaoH, et al. Contrastive regularization guided label refurbishment for fault diagnosis under label noise. Adv Eng Inform2024; 61: 102478.
23.
WangHLiY-F. Iterative error self-correction for robust fault diagnosis of mechanical equipment with noisy label. IEEE Trans Instrum Meas2022; 71: 1–13.
24.
DunkinFLiXWuG, et al. Wisdom via multiple perspectives: a multigranularity clusters fusion approach for fault diagnosis with noisy labels. IEEE/ASME Trans Mechatronics2025; 30(6): 1–11.
25.
QinGZhangKLaiX, et al. An adaptive symmetric loss in dynamic wide-kernel ResNet for rotating machinery fault diagnosis under noisy labels. IEEE Trans Instrum Meas2024; 73: 1–12.
26.
ZhangKTangBDengL, et al. A fault diagnosis method for wind turbines gearbox based on adaptive loss weighted meta-ResNet under noisy labels. Mech Syst Signal Process2021; 161: 107963.
27.
YuanXZhangZLiangP, et al. A fusion TFDAN-Based framework for rotating machinery fault diagnosis under noisy labels. Appl Acoust2024; 219: 109940.
28.
WangHLiCDingP, et al. A novel transformer-based few-shot learning method for intelligent fault diagnosis with noisy labels under varying working conditions. Reliab Eng Syst Saf2024; 251: 110400.
29.
ChengCLiuXZhouB, et al. intelligent fault diagnosis with noisy labels via semisupervised learning on industrial time series. IEEE Trans Ind Inform2023; 19: 7724–7732.
30.
PuXLiC. Meta-self-training based on teacher–student network for industrial label-noise fault diagnosis. IEEE Trans Instrum Meas2023; 72: 1–11.
31.
SunWYanRJinR, et al. Curriculum-based federated learning for machine fault diagnosis with noisy labels. IEEE Trans Ind Inform2024; 20: 13820–13830.
32.
MaJZhangKZhengQ, et al. A fault diagnosis method for elevator traction machine drum brakes based on co-correction with complementary labels. Measurement2025; 253: 117524.
33.
ZhengWZhangHLiY-F. A systematic approach to the high-level maintenance scheduling for high-speed trains in China. IEEE Trans Reliab2024; 73: 142–155.
34.
AlganGUlusoyİ. Label noise types and their effects on deep learning. Epub ahead of print 23March2020. DOI: 10.48550/arXiv.2003.10471.
35.
LiPPeiYLiJ. A comprehensive survey on design and application of autoencoder in deep learning. Appl Soft Comput2023; 138: 110176.
36.
CreswellAWhiteTDumoulinV, et al. Generative adversarial networks: an overview. IEEE Signal Process Mag2018; 35: 53–65.
37.
MakhzaniAShlensJJaitlyN, et al. Adversarial autoencoders. Epub ahead of print 25May2016. DOI: 10.48550/arXiv.1511.05644.
38.
BerthelotDCarliniNGoodfellowI, et al. MixMatch: a holistic approach to semi-supervised learning. Epub ahead of print 23October2019. DOI: 10.48550/arXiv.1905.02249.
JangKHongSKimM, et al. Adversarial autoencoder based feature learning for fault detection in industrial processes. IEEE Trans Ind Inform2022; 18: 827–834.
41.
TackJYuSJeongJ, et al. Consistency regularization for adversarial robustness. Proc AAAI Conf Artif Intell2022; 36: 8414–8422.
42.
NishiKDingYRichA, et al. Augmentation strategies for learning with noisy labels. In: 2021 IEEE/CVF conference on computer vision and pattern recognition (CVPR), Nashville, TN, 2021, pp. 8018–8027. Nashville, TN: IEEE.
43.
ZhangHCisseMDauphinYN, et al. mixup: beyond empirical risk minimization. Epub ahead of print 27April2018. DOI: 10.48550/arXiv.1710.09412.
44.
LiZZhangKZhengQ, et al. An open-set faults diagnosis method for bogie mechanical transmission components based on multi-channel feature-enhanced placeholder learning. IEEE Trans Ind Inform2025; 21(11): 8373–8383.
45.
LiZZhangKZhengQ, et al. Unsupervised fault detection with multi-source anomaly sensitivity enhancing convolutional autoencoder for high-speed train bogie bearings. Expert Syst Appl2025; 281: 127570.
46.
ZhangKWangBZhengQ, et al. A novel fault diagnosis of high-speed train axle box bearings with adaptive curriculum self-paced learning under noisy labels. Struct Health Monit2025.
WeiHFengLChenX, et al. Combating noisy labels by agreement: a joint training method with co-regularization. In: 2020 IEEE/CVF conference on computer vision and pattern recognition (CVPR), 2020, pp. 13723–13732. Seattle, WA: IEEE.
49.
AminiM-RFeofanovVPaulettoL, et al. Self-training: a survey. Neurocomputing2025; 616: 128904.
50.
KhoslaPTeterwakPWangC, et al. Supervised contrastive learning. Epub ahead of print 10March2021. DOI: 10.48550/arXiv.2004.11362.
51.
LiJSocherRHoiSCH. DivideMix: Learning with noisy labels as semi-supervised learning. Epub ahead of print 18February2020. DOI: 10.48550/arXiv.2002. 07394.
52.
ChenTKornblithSNorouziM, et al. A simple framework for contrastive learning of visual representations. In: International conference on machine learning, Virtual, 2020, pp. 1597–1607. PMLR.
