The results show that vibration-based machine learning can support robust, near-real-time fault identification in mechanical manufacturing environments and highlights the importance of chronological validation, feature engineering over multiple time windows, and the trade-off between predictive performance and deployment efficiency.
Abstract
Artificial intelligence is increasingly used to improve industrial quality control, but its practical value depends on whether models remain accurate under different operating conditions and fault classes. This study evaluates an artificial-intelligence-based workflow for gear quality control using vibration signals measured on a real two-stage reduction gearbox. Two orthogonal vibration channels were analyzed for six health states, three shaft speeds, and two load levels. Because the time-series data were only partly stationary, the dataset was divided chronologically into training and test segments. A 54-feature representation was built from rolling-window statistics and operating variables, and six classifiers were compared: Light Gradient Boosting Machine (LGBM), Extreme Gradient Boosting (XGBM), random forest, decision tree, multilayer perceptron (MLP), and logistic regression. LGBM achieved the best overall accuracy (0.9728) while maintaining substantially lower training time than several competing nonlinear models. Class-wise precision, recall, and F1-score ranged from 0.95 to 1.00, and the nominal response time for most operating-condition transitions was approximately 0.0998 s. The results show that vibration-based machine learning can support robust, near-real-time fault identification in mechanical manufacturing environments. The study also highlights the importance of chronological validation, feature engineering over multiple time windows, and the trade-off between predictive performance and deployment efficiency. Because the validation dataset originates from one gearbox platform, the results should be interpreted as promising internal evidence rather than as proof of universal industrial robustness.
A Composite Health Index (CHI) is developed to transform multi-motor sensor data into an interpretable machine-level degradation indicator and is used to train ensemble machine learning models including Random Forest, Extra Trees, and XGBoost.
Ahmet Pişmişoğlu, Erkan Caner Ozkat, M. Konar· Eksploatacja I Niezawodnosc-...· 0 citations
The results show that the proposed Condition Monitoring (CM) approach significantly reduces resource waste and prevents costly downtime, offering a practical and scalable asset management model for industrial applications.
A. Oner, Meral Bayraktar· Italian National Conference...· 0 citations
The near-perfect linear separability indicates that the dataset’s binary, controlled-laboratory labelling rather than intrinsic bearing-degradation physics drives the clean classification, and validation on 500–1000 or more samples with progressive-degradation labelling is essential before any operational claim can be supported.
P. Pugazhendi, Vinoth Vishwanathan, Aadil Arshad Ferhath et al.· Engineering Research Express· 0 citations
A machine-learning-based system for fault identification and classification based on vibration, slip ratio, and belt-tension measurements obtained from a controlled experimental setup demonstrates machine learning’s potential for scaled implementation in industrial predictive maintenance settings and establishes its applicability for belt-pulley systems’ real-time health evaluation.
A. Sivathanu· Proceedings of the Instituti...· 0 citations
Reliable fault diagnosis in rotating machinery is challenging due to the nonlinear and non-stationary nature of vibration signals. Although time–frequency analysis is widely used, it cannot capture the cross-scale coupling between amplitude-modulated (AM) and frequency-modulated (FM) components that carry essential diagnostic information. This study applies Holo-Hilbert Spectrum Analysis (HHSA) to extract amplitude–frequency modulation features and integrates them with six machine learning classifiers to identify four fault conditions. Random Forest, K-Nearest Neighbors, and Logistic Regression achieve accuracies of up to 99.95%, yielding higher accuracy than Fast Fourier Transform-based features. The proposed framework employs an HHSA-based feature extraction pipeline that effectively captures AM–FM coupling in nonlinear vibration signals. It also provides higher discriminative capability than traditional spectral approaches and maintains robustness across multiple classifiers. This method offers high diagnostic accuracy and strong potential for industrial predictive maintenance. Future work will focus on improving computational efficiency and evaluating the framework under more diverse and realistic operating conditions.
Received: 10 September 2025 | Revised: 20 April 2026 | Accepted: 10 June 2026
Conflicts of Interest
The authors declare that they have no conflicts of interest to this work.
Data Availability Statement
The VBL-VA001 datasets that support the findings of this study are openly available at https://doi.org/10.1007/s42417-023-00959-9, reference number [44].
Author Contribution Statement
Van-Trung Nguyen: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing - original draft, Writing - review & editing, Visualization. Ba-Tan Le: Investigation, Data curation, Writing - original draft, Writing - review & editing. Van-Phuong Dao: Methodology, Validation, Writing - review & editing.
Van-Trung Nguyen, Ba-Tan Le, van-Phuong Dao· Journal of Computational and...· 0 citations
In computer numerical control (CNC) systems, the remaining useful life (RUL) of cutting tools is an important factor to be considered to prevent unplanned tool stoppages, premature tool replacement, and to ensure machining quality. This research suggests a group-aware and explainable machine learning-based approach to normalized RUL prediction using vibration, current and operating condition features. In contrast to the traditional CNC RUL studies performed by observation level random splitting, the proposed method ensures complete separation of tools between training, validation, and independent test partitions to avoid any information leakage between partitions at the level of the tools. Seven stable predictors were identified: five vibration variables, one current variable and the hardness of the workpiece, by using the stability-based feature selection. Four methods of repeated group validation, formal validation, macro-tool error and worst-tool performance were used to compare baseline and nonlinear regression models. Random Forest was chosen as the final model, with the validation RMSE value of 0.1676 and test RMSE value of 0.2377 and test R^2of 0.3417. The framework showed that unseen-tool generalization could be measured, but there was a variation between test tools that showed the difficulty of cross-tool degradation variability. The features related to vibration accounted for 66.29% of the total model importance, while the workpiece hardness and current variability accounted for 26.72% and 6.99% respectively. When the dominant predictive information was extracted from the vibration, the best performance was achieved by sensor-group ablation, and the complete feature configuration gave the best performance in terms of validation.
Dr. Jugulkar Lalitkumar Maikulal, Avinash H. Kolekar, Dr. Sushil Kumar Choudhary et al.· Journal of Intelligent Decis...· 0 citations
