Acoustic emission signals characterization and damage source localization in composite heterogeneous panels

Geometry and materials of wind turbine blades are becoming more and more complex, leading to great challenges in the structural health monitoring of wind turbine blades. In particular, a large number of composite heterogeneous structures are used in wind turbine blades. In this paper, a novel acoustic emission (AE) based method is proposed for structural damage localization in such composite heterogeneous panels. Firstly, the attenuation and frequency propagation characteristics of AE signals are systematically investigated. Subsequently, AE signals undergo a transformation into graph-structured data utilizing graph theory and wavelet coefficients to extract intricate signal features. Then, a graph convolutional network (GCN)-based method is proposed to learn the features of the constructed graphs and to predict the coordinates of AE sources. The effectiveness of the proposed method is validated by pencil lead break (PLB) experiments conducted on a composite heterogeneous panel. The results demonstrate that the proposed method can accurately locate the position of AE sources and it outperforms traditional convolutional neural network (CNN) approaches.