Searching the optimal order for high order models - MIMO case

Author

Alves, V.A.O. ; de Godoy, R.J.C. ; Garcia, Christophe

fYear

2012

fDate

3-5 Oct. 2012

Firstpage

98

Lastpage

103

Abstract

In some system identification methods, high order ARX models are adopted as true descriptions of the process to be identified. Afterwards, such high order model is used as a basis for a model order reduction procedure. The optimization strategy for the search of the best high order (in the fit index sense) for SISO ARX models, presented in a related paper [1], is extended to the MIMO case. The analyzed system is the Wood & Berry binary distillation column [2]. In this study, the search for the best dead times associated with the high order ARX model is also addressed. The results obtained by the optimization strategy are compared with those derived by exhaustive search.

Keywords

MIMO systems; autoregressive processes; distillation equipment; optimal control; optimisation; reduced order systems; search problems; MIMO; SISO ARX model; Wood-Berry binary distillation column; best dead time; exhaustive search; fit index sense; high order ARX model; high order model; model order reduction; optimal order searching; optimization strategy; process description; system identification method; Computational modeling; Indexes; MIMO; Noise measurement; Optimization; Predictive models; Signal to noise ratio;

fLanguage

English

Publisher

ieee

Conference_Titel

Control Applications (CCA), 2012 IEEE International Conference on

Conference_Location

Dubrovnik

ISSN

1085-1992

Print_ISBN

978-1-4673-4503-3

Electronic_ISBN

1085-1992

Type

conf

DOI

10.1109/CCA.2012.6402422

Filename

6402422