Acoustic emission pattern recognition method utilizing elastic wave simulation

Abstract

Fracture mode analysis via acoustic emission (AE) has attracted research attention as a means of identifying the damage mechanism in machines and engineering structures. However, the estimation of relevant parameters from waveforms detected by AE sensors requires complex analysis because the waveforms are often distorted by the sensor response. Therefore, in this study, we develop a method for the classification of AE signals based on simulated AE waveforms. The simulated AE waveform is a convolution of the simulated elastic wave calculated via the 3D finite-difference time-domain (FDTD) method and the sensor response acquired with the use of the long bar method. The AE signals with different propagation angles obtained with the use of an epoxy-glass fiber composite specimen are classified, and wavelet contour maps and correlation coefficients are used for AE classification. The success of the AE classification is validated by the similarity of waveform features between the experiment and simulation.