An improved approach for frequency empirical mode decomposition

摘要

Empirical mode decomposition (EMD) is an adaptive signal processing method. However as an empirical method, EMD not only lacks the mathematical theoretical foundation, but also has end effect, overshoots or undershoots, IMF criteria problem during the sifting process. Especially because of the special sift method, EMD always decomposes wideband signal from high frequency to low frequency, the decomposition sequence may not be chosen. These characteristics bring on the incapability of separating components in structural health monitoring. This paper firstly proposed a modified FFT approach which can make the effective bandwidth up to twice Nyquist frequency, then apply this tools to improve frequency EMD method. From simulated example, it is can be seen that the analysis effect of frequency EMD was improved by this way.