Research on clustering identification of acoustic emission events in the process of wood crack propagation using PCA
This study presents a methodology for feature extraction and identification of acoustic emission (AE) events during wood crack propagation utilizing Principal component analysis (PCA) and enhanced K-means clustering algorithm. Experimental setups included double cantilever beam (DCB) for mode I crack propagation analysis and three-point bending test for mixed-mode crack propagation assessment. Various AE parameters, such as amplitude, duration, absolute mean value, peak frequency, and frequency centroid, were computed. PCA applied for dimensionality reduction to extract principal components and eliminate redundant information. The optimal number of clusters was determined using a combination of the elbow method and the Davies-Bouldin index to classify damage modes. Results indicate that the principal components contribute to 88.5% and 92% of the variance in the two tests, respectively, yielding three distinct types of AE events in both crack propagation scenarios. Specifically, high-frequency, low-amplitude signals correspond to microcrack initiation; low-frequency, low-amplitude signals signify interface delamination; and high-amplitude, long-duration events indicate mode I opening macroscopic damage (high frequency) and mixed-mode macroscopic failure (low frequency).