On the initialization of nonlinear LFR model identification with the best linear approximation

Maarten Schoukens; Roland Tóth

doi:10.1016/j.ifacol.2020.12.142

On the initialization of nonlinear LFR model identification with the best linear approximation

Maarten Schoukens, Roland Tóth

Research output: Contribution to journal › Conference article › peer-review

9 Citations (Scopus)

89 Downloads (Pure)

Abstract

Balancing the model complexity and the representation capability towards the process to be captured remains one of the main challenges in nonlinear system identification. One possibility to reduce model complexity is to impose structure on the model representation. To this end, this work considers the linear fractional representation framework. In a linear fractional representation the linear dynamics and the system nonlinearities are modeled by two separate blocks that are interconnected with one another. This results in a structured, yet flexible model structure. Estimating such a model directly from input-output data is not a trivial task as the involved optimization is nonlinear in nature. This paper proposes an initialization scheme for the model parameters based on the best linear approximation of the system and shows that this approach results in high quality models on a set of benchmark data sets.

Original language	English
Pages (from-to)	310-315
Number of pages	6
Journal	IFAC-PapersOnLine
Volume	53
Issue number	2
DOIs	https://doi.org/10.1016/j.ifacol.2020.12.142
Publication status	Published - 2020
Event	21st World Congress of the International Federation of Aufomatic Control (IFAC 2020 World Congress) - Berlin, Germany Duration: 12 Jul 2020 → 17 Jul 2020 Conference number: 21 https://www.ifac2020.org/

Bibliographical note

Publisher Copyright:
Copyright © 2020 The Authors. This is an open access article under the CC BY-NC-ND license

Copyright:
Copyright 2021 Elsevier B.V., All rights reserved.

Funding

Funders	Funder number
European Union's Horizon 2020 - Research and Innovation Framework Programme	714663, 798627

Keywords

Best Linear Approximation
Linear Fractional Representation
Neural Network
Nonlinear Identification

Access to Document

10.1016/j.ifacol.2020.12.142

pdfFinal published version, 916 KB

Cite this

@article{dd373899644948d28ef3c8519c7124a6,

title = "On the initialization of nonlinear LFR model identification with the best linear approximation",

abstract = "Balancing the model complexity and the representation capability towards the process to be captured remains one of the main challenges in nonlinear system identification. One possibility to reduce model complexity is to impose structure on the model representation. To this end, this work considers the linear fractional representation framework. In a linear fractional representation the linear dynamics and the system nonlinearities are modeled by two separate blocks that are interconnected with one another. This results in a structured, yet flexible model structure. Estimating such a model directly from input-output data is not a trivial task as the involved optimization is nonlinear in nature. This paper proposes an initialization scheme for the model parameters based on the best linear approximation of the system and shows that this approach results in high quality models on a set of benchmark data sets.",

keywords = "Best Linear Approximation, Linear Fractional Representation, Neural Network, Nonlinear Identification",

author = "Maarten Schoukens and Roland T{\'o}th",

note = "Publisher Copyright: Copyright {\textcopyright} 2020 The Authors. This is an open access article under the CC BY-NC-ND license Copyright: Copyright 2021 Elsevier B.V., All rights reserved.; 21st World Congress of the International Federation of Aufomatic Control (IFAC 2020 World Congress), IFAC 2020 ; Conference date: 12-07-2020 Through 17-07-2020",

year = "2020",

doi = "10.1016/j.ifacol.2020.12.142",

language = "English",

volume = "53",

pages = "310--315",

journal = "IFAC-PapersOnLine",

issn = "2405-8963",

publisher = "Elsevier",

number = "2",

url = "https://www.ifac2020.org/",

}

TY - JOUR

T1 - On the initialization of nonlinear LFR model identification with the best linear approximation

AU - Schoukens, Maarten

AU - Tóth, Roland

N1 - Conference code: 21

PY - 2020

Y1 - 2020

N2 - Balancing the model complexity and the representation capability towards the process to be captured remains one of the main challenges in nonlinear system identification. One possibility to reduce model complexity is to impose structure on the model representation. To this end, this work considers the linear fractional representation framework. In a linear fractional representation the linear dynamics and the system nonlinearities are modeled by two separate blocks that are interconnected with one another. This results in a structured, yet flexible model structure. Estimating such a model directly from input-output data is not a trivial task as the involved optimization is nonlinear in nature. This paper proposes an initialization scheme for the model parameters based on the best linear approximation of the system and shows that this approach results in high quality models on a set of benchmark data sets.

AB - Balancing the model complexity and the representation capability towards the process to be captured remains one of the main challenges in nonlinear system identification. One possibility to reduce model complexity is to impose structure on the model representation. To this end, this work considers the linear fractional representation framework. In a linear fractional representation the linear dynamics and the system nonlinearities are modeled by two separate blocks that are interconnected with one another. This results in a structured, yet flexible model structure. Estimating such a model directly from input-output data is not a trivial task as the involved optimization is nonlinear in nature. This paper proposes an initialization scheme for the model parameters based on the best linear approximation of the system and shows that this approach results in high quality models on a set of benchmark data sets.

KW - Best Linear Approximation

KW - Linear Fractional Representation

KW - Neural Network

KW - Nonlinear Identification

UR - http://www.scopus.com/inward/record.url?scp=85100475690&partnerID=8YFLogxK

U2 - 10.1016/j.ifacol.2020.12.142

DO - 10.1016/j.ifacol.2020.12.142

M3 - Conference article

AN - SCOPUS:85100475690

SN - 2405-8963

VL - 53

SP - 310

EP - 315

JO - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

IS - 2

T2 - 21st World Congress of the International Federation of Aufomatic Control (IFAC 2020 World Congress)

Y2 - 12 July 2020 through 17 July 2020

ER -

On the initialization of nonlinear LFR model identification with the best linear approximation

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this