Obtaining Uncertainty Estimates Compatible with Estimates of Monte Carlo Method | IEEE Conference Publication | IEEE Xplore

Obtaining Uncertainty Estimates Compatible with Estimates of Monte Carlo Method


Abstract:

The proposed approaches to the implementation of the algorithm for processing the results and the measurement uncertainty evaluation based on the Bayesian method are cons...Show More

Abstract:

The proposed approaches to the implementation of the algorithm for processing the results and the measurement uncertainty evaluation based on the Bayesian method are considered. Expressions to obtain unbiased estimates of the measurand, its standard and expanded uncertainties are given. It is shown that to obtain standard and expanded uncertainty, it is necessary to take into account the kurtosis of the input quantities distributions.
Date of Conference: 27-29 May 2019
Date Added to IEEE Xplore: 01 August 2019
ISBN Information:
Conference Location: Smolenice, Slovakia

1. Introduction

Standardization of measurement uncertainty evaluation is one of the most important tasks of traceability assurance at the international level. To solve this problem, in 1993, the Guide to the expression of uncertainty in measurement (GUM) [1] developed by leading international organizations was promulgated. The GUM is based on the implementation of a model approach to the measurement uncertainty evaluation based on the following methods: the law of propagation of uncertainty (LPU); the central limit theorem of the theory of probability (CLT) and the apparatus of the number of degrees of freedom. The main disadvantages of the GUM are: a bias in the estimates of the numerical value and the uncertainty of the measurand for a non-linear measurement model and large standard uncertainties of the input quantities; independence of estimates of expanded uncertainty from the distributions of input quantities. The elimination of the above disadvantages of CLT led to the improvement of the model approach based on the law of propagation of distributions (LPD) using the Monte Carlo method (MCM) [2]–[3], which is a numerical implementation of the Bayesian approach.

References

References is not available for this document.