Clustering Techniques and Artificial Neural Network for Acoustic Emission Data Analysis

Abstract

Acoustic emission (AE) sensor technology is commonly used for real-time monitoring of fatigue sensitive details. This is mainly due to its ability to detect fatigue events (crack initiation and opening) by mounting sensors in the vicinity of potential crack location. Also, AE data can be used for damage location detection. Even though AE provides many capabilities with regard to fatigue monitoring, many implementation challenges exist. A majority of the challenges is associated with noise elimination, AE signal analysis, and interpretation of the results. This article describes AE implementation for monitoring a fatigue-sensitive detail and use of data analysis techniques such as cluster analysis, non-linear mapping (NLM), and three-class classifiers to identify the relationship of each cluster to the characteristics of crack opening signals, background noise, and structural resonance.