Remaining Useful Life Prediction for Anti-friction Bearings Based on Envelope Spectrum and Extended Kalman Filter

Anti-friction bearings (AFB) are essential parts of many rotating machines. It is also well known that the fault in the bearings keeps developing during machine operation. This bearing fault can propagate further and trigger other faults within the machine, and eventually lead to the machine failures and shutdown. Hence, the early prediction of remaining useful life (RUL) plays a significant role to optimize maintenance schedule for overhauls and avoiding failures. Recent research studies in the literature have used data-driven models using vibration-based indicators (HI) to estimate the RUL. In the current study, an envelope analysis-based indicator is used that truly reflects bearing conditions only. Then only a few initial vibration measurements are used once the bearing defect is identified to estimate RUL using the extended Kalman filter (EKF). The RUL prediction method is applied to the experimental vibration data on a rig.