Auxiliary model gradient-based iterative identification with moving data window for Wiener nonlinear output-error systems

Abstract

This paper is concentrated on the identification issue regarding the Wiener nonlinear output-error (OE) system. The system is broken down into multiple subsystems having a smaller number of variables by employing the key term separation. Employing the idea of auxiliary model identification resolves the non-measurable variables in the information vector. With the employment of the auxiliary model identification idea along with the negative gradient search method, an auxiliary model gradient-based iterative algorithm is attained. The accuracy of parameter identification is improved by the addition of the moving data window, which updates the dynamic data by deleting the past data as well as appending the most recent measurement data. The moving data window auxiliary model gradient-based iterative algorithm is presented. Finally, a numerical example is used to examine and compare the performance of the proposed algorithms, and an application example of the continuous stirred tank reactor is used for validating the practicability of the proposed algorithm.