Karpenko O. Development of interval methods and increasing precision of measurement results estimation with using uncertainty conception.

Paper is devoted to development of new methods of data processing, which allow to make the procedure of estimation of expanded uncertainty by В type of evaluation easier and to increase precision of estimation of expanded uncertainty by A type of evaluation. The probability-to-fuzziness conversion is researched which allows to build membership function from the probability distribution according to the connection between confidence intervals of level 1-а with each a -cut of fuzzy subset, which thus gathers the whole set of confidence intervals in its membership function. Proved, that using extension principle by Zade doesn't provide results which possess intermediate position between interval and probabilistic methods as it represents fuzzy generalization of interval analysis. A new statistics based on the middle of range was proposed and researched for processing uniformly distributed results of measurement. Using it for evaluation of expanded uncertainty by type A leads to decreasing of uncertainty interval. Level of decreasing depends on the size of sample. Analytical equations were obtained which can estimate efficiency comparing to traditional methods of processing. Also offered applying correction to the interval analysis results for estimation of uncertainty by type B. It became possible because of preservation of 95% quantile property in membership function.