A Fast Smoothing-Based Algorithm to Generate l∞-Norm Constrained Signals for Multivariable Experiment Design

Nic Dirkx; Marcel Bosselaar; Tom Oomen

doi:10.1109/LCSYS.2021.3133655

A Fast Smoothing-Based Algorithm to Generate l_∞-Norm Constrained Signals for Multivariable Experiment Design

Nic Dirkx (Corresponding author), Marcel Bosselaar, Tom Oomen

Research output: Contribution to journal › Article › Academic › peer-review

82 Downloads (Pure)

Abstract

Handling peak amplitude constraints, or equivalently l∞-norm constraints, is an important application demand in experiment design for system identification. The aim of this letter is to present a method for the design of excitation signals with prescribed power spectrum under l∞-norm constraints for systems with many inputs and outputs. The method exploits an exponential smoothing function in an iterative algorithm. Fast convergence is achieved by a computationally efficient construction of the gradient and the Hessian matrix. Experimental results show excellent convergence behavior that overcomes local minima, while significantly reducing computation time compared to existing techniques.

Original language	English
Pages (from-to)	1784-1789
Number of pages	6
Journal	IEEE Control Systems Letters
Volume	6
DOIs	https://doi.org/10.1109/LCSYS.2021.3133655
Publication status	Published - 2022

Keywords

Crest-factor optimization
experiment design
system identification

Access to Document

10.1109/LCSYS.2021.3133655

published versionFinal published version, 3.33 MBLicence: TAVERNE

Cite this

@article{5bb5fdf2ec1d4fc0b86a71e0b8591ac9,

title = "A Fast Smoothing-Based Algorithm to Generate l∞-Norm Constrained Signals for Multivariable Experiment Design",

abstract = "Handling peak amplitude constraints, or equivalently l∞-norm constraints, is an important application demand in experiment design for system identification. The aim of this letter is to present a method for the design of excitation signals with prescribed power spectrum under l∞-norm constraints for systems with many inputs and outputs. The method exploits an exponential smoothing function in an iterative algorithm. Fast convergence is achieved by a computationally efficient construction of the gradient and the Hessian matrix. Experimental results show excellent convergence behavior that overcomes local minima, while significantly reducing computation time compared to existing techniques. ",

keywords = "Crest-factor optimization, experiment design, system identification",

author = "Nic Dirkx and Marcel Bosselaar and Tom Oomen",

note = "Publisher Copyright: {\textcopyright} 2017 IEEE.",

year = "2022",

doi = "10.1109/LCSYS.2021.3133655",

language = "English",

volume = "6",

pages = "1784--1789",

journal = "IEEE Control Systems Letters",

issn = "2475-1456",

publisher = "Institute of Electrical and Electronics Engineers",

}

TY - JOUR

T1 - A Fast Smoothing-Based Algorithm to Generate l∞-Norm Constrained Signals for Multivariable Experiment Design

AU - Dirkx, Nic

AU - Bosselaar, Marcel

AU - Oomen, Tom

PY - 2022

Y1 - 2022

N2 - Handling peak amplitude constraints, or equivalently l∞-norm constraints, is an important application demand in experiment design for system identification. The aim of this letter is to present a method for the design of excitation signals with prescribed power spectrum under l∞-norm constraints for systems with many inputs and outputs. The method exploits an exponential smoothing function in an iterative algorithm. Fast convergence is achieved by a computationally efficient construction of the gradient and the Hessian matrix. Experimental results show excellent convergence behavior that overcomes local minima, while significantly reducing computation time compared to existing techniques.

AB - Handling peak amplitude constraints, or equivalently l∞-norm constraints, is an important application demand in experiment design for system identification. The aim of this letter is to present a method for the design of excitation signals with prescribed power spectrum under l∞-norm constraints for systems with many inputs and outputs. The method exploits an exponential smoothing function in an iterative algorithm. Fast convergence is achieved by a computationally efficient construction of the gradient and the Hessian matrix. Experimental results show excellent convergence behavior that overcomes local minima, while significantly reducing computation time compared to existing techniques.

KW - Crest-factor optimization

KW - experiment design

KW - system identification

UR - http://www.scopus.com/inward/record.url?scp=85121371946&partnerID=8YFLogxK

U2 - 10.1109/LCSYS.2021.3133655

DO - 10.1109/LCSYS.2021.3133655

M3 - Article

AN - SCOPUS:85121371946

SN - 2475-1456

VL - 6

SP - 1784

EP - 1789

JO - IEEE Control Systems Letters

JF - IEEE Control Systems Letters

ER -

A Fast Smoothing-Based Algorithm to Generate l_∞-Norm Constrained Signals for Multivariable Experiment Design

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this