Анализ МНК-оценок для идентификации разностных уравнений нестационарных временных рядов

Abstract

The aim of the present paper is to investigate the conditions under which it is possible to expect the derivation of acceptable LSM estimates of the parameters of the difference equation from the results of observations of non-stationary time series of its variables. In particular, the problem of identifying thermodynamic processes is being solved. In the paper, a comparative analysis of the displacement of LSM estimators of the parameters of the investigated autoregression equations and standard deviation errors is carried out. Conclusions are made about the possibility of applying direct LSM estimates without a significant loss of their adequacy.