Expanded Uncertainty Model Based on Whitened Weighting Function

摘要

Numerous measurements and known or typical probability distribution are required for the evaluation of measurement uncertainty by conventional statistics theory. It is difficult to estimate the uncertainty of a measurand if the number of measurements is small and the probability distribution is unclear. A novel method using the whitened weighting function to evaluate expanded uncertainty is proposed to solve these problems. The method permits a very small number of measurements and an unknown probability distribution, and an expanded uncertainty can be easily obtained without calculating the standard uncertainty and the coverage factor. The feasibility of the method is validated with computer-simulation experiments made with different distributions such as normal, Rayleigh, triangular and uniform distributions.