Enhanced Fault Detection in High-Speed Train Bearings Using Empirical Mode Decomposition (EMD) and Kurtosis-Based IMF Selection: A Test Bench Approach

¹ Complex Systems and Interactions, Ecole Centrale of Casablanca, Bouskoura, Morocco
² Université de Pau et des Pays de l’Adour, E2S UPPA, CNRS UMR 5254, IPREM, Pau, France
³ Société Marocaine de Maintenance des Rames à Grande Vitesse (SIANA), Tanger, Morocco
⁴ Office National des Chemins de Fer (ONCF), Rabat-Salé, Morocco

^* e-mail: meryem.abtane@centrale-casablanca.ma

Published online: 30 June 2025

Abstract

The critical operating conditions of high-speed trains (HSTs) increase the occurrence of mechanical faults, particularly in key components such as axle bearings. To enhance fault detection and prevention, our study begins with controlled experimental simulations performed on a dedicated test rig at the Complex Systems and Interactions (CSI) Laboratory of Ecole Centrale Casablanca. This setup enables a systematic investigation of various types of bearing faults under well-defined conditions. The proposed methodology utilizes Empirical Mode Decomposition (EMD) to decompose vibration signals into Intrinsic Mode Functions (IMFs). A kurtosis- based selection criterion is then applied to identify the IMF that best highlights fault-related features. This approach enhances the precision of fault detection and the characterization of bearing defects. It demonstrates strong potential for improving diagnostic capabilities in both conventional rolling-element bearings and future applications involving axle bearing fault detection in high-speed rail systems.