基于夹角余弦和模糊阈值的EEMD去噪方法

发布时间：2018-04-02 04:26

本文选题：信号去噪　切入点：聚合经验模态分解　出处：《传感技术学报》2016年06期

【摘要】：针对非线性非平稳信号的去噪问题,结合EEMD分解信号的自适应特性,提出一种基于夹角余弦和模糊阈值的去噪方法。首先用夹角余弦法计算各个本征模态函数(IMF)与观测信号之间的相似度,以相似度曲线的首个极小值的后一个位置为分界点将分解出的IMF分为噪声主导模态和信号主导模态;然后根据Visu Shrink阈值易"过扼杀"细节系数和SUREShrink阈值易"过保留"噪声系数的特点,利用模糊阈值对噪声主导的IMF进行处理;最后将所有的IMF重构得到消噪信号。分别采用仿真信号和真实ECG信号进行去噪实验。结果表明,所提方法在整体性能上优于小波半软阈值方法、基于EMD的软阈值(EMD-Soft)和间隔阈值(EMD-IT)方法,是一种有效的去噪方法。
[Abstract]:A new denoising method based on angle cosine and fuzzy threshold is proposed to solve the problem of nonlinear non-stationary signal de-noising combined with the adaptive characteristic of EEMD decomposition signal.Firstly, the similarity between each intrinsic mode function (IMF) and the observed signal is calculated by the inclusion angle cosine method. The decomposed IMF is divided into noise-dominated mode and signal-dominated mode with the position of the first minimum value of the similarity curve as the boundary point.Then, according to the characteristics of Visu Shrink threshold which is easy to "kill" detail coefficient and SUREShrink threshold value easy to "over-retain" noise coefficient, fuzzy threshold is used to deal with the noise-dominated IMF. Finally, all IMF signals are reconstructed to get denoising signals.The simulation signal and the real ECG signal are used for denoising experiment.The results show that the proposed method is superior to the wavelet semi-soft threshold method in overall performance. The soft threshold method based on EMD and the interval threshold method based on EMD-ITT method are an effective denoising method.
【作者单位】：湖南大学电气与信息工程学院;
【基金】：中央国有资本经营预算项目(财企[2013]470号) 中央高校基本科研项目(2014-004) 国家自然科学基金项目(61172089) 湖南省科技计划项目(2014WK3001) 中国博士后科研基金项目(2014M562100) 湖南省科技计划重点项目(2015JC3053) 湖南省自然科学基金项目(14JJ4026)
【分类号】：TN911.4

【参考文献】