Vodotyka S. Improvement of methods for measurement uncertainty estimation during calibration of measuring instruments

Research object - the process of calibration of measuring instruments and transducers. A research objective - increase of accuracy of measurement uncertainty estimates through the use of information about uncertainties and laws of distribution of reproducible values of the reference quantities and of the measuring instrument readings. Research methods - analytical methods based on probability theory and mathematical statistics, theory of measurement, simulation techniques, optimization methods for finding the minimum of functions, theory of artificial neural networks. Equipment for researches - the personal computer. Theoretical and practical outcomes - formulated provisions allow increasing of accuracy of measurement uncertainty estimates during calibration of measuring instruments. Developed and improved methods are the basis for the development of reference documents about the procedure for processing of the measurement results and of the measurement uncertainty estimation during calibration of measuring instruments according to specific models of the transfer of the size of physical quantities. At the state level development and implementation of those documents will provide the competence of conformity assessment bodies. Scientific novelty - have been further improved methods of measurement uncertainty estimation during calibration of measuring instruments and transducers in a controlled point scale, which take into account finite resolution and variation of the measurement instrument readings, as well as the correlation between the readings of the reference and calibrated measuring devices; first time the method based on implementation of the uncertainty management procedures through optimization methods for finding the maximum likelihood function, which takes into account the combined uncertainties of reproducible values of the reference quantities and of the measurement instrument readings, for measurement results processing and measurement uncertainties estimation is proposed for evaluation of the calibration characteristic of the measurement instrument or transducer; first time the method based on the using of artificial neural network, which eliminates the stage of the choice of calibration characteristic model that fulfill the adaptation criterion is proposed for evaluation of the calibration characteristic of the measurement instrument or transducer; first time the method based on the using of artificial neural network, which is based on interval arithmetic algorithm with Levenberg-Marquardt training and takes into account the combined uncertainties of reproducible values of the reference quantities and the measurement instrument readings, is proposed for evaluation of the calibration characteristic of the measurement instrument or transducer. The outcomes of dissertational work are implemented during work execution on State enterprise "Scientific, Research, Design and Technology Institute of Municipal Economy" (Kiev), State enterprise "Donetsk Scientific and Research Center of Standardization, Metrology and Certification" (Donetsk), National Scientific Center "Institute of Metrology" (Kharkiv) and during educational process at the Department of metrology and measurement techniques of Kharkov National University of Radioelectronics (Kharkiv). If confirmed by the acts of implementation. The scientific and practical results of the thesis can be used to provide needs of measurements traceability during using of reference samples of organic and biological origin, the bodies of state metrological system for the creation of methods of measurement results processing including the quantitative chemical analysis of samples of substances and materials; in the educational process in metrological disciplines in higher education.