变分分解法在传感器信号信噪分离中的应用

60 浏览量更新于2024-08-27 收藏 491KB PDF 举报

"这篇研究论文探讨了利用变分模式分解（VMD）技术对传感器信号进行信噪分离的方法，特别是针对微电子机械系统（MEMS）水听器数据采集中的强噪声干扰问题。作者指出，VMD是一种非递归的变分方法，能够将信号的频率分段，提取出中心频率和内在模态函数（IMF）。通过对比EMD（经验模态分解）、EEMD（集合经验模态分解）和VMD作为带通滤波器的效果，实验结果显示VMD在去噪效果上优于前两者，并且在处理北京大学‘Fenji’实验数据时表现出色。" 本文的核心知识点包括： 1. **变分模式分解（VMD）**：VMD是一种非线性、非递归的数据分析工具，用于信号的频谱分解。它能够将复杂信号分解为一系列内在模态函数（IMFs），每个IMF具有单一或有限数量的频率成分，有助于识别信号的不同特征频率成分。 2. **微电子机械系统（MEMS）水听器**：MEMS水听器是微型化的声学传感器，常用于水下声学监测，如海洋环境监测、水下通信等。由于其小型化和高灵敏度，数据采集过程中容易受到强噪声干扰。 3. **信噪分离**：在信号处理中，信噪分离是关键步骤，旨在提取信号的有效信息，去除噪声。对于MEMS水听器信号，有效的信噪分离能提高数据的准确性和后续分析的可靠性。 4. **内在模态函数（IMF）**：IMF是EMD和VMD等分解方法中的基本元素，代表信号的局部振荡模式。每个IMF都具有清晰的局部频率和相位特性，便于分析信号的动态行为。 5. **经验模态分解（EMD）与集合经验模态分解（EEMD）**：EMD是VMD的基础，通过迭代过程自适应地将信号分解为多个IMFs。EEMD是EMD的改进版，通过添加白噪声来克服EMD的局部极值点依赖性，提高分解的稳定性。 6. **带通滤波器**：VMD、EMD和EEMD被视作带通滤波器，用于选择特定频率范围内的信号，消除其他不相关的频率成分。VMD在实验中表现出了更好的滤波性能。 7. **性能指数与实验结果**：通过比较不同信噪比下的去噪效果和性能指标，如信噪比提升、均方根误差等，VMD在信噪分离上的优势得到了证实。 8. **应用实例**：论文中提到了VMD在处理北京大学‘Fenji’实验数据时的应用，这表明VMD在实际信号处理中的有效性，尤其是在复杂或有噪声的环境中。这些知识点为理解和应用VMD进行信号处理提供了基础，特别是在面临强噪声干扰的传感器信号处理场景中，VMD可能是一种极具潜力的解决方案。

Signal-Noise Separation of Sensor Signal Based on Variational Mode Decomposition

Bin He, Yanping Bai

School of Science

North University of China

Shanxi, China

e-mail: aihebin19@163.com

Abstract—For the problem of strong noisy interference in the

data acquisition of the micro electro mechanical system

(MEMS) hydrophone, signal-noise separation of the MEMS

hydrophone receiving signal make use of the variational mode

decomposition (VMD), which is a completely non-recursive

variational method. The frequency of the signal is segmented

by the default parameters, then the center frequency and the

intrinsic mode function (IMF) are obtained. Firstly, the

empirical mode decomposition (EMD), the ensemble empirical

mode decomposition (EEMD) and the VMD are regarded as a

bandpass filter, respectively. And the simulation signal of

different signal-to-noise ratio is going to be denoised.

According to the denoising effect and performance index, the

experiment shows that the denoising effect of the VMD is much

better than the previous two methods. Secondly, the VMD is

applied to the Fenji experimental data of the North University

of China in the lake trial. Finally, the results show that the

VMD is one of the best estimation algorithms of MEMS

hydrophone original signal.

Keywords-VMD; sensor signal; EMD; EEMD; signal-noise

separation

I. INTRODUCTION

According to various energy forms of ocean exploration,

acoustic waves have the most distant transmission

performance in the ocean. Therefore, underwater acoustic

waves have become the main carrier of information

transmission in the ocean and underwater acoustic

technology has become the main method about research and

exploration. An acoustic source launches an acoustic wave

which carries the acoustic source. The acoustic waves arrive

in acoustic receivers or hydrophone array through the sea.

Due to the complex ocean environment, the acoustic waves

are mixed with different noise in the process of transmitting

and receiving, which affects the analysis of acoustic source

information, the position and number of acoustic source.

While the researchers are analysing the acoustic source

information, they should denoise the received signal

primarily [1-3]. Nowadays, there are a lot of methods being

used to deal with the noise reduction of underwater acoustic

signals, such as fouier filtering, wavelet transform, adaptive

filtering and empirical mode decomposition et al. Although

these methods have some noise reduction effect, there are

some shortcomings. The known frequency signal processing

has obvious denoising effect in the Fourier filtering method,

but denoising effect of some multi frequency signal is

insufficient. Donoho and other scientists have proposed

wavelet transform method, which has good effect on non-

stationary signal denoising [4]. But the shortcoming of

wavelet transform is the choice of wavelet base and

decomposition level. An empirical mode decomposition

method is proposed by Norden E. Huang et al. Empirical

mode decomposition is based on the characteristics of

adaptive basis selection of the signal decomposition in order

to overcome the deficiency of the basis selection and the

mode mixing of intrinsic mode component [5-6]. To reduce

the mode mixing, Zhaohua Wu and other scientists proposed

ensemble empirical mode decomposition method, which

achieved good results in the experiments. The method is the

way to use the added auxiliary white noise to reduce the

impact of the mode mixing, but the mode mixing of EEMD

is serious in the case of low signal-to-noise ratio [7].

Konstantin Dragomiretskiy and Dominique Zosso have

proposed VMD to solve the influence of mode mixing well

in 2014. Aiming at the problem of strong noise interference

in the data acquisition of MEMS hydrophone, we use the

VMD to separate the signal and noise of the MEMS

hydrophone signal [8-10]. In the simulation experiment, this

method is superior to the signal-noise separation of EMD

and EEMD in the noise reduction effect and the performance

index. And the method is applied to the Fenji experiments of

the North University of China. The results show that the

method has good feasibility and superiority.

The paper is organized as follows. In Section 2, the

principle and procedure of the EMD, EEMD and VMD are

introduced in the experiment. In Section 3, the EMD, EEMD

and VMD are used to separate the signal and noise of the

simulated signal, and the comparison of the denoising effect

and the performance index is also carried out. In Section 4,

we apply the VMD algorithm to denoise the Fenji lake trial

data of the North University of China. In Section 5, the

conclusion is explained.

II. B

ASIC PRINCIPLE AND PROCESS

A. Empirical Mode Decomposition

EMD algorithm was firstly proposed by Norden E.Huang

in 1998, which is a good method to deal with nonlinear and

non-stationary signals [11-12]. The method is based on the

time scale characteristics of complex signal and the noisy

signal is decomposed into a set of data series which retain

the local feature information, namely IMF. Then, the

132

2016 8th IEEE International Conference on Communication S oftw are and N etw ork s

下载后可阅读完整内容，剩余6页未读，立即下载

weixin_38690149

粉丝: 7
资源: 909

变分分解法在传感器信号信噪分离中的应用

行业分类-设备装置-基于变分模态分解和小波奇异分解的轴承故障诊断方法.zip

mvmd_modedecomposition_信号分解_模态分解_变分模态分解

变分模态分解vmd心电信号降噪matlab代码

变分模态分解降噪的原理

变分模态分解后信号重构

变分模态分解后信号重构matlab代码

matlab 逐次变分模态分解

变分模态分解后利用相关系数计算出有效信号,并将有效信号进一步去噪后进行重构

vmd变分模态分解特征提取

变分模态分解中的变分是什么意思

最新资源