Research on signal analysis method of acoustic emission of material 2.25Cr-1Mo based on wavelet filter and clustering

Abstract

Extracting acoustic emission (AE) signals of cracks and micro-cracks after material yield phase from noise is of great significance for the study of AE character of material 2.25Cr-1Mo. In this paper wavelet was used to find out the main frequency bands of the burst cracking AE signals from the tensile test of material 2.25Cr-1Mo. Burst cracking waveform was successfully extracted from mixed waveform by reconstructing wavelet coefficients with the main frequency bands. Then the new descriptors of the waveform were determined by using of a 30% floating threshold. The DBSCAN clustering was applied to separate noise signals of electric and vibration successfully. At last the k-mean clustering was used to separate burst cracking signal data from data set effectively and accurately. According to the Analysis of the material of 2.25Cr-1Mo tensile test, the cumulative energy of burst AE signals could reflect the yield point and the degree of material damage. © Springer-Verlag Berlin Heidelberg 2012.