Study On Fault Feature Extraction Algorithm For Rotating Machinery Based On Morlet Wavelet And SVD

In the process of fault diagnosis, how to extract fault feature from original signal components is the emphasis and difficulty. Morlet wavelet transform and Singular Value Decomposition(SVD) are two kinds of modern numerical analysis method and have each unique effect on fault feature extraction. Based on the basic principle and important properties of Morlet wavelet and SVD,the application in signal feature extraction of these two methods is deeply researched and the combination algorithm used in fault feature extraction is proposed.Firstly,the influence of scale parameter and displacement paremeter on Morlet wavelet is analyzed.The calculation method of Morlet wavelet transform is derived. According to the problem of parameter optimization for Morlet wavelet,the improved optimization method based on Shannon entropy is proposed.Secondly, the signal separation principle based on Hankel matrix when using SVD is studied.The relation between singular values and different components in original signal are studied. The change law of the positions of singular values representing periodic components and noise components is studied.Then, the SVD method is used in the coefficient matrix after Morlet wavelet decomposition and the result is analyzed. A method based on energy spectrum of singular value is proposed.The location of the fault characteristic frequency is obtained by choosing the maximum in energy spectrum.This method has achieved good effect in the feature extration of simulation signal.Fanally, the method is applied to fault diagnosis of rotor system. In practical applications, the validity and practicability of the proposed method are verified.