A new remaining useful life prediction approach for independent component based on the Wiener process and Bayesian estimating paradigm

Abstract

Online remaining useful life (RUL) prediction is the key of prognostics and health management. Aiming at the problem that the online RUL prediction methods in existing literature usually rely on the history information of the other specimens of the same kind, a new online RUL prediction approach for independent component is proposed in this paper. The offline information for a certain independent component is collected and utilized to confirm the estimates of the parameters by maximum likelihood estimation (MLE) method, and then the obtained estimates are updated employing the Bayesian mechanism with the real time condition monitoring data. The RUL is defined on the concept of first hitting time; furthermore, the exact analytical solution for RUL distribution is deduced. For the validation of our proposed RUL prediction approach, a numerical example is provided. The results reflect that our approach gains a higher RUL prediction accuracy than some other online RUL predicting method in the existing literature.