Model order selection of nonlinear input-output models - A clustering based approach

Balazs Feil, J. Abonyi, F. Szeifert

Research output: Contribution to journalArticle

36 Citations (Scopus)

Abstract

Selecting the order of an input-output model of a dynamical system is a key step toward the goal of system identification. The false nearest neighbors algorithm (FNN) is a useful tool for the estimation of the order of linear and nonlinear systems. While advanced FNN uses nonlinear input-output data-based models for the model-based selection of the threshold constant that is used to compute the percentage of false neighbors, the computational effort of the method increases along with the number of data and the dimension of the model. To increase the efficiency of this method, in this paper we propose a clustering-based algorithm. Clustering is applied to the product space of the input and output variables. The model structure is then estimated on the basis of the cluster covariance matrix eigenvalues. The main advantage of the proposed solution is that it is model-free. This means that no particular model needs to be constructed in order to select the order of the model, while most other techniques are 'wrapped' around a particular model construction method. This saves the computational effort and avoids a possible bias due to the particular construction method used. Three simulation examples are given to illustrate the proposed technique: estimation of the model structure for a linear system, a polymerization reactor and the van der Vusse reactor.

Original languageEnglish
Pages (from-to)593-602
Number of pages10
JournalJournal of Process Control
Volume14
Issue number6
DOIs
Publication statusPublished - Sep 2004

Fingerprint

Order Selection
Model Selection
Clustering
Output
Model structures
Model
Linear systems
Reactor
Nearest Neighbor
Linear Systems
Covariance matrix
Product Space
Polymerization
System Identification
Nonlinear systems
Identification (control systems)
Dynamical systems
Percentage
Nonlinear Systems
Dynamical system

Keywords

  • False nearest neighbors
  • Fuzzy clustering
  • Minimum description length (MDL)
  • Model order selection
  • System identification

ASJC Scopus subject areas

  • Process Chemistry and Technology
  • Control and Systems Engineering
  • Industrial and Manufacturing Engineering

Cite this

Model order selection of nonlinear input-output models - A clustering based approach. / Feil, Balazs; Abonyi, J.; Szeifert, F.

In: Journal of Process Control, Vol. 14, No. 6, 09.2004, p. 593-602.

Research output: Contribution to journalArticle

@article{17a6bc5921bf4f9388d9ef4fc137a33e,
title = "Model order selection of nonlinear input-output models - A clustering based approach",
abstract = "Selecting the order of an input-output model of a dynamical system is a key step toward the goal of system identification. The false nearest neighbors algorithm (FNN) is a useful tool for the estimation of the order of linear and nonlinear systems. While advanced FNN uses nonlinear input-output data-based models for the model-based selection of the threshold constant that is used to compute the percentage of false neighbors, the computational effort of the method increases along with the number of data and the dimension of the model. To increase the efficiency of this method, in this paper we propose a clustering-based algorithm. Clustering is applied to the product space of the input and output variables. The model structure is then estimated on the basis of the cluster covariance matrix eigenvalues. The main advantage of the proposed solution is that it is model-free. This means that no particular model needs to be constructed in order to select the order of the model, while most other techniques are 'wrapped' around a particular model construction method. This saves the computational effort and avoids a possible bias due to the particular construction method used. Three simulation examples are given to illustrate the proposed technique: estimation of the model structure for a linear system, a polymerization reactor and the van der Vusse reactor.",
keywords = "False nearest neighbors, Fuzzy clustering, Minimum description length (MDL), Model order selection, System identification",
author = "Balazs Feil and J. Abonyi and F. Szeifert",
year = "2004",
month = "9",
doi = "10.1016/j.jprocont.2004.01.005",
language = "English",
volume = "14",
pages = "593--602",
journal = "Journal of Process Control",
issn = "0959-1524",
publisher = "Elsevier Limited",
number = "6",

}

TY - JOUR

T1 - Model order selection of nonlinear input-output models - A clustering based approach

AU - Feil, Balazs

AU - Abonyi, J.

AU - Szeifert, F.

PY - 2004/9

Y1 - 2004/9

N2 - Selecting the order of an input-output model of a dynamical system is a key step toward the goal of system identification. The false nearest neighbors algorithm (FNN) is a useful tool for the estimation of the order of linear and nonlinear systems. While advanced FNN uses nonlinear input-output data-based models for the model-based selection of the threshold constant that is used to compute the percentage of false neighbors, the computational effort of the method increases along with the number of data and the dimension of the model. To increase the efficiency of this method, in this paper we propose a clustering-based algorithm. Clustering is applied to the product space of the input and output variables. The model structure is then estimated on the basis of the cluster covariance matrix eigenvalues. The main advantage of the proposed solution is that it is model-free. This means that no particular model needs to be constructed in order to select the order of the model, while most other techniques are 'wrapped' around a particular model construction method. This saves the computational effort and avoids a possible bias due to the particular construction method used. Three simulation examples are given to illustrate the proposed technique: estimation of the model structure for a linear system, a polymerization reactor and the van der Vusse reactor.

AB - Selecting the order of an input-output model of a dynamical system is a key step toward the goal of system identification. The false nearest neighbors algorithm (FNN) is a useful tool for the estimation of the order of linear and nonlinear systems. While advanced FNN uses nonlinear input-output data-based models for the model-based selection of the threshold constant that is used to compute the percentage of false neighbors, the computational effort of the method increases along with the number of data and the dimension of the model. To increase the efficiency of this method, in this paper we propose a clustering-based algorithm. Clustering is applied to the product space of the input and output variables. The model structure is then estimated on the basis of the cluster covariance matrix eigenvalues. The main advantage of the proposed solution is that it is model-free. This means that no particular model needs to be constructed in order to select the order of the model, while most other techniques are 'wrapped' around a particular model construction method. This saves the computational effort and avoids a possible bias due to the particular construction method used. Three simulation examples are given to illustrate the proposed technique: estimation of the model structure for a linear system, a polymerization reactor and the van der Vusse reactor.

KW - False nearest neighbors

KW - Fuzzy clustering

KW - Minimum description length (MDL)

KW - Model order selection

KW - System identification

UR - http://www.scopus.com/inward/record.url?scp=1842865087&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=1842865087&partnerID=8YFLogxK

U2 - 10.1016/j.jprocont.2004.01.005

DO - 10.1016/j.jprocont.2004.01.005

M3 - Article

AN - SCOPUS:1842865087

VL - 14

SP - 593

EP - 602

JO - Journal of Process Control

JF - Journal of Process Control

SN - 0959-1524

IS - 6

ER -