Acoustic emission technique for damage identification of large-scale CFRP tubes under complex damage mechanisms

The purpose of this paper is to investigate the performance of acoustic emission (AE) detection technology applied in damage identification of large-scale, complex-layup carbon fiber-reinforced polymer (CFRP) components. Two sets of CFRP tubes were designed and subjected to full-scale axial compression tests. The collected AE signals were clustered using pattern recognition techniques. A total of six algorithms—k-means, hierarchical, self-organizing mapping (SOM) + k-means, principal component analysis (PCA) + k-means, SOM + hierarchical, and PCA + hierarchical—were compared in terms of their clustering efficacy. The results illustrated that PCA + k-means has the best performance in clustering. Four damage mechanisms—matrix cracking, fiber-matrix debonding, delamination, and fiber failure—can be efficiently identified. By analyzing the typical parameters of AE signals based on the clustering results, it was determined that peak frequency and energy are the most effective for damage identification. The four damage types were characterized by frequency bands of 0–50 kHz (matrix cracking), 75–125 kHz (delamination), and 150–250 kHz (debonding), and energy greater than 10⁴ μvolt‑sec (fiber failure). Using the AE technique with a suitable clustering method is an effective tool for the health monitoring of composite structures.