De-noising of rail crack AE signal based on wavelet modulus maxima

On the basis of wavelet modulus maxima characterization of local regularity theory and the advantages of wavelet transform, an optimized wavelet modulus maxima de-noising method applied on rail crack AE signal is presented in this paper. In the background of the new real-time rail crack detection method by AE, wavelet modulus maxima de-noising is proved to be an effective way to extract the crack signal. Different parameters choosing principles in different noise conditions are discussed, in order to get the best de-noise effectiveness at different speed. The Segmented multi-frequency damping oscillation model is proposed, and the relation between the simulate signal and the real ones are found. Through the experiments of simulate signals, the principles of selecting proper parameters and the de-noising abilities at different speed are demonstrated, which give a strong evidence of the effectiveness of this method.