Optimization of variational mode decomposition using stationary wavelet transform and its application to transient electromagnetic signal noise reduction

Abstract

To solve the problem of signal loss due to local reconstruction in the variational mode decomposition (VMD) method, this study proposes to use the stationary wavelet transform (SWT) to extract the effective signal in the mixed noise modes and reconstruct the noise-reduced signal. First the slime mold algorithm (SMA) takes to realize the adaptive difficulty of selecting the important parameters K (the number of eigenmode decompositions) and a (the quadratic penalty coefficient) in the VMD. Then, the VMD decomposed modes are divided into the basic signal and noise signal according to the definition of Euclidean distance, finally the noise signal is decomposed in a new step by using SWT, and the basic signal is reconstructed with the effective signal to get the final noise reduced signal. Through the establishment of simulation tests and transient electromagnetic field tests in the mined-out area, the results show that the VMD-SWT method exhibits a better denoising effect and higher inversion accuracy for the transient electromagnetic signals, proving the superiority and applicability.