Structure selection for nonlinear input-output models based on fuzzy cluster analysis

J. Abonyi, Robert Babuška, Balazs Feil

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Selecting the structure (the input variables or regressors) of an input-output dynamic model is a crucial step in system identification. In this paper, a method is proposed that uses fuzzy clustering to select the structure of a nonlinear input-output model. 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 structure, 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. Two simulation examples are given to illustrate the proposed technique: estimation of the model structure for a polymerization reactor and the van der Vusse reactor.

Original languageEnglish
Title of host publicationIEEE International Conference on Fuzzy Systems
EditorsO Nasraoui, H Frigui, J.M. Keller
Pages464-469
Number of pages6
Volume1
Publication statusPublished - 2003
EventThe IEEE International conference on Fuzzy Systems - St. Louis, MO, United States
Duration: May 25 2003May 28 2003

Other

OtherThe IEEE International conference on Fuzzy Systems
CountryUnited States
CitySt. Louis, MO
Period5/25/035/28/03

Fingerprint

Cluster analysis
Model structures
Fuzzy clustering
Covariance matrix
Dynamic models
Identification (control systems)
Polymerization

Keywords

  • False-nearest neighbor
  • Fuzzy clustering
  • Minimum description length (MDL)
  • Model structure selection
  • System identification

ASJC Scopus subject areas

  • Software
  • Safety, Risk, Reliability and Quality
  • Chemical Health and Safety

Cite this

Abonyi, J., Babuška, R., & Feil, B. (2003). Structure selection for nonlinear input-output models based on fuzzy cluster analysis. In O. Nasraoui, H. Frigui, & J. M. Keller (Eds.), IEEE International Conference on Fuzzy Systems (Vol. 1, pp. 464-469)

Structure selection for nonlinear input-output models based on fuzzy cluster analysis. / Abonyi, J.; Babuška, Robert; Feil, Balazs.

IEEE International Conference on Fuzzy Systems. ed. / O Nasraoui; H Frigui; J.M. Keller. Vol. 1 2003. p. 464-469.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abonyi, J, Babuška, R & Feil, B 2003, Structure selection for nonlinear input-output models based on fuzzy cluster analysis. in O Nasraoui, H Frigui & JM Keller (eds), IEEE International Conference on Fuzzy Systems. vol. 1, pp. 464-469, The IEEE International conference on Fuzzy Systems, St. Louis, MO, United States, 5/25/03.
Abonyi J, Babuška R, Feil B. Structure selection for nonlinear input-output models based on fuzzy cluster analysis. In Nasraoui O, Frigui H, Keller JM, editors, IEEE International Conference on Fuzzy Systems. Vol. 1. 2003. p. 464-469
Abonyi, J. ; Babuška, Robert ; Feil, Balazs. / Structure selection for nonlinear input-output models based on fuzzy cluster analysis. IEEE International Conference on Fuzzy Systems. editor / O Nasraoui ; H Frigui ; J.M. Keller. Vol. 1 2003. pp. 464-469
@inproceedings{1046ca0525dc4587bf75289cc43f586b,
title = "Structure selection for nonlinear input-output models based on fuzzy cluster analysis",
abstract = "Selecting the structure (the input variables or regressors) of an input-output dynamic model is a crucial step in system identification. In this paper, a method is proposed that uses fuzzy clustering to select the structure of a nonlinear input-output model. 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 structure, 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. Two simulation examples are given to illustrate the proposed technique: estimation of the model structure for a polymerization reactor and the van der Vusse reactor.",
keywords = "False-nearest neighbor, Fuzzy clustering, Minimum description length (MDL), Model structure selection, System identification",
author = "J. Abonyi and Robert Babuška and Balazs Feil",
year = "2003",
language = "English",
volume = "1",
pages = "464--469",
editor = "O Nasraoui and H Frigui and J.M. Keller",
booktitle = "IEEE International Conference on Fuzzy Systems",

}

TY - GEN

T1 - Structure selection for nonlinear input-output models based on fuzzy cluster analysis

AU - Abonyi, J.

AU - Babuška, Robert

AU - Feil, Balazs

PY - 2003

Y1 - 2003

N2 - Selecting the structure (the input variables or regressors) of an input-output dynamic model is a crucial step in system identification. In this paper, a method is proposed that uses fuzzy clustering to select the structure of a nonlinear input-output model. 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 structure, 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. Two simulation examples are given to illustrate the proposed technique: estimation of the model structure for a polymerization reactor and the van der Vusse reactor.

AB - Selecting the structure (the input variables or regressors) of an input-output dynamic model is a crucial step in system identification. In this paper, a method is proposed that uses fuzzy clustering to select the structure of a nonlinear input-output model. 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 structure, 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. Two simulation examples are given to illustrate the proposed technique: estimation of the model structure for a polymerization reactor and the van der Vusse reactor.

KW - False-nearest neighbor

KW - Fuzzy clustering

KW - Minimum description length (MDL)

KW - Model structure selection

KW - System identification

UR - http://www.scopus.com/inward/record.url?scp=0037523487&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0037523487&partnerID=8YFLogxK

M3 - Conference contribution

VL - 1

SP - 464

EP - 469

BT - IEEE International Conference on Fuzzy Systems

A2 - Nasraoui, O

A2 - Frigui, H

A2 - Keller, J.M.

ER -