Identification of fatigue damage evaluation using entropy of acoustic emission waveform

摘要

Acoustic emission (AE) is a passive nondestructive testing (NDT) technique which is employed to identify critical damagein structures before failure can occur. Currently, AE monitoring is carried out by calculating the features of the signalreceived by the AE sensor. User-defined acquisition settings (i.e., timing and threshold) significantly affect many traditionalAE features such as count, energy, centroid frequency, rise time and duration. In AE monitoring, AE features are stronglyrelated to the damage sources. Therefore, AE features that are calculated due to inaccurate user-defined acquisition settingscan result in inaccurately classified damage sources. This work presents a new feature of the signal based on themeasure of randomness calculated using second-order Renyi’s entropy. The new feature is computed from its discreteamplitude distribution making it independent of acquisition settings. This can reduce the need for human judgement inmeasuring the feature of the signal. To investigate the effectiveness of the presented feature, fatigue testing is conductedon an un-notched steel sample with simultaneous AE monitoring. Digital image correlation (DIC) is measured alongsideAE monitoring to correlate both monitoring methods with material damage. The results suggest that the new feature issensitive in identifying critical damages in the material.