Damage identification in beam-like structures by vibration-based analysis and artificial neural networks

Abstract

Laminated composites and sandwich structures are increasingly being used in different engineering applications such as in aeronautical, marine and offshore structures where high stiffness, light weight, good corrosion resistance and temperature stability are the primary issues. During their service life, these structures experience extreme loadings and harsh environmental conditions potentially leading to structural damage. This could significantly reduce mechanical strength and result in performance degradation of the structure. Therefore, in order to maintain the performance of the structure, localisation and quantification of the damage is a promising research area. Since the determination of the severity and the location of the damage is an inverse and non-unique problem, an intelligent algorithm is needed to perform the damage detection analysis. This study presents a damage detection algorithm, which uses vibration-based analysis data obtained from beam-like structures to locate and quantify the damage by using artificial neural networks. The inputs and the corresponding outputs required to train the neural networks are obtained from the finite element analyses for different vibration modes of the beams. Multi- layer feedforward backpropogation neural networks have been designed and trained by using different damage scenarios. After validation of the neural networks, new damage cases obtained from finite element and experimental analyses have been introduced and neural networks have been tested for location and severity predictions. The results from the neural networks depict that severity and location of the damage can be predicted by using as input the global (natural frequencies) and the local (strain or curvature mode shapes) dynamic behaviour of the beam-like structures.

More information

Identifiers

Catalogue record

Export record

ASCII CitationAtomBibTeXData Cite XMLDublin CoreDublin CoreEP3 XMLEndNoteHTML CitationHTML CitationHTML ListJSONMETSMODSMPEG-21 DIDLOpenURL ContextObjectOpenURL ContextObject in SpanRDF+N-TriplesRDF+N3RDF+XMLRIOXX2 XMLReferReference ManagerSimple Metadata