Feature selection and clustering of damage for pseudo-ductile unidirectional carbon/glass hybrid composite using acoustic emission

Abstract

Hybrid composites (made up of carbon and glass) have widely adopted acoustic emission (AE) as a technique to study their structural behavior. However, the fiber breakage damage mechanisms from the AE signal are only limited to bulk fiber, which prevents the analysis of the constituent fiber behavior towards failure. To solve this problem, AE analysis was done on tensile loaded thin-ply carbon and glass hybrid composite (exhibiting pseudo-ductility) with new derived AE parameters and incorporating the Cronbach’s alpha test in the feature selection process. The feature selection process obtained the desired parameters that enabled signal clustering of the constituent fibers breakage mechanism separately and the delamination, pull-out, and matrix cracking. Pseudo-ductility is discussed using the clustered AE signals explaining the constituent fibers fracture behavior towards failure. This work is essential in evaluating hybrid composites by studying the behavior of the constituent fibers using AE analysis.