Handling Noisy Labels in Industrial Scenarios: A Data-Centric Approach for Lubrication Fault Diagnosis

• Bruno F. L. Pinto UFSC
• Thiago C. Bittencourt UFSC
• Victor Afonso Bauler UFSC
• João Paulo Vieira UFSC
• Karoline da Rocha UFSC
• Diego Marques do Carmo UFSC
• Danilo Silva UFSC

DOI: https://doi.org/10.5753/eniac.2026.31059

Resumo

O diagnóstico de falhas industriais esbarra frequentemente na inconsistência de rótulos extraídos de laudos de manutenção. Para contornar este problema, este trabalho investiga uma pipeline centrada nos dados para o diagnóstico de falhas de lubrificação em mancais de transportadores de correia, usando sinais reais de vibração. A abordagem adapta ao domínio de vibração um método de detecção de rótulos ruidosos previamente proposto na literatura, usando uma rede neural convolucional para ranquear rótulos suspeitos antes da correção e filtragem automáticas. A melhor configuração elevou a AUROC da Random Forest de 87,8% para 91,3%, mostrando que o tratamento dos rótulos pode melhorar o diagnóstico de falhas de lubrificação em dados industriais.

Referências

Alharbi, F., Luo, S., Zhang, H., Shaukat, K., Yang, G., Wheeler, C. A., and Chen, Z. (2023). A brief review of acoustic and vibration signal-based fault detection for belt conveyor idlers using machine learning models. Sensors, 23(4).

Bauler, V., Cordioli, J., Silva, D., and Braga, D. (2024). Bearing fault detection in conveyor belt drums using machine learning. Annual Conference of the PHM Society, 16.

DeLong, E. R., DeLong, D. M., and Clarke-Pearson, D. L. (1988). Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics, pages 837–845.

Duan, J., Xia, Z., Xu, G., Dan, Z., Zhang, S., and Liang, L. (2018). Lubrication condition monitoring and evaluation of rolling bearing based on acoustic emission. In 2018 IEEE International Conference on Prognostics and Health Management (ICPHM), pages 1–7.

Elforjani, M. (2018). Diagnosis and prognosis of slow speed bearing behavior under grease starvation condition. Structural Health Monitoring, 17(3):532–548.

Jardine, A. K. S., Lin, D., and Banjevic, D. (2006). A review on machinery diagnostics and prognostics implementing condition-based maintenance. Mechanical Systems and Signal Processing, 20(7):1483–1510.

Jiang, G., Zhang, J., Bai, X., Wang, W., and Meng, D. (2024). Which is more effective in label noise cleaning, correction or filtering? Proceedings of the AAAI Conference on Artificial Intelligence, 38:12866–12873.

Loshchilov, I. and Hutter, F. (2019). Decoupled weight decay regularization. In International Conference on Learning Representations (ICLR).

Northcutt, C. G., Jiang, L., and Chuang, I. L. (2021). Confident learning: Estimating uncertainty in dataset labels. Journal of Artificial Intelligence Research, 70:1373–1411.

Pickler, H., Kamassury, J. K., and Silva, D. (2025). Benchmarking noisy label detection methods. arXiv preprint arXiv:2510.16211.

Randall, R. B. and Antoni, J. (2011). Rolling element bearing diagnostics—a tutorial. Mechanical Systems and Signal Processing, 25(2):485–520.

Sikora, M. V. (2024). Comparison of machine learning binary classifiers for detection of gear defects. Master’s thesis, Universidade Federal de Santa Catarina, Florianópolis, Brazil.

Song, H., Kim, M., Park, D., Shin, Y., and Lee, J.-G. (2023). Learning from noisy labels with deep neural networks: A survey. IEEE Transactions on Neural Networks and Learning Systems, 34:8135–8153.

Vieira, J. P., Bauler, V. A., Rosa, R. K., and Silva, D. (2026). Towards a more realistic evaluation of machine learning models for bearing fault diagnosis. Mechanical Systems and Signal Processing, 258:114640.

Zha, D., Bhat, Z. P., Lai, K.-H., Yang, F., Jiang, Z., Zhong, S., and Hu, X. (2025). Datacentric artificial intelligence: A survey. ACM Comput. Surv., 57(5).

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, 17:425.