Automatic Predictive Maintenance System for Machining Tools using Semi-Supervised Learning
DOI:
https://doi.org/10.17979/ja-cea.2026.47.13802Keywords:
Machine Learning, Condition Monitoring, Fault detection and diagnosis, Bayesian methods, Intelligent maintenance systems, Statistical methods/signal analysis for FDIAbstract
Tool condition monitoring is fundamental for predictive maintenance in machining equipment. However, the lack of labeled historical data and the severe imbalance between normal operating cycles and failures hinder the implementation of supervised models. This paper presents an intelligent system with semi-supervised adjustment for classification and labeling of machining cycles based on a set of statistical features extracted from the torque signal. The algorithm reduces dimensionality via Principal Component Analysis (PCA) and employs Gaussian Mixture Models (GMM) to identify nominal behavior. Subsequently, it isolates anomalies using a Novelty Score, generates synthetic data to balance these minority classes, and applies a final Bayesian classifier. This results in a model that is ready for use in production monitoring or for labelling datasets during the model training phase. The results demonstrate the algorithm’s capability to cluster and classify, as well as the potential for semi-supervision
provided by adjusting the Novelty Score threshold.
References
Belmokhtar, S., Bratcu, A., Dolgui, A., 2006. Modular machining line design and reconfiguration: Some optimization methods. In: Kordic, V., Lazinica, A., Merdan, M. (Eds.), Manufacturing the Future. IntechOpen, London, Ch. 5. URL: https://doi.org/10.5772/5047 DOI: 10.5772/5047
Bruneau, P., Gelgon, M., Picarougne, F., 2010. Parsimonious reduction of gaussian mixture models with a variational-bayes approach. Pattern Recognition 43 (3), 850–858. URL: https://www.sciencedirect.com/science/article/pii/S0031320309003021 DOI: https://doi.org/10.1016/j.patcog.2009.08.006
Chai, C., Li, G., 2020. Human-in-the-loop techniques in machine learning. IEEE Data Eng. Bull. 43, 37–52. URL: https://api.semanticscholar.org/CorpusID:240291940
Dhasaian, M. R. S., Parnitha, Prasana, Naini, 11 2025. Unsupervised and semi supervised machine learning frameworks for multi class tool wear recognition. American Journal of Management and IOT Medical Computing 4, 22–28. URL: https://ajmimc.com/journal/index.php/ajmimc/ article/view/129 DOI: 10.64751/AJMIMC.2025.V4.N4(1).PP22-28
Fomin, D., Lalanda, P., Morand, D., 2024. Automating the production of machine learning models for imbalanced datasets - application to industry 4.0. 2024 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events, PerCom Workshops 2024, 39–44. URL: https://ieeexplore.ieee.org/document/10503057 DOI: 10.1109/PERCOMWORKSHOPS59983.2024.10503057
Kuntoğlu, M., Aslan, A., Pimenov, D. Y., Üsame Ali Usca, Salur, E., Gupta, M. K., Mikolajczyk, T., Giasin, K., Kapłonek, W., Sharma, S., 12 2020. A review of indirect tool condition monitoring systems and decision-making methods in turning: Critical analysis and trends. Sensors 2021 21, 108. URL: https://www.mdpi.com/1424-8220/21/1/108/ DOI: 10.3390/S21010108
Mallioris, P., Aivazidou, E., Bechtsis, D., 6 2024. Predictive maintenance in industry 4.0: A systematic multi-sector mapping. CIRP Journal of Manufacturing Science and Technology 50, 80–103. URL: https://www.sciencedirect.com/science/article/pii/S1755581724000221#ab0010 DOI: 10.1016/J.CIRPJ.2024.02.003
Sama, A., Pardo-Ayala, D. E., Cabestany, J., Rodr´ıguez-Molinero, A., 2010. Time series analysis of inertial-body signals for the extraction of dynamic properties from human gait. In: The 2010 International Joint Conference on Neural Networks (IJCNN). pp. 1–5. DOI: 10.1109/IJCNN.2010.5596663
Schafer, R., Jul. 2011. What Is a Savitzky-Golay Filter? [Lecture Notes]. IEEE Signal Processing Magazine 28 (4), 111–117. DOI: 10.1109/MSP.2011.941097
Thomas, D.,Weiss, B., 12 2021. Maintenance costs and advanced maintenance techniques in manufacturing machinery: Survey and analysis. International journal of prognostics and health management 12. URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC9890517/ DOI: 10.36001/IJPHM.2021.V12I1.2883
Tran, M. Q., Doan, H. P., Vu, V. Q., Vu, L. T., 2 2023. Machine learning and iot-based approach for tool condition monitoring: A review and future prospects. Measurement 207, 112351. URL: https://www.sciencedirect.com/science/article/pii/ S0263224122015470#s0005 DOI: 10.1016/J.MEASUREMENT.2022.112351
Wang, X., Wirth, A., Wang, L., 10 2007. Structure-based statistical features and multivariate time series clustering. Proceedings - IEEE International Conference on Data Mining, ICDM, 351–360. URL: https://ieeexplore.ieee.org/document/4470259 DOI: 10.1109/ICDM.2007.103
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Pablo Sanchis, Sergio Garcia-Nieto, Ignacio Cebreiro, Juan José Sanz
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
