Uncertainty Quantification in Prognostic Health Management Systems

Abstract

Uncertainty plays a role in nearly all aspects of prognostic health management (PHM) systems. Aleatory uncertainty from inherently-variable inputs such as material properties, epistemic uncertainty from a lack of knowledge about the system and its inputs, and ontological uncertainty due to completely unknown factors must all be accounted for in order to provide the most accurate assessment of the health of the monitored system. Northrop Grumman Innovation Systems (NGIS) develops, produces, and provides sustainment of solid rocket motor systems for the aerospace and defense industry and has extensive experience in applying uncertainty quantification (UQ) principles to complicated numerical simulations and analyses. In this paper, lessons learned by NGIS on UQ simulations and analyses are presented, and their applicability to PHM systems is explored. Methods for measuring and tracking uncertainty through the PHM predictive train are presented, as is a Monte-Carlo-based method for performing prognostic numerical calculations, which accounts for and quantifies epistemic, ontological, and aleatory uncertainty.