Methodology for Selecting Robust Vibration Diagnostics Models Under Domain Shift and Label Noise
DOI:
https://doi.org/10.32515/2664-262X.2026.13(44).230-240Keywords:
mobile machines, vibration diagnostics, domain shift, label noise, robust learning, pseudo-labeling, domain adaptation, rolling bearings; gearboxAbstract
This paper addresses a practical issue in vibration-based condition monitoring of mobile machines: diagnostic models that perform well on training data often degrade in the field due to domain shift (changes in operating regime, load, sensor mounting, and test-rig vs. real operation) and label noise (imprecise service records, ambiguous transition states, and human annotation errors).
We propose a conditional model-selection methodology formulated as a decision matrix that links domain-shift type and severity, labeled data volume and expected label reliability, and deployment constraints (validation feasibility and reproducibility requirements) to an appropriate class of models and training strategies.
The methodology is validated on two public scenarios with injected symmetric label noise: a moderate severity shift on the CWRU bearing dataset and a severe cross-condition shift on the SEU gearbox dataset. For the moderate shift, a two-stage robustification with confidence-based pseudo-label filtering achieves the best performance under high label noise, reaching Macro-F1 = 0.9368 at 20% noise. For the severe shift, a classical pipeline with simple feature-level domain alignment remains the most stable option, while ensemble strategies provide lower and less consistent gains. The results yield actionable guidance for integrating robust diagnostics into an intelligent technical service system under real-world constraints.
References
Список літератури
1. Аулін В. В., Гриньків А. В., Головатий А. О., Лисенко С. В., Голуб Д. В., Кузик О.В., Тихий А. А. Методологічні основи проектування та функціонування інтелектуальних транспортних і виробничих систем: монографія під заг. ред. д.т.н., проф. Ауліна В.В. Кропивницький: Видавець Лисенко В.Ф., 2020. 428с.
2. Матвієнко О. О., Аулін В. В. Класифікація типів сигналів та методів машинного навчання для інтелектуальної оцінки технічного стану мобільних машин підприємств агропромислового виробництва. Збірник наукових праць. Науковий вісник. Технічні науки. 2025. № 11(42)_ІІ. С. 298–312. DOI: 10.32515/2664-262X.2025.11(42).2.298-312.
3. Матвієнко О. О., Аулін В. В., Гриньків А.В. Стан та напрями розвитку архітектури даних для інтелектуальної оцінки технічного стану мобільних машин підприємств агропромислового виробництва. Збірник наукових праць. Науковий вісник. Технічні науки. 2025. № 12(43)_І. С. 227–237. DOI: 10.32515/2664-262X.2025.12(43).1.227-237.
4. Loutas T. H., Roulias D., Pauly E., Kostopoulos V. The combined use of vibration, acoustic emission and oil debris on-line monitoring towards a more effective condition monitoring of rotating machinery // Mechanical Systems and Signal Processing. 2011. Vol. 25, no. 4. P. 1339–1352. DOI: 10.1016/j.ymssp.2010.11.007.
5. Oyedoja K. O. Diagnostics of Bearing Defects Using Vibration Signal // International Journal of Computer and Electrical Engineering. 2012. Vol. 4, no. 6. P. 821–825. DOI: 10.7763/IJCEE.2012.V4.612.
6. Sen A., Majumder M. C., Mukhopadhyay S., Biswas R. K. Condition Monitoring of Rotating Equipment Considering the Cause and Effects of Vibration: A Brief Review // International Journal of Modern Engineering Research (IJMER). 2017. Vol. 7, Iss. 1. P. 36–49.
7. Ciaburro G., Iannace G. Machine learning based methods for acoustic emission testing: A review. Applied Sciences. 2022. Vol. 12, No. 20. 10476. DOI: 10.3390/app122010476.
8. Da Silva R. R., Da S. Costa E., De Oliveira R. C. L., Mesquita A. L. A. Fault Diagnosis in Rotating Machine Using Full Spectrum of Vibration and Fuzzy Logic // Journal of Engineering Science and Technology. 2017. Vol. 12, no. 11. P. 2952–2964.
9. Saberi A. N., Belahcen A., Sobra J., Vaimann T. LightGBM-Based Fault Diagnosis of Rotating Machinery Under Changing Working Conditions Using Modified Recursive Feature Elimination // IEEE Access. 2022. Vol. 10. P. 81910–81925. DOI: 10.1109/ACCESS.2022.3195939.
10. Zhang B., Li W., Tong Z., Zhang M. Bearing fault diagnosis under varying working condition based on domain adaptation // arXiv preprint. 2017. arXiv:1707.09890.
11. Wang Q., Michau G., Fink O. Domain Adaptive Transfer Learning for Fault Diagnosis // 2019 Prognostics and System Health Management Conference (PHM-Paris). IEEE, 2019. DOI: 10.1109/PHM-Paris.2019.00054.
12. Zhang X., Gu G. Fault Diagnosis for Rolling Bearings Under Complex Working Conditions Based on Domain-Conditioned Adaptation // Machines. 2024. Vol. 12, no. 11. Art. 787. DOI: 10.3390/machines12110787.
13. Zhang Q., Lv Z., Hao C., Yan H., Fan Q. Intelligent Fault Diagnosis of Bearings in Unsupervised Dynamic Domain Adaptation Networks Under Variable Conditions // IEEE Access. 2024. Vol. 12. P. 82911–82925. DOI: 10.1109/ACCESS.2024.3413087.
14. Zhong X. Failure Mechanism Information-Assisted Multi-Domain Adversarial Transfer Fault Diagnosis Model for Rolling Bearings under Variable Operating Conditions // Sensors. 2024. Vol. 24, no. 3. Art. 1036. DOI: 10.3390/s24031036.
15. Li X., Wang J., Wang J., Wang J., Li Q., Yu X., Chen J. Research on Unsupervised Domain Adaptive Bearing Fault Diagnosis Method Based on Migration Learning Using MSACNN-IJMMD-DANN // Machines. 2025. Vol. 13. Art. 618. DOI: 10.3390/machines13070618.
16. Jalayer M., Kaboli A., Orsenigo C., Vercellis C. Fault Detection and Diagnosis with Imbalanced and Noisy Data: A Hybrid Framework for Rotating Machinery // Machines. 2022. Vol. 10. Art. 237. DOI: 10.3390/machines10040237.
17. Duman T. B., Bayram B., İnce G. Acoustic Anomaly Detection Using Convolutional Autoencoders in Industrial Processes // Advances in Intelligent Systems and Computing. 2020. Vol. 1028. P. 397–406. DOI: 10.1007/978-3-030-20055-8_41.
18. Nasim F., Masood S., Jaffar A., Ahmad U., Rashid M. Intelligent Sound-Based Early Fault Detection System for Vehicles // Computer Systems Science and Engineering. 2023. Vol. 46, no. 3. P. 3175–3190. DOI: 10.32604/csse.2023.034550.
19. Kawaguchi Y., Imoto K., Koizumi Y., Harada N., Niizumi D., Dohi K., Tanabe R., Purohit H., Endo T. Description and Discussion on DCASE 2021 Challenge Task 2: Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions // Detection and Classification of Acoustic Scenes and Events 2021 : Proceedings. Online, 15–19 Nov 2021. P. 186–190. ISBN 978-84-09-36072-7.
20. Ignjatovska A., Shishkovski D., Pecioski D. Classification of present faults in rotating machinery based on time and frequency domain feature extraction. Vibroengineering Procedia. 2023. Vol. 51. P. 22–28. DOI: 10.21595/vp.2023.23667.
21. Tuleski B. L., Yamaguchi C. K., Stefenon S. F., Coelho L. S., Mariani V. C. Audio-Based Engine Fault Diagnosis with Wavelet, Markov Blanket, ROCKET, and Optimized Machine Learning Classifiers // Sensors. 2024. Vol. 24. Art. 7316. DOI: 10.3390/s24227316.
22. Lüttenberg H., Bartelheimer C., Beverungen D. Designing Predictive Maintenance for Agricultural Machines // Proceedings of the Twenty-Sixth European Conference on Information Systems (ECIS 2018). Portsmouth, UK, 2018. Research Paper No. 153. URL: https://aisel.aisnet.org/ecis2018_rp/153
23. Case Western Reserve University Bearing Data Center: веб-сайт. URL: https://engineering.case.edu/bearingdatacenter (дата звернення: 01.03.2026).
24. Cathy Siyu. Mechanical-datasets. [Dataset]. GitHub. URL: https://github.com/cathysiyu/Mechanical-datasets (дата звернення: 01.03.2026).
References
1. Aulin, V. V., Hrynkiv, A. V., Holovatyi, A. O., Lysenko, S. V., Holub, D. V., Kuzyk, O. V., & Tykhyi, A. A. (2020). Methodological foundations of design and operation of intelligent transportation and manufacturing systems. Lysenko V.F. [in Ukrainian].
2. Matviienko, O. O., Aulin V.V. (2025). Classification of signal types and machining methods for intelligent assessment of the technical mill of mobile machines for agro-industrial production. Central Ukrainian Scientific Bulletin. Technical Sciences, (11(42)_II), 298-312. https://doi.org/10.32515/2664-262X.2025.11(42).2.298-312 [in Ukrainian].
3. Matviienko, O. O., Aulin V.V. (2025). Status and development directions of data architecture for intelligent assessment of the technical condition of mobile machines of agro-industrial enterprises. Central Ukrainian Scientific Bulletin. Technical Sciences, (12(43)_I), 227-237. https://doi.org/10.32515/2664-262X.2025.12(43).1.227-237 [in Ukrainian].
4. Loutas, T. H., Roulias, D., Pauly, E., & Kostopoulos, V. (2011). The combined use of vibration, acoustic emission and oil debris on-line monitoring towards a more effective condition monitoring of rotating machinery. Mechanical Systems and Signal Processing, 25(4), 1339–1352. https://doi.org/10.1016/j.ymssp.2010.11.007
5. Oyedoja, K. O. (2012). Diagnostics of bearing defects using vibration signal. International Journal of Computer and Electrical Engineering, 4(6), 821–825. https://doi.org/10.7763/IJCEE.2012.V4.612
6. Sen, A., Majumder, M. C., Mukhopadhyay, S., & Biswas, R. K. (2017). Condition monitoring of rotating equipment considering the cause and effects of vibration: A brief review. International Journal of Modern Engineering Research (IJMER), 7(1), 36–49.
7. Ciaburro, G., & Iannace, G. (2022). Machine learning based methods for acoustic emission testing: A review. Applied Sciences, 12(20), 10476. https://doi.org/10.3390/app122010476
8. Da Silva, R. R., Da S. Costa, E., De Oliveira, R. C. L., & Mesquita, A. L. A. (2017). Fault diagnosis in rotating machine using full spectrum of vibration and fuzzy logic. Journal of Engineering Science and Technology, 12(11), 2952–2964.
9. Saberi, A. N., Belahcen, A., Sobra, J., & Vaimann, T. (2022). LightGBM-based fault diagnosis of rotating machinery under changing working conditions using modified recursive feature elimination. IEEE Access, 10, 81910–81925. https://doi.org/10.1109/ACCESS.2022.3195939
10. Zhang, B., Li, W., Tong, Z., & Zhang, M. (2017). Bearing fault diagnosis under varying working condition based on domain adaptation (arXiv:1707.09890). arXiv.
11. Wang, Q., Michau, G., & Fink, O. (2019). Domain adaptive transfer learning for fault diagnosis. In 2019 Prognostics and System Health Management Conference (PHM-Paris). IEEE. https://doi.org/10.1109/PHM-Paris.2019.00054
12. Zhang, X., & Gu, G. (2024). Fault diagnosis for rolling bearings under complex working conditions based on domain-conditioned adaptation. Machines, 12(11), 787. https://doi.org/10.3390/machines12110787
13. Zhang, Q., Lv, Z., Hao, C., Yan, H., & Fan, Q. (2024). Intelligent fault diagnosis of bearings in unsupervised dynamic domain adaptation networks under variable conditions. IEEE Access, 12, 82911–82925. https://doi.org/10.1109/ACCESS.2024.3413087
14. Zhong, X. (2024). Failure mechanism information-assisted multi-domain adversarial transfer fault diagnosis model for rolling bearings under variable operating conditions. Sensors, 24(3), 1036. https://doi.org/10.3390/s24031036
15. Li, X., Wang, J., Wang, J., Wang, J., Li, Q., Yu, X., & Chen, J. (2025). Research on unsupervised domain adaptive bearing fault diagnosis method based on migration learning using MSACNN-IJMMD-DANN. Machines, 13, 618. https://doi.org/10.3390/machines13070618
16. Jalayer, M., Kaboli, A., Orsenigo, C., & Vercellis, C. (2022). Fault detection and diagnosis with imbalanced and noisy data: A hybrid framework for rotating machinery. Machines, 10, 237. https://doi.org/10.3390/machines10040237
17. Duman, T. B., Bayram, B., & İnce, G. (2020). Acoustic anomaly detection using convolutional autoencoders in industrial processes. In Advances in Intelligent Systems and Computing (Vol. 1028, pp. 397–406). Springer. https://doi.org/10.1007/978-3-030-20055-8_41
18. Nasim, F., Masood, S., Jaffar, A., Ahmad, U., & Rashid, M. (2023). Intelligent sound-based early fault detection system for vehicles. Computer Systems Science and Engineering, 46(3), 3175–3190. https://doi.org/10.32604/csse.2023.034550
19. Kawaguchi, Y., Imoto, K., Koizumi, Y., Harada, N., Niizumi, D., Dohi, K., Tanabe, R., Purohit, H., & Endo, T. (2021). Description and discussion on DCASE 2021 Challenge Task 2: Unsupervised anomalous sound detection for machine condition monitoring under domain shifted conditions. In Proceedings of the Detection and Classification of Acoustic Scenes and Events 2021 (DCASE 2021 Workshop) (pp. 186–190).
20. Ignjatovska, A., Shishkovski, D., & Pecioski, D. (2023, October 20–21). Classification of present faults in rotating machinery based on time and frequency domain feature extraction. Vibroengineering Procedia, 51, 22–28. https://doi.org/10.21595/vp.2023.23667
21. Tuleski, B. L., Yamaguchi, C. K., Stefenon, S. F., Coelho, L. S., & Mariani, V. C. (2024). Audio-based engine fault diagnosis with wavelet, Markov blanket, ROCKET, and optimized machine learning classifiers. Sensors, 24, 7316. https://doi.org/10.3390/s24227316
22. Lüttenberg, H., Bartelheimer, C., & Beverungen, D. (2018). Designing predictive maintenance for agricultural machines. In Proceedings of the Twenty-Sixth European Conference on Information Systems (ECIS 2018). https://aisel.aisnet.org/ecis2018_rp/153
23. Case Western Reserve University. (n.d.). Bearing Data Center. Case School of Engineering. https://engineering.case.edu/bearingdatacenter
24. Cathy Siyu. (n.d.). Mechanical-datasets [Dataset]. GitHub. https://github.com/cathysiyu/Mechanical-datasets
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Oleksandr Matviienko, Oleksander Livitskyi Andriy Grynkiv
This work is licensed under a Creative Commons Attribution 4.0 International License.
