An example and introduction to EMD (Empirical mode decomposition) algorithm. EMD is the basis for HHT and is very suitable for work with non-stationary signals.
Empirical Mode Decomposition is a simple iterative process that breaks the signal into components called intrinsic mode functions (IMF). Every IMF contains the highest frequency of the signal in the previous iteration, thus enabling high-frequency noise rejection.
IMF is defined as the function that satisfies the following two requirements:
- The number of extrema and the number of zero-crossings in the dataset must either be equal or differ at most by one
- The mean value of the envelope defined by the local maxima and the envelope defined by the local minima is zero.
This script will show the implementation of classical EMD algorithm. For other kinds of EMD, such as ensemble EMD, please refer to:
To install emd package, insert the command:
> pip install emd