Comparative analysis of metrological certification methods of mathematical models
Andriy Tevyashev, Юрий Сергеевич Асаенко, Анатолий Михайлович Кобылин · Eastern-European Journal of Enterprise Technologies · 2015
Additional selection criteria of mathematical models for use in real-time systems based on the metrological certification results were considered in the paper. This allows not only to select adequate models, but also the models that provide the minimum width of the uncertainty intervals of the dependent variables (calculation results) at a given initial data uncertainty. The comparative analysis of the two classical methods for metrological certification: simulation method, statistical linearization method and three variants of the real interval method: classical interval mathematics method, Kaucher interval mathematics method, centered interval method was given.It was shown that metrological certification results by the considered methods are virtually identical for the considered models, and the centered interval method is the most efficient.