Novel Model-Based Framework for Multi-Fault Analysis in Ball Bearings

This paper presents a novel framework integrating fault models into a semi-physical, non-linear model calibrated with real sensor data to enhance fault analysis in ball- bearings. Unlike conventional methods that separate physics-based modeling and data-driven diagnostics, this approach merges both to improve fault assessment and predictive maintenance. The model accounts for degradation mechanisms such as wear, crack propagation, and creep, capturing their effects on mass, stiffness, and damping. This framework offers a reliable tool for bearing health monitoring, facilitating early fault detection and reducing industrial operational risks.