基于PSO-VMD的工频磁异常信号去噪算法  

Power frequency magnetic anomaly signal denoising algorithm based on PSO-VMD

作　　者：田斌[1,2] 赵晨 杨超[1,2] 洪汉玉 Tian Bin;Zhao Chen;Yang Chao;Hong Hanyu(Hubei Key Laboratory of Optical Information and Pattern Recognition,Wuhan University of Engineering,Wuhan 430025,China;College of Electrical Information,Wuhan University of Engineering,Wuhan 430025,China)

机构地区：[1]武汉工程大学光学信息与模式识别湖北省重点实验室,湖北武汉430025 [2]武汉工程大学电气信息学院,湖北武汉430025

出　　处：《华中科技大学学报（自然科学版）》2024年第3期47-51,64,共6页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(62171329)。

摘　　要：为解决工频水下磁目标信号特征在强背景噪声干扰下提取困难的问题,提出一种基于粒子群(PSO)优化变分模态分解(VMD)降噪方法.对VMD优化时选取包络谱峰值因子作为适应度函数,该算法不仅能有效克服经验模态分解(EMD)算法的模态混叠和端点效应问题,还能克服了VMD依赖人工经验调参造成分解效果存在偏差的问题.并将其应用于仿真与实测信号的降噪算例中,结果表明相较于集合经验模态分解(EEMD)与VMD算法,PSO-VMD算法不仅将信噪比提升了22 dB左右,还最大限度地保留了磁异常信号的原始特征,提取到了水下目标磁扰动信号,为水下磁异常检测提供一种新思路.In order to solve the problem that it is difficult to extract the characteristics of power frequency underwater magnetic target signal under strong background noise interference,a noise reduction method based on particle swarm optimization(PSO)optimized variational mode decomposition(VMD)was proposed.When optimizing VMD,the envelope spectrum peak factor was selected as the fitness function.This algorithm can not only effectively overcome the modal aliasing and endpoint effect of the empirical mode decomposition(EMD)algorithm,but also overcome the problem that VMD relies on artificial experience to adjust the parameters,resulting in a deviation in the decomposition effect.It was applied to the noise reduction examples of simulated and measured signals.The results show that compared with the ensemble empirical mode decomposition(EEMD)and VMD algorithm,the PSO-VMD algorithm not only improves the signal-to-noise ratio by about 22 dB,but also retains the original characteristics of the magnetic anomaly signal to the maximum extent.The magnetic disturbance signal of the underwater target is extracted,which provides a new idea for underwater magnetic anomaly detection.

关 键 词：变分模态分解 粒子群优化算法 降噪 参数优化 工频磁场 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]