Практичні аспекти застосування методу та алгоритму параметричної ідентифікації дискретних динамічних систем з інтервальною невизначеністю

Abstract

This study is devoted to modeling dynamical nonlinear non-stationary systems, whose uncertainty lies in a priori known bounds. The convergence and computational properties of a new method for identifying models of such systems is studied. The method provides specified prediction accuracy for dynamic processes within the bounds of non-stationary and measurement errors. A research for a class of chemical-engineering systems, which are described by models of three levels of difficulty and for typical prediction errors (15% -30%) is conducted.