DEVELOPMENT OF A HYBRID METHOD FOR SELECTING AND DETERMINING THE DISTRIBUTION TYPE OF A PRIOR INFORMATION IN THE EVALUATION OF TYPE B STANDARD MEASUREMENT UNCERTAINTY

Abstract

This article investigates theoretical and practical aspects of one of
the urgent metrological issues — the concept of measurement uncertainty for
physico-chemical quantities. A hybrid method is developed for selecting and
determining the distribution type of a priori information in evaluating type B
standard measurement uncertainty, which is a key task in metrological assurance.
The relevance of the study lies in the need to improve the reliability of uncertainty
assessment under conditions of limited a priori information, as existing approaches
often rely on subjective expert judgments, which can lead to inaccurate results. The
scientific novelty includes the development of a hybrid method and mathematical
models for selecting the distribution of the information base in assessing type B
standard uncertainty; the method is based on the combination of the maximum
entropy principle and Bayesian theory. The research methodology includes analysis
of existing approaches, formalization of metrological criteria for distribution
selection, and development of a decision-making algorithm. The adequacy and
effectiveness of the method were tested at the Scientific Laboratory of Physico-
Chemical Quantities of the “Uzbek National Institute of Metrology” and accredited measurement (testing) laboratories compliant with ISO/IEC 17025:2017. The
results can be applied in improving the accuracy of measurement uncertainty
assessments and refining metrological control procedures, and are required in ISO/
IEC 17025:2017 accredited laboratories, interlaboratory comparison providers
under ISO 17043:2023, and international comparisons within the BIPM-KCDB
database on calibration and measurement capabilities (CMC). The developed
method may be used by scientific centers and laboratories involved in uncertainty
evaluation, metrology, and testing. Its implementation will enhance the objectivity
of uncertainty evaluation, improve the reliability of expanded uncertainty, and
reduce risks associated with type B standard uncertainty.