Damage Recognition of Acoustic Emission and Micro-CT Characterization of Bi-adhesive Repaired Composites Based on the Machine Learning Method

Bi-adhesive repair method is one of several repair technologies that use the adhesive bonding approach for patch-repaired composites. However, these repairs are subject to matrix-cracking and interface debonding damage. Furthermore, a change in the length ratio (the length of the rigid adhesive region divided by the length of the overall repaired region) also produces a change in the damage modes, which has a significant impact on the repair performance. Hence, this study aims to evaluate the effects of four different length ratios (0, 0.2, 0.5, 1) on the behavior of damage evolution in bi-adhesive repaired composites. The acoustic emission damage identification and micro-CT characterization are carried out based on the machine learning method. A simple prediction method is employed to distinguish damage modes in bi-adhesive repaired composites, achieving a prediction accuracy over 90%. The results demonstrated that the length ratio has a substantial effect on matrix-cracking, fiber-matrix debonding, and their interaction in bi-adhesive repaired composites. These acquired characteristics information of acoustic emission signals provide insights into the impact of length ratio on the progression of damage evolution. Additionally, the visualization of interior damage offers insights into the variations in failure characteristics within distinct bi-adhesive repaired composites, thereby supporting the conclusions gained from acoustic emission studies. This research effectively achieves the real-time monitoring of damage modes in bi-adhesive repaired composites, contributing to the comprehension of the relationship between length ratio and damage mechanism.