A semiempirical reduced-order identification modeling tool for partially unknown discrete-time plants by using a multi-estimation scheme

S. Alonso-Quesada, M. De La Sen, A. Bilbao-Guillerna, A. Ibeas

Research output: Contribution to journalArticleResearchpeer-review

1 Citation (Scopus)

Abstract

A multi-estimation scheme is presented to recursively identify a partially unknown plant. The multi-estimation scheme is composed of a set of identifiers, each of them for identifying each possible linear nominal model of the plant referred to distinct operation points of the system or being each relevant for inputs containing distinct frequency spectra. Each nominal model is built as a parallel connection of first-order filters and contains some, but not all, natural modes which are supposed known, of the true plant to be identified. The assumption that the elementary filters are of first-order is identical to the assumption that all poles are real and distinct, which is feasible in many practical situations and may be taken as an acceptable approximation in more general cases. The assumption that the modes are known may work in an acceptable way when nominal values and a small range of uncertainty are known. The multi-estimation scheme is completed with a supervisor that selects the most appropriate estimation model of the plant by means of a switching law that operates on the various estimators of the whole multi-estimation scheme. Such a law evaluates a performance index related to the identification error associated with each estimator over a sliding time interval and selects the estimation model which supplies the smallest index for each current time interval. In this way, a system identification scheme which incorporates model order-reduction issues, as well as switches between the various identifiers, can be designed according to the relevant frequency spectrum of the externally applied input.
Original languageEnglish
Pages (from-to)419-436
JournalInstrumentation Science and Technology
Volume35
DOIs
Publication statusPublished - 1 Jul 2007

Keywords

  • Model order-reduction
  • Multi-estimation technique
  • Relative adaptation dead-zone
  • Switching law
  • System identification
  • Unmodeled dynamics

Fingerprint Dive into the research topics of 'A semiempirical reduced-order identification modeling tool for partially unknown discrete-time plants by using a multi-estimation scheme'. Together they form a unique fingerprint.

Cite this