DESIGN OF EXPERIMENTS IN A TRACEABLE INTERVAL VARIABLES AS A REASONABLE TOOL
FOR SOLVING TASKS OF MEASUREMENT

Ordinartseva Natalia Pavlovna, candidate of technical sciences, associate professor, sub-department of information and measuring equipment and metrology, Penza State University (40 Krasnaya street, Penza, Russia), np_ordinartseva@mail.ru

006.91

10.21685/2307-5538-2018-3-6

Background. The necessity of further development of the classical theory design of experience on interval variables is substantiated, which allows to have a correct mathematical tool for solving measurement and metrology problems.
Materials and methods. Design of experiments on the interval data of the guaranteed finding of the measurement result using
such important properties of the interval as the inclusion monotony, two-sided approximation, narrowing of the interval function is most adapted for solving the problems of measurement and metrology. The method is based on the construction of an interpolation function within the interval of its guaranteed finding; the values of the function in the interpolation nodes coincide with robust, efficient and consistent estimates of the obtained measurement results, and at points far from the nodes, by means of a compressive internal approximation operator, they tend to the values obtained by the least squares method (LSM) within the guaranteed interval.
Conclusions. The development of the GUM approach, the construction of regression curve on the interval variables. The transition to interval variables allows to overcome some of the shortcomings identified in the GUM.

design of experiments, interval variables, interval guaranteed of finding the result, the internal approximation operator