Interval approach to processing the noised thermophysical data

Abstract

The paper deals with application of numerical methods to processing the experimental data on the thermophysical properties of several chemical substances and their compounds (cryolites, rare earth compounds, and alkali halides). The main aim of investigations is in estimating the parameters of dependencies between the heat of fusion and the melting temperature of these chemical substances. The data are corrupted by the measuring errors. Procession is implemented under conditions of uncertainty: there is no any information on probabilistic properties of the corrupting factors, samples of measurements are short, and only approximate functions are known that describes mentioned dependencies. Under such conditions, the standard statistical methods can be applied formally. To obtain guarantied results in parameters estimation, the interval analysis methods and procedures are used.