New approach for the damage mechanisms identification in the glass-epoxy composites
Résumé
The aim of this work is to introduce a new approach for clustering the acoustic emission data occurring during the fracture process of glass fibre reinforced polymer (GFRP) plates. In particular, we have developed signal processing techniques based on continues wavelet transform in order to isolate the acoustic signature corresponding to each damage mechanism (fibre breakage, matrix cracking, etc.). Models are then established in the form of dictionaries, whose elements are the waveforms collected during specific experiments in which the creates damage mechanisms are well known. The conducted experiments allowed isolating not only strongly excited mechanisms but also those whose amplitude is below the detection threshold (which are usually considered as noise and hence systematically eliminated by using automatic threshold). This valuable decrease in the threshold (from 30dB to 25dB) will be used in the study of precursor mechanisms as well as slow dynamic experiments.
Origine : Fichiers éditeurs autorisés sur une archive ouverte