基于改进的经验模态方法脑电信号分解

EEG signals decomposition based on improved ensemble empirical mode decomposition method

针对经验模态分解(empirical mode decomposition,EMD)过程中存在的包络拟合问题,提出了一种消减欠冲现象的改进算法。该算法通过引入伪极值点增加了极值点的数目,构成了新的极值序列;然后利用新的极值序列插值拟合得到新的包络线;最后通过仿真实验对比所提算法和经典拟合算法包络拟合产生的欠冲点数目。实验结果显示,与经典拟合算法相比,改进的算法产生的欠冲点数目减少了大约77.5%。实验结果表明,此算法可以有效地消减欠冲点的数目,拟合出的包络线更加贴近原始信号,拥有更好的平滑性。

Due to the envelope fitting problem exists in the process of empirical mode decomposition,this paper proposed an improved algorithm which could eliminate the undershoot phenomenon exactly.By introducing pseudo-extreme points,this algorithm increased the number of extreme points and formed new extreme value sequence.Then it got new envelope by using the new extreme value sequence interpolation.The envelope fitted by this method was closer to the original signal and had better smoothness.Finally,a contrast result between cubic spline interpolation and this algorithm show that the number of undershoots decreases by about 77.5%,which proves that this algorithm can effectively reduce the number of undershoot points.In addition,fitting the envelope can tightly wrap the original signal and have a better envelope.