README.md

NOTE: vmdpy is officially distributed with, and maintained in, sktime, since August 2023.

Please ask questions or open issues or PR on the sktime repository.

vmdpy: Variational mode decomposition in Python

Function for decomposing a signal according to the Variational Mode Decomposition (Dragomiretskiy and Zosso, 2014) method.

This package is a Python translation of the original VMD MATLAB toolbox

Installation

vmdpy is distributed with sktime.

Install sktime via pip or conda, i.e.,

pip install sktime

or

conda install sktime

For further details, see the sktime installation guide

Citation and Contact

Paper available at: https://doi.org/10.1016/j.bspc.2020.102073

If you find this package useful, we kindly ask you to cite it in your work: Vinícius R. Carvalho, Márcio F.D. Moraes, Antônio P. Braga, Eduardo M.A.M. Mendes, Evaluating five different adaptive decomposition methods for EEG signal seizure detection and classification, Biomedical Signal Processing and Control, Volume 62, 2020, 102073, ISSN 1746-8094, https://doi.org/10.1016/j.bspc.2020.102073.

For contributing new functionality or fixing anything in the package, kindly make a PR to the sktime repository (libs.vmdpy module).

For suggestions, questions, comments:

• sktime issue tracker or discussion forum, please ping vrcarva
• sktime discord

Example script

#%% Simple example: generate signal with 3 components + noise
import numpy as np
import matplotlib.pyplot as plt
from sktime.libs.vmdpy import VMD

# Time Domain 0 to T
T = 1000
fs = 1 / T
t = np.arange(1, T + 1) / T
freqs = 2 * np.pi * (t - 0.5 - fs) / (fs)

# center frequencies of components
f_1 = 2
f_2 = 24
f_3 = 288

# modes
v_1 = np.cos(2 * np.pi * f_1 * t)
v_2 = 1 / 4 * (np.cos(2 * np.pi * f_2 * t))
v_3 = 1 / 16 * (np.cos(2 * np.pi * f_3 * t))

f = v_1 + v_2 + v_3 + 0.1 * np.random.randn(v_1.size)

# some sample parameters for VMD
alpha = 2000  # moderate bandwidth constraint
tau = 0.0  # noise-tolerance (no strict fidelity enforcement)
K = 3  # 3 modes
DC = 0  # no DC part imposed
init = 1  # initialize omegas uniformly
tol = 1e-7

# Run VMD
u, u_hat, omega = VMD(f, alpha, tau, K, DC, init, tol)

# Visualize decomposed modes
plt.figure()
plt.subplot(2, 1, 1)
plt.plot(f)
plt.title("Original signal")
plt.xlabel("time (s)")
plt.subplot(2, 1, 2)
plt.plot(u.T)
plt.title("Decomposed modes")
plt.xlabel("time (s)")
plt.legend(["Mode %d" % m_i for m_i in range(u.shape[0])])
plt.tight_layout()