Application of the LeJEPA self-supervised learning technique for the diagnosis of rotating looseness faults using vibration data

• João Paulo Vieira UFSC
• Victor Afonso Bauler UFSC
• Thiago Carvalho Bittencourt UFSC
• Bruno Forte de Lunardi Pinto UFSC
• Karoline da Rocha UFSC
• Diego Marques do Carmo UFSC
• Danilo Silva UFSC

DOI: https://doi.org/10.5753/eniac.2026.29905

Resumo

Rotating looseness is an important industrial fault, yet the limited availability of public datasets has restricted the investigation of machine learning (ML) approaches for its diagnosis. This work investigates self-supervised learning (SSL) for rotating looseness diagnosis in bearing housings using an industrial vibration dataset. A LeJEPA-based framework is evaluated under a machine-wise protocol, where models are tested on unseen machines. Results show that SSL consistently improves deep learning (DL) performance in low-label settings, with gains of up to 13% to the DL baseline and a mean AUROC of 89.12% ± 4.78%. As more labeled data becomes available, a strong classical ML baseline prevails, reaching a mean AUROC of 90.73% ± 3.98%. These findings highlight the potential of SSL for industrial fault diagnosis under realistic deployment conditions, specially when few data is available.

Referências

Abburi, H., Chaudhary, T., Ilyas, S. H. W., Manne, L., Mittal, D., Williams, D., Snaidauf, D., Bowen, E., and Veeramani, B. (2023). A closer look at bearing fault classification approaches. In Annual Conference of the PHM Society, volume 15.

Balestriero, R. and LeCun, Y. (2025). LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics. Version Number: 3.

González, M., Díaz, V. G., Pérez, B. L., G-Bustelo, B. C. P., and Anzola, J. P. (2023). Bearing fault diagnosis with envelope analysis and machine learning approaches using cwru dataset. IEEE Access, 11:57796–57805.

Guillame-Bert, M., Bruch, S., Stotz, R., and Pfeifer, J. (2023). Yggdrasil decision forests: A fast and extensible decision forests library. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, KDD ’23, page 4068–4077. ACM.

Jung, J., Lee, S. B., Lim, C., Cho, C.-H., and Kim, K. (2016). Electrical Monitoring of Mechanical Looseness for Induction Motors With Sleeve Bearings. IEEE Transactions on Energy Conversion, 31(4):1377–1386.

Kapoor, S. and Narayanan, A. (2023). Leakage and the reproducibility crisis in machine-learning-based science. Patterns, 4(9):100804.

Liang, Q., Liang, X., Zhang, M., Liu, C., and Zhang, T. (2025). A Novel Comprehensive Semi-Supervised Learning Method for Fault Diagnosis Under Extremely Low Label Rate. IEEE Transactions on Instrumentation and Measurement, 74:1–14.

Lu, W., Wang, Y., Wang, W., Zhang, S., and Cai, Z. (2026). A Multi-Task Self-Supervised Fault Diagnosis Method for Rotating Machinery Under Limited Labeled Data. Quality and Reliability Engineering International, page qre.70272.

Randall, R. B. (2021). Vibration-based condition monitoring: industrial, automotive and aerospace applications. John Wiley & Sons.

Smith, W. A. and Randall, R. B. (2015). Rolling element bearing diagnostics using the case western reserve university data: A benchmark study. Mechanical Systems and Signal Processing, 64-65:100–131.

Suechoey, B., Siriporananon, S., Chupun, P., Boonkhun, C., and Chompooinwai, C. (2021). Performance analysis and fault classification in a large electric motor using vibration assessment technique. International Journal of Intelligent Engineering and Systems, 14:124–133.

Tang, J., Xiao, J., Chen, W., Li, X., Wei, C., Ding, X., and Huang, W. (2024). A prior knowledge-enhanced self-supervised learning framework using time-frequency invariance for machinery intelligent fault diagnosis with small samples. Engineering Applications of Artificial Intelligence, 133:108503.

Tsamardinos, I., Rakhshani, A., and Lagani, V. (2015). Performance-estimation properties of cross-validation-based protocols with simultaneous hyper-parameter optimization. International Journal on Artificial Intelligence Tools, 24(05):1540023.

Varejão, I. M. D. S., Costa, L. G. D. O., Silva, L. H. P. D., Rodrigues, A., Ribeiro, M. P., Varejão, F. M., and Oliveira-Santos, T. (2025). The similarity bias problem: What it is and how it impacts vibration based intelligent fault diagnosis. Mechanical Systems and Signal Processing, 235:112822. Publisher: Elsevier BV.

Vieira, J. P., Bauler, V. A., Rosa, R. K., and Silva, D. (2025). Towards a more realistic evaluation of machine learning models for bearing fault diagnosis. Version Number: 4.

Vieira, J. P., Rosa, R. K., Bauler, V. A., Braga, D., and Silva, D. (2024). Bearing fault diagnosis using machine learning and a novel set of fault-related spectral features. In Anais do XXI Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2024), pages 120–131, Brasil. Sociedade Brasileira de Computação - SBC.

Wang, H., Wang, X., Yang, Y., Gryllias, K., and Liu, Z. (2024). A Few-Shot Machinery Fault Diagnosis Framework Based on Self-Supervised Signal Representation Learning. IEEE Transactions on Instrumentation and Measurement, 73:1–14.

Wei, M., Liu, Y., Zhang, T., Wang, Z., and Zhu, J. (2021). Fault Diagnosis of Rotating Machinery Based on Improved Self-Supervised Learning Method and Very Few Labeled Samples. Sensors, 22(1):192.

Wheat, L., Mohrenschildt, M. V., Habibi, S., and Al-Ani, D. (2024). Impact of Data Leakage in Vibration Signals Used for Bearing Fault Diagnosis. IEEE Access, 12:169879–169895. Publisher: Institute of Electrical and Electronics Engineers (IEEE).

Yan, Z. and Liu, H. (2022). SMoCo: A Powerful and Efficient Method Based on Self-Supervised Learning for Fault Diagnosis of Aero-Engine Bearing under Limited Data. Mathematics, 10(15):2796. Publisher: MDPI AG.

Yan, Z., Liu, H., Tao, L., Ma, J., and Cheng, Y. (2023). A Universal Feature Extractor Based on Self-Supervised Pre-Training for Fault Diagnosis of Rotating Machinery under Limited Data. Aerospace, 10(8):681.

Zaidi, S. H. H., Shenfield, A., Zhang, H., and Ikpehai, A. (2026). A Systematic Review of Anomaly and Fault Detection Using Machine Learning for Industrial Machinery. Algorithms, 19(2):108.

Zhang, W., Peng, G., Li, C., Chen, Y., and Zhang, Z. (2017). A New Deep Learning Model for Fault Diagnosis with Good Anti-Noise and Domain Adaptation Ability on Raw Vibration Signals. Sensors, 17(2):425. Publisher: MDPI AG.

Zhou, A. Y. and Barati Farimani, A. (2024). FaultFormer: Pretraining Transformers for Adaptable Bearing Fault Classification. IEEE Access, 12:70719–70728.