Hilbert-Huang变换应用中的预处理方法研究

Study on signal pre-processing method for Hilbert-Huang transform

为提高Hilbert-Huang变换(HHT)中瞬时频率计算的真实性和稳定性,提出了一种信号的预处理方法.该方法首先对信号作傅里叶变换,然后根据插值点数作补零处理,再进行傅里叶反变换,完成傅里叶插值功能.对傅里叶插值处理后的信号作经验模态分解(EMD)得到一组固有模态函数(IMF),计算所有IMF的瞬时频率和幅值,最终获得信号时频分布的Hilbert谱.结果表明,该傅里叶插值的预处理方法能够有效消除和抑制HHT分析中的瞬时频率波动和虚假成分产生,增强了瞬时频率的准确性,提高了HHT方法的信号分析频率,该方法能有效应用于实际信号处理的HHT时频分析中.

For improving the facticity and stability of the instantaneous frequencies in Hilbert-Huang transform （HHT）, a signal pre-processing method was proposed. Signals after FFT were interpolated by adding zeros according to the interpolation number. Signals were processed using IFFT and the FFT interpolation function was accomplished. The FFT interpolated signals were decomposed into a set of intrinsic mode function （IMF） using empirical mode decomposition （EMD）. The instantaneous frequencies and amplitudes of all IMFs were calculated, and the Hilbert spectrum for a time-frequency distribution of the signals was formed. The results show that the pre-prossing method using FFT interpolation technology can effectively suppress and eliminate the fluctuation of the instantaneous frequencies and the production of spurious components in HHT, improve the fidelity of the instantaneous frequencies, and increase the signal analysis frequency of HHT method. The presented method is useful for the signal processing by HHT in practice.