Online updating with a wiener-process-based prediction model using UKF algorithm for remaining useful life estimation

Remaining useful life (RUL) prediction is an important part in prognostics and health management. This paper addresses the problem of estimating RUL from observed degradation data. A Wiener-process-based model for online lifetime prediction is developed to achieve this aim. In this paper a Wiener process with non-linear drift and measurement error is used to characterize the degradation process and for online RUL prediction, we exploit an unscented Kalman filter (UKF) based approach for lifetime and parameter jointly estimation. We treat the parameters of the Wiener process as the state of UKF so it can be updated once new data come. A practical case study for gyros in inertial navigation system is provided and results show that our developed model performs well in real time RUL estimation.