Abstract
Condition monitoring is an important prognostic tool to determine the current operation status of a system/device and to estimate the distribution of the remaining useful life. This article proposes a two-phase model to characterize the degradation process of rotational bearings. A Bayesian framework is used to integrate historical data with up-to-date in situ observations of new working units to improve the degradation modeling and prediction. A new approach is developed to compute the distribution of the remaining useful life based on the degradation signals, which is more accurate compared with methods reported in the literature. Finally, extensive numerical results demonstrate that the proposed framework is effective and efficient.