I got IMFs using EMD algorithm. As all the IMFs might not contain significance and may be fully noise. To calculate the spread function, the spectrum-weighted mean period has to be calculated using the Fourier spectrum as a function of the natural log of the period. However, the calculation of spectrum-weighted mean period as integral form according to "Wu and Huang" is not clear to me. (The publication is attached here.) Can anybody suggest me about any detailed methodological steps or code in this regard?