A new method to classify railway vehicle axle fatigue crack AE signal

Abstract

This paper presents a new method for classifying railway vehicle axle fatigue crack acoustic emission (AE) signal. The method is developed by integrating self-adaptive empirical mode decomposition (EMD) with Elman neural network (ENN). The method first uses EMD to decompose the signals into six intrinsic mode functions (IMFs) and one residual. From the IMFs and the residual obtained by EMD, a three-dimensional energy (TDE) feature vector consisting of energy entropy, energy distribution ratio, and interval average energy are computed by Hilbert-Huang transform. An ENN will be trained using the TDE feature vector to classify the AE signal caused by railway vehicle axle fatigue crack. The result shows that this method is better than other EMD energy domain classification method on identifying railway vehicle axle fatigue crack AE signal.