Особливості комп’ютерної реалізації методу локалізації параметрів інтервальних моделей із виділенням «насиченого блоку»

Abstract

Parameter identification is one from the stages of construction the models of the static systems. One from the most effective methods of parameter identification based on data with an interval uncertainty there is a method of localization (approximation) the area of interval model parameters with selection the «saturated block» in the basic interval system of linear algebraic equations (ISLAE). This method has some advantages in comparison with existing methods. In particular it is the best by the criterion of simplicity of operation by corridor of interval models for the prediction tasks and by the criterion of minimization the computing complexity of finding the multiple estimations of models parameters. However the conditions of method’s application are not research. There is the efficiency of computer realization the method and the conditions at which the known algorithms of its realization do not enable to get the solution are research in the article.