Acoustic emission signatures of damage mechanisms occuring during fatigue tests on CMC. Unsupervised and supervised classification of AE data
Identification de la signature acoustique des différents mécanismes d'endommagement lors d'essais de fatigue sur CMC. Application de classificateurs supervisé et non supervisé
Résumé
The intent of this paper is to identify the damage phenomena controlling the life of composites at high temperatures (> 600°C), and to measure the kinetics of these phenomena during mechanical tests. The analysis of AE data provides a classification into groups of signals having similar characteristics that can be afterward associated to one or to several damage mechanisms. A method of unsupervised classification was used to differentiate the signals resulting from various mechanisms of damage and from failures. Then, a method of supervised classification was developed, so as to establish a classification in real-time whatever the temperature, the applied constraint and the mode loading (static or cyclic).