Output-only modal parameter estimation of time-varying structures based on the support vector machine and vector time-dependent autoregressive models

S. D. Zhou, B. Y. Cao, L. Yu, L. Liu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper proposes a method of output-only modal identification of TV structures via combining the SVM and vector time-dependent autoregressive (VTAR) models for the cases of multiple outputs. Firstly, the formulation of the SVM-based modal estimator for the parameters of VTAR models is derived. Secondly, the bounded quadratic programming (QP) solution of SVM is introduced. Thirdly, the transformation from the estimated parameters to the modal parameters are presented. Finally, the proposed method a series numerical examples validate the proposed method.

Original languageEnglish
Title of host publicationProceedings of ISMA 2016 - International Conference on Noise and Vibration Engineering and USD2016 - International Conference on Uncertainty in Structural Dynamics
EditorsPaul Sas, David Moens, Axel van de Walle
PublisherKU Leuven, Departement Werktuigkunde
Pages2885-2894
Number of pages10
ISBN (Electronic)9789073802940
Publication statusPublished - 2016
Event27th International Conference on Noise and Vibration Engineering, ISMA 2016 and International Conference on Uncertainty in Structural Dynamics, USD2016 - Leuven, Belgium
Duration: 19 Sept 201621 Sept 2016

Publication series

NameProceedings of ISMA 2016 - International Conference on Noise and Vibration Engineering and USD2016 - International Conference on Uncertainty in Structural Dynamics

Conference

Conference27th International Conference on Noise and Vibration Engineering, ISMA 2016 and International Conference on Uncertainty in Structural Dynamics, USD2016
Country/TerritoryBelgium
CityLeuven
Period19/09/1621/09/16

Fingerprint

Dive into the research topics of 'Output-only modal parameter estimation of time-varying structures based on the support vector machine and vector time-dependent autoregressive models'. Together they form a unique fingerprint.

Cite this