Damage detection of bridge structures under unknown seismic excitations using support vector machine based on transmissibility function and wavelet packet energy
{"title":"Damage detection of bridge structures under unknown seismic excitations using support vector machine based on transmissibility function and wavelet packet energy","authors":"Lijun Liu, Jianan Mi, Yixiao Zhang, Y. Lei","doi":"10.12989/SSS.2021.27.2.257","DOIUrl":null,"url":null,"abstract":"Since it may be hard to obtain the exact external load in practice, damage identification of bridge structures using only structural responses under unknown seismic excitations is an important but challenging task. Since structural responses are determined by both structural properties and seismic excitation, it is necessary to remove the effects of external excitation and only retain the structural information for structural damage identification. In this paper, a data-driven approach using structural responses only is proposed for structural damage alarming and localization of bridge structures. The transmissibility functions (TF) of structural responses are used to eliminate the influence of unknown seismic excitations. Moreover, the inverse Fourier transform of TFs and wavelet packet transform are used to reduce the influence of frequency bands and to extract the damagesensitive feature, respectively. Based on Support vector machines (SVM), structural responses under ambient excitations are used for training SVM. Then, structural responses under unknown seismic excitations are also processed accordingly and used for damage alarming and localization by the trained SMV. The numerical simulation examples of beam-type bridge and a cablestayed bridge under unknown seismic excitations are studied to illustrate the performance of the proposed approach.","PeriodicalId":51155,"journal":{"name":"Smart Structures and Systems","volume":null,"pages":null},"PeriodicalIF":2.1000,"publicationDate":"2021-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Smart Structures and Systems","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.12989/SSS.2021.27.2.257","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, CIVIL","Score":null,"Total":0}
引用次数: 0
Abstract
Since it may be hard to obtain the exact external load in practice, damage identification of bridge structures using only structural responses under unknown seismic excitations is an important but challenging task. Since structural responses are determined by both structural properties and seismic excitation, it is necessary to remove the effects of external excitation and only retain the structural information for structural damage identification. In this paper, a data-driven approach using structural responses only is proposed for structural damage alarming and localization of bridge structures. The transmissibility functions (TF) of structural responses are used to eliminate the influence of unknown seismic excitations. Moreover, the inverse Fourier transform of TFs and wavelet packet transform are used to reduce the influence of frequency bands and to extract the damagesensitive feature, respectively. Based on Support vector machines (SVM), structural responses under ambient excitations are used for training SVM. Then, structural responses under unknown seismic excitations are also processed accordingly and used for damage alarming and localization by the trained SMV. The numerical simulation examples of beam-type bridge and a cablestayed bridge under unknown seismic excitations are studied to illustrate the performance of the proposed approach.
期刊介绍:
An International Journal of Mechatronics, Sensors, Monitoring, Control, Diagnosis, and Management airns at providing a major publication channel for researchers in the general area of smart structures and systems. Typical subjects considered by the journal include:
Sensors/Actuators(Materials/devices/ informatics/networking)
Structural Health Monitoring and Control
Diagnosis/Prognosis
Life Cycle Engineering(planning/design/ maintenance/renewal)
and related areas.