Processing multi-axial signals by POD for damage detection

Georgiou, IT; Adams, DE; Bajaj, AK

dc.contributor.author	Georgiou, IT	en
dc.contributor.author	Adams, DE	en
dc.contributor.author	Bajaj, AK	en
dc.date.accessioned	2014-03-01T02:51:10Z
dc.date.available	2014-03-01T02:51:10Z
dc.date.issued	2007	en
dc.identifier.issn	21915644	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/35411
dc.relation.uri	http://www.scopus.com/inward/record.url?eid=2-s2.0-84861569761&partnerID=40&md5=730f3de925f9412a8a5173911ea382ce	en
dc.subject.other	Damage location	en
dc.subject.other	Damage quantification	en
dc.subject.other	Fixed points	en
dc.subject.other	Fundamental operations	en
dc.subject.other	Impact force	en
dc.subject.other	Impact response	en
dc.subject.other	Initial conditions	en
dc.subject.other	Modal properties	en
dc.subject.other	Multi-field	en
dc.subject.other	Network of sensors	en
dc.subject.other	Nonlinear structure	en
dc.subject.other	Novel applications	en
dc.subject.other	Proper orthogonal decompositions	en
dc.subject.other	Spatio-temporal	en
dc.subject.other	Triaxial accelerometer	en
dc.subject.other	Turbulence dynamics	en
dc.subject.other	Wave analysis	en
dc.subject.other	Damage detection	en
dc.subject.other	Flow fields	en
dc.subject.other	Modal analysis	en
dc.subject.other	Principal component analysis	en
dc.subject.other	Random processes	en
dc.subject.other	Signal processing	en
dc.subject.other	Structural dynamics	en
dc.subject.other	Time series	en
dc.subject.other	Data handling	en
dc.title	Processing multi-axial signals by POD for damage detection	en
heal.type	conferenceItem	en
heal.publicationDate	2007	en
heal.abstract	There is a plethora of signal processing transforms ranging from the classical FFT to the very popular wavelet transform used for modal analysis and damage detection in structures. These transforms operate on a signal at a time. But in practice a set of signals is usually available (network of sensors). To detect the modal properties of a structure from a given spatially distributed set of time series, the whole information must be somehow fused. Time and space auto-correlations are fundamental operations that correlate or fuse distributed (spatio-temporal) data: they give rise to the Proper Orthogonal Decomposition (POD) Transform. Originally introduced to study turbulence dynamics in fluids and in general stochastic processes, the POD transform has been advanced into a powerful data processing tool especially for vibration and wave analysis of nonlinear structures with multi-field response. In this work, we present a novel application of the multi-field POD transform for signal processing. In particular, POD is applied to transform a set of time series recorded by a single tri-axial accelerometer at a fixed point as an impact force ranges over a grid of locations. The technique was applied to study the impact response of a canister-like composite structure: The POD transform reveals that the autocorrelation energy is distributed almost exponentially over a small set of POD modes. The distribution of the POD spectrum is robust but the shapes of the POD modes are sensitive to small changes in initial conditions and imperfections. These features seem to be promising indicators for damage location and damage quantification.	en
heal.journalName	Conference Proceedings of the Society for Experimental Mechanics Series	en