Variation mode decomposition algorithm parameter optimization method based on particle swarm optimization algorithm

Abstract

The invention discloses a variation mode decomposition algorithm parameter optimization method based on a particle swarm optimization algorithm and belongs to the scope of non-stationary signal time-frequency analysis methods. According to the main process of the method, first, a characteristic quantity, namely entropy of an IMF Fourier spectrum, capable of reflecting a decomposition effect of an Intrinsic Mode Function (IMF) of non-stationary signal characteristics is defined, and a Fourier spectrum entropy function of the IMF is used as a fitness function of the particle swarm optimization algorithm; second, the PSO algorithm is initialized, and then an optimal solution of the fitness function starts to be searched for; and last, the obtained optimal solution is used as a parameter of a variation mode decomposition algorithm. According to the method, an optimal parameter combination of the parameters K and alpha of the variation mode decomposition algorithm is solved automatically through the particle swarm optimization algorithm, the solved parameters K and alpha are optimal, and the process is fast and efficient.

Description

technical field [0001] The invention belongs to the category of non-stationary signal time-frequency analysis method, and is applied to determine the optimal combination of parameters K and α of the corresponding variational mode decomposition algorithm when performing variational mode decomposition on non-stationary signals. Among them, K represents the number of decomposing the non-stationary signal into single component IMF (eigenmode function), and α is the quadratic penalty parameter. Background technique [0002] For a long time, the Empirical Mode Decomposition (EMD) algorithm has been widely used to decompose non-stationary signals to obtain its Intrinsic Mode Function (IMF). Time-frequency division of stationary signals. Because the EMD algorithm adopts the processing method of cyclic screening and peeling to obtain the IMF component, the EMD algorithm has the following defects: it lacks strong mathematical theory support; the algorithm uses recursive screening to ...

AI Technical Summary

Problems solved by technology

Because the EMD algorithm adopts the processing method of cyclic screening and peeling to obtain the IMF component, the EMD algorithm has the following defects: it lacks strong mathematical theory support; the algorithm uses recursive screening to decompose successively, and cannot reverse error correction; Mode loss, etc., these defects seriously restrict the application of EMD

The parameters K and α have a great influence on the decomposition results and are difficult to determine

Images