A Novel Computational Methodology for Uncertainty Quantification in Prognostics Using The Most Probable Point Concept

Abstract

This paper develops a novel computational approach to quantify the uncertainty in prognostics in the context of condition- based monitoring. Prognostics consists of two major steps; first, it is necessary to estimate the state of health at any time instant, and then, it is required to predict the remaining useful life of the engineering component/system of interest. While the topic of estimation has been addressed through different types of Bayesian tracking techniques, this paper primarily focuses on the second aspect of future prediction and remain- ing useful life computation, which is influenced by several sources of uncertainty. Therefore, it is important to identify these sources of uncertainty, and quantify their combined ef- fect on the remaining useful life prediction. The computation of uncertainty in remaining useful life can be treated as an uncertainty propagation problem which can be solved using probabilistic techniques. This paper investigates the use of the Most Probable Point approach (which was originally developed to estimate the failure probability of structural systems) for calculating the probability distribution of the remaining useful life prediction. The proposed methodology is illustrated using a battery which is used to power an un- manned aerial vehicle.