Botsiura O. Improvement of the methods for the measurement uncertainty evaluation with considering of Bayesian approach

The object of research is the process of measuring the results of measurements in test and calibration laboratories accredited to meet the requirements of DSTU ISO / IEC 17025: 2006; The purpose of the dissertation is to increase the accuracy of the estimates of numerical values and the uncertainty of the measurement results, taking into account the Bayesian approach; The basis of the work are theoretical research methods based on the theory of probability and mathematical statistics - for improving existing methods of processing the results and estimating uncertainty of measurements; The theory of measurements - for the development of applied aspects of measurement of uncertainty of measurements during metrological works; Methods of statistical simulation - to check the reliability of the developed provisions; Scientific novelty of the obtained results: For the first time mathematical dependences were used which were used to determine the coefficient of coverage in the estimation of the expanded uncertainty of the combination of uncertainty deposits of type A taking into account the Bayesian approach in the absence or presence of a correlation between the results of measurements of the input quantities. For the first time, the effect of the law of the distribution of independent observations of the measured value on the estimation of standard uncertainty of measurements of type A has been investigated, which enables to increase the accuracy of obtaining these estimates under abnormal laws of the distribution of independent observations. On the basis of the finite increment method, for the first time, a dependence was obtained for calculating the numerical value of the measured quantity and the procedure for processing the measurement results for the estimation of the total standard uncertainty of type B in the nonlinear model equation was improved, which allows to reduce the shift of estimates of the measurement result. For the first time mathematical dependences were obtained for different methods of calculating coverage coefficients under the Bayesian approach to the measurement of uncertainty of measurements. The procedure for obtaining the estimation of the coefficient of coverage is substantiated, which makes it possible to obtain the most accurate estimates of the expanded uncertainty of measurements.