Application of Parallel-Sequential Identification Algorithms in Diagnostic Tasks

Abstract

Ensuring the reliability of automatic control systems is directly related to solving diagnostic problems that require the identification of current parameter values in the presence of uncertain impacts or changing operating conditions. This determines the high importance of an effective solution to the problem of parametric identification. Currently, there are many alternatives for solving identification problems, one of which is the joint use of classical identification methods and artificial neural networks. The article proposes a method of systems’ synthesis for identifying the dynamic objects’ parameters based on the complex use of artificial neural networks and recurrent algorithms, which is a parallel-sequential information processing procedure. The efficiency of the recurrent identification algorithm using a neural network is shown. The results of the numerical simulation are presented.