Acoustic emission applied to mode I fatigue damage monitoring of adhesively bonded joints

The use of adhesively bonded joints has increased considerably due to their lightweight, relevant strength-weight ratio and possibility to join multi-materials. Nevertheless, there are still some challenges in the application of this kind of joints in primary structures, such as guaranteeing their reliability during the components’ useful life. Structural health monitoring methods are suggested to ensure in-service safety and reliability of adhesive joints. The acoustic emission appears promising because it can detect the elastic waves produced within the material when it is under damage or straining. This research focuses on mode I fatigue damage monitoring metallic double cantilever beam adhesively bonded joints using the acoustic emission method. Digital image correlation and visual evaluation were applied during fatigue interruptions to track the crack-tip position within the adhesive and correlate them with the acoustic emission outcomes. The acoustic emission method is susceptible and different kinds of waves (background, friction and damage) can be easily assessed during the tests, producing an immense amount of data. So, unsupervised artificial neural networks for patterning recognition were proposed. Self-organising maps and k-means algorithms were used for data clustering and then classified regarding their sources. Finally, the acoustic emission results, digital image correlation and visual evaluations were compared