FAST ESTIMATION AND UNCERTAINTY QUANTIFICATION IN ELECTRICAL CAPACITANCE TOMOGRAPHY USING SURROGATE TECHNIQUES

摘要

The need for uncertainty quantification (UQ) in metrology has seen serious research efforts and is of ever-growing interest in order to quantify the quality of measurement results. This especially holds for indirect measurement problems. Bayesian methods allow a natural and universal access for UQ but become computationally expensive for scenarios with a complex interaction between the unknown quantity x and the raw measurements d. In this paper we present a surrogate approach for fast estimation and UQ of material distributions in the inverse problem of electrical capacitance tomography using a recursive Bayesian estimator.