Research On Signal Method Based On Variational Modal Decomposition

Variational Mode Decomposition(VMD)is an adaptive signal decomposition method,which is not only suitable for processing linear and stationary signals,but also suitable for processing non-linear and non-stationary signals.The traditional Fourier transformation can represent the physical meaning of the signal well and;it is an ideal tool for studying the stationary signal.However,the Fourier transform has high requirements on the linearity and stability of the signal.Moreover,the signals we generate in life are often nonlinear and nonstationary.Therefore,the VMD has a broader application prospect and research value.At present,VMD has great application value in mechanical fault diagnosis,feature extraction,signal detection,seismic exploration and other fields.In speech signal analysis and image processing technology,noise reduction has always been a classic research content.In this paper,the application of VMD in signal de-noising is studied in depth.In combination with wavelet threshold and mean value filtering,a noise reduction method for one-dimensional speech signals and two-dimensional image signals is proposed.The specific research contents is as follows:For the noise reduction problem of one-dimensional speech signals,according to the instability and non-continuity of the speech signal,this paper proposes a de-noising method based on one-dimensional VMD and wavelet threshold de-noising.In this method,the input one-dimensional speech signal is first decomposed by the method of VMD,and is decomposed into a finite number of intrinsic modal components.Subsequently,each intrinsic modal component obtained by the decomposition is de-noised by the wavelet threshold.The method performs noise reduction processing.Finally,the noise-reduced intrinsic modal components are reconstructed to obtain a de-noised speech signal.This paper uses seven different signal-to-noise ratios,one-dimensional simulation signals and noise-containing speech signals to carry out simulation experiments;and compare the experimentsal results with a variety of algorithms.The experimental results show that this method can better reduce the Gaussian white noise contained in the speech signal.For the problem of noise reduction of 2D image signals,a noise reduction method based on the 2D-VMD and mean filtering is proposed in this paper.In this method,the input twodimensional image signal is first decomposed by a two-dimensional VMD method into a finite number of intrinsic modal components.These components are arranged according to the frequency of the signal,and are mainly distributed on the basis of the noise signal.In the highfrequency signal,the low-frequency signal is mainly composed of useful signals.The lowfrequency natural modal components are retained,the high-frequency natural modal components are filtered by averaging,and the high-frequency natural modal components after de-noising are obtained;finally,the noise is reduced.The high-frequency natural modal components and the retained low-frequency natural modal components are then reconstructed to obtain a de-noised image signal.In this paper,three images with seven different SNRs are used for simulation experiments,and compared with three traditional image de-noising algorithms.The experimental results show that the proposed method can reduce Gaussian white noise in images.