DIAGNOSTICS OF THE HIGH-PRECISE BALLISTIC MEASURED GRAVITY ACCELERATION BY METHODS OF NON-CLASSICAL ERRORS THEORY

The purpose of the investigation is to show the necessity of using modern ideas about the law of error distribution for observations involved in the categories of the "Non-classical error theory of measurements" (NETM) in the process of performing high-precision ballistic definitions of gravitational acceleration. These definitions are characterized by large volumes, which according to the H. Jeffreys' theory, professor at the University of Cambridge, automatically takes them beyond the bounds of the classical concepts about the errors of measurements law. These outdated views about the distribution law of errors of large volume measurements are the main obstacles to improve the methodology of these highly precise and important definitions. The research methodology is provided by the NETM-procedures that was designed to control the probabilistic from of the statistical distribution of absolute high-precise ballistic measurements g with large sample volumes based on H. Jeffreys' recommendations and on the principles of hypothesis testing theory. The main result of the research is to carry out NETM-diagnostics of a metrological situation with the ballistic gravimeter FG-5 after some improvements of the program of the observation. This method of diagnostics is based on the use confidence intervals to the estimates of asymmetry and kurtosis of the obtained samples of measurements g with the following application of the Pearson's chi(2)-test to determine the significance of the deviations of its distribution from the established norms. In accordance with the categories of the NETM, such norms are the Gauss's and Person-Jeffreys' s laws, since only they ensure the non-singularity of the weight function of the sample, and therefore the possibility of obtaining non generate estimates g during the mathematical processing of measurements. Scientific novelty: using the possibilities of the new important tool in the field "Data analysis" using the NETM to improve the technique of the high-precise measurements g, which are performed in a complicated metrological situation with the necessity of taking into account a number of non-stationary sources of systematic errors. The practical significance of the research is in use of NETM-diagnostics of the probabilistic form of the distribution of measurements g in order to improve the methodology of these highly precise determinations. The investigation seeks reasons for the deviations of errors distributions from established norms providing metrological literacy of the high-precise large-scale measurements.