Investigation of VMD denoising method based on Monte Carlo simulation: a comparative study between newly introduced autocorrelation-based method and PDF distance based method

Abstract

A signal with low signal to noise ratio is always difficult to be analysed by the traditional signal processing methods, especially the vibration and acoustic signals that contain non-linear, non-stationary, modulation phenomenon. Extracting features contaminated in heavy background noise requires an effective denoising tool and hence variational mode decomposition based denoising method has been considered in this paper. An initial investigation has been carried out for a simulation signal with very low signal to noise ratio.Firstly, VMD is introduced to decompose the signal into a number of intrinsic mode functions.The selection of IMFs is very important to get the reconstructed denoised signal. For this purpose, a noble method based on autocorrelation has been proposed along with the frequency domain denoising technique. Use of Monte-Carlo method proves the effectiveness of the proposed Autocorrelation based method and provides a comparative analysis between probability distribution function-based method and the proposed method.