Vibration-Based Condition Monitoring of Tapered Roller Bearings Using Kurtosis and ANOVA

This study develops a robust diagnostic framework for early fault detection in tapered roller bearings (TRBs) using vibration-based analysis. It explores the effect of operating parameters on vibration kurtosis to enable fault diagnosis in the 30205J2/Q, 30206J2/Q, and 30207J2/Q TRBs. The proposed Taguchi L27 orthogonal experimental design analyzes the effects of speed, load, unbalance, bearing type, and defect severity on kurtosis. Both the Inner Race Defect Model (IRDM) and Outer Race Defect Model (ORDM) demonstrate high predictive accuracy with R² values of 98.68% and 97.61% respectively. Analysis of variance (ANOVA) results indicate that the bearing type significantly impacts IRDM, while speed and defect geometry dominate ORDM. Time and frequency-domain analysis reveal distinct vibration patterns for effective fault identification. The interaction of speed, load, defect type, unbalance and bearing type significantly influenced kurtosis, highlighting the need for diagnostic strategies. Predictive kurtosis thresholds were established to enable effective condition monitoring and reducing failures.