Cluster Analysis of Acoustic Emission Signals on Tensile Damage Process of C/SiC Using an Improved K-Means Algorithm

C/SiC is one of most important thermal-structural materials for future aeronautical applications, but there is no consensus on the damage mechanism and pattern of it. To identify the failure processes under mechanical load is important to evaluate this material. In this paper, acoustic emission (AE), an effective continuous damage monitoring technique, is used to monitor tensile tests of 2D-C/SiC. An unsupervised clustering method based on the deep fusion of a genetic algorithm and a K-means algorithm is proposed to analyze the AE signals. Pattern recognition analysis of AE data was carried out to describe the evolution of various damage mechanisms during the failure process. According to the features of each class and associated SEM images of the fracture surfaces, a match between AE clusters and fracture mechanisms involved in the process, including matrix cracking, interface failure, interlaminar delamination, fiber breakage and bundle breakage, is established. It is found that the damage evolution of 2D-C/SiC under tensile loading can be divided into four stages at room temperature. The first stage and the third stage are the main development stages of matrix cracking and interface damage. The second stage and the fourth stage are the main damage periods of fiber, fiber bundle and interlaminar delamination. The damage pattern recognition and damage evolution process of 2D-C/SiC composites during tensile test are described.