pyNNST 0.2

Definition of non-stationary index for time-series

Index of non-stationarity

Obtaining non-stationary index for time-series.

Installing this package

Use pip to install it by:

$ pip install pyNNST

Simple examples

Here is a simple example on how to use the code:

# Import packages
import pyNNST
import numpy as np

# Define a sample signal x
T = 20                                # Time length of x
fs = 400                              # Sampling frequency of x
dt = 1 / fs                           # Time between discreete signal values
x = np.random.rand(T * fs)            # Signal
time = np.linspace(0, T - dt, T * fs) # Time vector
std = np.std(x, ddof = 1)             # Standard deviation of x
mean = np.mean(x)                     # Mean value of x

# Class initialization
example = pyNNST.nnst(x, nperseg = 100, noverlap = 0, confidence = 95)

# Compute the run test for non-stationarity
example.idns()
outcome = example.get_outcome()  # Get the results of the test as a string
index = example.get_index()      # Get the index of non-stationarity
limits = example.get_limits()    # Get the limits outside of which the signal is non-stationary

Reference:

Non-stationarity index in vibration fatigue: Theoretical and experimental research; L. Capponi, M. Cesnik, J. Slavic, F. Cianetti, M. Boltezar; International Journal of Fatigue 104, 221-230 https://www.sciencedirect.com/science/article/abs/pii/S014211231730316X