A sequential least squares algorithm for ARMAX dynamic network identification

Harm H.M. Weerts, Miguel Galrinho, Giulio Bottegal, Håkan Hjalmarsson, Paul M.J.Van den Hof

Research output: Contribution to journalConference articleAcademicpeer-review

Abstract

Identification of dynamic networks in prediction error setting often requires the solution of a non-convex optimization problem, which can be difficult to solve especially for large-scale systems. Focusing on ARMAX models of dynamic networks, we instead employ a method based on a sequence of least-squares steps. For single-input single-output models, we show that the method is equivalent to the recently developed Weighted Null Space Fitting, and, drawing from the analysis of that method, we conjecture that the proposed method is both consistent as well as asymptotically efficient under suitable assumptions. Simulations indicate that the sequential least squares estimates can be of high quality even for short data sets.

LanguageEnglish
Pages844-849
Number of pages6
JournalIFAC-PapersOnLine
Volume51
Issue number15
DOIs
StatePublished - 1 Jan 2018

Fingerprint

Large scale systems

Keywords

  • dynamic networks
  • identification algorithm
  • least squares
  • System identification

Cite this

@article{2fe0037221b54321858b0b9aa01c1dd6,
title = "A sequential least squares algorithm for ARMAX dynamic network identification",
abstract = "Identification of dynamic networks in prediction error setting often requires the solution of a non-convex optimization problem, which can be difficult to solve especially for large-scale systems. Focusing on ARMAX models of dynamic networks, we instead employ a method based on a sequence of least-squares steps. For single-input single-output models, we show that the method is equivalent to the recently developed Weighted Null Space Fitting, and, drawing from the analysis of that method, we conjecture that the proposed method is both consistent as well as asymptotically efficient under suitable assumptions. Simulations indicate that the sequential least squares estimates can be of high quality even for short data sets.",
keywords = "dynamic networks, identification algorithm, least squares, System identification",
author = "Weerts, {Harm H.M.} and Miguel Galrinho and Giulio Bottegal and H{\aa}kan Hjalmarsson and {den Hof}, {Paul M.J.Van}",
year = "2018",
month = "1",
day = "1",
doi = "10.1016/j.ifacol.2018.09.119",
language = "English",
volume = "51",
pages = "844--849",
journal = "IFAC-PapersOnLine",
issn = "2405-8963",
publisher = "Elsevier",
number = "15",

}

A sequential least squares algorithm for ARMAX dynamic network identification. / Weerts, Harm H.M.; Galrinho, Miguel; Bottegal, Giulio; Hjalmarsson, Håkan; den Hof, Paul M.J.Van.

In: IFAC-PapersOnLine, Vol. 51, No. 15, 01.01.2018, p. 844-849.

Research output: Contribution to journalConference articleAcademicpeer-review

TY - JOUR

T1 - A sequential least squares algorithm for ARMAX dynamic network identification

AU - Weerts,Harm H.M.

AU - Galrinho,Miguel

AU - Bottegal,Giulio

AU - Hjalmarsson,Håkan

AU - den Hof,Paul M.J.Van

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Identification of dynamic networks in prediction error setting often requires the solution of a non-convex optimization problem, which can be difficult to solve especially for large-scale systems. Focusing on ARMAX models of dynamic networks, we instead employ a method based on a sequence of least-squares steps. For single-input single-output models, we show that the method is equivalent to the recently developed Weighted Null Space Fitting, and, drawing from the analysis of that method, we conjecture that the proposed method is both consistent as well as asymptotically efficient under suitable assumptions. Simulations indicate that the sequential least squares estimates can be of high quality even for short data sets.

AB - Identification of dynamic networks in prediction error setting often requires the solution of a non-convex optimization problem, which can be difficult to solve especially for large-scale systems. Focusing on ARMAX models of dynamic networks, we instead employ a method based on a sequence of least-squares steps. For single-input single-output models, we show that the method is equivalent to the recently developed Weighted Null Space Fitting, and, drawing from the analysis of that method, we conjecture that the proposed method is both consistent as well as asymptotically efficient under suitable assumptions. Simulations indicate that the sequential least squares estimates can be of high quality even for short data sets.

KW - dynamic networks

KW - identification algorithm

KW - least squares

KW - System identification

UR - http://www.scopus.com/inward/record.url?scp=85054462289&partnerID=8YFLogxK

U2 - 10.1016/j.ifacol.2018.09.119

DO - 10.1016/j.ifacol.2018.09.119

M3 - Conference article

VL - 51

SP - 844

EP - 849

JO - IFAC-PapersOnLine

T2 - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

SN - 2405-8963

IS - 15

ER -