基于白噪声分离的集合经验模态分解心电信号去噪方法研究

Study on Electrocardiogram Signal De-noising Methods Based on Ensemble Empirical Mode Decomposition Decomposed by White Noise

集合经验模态分解(EEMD)是一种处理心电等非平稳信号的有效方法,但其参数白噪声比值系数与平均次数依靠经验设置,导致处理结果准确度低且对未知信号自适应性差。针对上述问题,本研究提出了基于白噪声分离的EEMD心电信号去噪方法。该方法通过经验模态分解(EMD)将心电信号分解至不同频带,基于白噪声能量密度和对应的平均周期的乘积趋向于一个常数的特性,提取信号高频分量重构信号高频成分;依据避免模态混叠参数准则实现针对不同信号的分解参数自适应获取。经过对心电信号的验证,结果表明该方法去噪效果明显,自适应性强,是一种有效的去噪方法。

Ensemble empirical mode decomposition（EEMD）is an effective method for non-stationary signal analysis,such as electrocardiogram（ECG）signals.However,the precision and correctness of EEMD are affected by the two parameters,ratio of the added noise and ensemble number.The values of two parameters are set relying on experience and lacking of adaptability for uncertain signals.In order to solve these problems,we proposed a method based on white noise decomposed by EEMD in the present study shown in this paper.Empirical mode decomposition（EMD）was applied to decompose the signal to different intrinsic mode functions（IMFs）in the de-noising process.The white noise IMFs were selected to constitute high frequency part based on the character that the product of the energy density of white noise and its average period tended to be a constant.Then the two parameters of EEMD were adaptively obtained according to the criterion which was used to avoid modal aliasing.Experimental results showed that the method was an effective one for ECG signal de-noising.