Significance, interpretation, and quantification of uncertainty in prognostics and remaining useful life prediction

This paper analyzes the significance, interpretation, and quantification of uncertainty in prognostics, with an emphasis on predicting the remaining useful life of engineering systems and components. Prognostics deals with predicting the future behavior of engineering systems, and is affected by various sources of uncertainty. In order to facilitate meaningful prognostics-based decision-making, it is important to analyze how these sources of uncertainty affect prognostics, and thereby, compute the overall uncertainty in the remaining useful life prediction. This paper investigates the classical (frequentist) and subjective (Bayesian) interpretations of uncertainty and their implications on prognostics, and argues that the Bayesian interpretation of uncertainty is more suitable for condition-based prognostics and health monitoring. It is also demonstrated that uncertainty quantification in remaining useful life prediction needs to be approached as an uncertainty propagation problem. Several uncertainty propagation methods are discussed in this context, and the practical challenges involved in such uncertainty quantification are outlined.