摘要
针对小波阈值降噪方法中小波基和阈值缺乏选取依据的缺陷,提出了一种基于经验模态分解(EMD)的小波阈值降噪方法。首先将带噪信号进行EMD分解得到一系列本征模态分量(IMF),仅对带噪的高频IMF分量进行小波阈值降噪处理,将处理结果与不含噪声的低频IMF分量进行信号还原得到降噪后信号。方法有效避免了直接小波阈值降噪高频分量损失的问题,同时还可直接去除信号中可能存在的趋势项,比直接小波阈值降噪具有更好的效果。仿真数据处理证明了方法的有效性。
Wavelet base and threshold have no theoretical basis for choosing wavelet threshold de - noising. A new method based on empirical mode decomposition （EMD） is proposed. At first, noisy signal is decomposed to several intrinsic mode functions （IMF）. Secondly, wavelet threshold de -noising only acts on the high frequency IMF which contain noise, the results and the low frequency IMF can reconstructed to obtain the denoised signal. This method avoids the high frequency information lost during the de - nosing process, meanwhile it can eliminate the tendency part that might exist in the signal. The simulation results show that wavelet threshold de - noising based on EMD has advantage over the traditional wavelet threshold de - noising methods.
出处
《计算机仿真》
CSCD
北大核心
2009年第9期325-328,337,共5页
Computer Simulation
关键词
经验模态分解
降噪
本征模态分量
趋势项
Empirical mode decomposition（EMD）
De -nosing
Intrinsic mode function
Tendency part
