A PDEM-COM framework for uncertainty quantification of backward issues involving both aleatory and epistemic uncertainties

Abstract

Uncertainties that exist in nature or due to lack of knowledge have been widely recognized by researchers and engineering practitioners throughout engineering design and analysis for decades. Though great efforts have been devoted to the issues of uncertainty quantification (UQ) in various aspects, the methodologies on the quantification of aleatory uncertainty and epistemic uncertainty are usually logically inconsistent. For instance, the aleatory uncertainty is usually quantified in the framework of probability theory, whereas the epistemic uncertainty is quantified mostly by non-probabilistic methods. In the present paper, a probabilistically consistent framework for the quantification of both aleatory and epistemic uncertainty by synthesizing the probability density evolution method (PDEM) and the change of probability measure (COM) is outlined. The framework is then applied to the backward issues of uncertainty quantification. In particular, the uncertainty model updating issue is discussed in this paper. A numerical example is presented, and the results indicate the flexibility and efficiency of the proposed PDEM-COM framework.