Pattern recognition of acoustic emission signal during the mode I fracture mechanisms in carbon- epoxy composite

The aim of the paper is to use Acoustic Emission technique to distinguish the micro/macro failure mechanisms of carbon-epoxy composite laminates during Double Cantilever Beam (DCB) tests. In order to recognize and detect different damage mechanisms, Self-Organizing Map (SOM) method has been used to cluster the AE signals according with the fracture mode that originated them. In addition, most significate Learning vector quantization (LVQ) program has been applied to verify the signals. Five AE features were selected as main parameters: Rise-time, Counts, Energy, Duration and Amplitude. The results highlighted that different signals can be recognized and classified related to their origin. The failure mechanisms detected are Matrix cracking, delamination, and fiber breakage. Scanning Electron Microscopy (SEM) images validate the results. Mathematics data and experimental results confirmed a good converging of AE data