基于CEEMDAN和小波软阈值的电能质量扰动信号去噪新方法  

A new method for power quality disturbance signal denoising based on CEEMDAN and wavelet soft threshold

作　　者：张震[1] 刘明萍[1] 张镇涛 汪庆年[1] ZHANG Zhen;LIU Mingping;ZHANG Zhentao;WANG Qingnian(School of Information Engineering,Nanchang University,Nanchang 330031,China)

机构地区：[1]南昌大学信息工程学院,江西南昌330031

出　　处：《现代电子技术》2021年第18期63-68,共6页Modern Electronics Technique

基　　金：国家自然科学基金(61865011);国家自然科学基金(61665006);南昌大学研究生创新项目(CX2019092)。

摘　　要：为了有效降低噪声对电能质量扰动信号检测的影响,文中提出基于自适应加噪的完备集合经验模态分解(CEEMDAN)和小波软阈值相结合的去噪新方法。该方法利用CEEMDAN将含噪信号分解成为多个频率由高到低的固有模态分量(IMF),采用BP神经网络对含噪信号两端进行延拓来抑制端点效应,通过计算各IMF的多尺度排列熵(MPE)的均值来确定高频含噪分量和低频扰动信号分量,然后使用小波软阈值对高频分量进行去噪,最后将去噪后的IMF与低频IMF进行信号重构,这样既保留了高频分量的有效信息又滤除了噪声。仿真实验表明,利用所选定的新方法去噪效果相对于小波硬阈值去噪、小波软阈值等传统方法有一定的优势,去噪后的信号能够真实地反映电能质量扰动信号的特征,所得结果合理、有效。In order to effectively reduce the influence of noise on the detection of power quality disturbance signals,a new denoising method combining complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)and wavelet soft threshold is proposed.In the method,the CEEMDAN is used to decompose the noisy signals into multiple inherent mode components(IMF)with high to low frequencies,and BP neural network is adopted to extend the two ends of the noisy signal to suppress the end effect.The high⁃frequency noisy components and low⁃frequency disturbed signal components are determined by calculating the mean value of the multi⁃scale permutation entropy(MPE)of each IMF,and then the high⁃frequency components are denoised by using the wavelet soft threshold.Finally,the signals of the denoised IMF and low⁃frequency IMF are reconstructed,which can not only retain the effective information of the high⁃frequency components,but also filter out the noise.The simulation experiment results show that the denoising effect of the new method is better than that of the traditional methods such as wavelet hard threshold denoising and wavelet soft threshold denoising.The denoised signal can truly reflect the characteristics of power quality disturbance signal,which proves that the method is reasonable and effective.

关 键 词：电能质量 信号去噪 CEEMDAN 小波软阈值 IMF分解 噪声滤除 

分 类 号：TN957.54[电子电信—信号与信息处理]