Samoylichenko O. Methods of increase the accuracy of metrological characteristics estimation while reference materials certification

Thesis is devoted to the problems of methods development to increase the accuracy of expanded uncertainty estimation and uncertainty components because of nonhomogeneity and nonstability while reference materials certification. To estimate the expanded uncertainty normalization transformation based on Johnson's is offered a new method. Accuracy increasing is achieved due to more accurate estimation of certificated characteristics reciprocal distribution. The reciprocal distribution fit the given confidence probability. This method makes it possible to estimate the expanded uncertainty when the components were estimated as type A as type B and their combinations; also, when the distribution low is antimodal. The method to increase the accuracy of uncertainty components from nonhomogeneity and nonstability was developed. It based on the statistical criteria power function determining, used when mentioned component estimate. Developed methods were implemented as system of measurement experiments statistical data processing. It helps to automate selected approaches of the reference material certification in the measurement and tested laboratories.