PROJECT TITLE :

Benchmark Test Distributions for Expanded Uncertainty Evaluation Algorithms

ABSTRACT:

Expanded uncertainty estimation is normally required for mission-important applications, e.g., those involving health and safety. It helps to induce a distribution range of the specified confidence level for the uncertainty analysis of a system. There are a number of accessible techniques to estimate the expanded uncertainty. However, there's currently no commonly accepted benchmark check distribution set adopted to compare the performances of different techniques after they are used to estimate the expanded uncertainty. While not such a standard benchmarking platform, the relative reliability of a particular technique compared to alternative techniques will be untrustworthy. To address the shortcoming, this paper proposes a set of analytically derived benchmark check distributions. It goes on to point out the advantages of using them by comparing the performance of existing distribution fitting techniques when applied to the instant-primarily based expanded uncertainty analysis. The most commonly used moment-primarily based distribution fitting techniques, like Pearson, Tukey’s gh, Cornish–Fisher expansion, and extended generalized lambda distributions, are utilized as check cases during this paper. The test distribution set proposed during this paper provides a typical benchmarking platform for metrologists aiming to assess the performance of various expanded uncertainty estimation techniques. Results from the performance comparison would help practitioners to create a better choice of a distribution fitting technique that may best suit their respective systems.