A New Method for Obtaining the Local Mean in the Empirical Mode Decomposition Algorithm

This paper introduces a new definition for local mean to improve the empirical mode decomposition (EMD) algorithm. There is a lot of research going on the development of empirical mode decomposition algorithm. Due to the repetition of the local mean in each step of the sifting process of EMD algorithm, determination of the local mean is one of the important sections of this algorithm. In the proposed method, local mean has been calculated comparatively more accurate than common algorithm by considering the signal fluctuation between each two consecutive extrema. This modification which is made on the EMD method, controls the overshoot of the envelop curves in common EMD method. The modified EMD method can resolve the spectral components differing by less than one octave, more efficiently than common EMD method and gives better interpretation of the signal. The advantages of the present method have been proved through analyzing some simulated signals.