Study on the Frequency Tracking ability of Empirical Mode Decomposition

This paper aims to study and evaluate the potential of Empirical Mode Decomposition (EMD) algorithm as an online frequency tracker in the presence of sine noises (harmonics). The EMD algorithm works on the principle of time scale variations between the upper maximum values and lower minimum values of a waveform, and for a long time it has been considered as a good offline analysis approach. This paper studies the online frequency tracking ability of EMD with Hilbert Transform and brings forth the limitation of the Fourier Transform, S Transform and Hilbert Transform as frequency estimators. Here a signal section between two zero crossing is subjected to EMD which segregates the oscillating modes of the signal. Further, fundamental oscillating mode is identified and subjected to Hilbert Transform for frequency estimation. The identification of the fundamental component is done by evaluating a correlation parameter between each of the obtained oscillating components and the original section of the signal. The paper also investigates the effect the sampling frequency has on the efficiency of the algorithm.