Comparative Evaluation of Machine Learning Algorithms for Fault Diagnosis in Automotive Press Lines

Abstract

Highlights A machine learning-based condition monitoring framework was developed and validated using year-long data collected from industrial automotive transfer presses. Random Forest outperformed five competing classifiers, achieving 95% diagnostic accuracy with high precision (97%) and low false-positive rates under real production conditions. Synchronized vibration and encoder data enabled angular-domain localization of individual damaged gear teeth, allowing component-level fault identification. The proposed methodology provides a practical and scalable solution for predictive maintenance of transfer press gearboxes operating under non-stationary industrial conditions. What are the main findings? A data-driven fault diagnosis framework using multimodal sensor inputs achieved high-fidelity gear fault detection, with Random Forest outperforming all models (95% accuracy), followed by k-NN (86%). By synchronizing the diagnostic network directly with the encoder, this framework achieved a level of resolution sharp enough to pinpoint component-level degradation at the tooth-26 domain under actual industrial operations. What are the implications of the main findings? Our findings validate that deploying nonlinear, ensemble-based learning models provides a highly dependable barrier against false alarms and overlooked failures within intricate production settings. This strategy lays down a viable foundation for next-generation, self-governing maintenance pipelines, easing the transition into continuous monitoring and future remaining useful life (RUL) forecasting. Abstract Minimizing unplanned downtime is critical for maintaining productivity in modern manufacturing. While combining sensor networks with machine learning provides a practical way to detect mechanical failures early, conventional data-driven diagnostics often fail during highly transient stamping operations. This failure stems from severe spectral smearing and signal distortions caused by fluctuating process loads and variable operating speeds. To address these limitations, we present a field-tested fault diagnosis (FD) framework deployed in an active automotive components plant. Over a twelve-month observation period, we collected raw vibration and process data from two operational transfer presses, building a comparative dataset that captures both localized gear damage and healthy baseline dynamics. After preprocessing the data to isolate signal anomalies, we systematically evaluated the diagnostic performance of six algorithms: SVM, Random Forest, Naive Bayes, k-NN, Decision Trees, and Logistic Regression. By integrating angle-based position data from a high-resolution encoder, the developed framework successfully pinpointed specific defective gear teeth. Ultimately, the Random Forest model outperformed the others, delivering the most robust detection accuracy under real-world factory conditions. These 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.