Identification for acoustic emission signal of crack based on EMD approximate entropy and SVM

Abstract

When cracks appear, material of the pipeline will deform, which is leading to the phenomenon of acoustic emission (AE). It is a kind of nonlinear, nonstationary complex signal. Based on this feature, an effective method of the signal identification based on empirical mode decomposition (EMD) approximate entropy (ApEn) and support vector machine (SVM) is applied in this paper. First, the signals are decomposed into some intrinsic mode function (IMF) signals using EMD algorithm; secondly, with simple processing of IMF, ApEn is used to calculate feature information; and lastly, the results are used as feature vector. The experimental results show that the method is an effective and convenient way to identify the signals.