Research On Hilbert-Huang Transform And Its Improvements

Hilbert-Huang transform (HHT) is a new method for signal analysis proposed by Huang in 1998. The main innovations in the method are the intrinsic mode function (IMF) and the empirical mode decomposition (EMD). Since it is based on the local characteristic of a signal and adaptive, it is highly efficient and applicable to analyze non-linear and non-stationary signal.The research of this paper is mainly focus on three improvements of HHT. First the basic principle and the algorithm of HHT are elaborated in detail. Then four major problems in HHT, included curve fitting, end effects, mode mixing and stop criterion, are summarized.For the problem of local mean fitting, the curve interpolations such as cubic spline fitting used in the traditional EMD is sensitive to extrema. An effective empirical mode decomposition method using support vector regression machines (SVRM) predicts the local mean is proposed. The analysis results indicate that the proposed algorithm has higher performance in the ability of frequency separation and can eliminate mode mixing in some intermittences.For the importance of stop criterion in EMD method, there are some improved approaches were introduced. According the orthogonality of IMF, the energy difference tracking criterion introduced by Cheng is reinterpreted in another way. A stop criterion based on the entropy of Hilbert spectrum is proposed. The experimental results show the criterion is effective.As we known the Teager energy operator (TEO) can track the energy and identify the instantaneous amplitude and frequency, the TEO is proposed to estimate the instantaneous frequency of IMF. The simulation results indicate it has better time-frequency distribution than Hilbert transform.