A Sequential Variational Mode Decomposition Method

TL;DR

This paper presents a sequential variational mode decomposition method that accurately separates non-stationary signals without prior component knowledge, automatically determines the number of modes, and improves decomposition quality over existing methods.

Contribution

The proposed method introduces a sequential variational approach that automatically determines the number of modes and reduces end effects, enhancing signal separation accuracy.

Findings

Significantly improves separation accuracy compared to VMD and EMD.

Automatically determines the number of components during decomposition.

Reduces end effects through principal elongation and refinement processes.

Abstract

In this paper, we introduce a sequential variational mode decomposition method to separate non-stationary mixed signals successively. This method is inspired by the variational method, and can precisely recover the original components one by one from the raw mixture without prior knowing or assuming the number of components. And in such a way, the mode number also can be determined during the separation procedure. Such character brings great convenience for real application and differs from the current VMD method. Furthermore, we also conduct a principal elongation for the mixture signal before the decomposing operation. By applying such an approach, the end effect can be reduced to a low level compared with the VMD method. To obtain higher accuracy, a refinement process has been introduced after gross extraction. Combined these techniques together, the final decomposition result implies a significant improvement compared with the VMD method and EMD method.