Experimental study of accuracy, precision, stability and uncertainty of physical and chemical quantities using certified reference materials
Аннотация
In the course of this work, measurement processing software was developed. An analytical review has been carried out to identify the advantages of algorithms and software to assess uncertainty over other measurement processing methods. Functional and experimental testing of the software was carried out. To date, there are various statistical methods of processing data, which can be divided into two large groups: parametric and non-parametric. Using parametric methods, it is assumed that the take-off elements belong to a certain parametric distribution family: normal, Poisson, and so on. Assuming that we know the form of the distribution, we can conclude about unknown parameters. However, remember that there are random variables that do not obey the forms of standard distributions. If you make an error during the assumption of a given distribution law, then, during data processing, an offset occurs. When developing algorithms and their software providing uncertainty of measurements, it is difficult that the values of extended uncertainty in many depends on the mathematical model of measurements. Criteria are defined for comparison of software products for calculation of measurement uncertainty in the field of measurement model implementation. The results of comparative analysis of the most common universal software products for calculation of measurement uncertainty are presented.