Abstract

Lamb waves, characterized by slow attenuation, short wavelength, and sensitivity to damage, have been widely utilized in the field of composite material damage detection. In addressing the issue of damage localization in composite curved tank structures, this study proposes a composite tank localization algorithm based on convolutional neural networks (CNNs) using Lamb wave data. The approach transforms the damage localization regression problem into a classification problem through partitioning, achieving damage zone localization in composite tank structures with fewer sensors. The accuracy of the algorithm is more than 98% in the dataset constructed from physical experiments, which verifies its accuracy and applicability.