Expectation-Maximization Method for Data-Based Estimation of the Cantilever Beam End-to-End Problem

摘要

This paper describes an uncertainty quantification approach to address the cantilever beam end-to-end problem posed by Sandia National Laboratories. The variations of the deflection at the tip of a beam can be observed under various sources of uncertainties, which include uncertainties in geometry and material properties, systematic and random measurement errors, and loading variability. To this end, in this paper, the probability distribution for the deflection is characterized and its parameters are estimated using the expectation-maximization-based maximum likelihood method.