Continuous Wavelet Transform and Deep Learning for Accurate AE Zone Detection in Laminated Composite Structures

Abstract

Acoustic Emission (AE) source localization is a crucial area of research, particularly in the context of laminated composite structures. Existing AE localization techniques based on the Time Difference of Arrival are limited to homogeneous isotropic materials and are not suitable for the anisotropic characteristics of laminated composites. To address these limitations, this study introduces a novel approach using Continuous Wavelet Transform (CWT) to transform AE signal waveforms captured by a sensor network into scalograms. These scalograms were then utilized to train deep Convolutional Neural Network (CNN) models, resulting in exceptional prediction accuracy. The trained models achieved remarkable performance, with training, validation and testing accuracy exceeding 97%, 95%, and 96%, respectively. To further validate the model, additional AE experimental data was collected and tested, yielding an accuracy of 95% with only two misclassifications out of 40 test data points. Moreover, a user-friendly web application was developed using the Streamlit open-source framework, enabling the practical deployment of this industrial-grade AE localization system without necessitating advanced AI skills, achieving significant Technology Readiness Level (TRL) for widespread utilization in engineering applications.