改进变分模态分解的加速度信号降噪方法  

Noise reduction method for acceleration signal based on variational mode decomposition

作　　者：贾国庆[1] 霍真如 易辉跃[2,3] 房卫东 许晖[2,3] Jia Guoqing;Huo Zhenru;Yi Huiyue;Fang Weidong;Xu Hui(School of Physics Electronic Information Engineering,Qinghai Minzu University,Xining 810007,China;Shanghai Institute of Microsystems Information Technology,Chinese Academy of Sciences,Shanghai 201899,China;Shanghai Wireless Communication Research Center,Shanghai 201210,China)

机构地区：[1]青海民族大学物理与电子信息工程学院,青海西宁810007 [2]中国科学院上海微系统与信息技术研究所,上海201899 [3]上海无线通信研究中心,上海201210

出　　处：《黑龙江科技大学学报》2022年第4期506-511,共6页Journal of Heilongjiang University of Science And Technology

基　　金：青海省应用基础研究计划项目(2020-ZJ-724)。

摘　　要：变分模态分解算法重构信号的模态分量选取影响加速信号滤除噪声的效果。通过建立模态分解数与输入信号样本熵模型确定模态分解数,计算分解后各模态分量的样本熵大小选取模态分量,经数据平滑处理后重构信号,给出改进的变分模态分解算法,与卡尔曼滤波、小波变换、经验模态分解和传统变分模态分解对比分析滤除信号噪声的效果。结果表明,相比于传统变分模态分解算法,所提改进变分模态分解算法的均方误差降低了3%,对信噪比改善提升了12.35%,说明了改进算法对加速度信号滤除噪声效果更好,更有利于提取被测物体的特征参数信息。This paper proposes an improved variational model decomposition algorithm,which is designed to improve the effect of accelerating removing signal noise by selecting the modal components of the reconstructed signal using the variational model decomposition algorithm.The study involves determining model decomposition number by the established model components and the input signal sample entropy model;calculating the sample entropy number of each decomposed model component and selecting model components;reconstructing the signal after data smoothing;and comparing and analyzing the effect of removing signal noise with Kalman filter,wavelet transform,empirical model decomposition and traditional variational model decomposition.The results show that,compared with the traditional variational mode decomposition algorithmthe,the mean square error of the improved variational mode decomposition algorithm is reduced by 3%;and the improvement of the signal-to-noise ratio increases by 12.35%,which indicates that the improved algorithm could filter the noise better with acceleration signal,and could be more conductive to extracting the characteristic parameter information of the measured object.

关 键 词：加速度信号 变分模式分解 降噪 样本熵 

分 类 号：TN911.7[电子电信—通信与信息系统]