Repeatability assessment of FBG-based strain measurements for bearing fault diagnostics

Rolling element bearings are essential to rotating machinery, yet failures remain a leading cause of downtime and maintenance. Traditional electromechanical sensor-based methods for rotating machinery and bearing diagnostics often suffer from signal attenuation, poor fault localization, and susceptibility to electromagnetic interference (EMI). Fiber Bragg grating (FBG) sensors instead offer a compact, EMI-immune alternative capable of capturing localized strain at the fault site. However, the repeatability of the diagnostics capabilities under realistic conditions remains largely unexplored. In this work, we introduce a non-invasive, repeatable adhesive-based installation method for integrating FBG sensors directly onto bearing raceways. We evaluate the robustness of this approach using healthy and artificially damaged bearings, comparing strain signals from multiple sensor and bearing re-installations. Key fault indicators, such as ball pass frequency outer and Inner, are analyzed across test cases. Results show that while damage-induced features are detectable, substantial variability arises due to installation conditions, particularly on the rotating inner race. Strain signals from sensors in the load zone exhibited the lowest uncertainty and best fault discrimination. Our findings demonstrate the feasibility of repeatable, direct FBG-based bearing diagnostics and underscore the need for tailored metrics to maximize detection reliability in real-world applications.