Filter Principle of Hilbert-Huang Transform and Its Application in Time Series Analysis

Hilbert-Huang transform (HHT) is a typical realization of the local wave analysis (LWA) and especially powerful for analysis of nonlinear and non-stationary signal. In this paper, from the view of filter, the principle of HHT will be studied and a smoothing filter theory will be put forward for explaining empirical mode decomposition (EMD). And because intrinsic mode functions (IMFs) by EMD and the time-frequency-energy distribution by Hubert transform (HT) can reflect the inherent essential characteristics of time series, based on HHT, a new method is obtained about change-point detection of time series. Experiments test the validity of the proposed method