Advanced Search

ISSN1001-3806 CN51-1125/TN Map

Volume 46 Issue 1
Jan.  2022
Article Contents
Turn off MathJax

Citation:

Laser Doppler vibration signal processing based on wavelet denoising

  • Corresponding author: GAN Xuehui, xuehuig@dhu.edu.cn
  • Received Date: 2021-01-14
    Accepted Date: 2021-02-24
  • In order to reduce the interference of noise in laser Doppler vibration signal, a laser Doppler vibration signal processing method based on improved wavelet de-noising was proposed. The scale was introduced into the threshold function, and a new evaluation index was established to select the optimal decomposition level. The improved algorithm was used to process the vibration signal. The improved algorithm and the original algorithm were adopted for processing the vibration signal. The simulation analysis and experimental verification were then carried out, and the vibration data before and after processing were obtained. The results indicate that the signal-to-noise ratio of simulated signals processed by the improved algorithm is 19.4% higher than that of the original soft and hard threshold algorithm. The measured tuning fork vibration frequency is 515Hz, which is consistent with the actual tuning fork frequency. This result is helpful to reduce the influence of noise in the laser Doppler vibration signal and obtain the vibration state.
  • 加载中
  • [1]

    QIU S, LIU T, REN Y, et al. Detection of spinning objects at oblique light incidence using the optical rotational Doppler effect [J]. Optics Express, 2019, 27(17): 81-92.
    [2]

    LI W. Research and application of velocity measurement system based on laser Doppler effect[D]. Nanjing: Nanjing University of Science and Technology, 2018: 10-61(in Chinese).
    [3]

    PIERRE M, TRISTAN G, ARTHUR L, et al. The robotized laser doppler vibrometer: On the use of an industrial robot arm to perform 3-D full-field velocity measurements[J]. Optics and Lasers in Engineering, 2021, 137: 106363. doi: 10.1016/j.optlaseng.2020.106363
    [4]

    HASHEMINEJAD N, VUYE C, MARGARITIS A, et al. Identification of the viscoelastic properties of an asphalt mixture using a scanning laser Doppler vibrometer[J]. Materials and Structures, 2020, 53: 131. doi: 10.1617/s11527-020-01567-9
    [5]

    XU Y F, CHEN D M, ZHU W D. Modal parameter estimation using free response measured by a continuously scanning laser Doppler vibrometer system with application to structural damage identification[J]. Journal of Sound and Vibration, 2020, 485: 115536. doi: 10.1016/j.jsv.2020.115536
    [6]

    ABBAS S H, JANG J K, KIM D H, et al. Underwater vibration ana-lysis method for rotating propeller blades using laser Doppler vibrometer[J]. Optics and Lasers in Engineering, 2020, 132: 106133. doi: 10.1016/j.optlaseng.2020.106133
    [7]

    WANG T, SHEN Y H, YAO J Q. Research on laser radar echo signal denoising based on wavelet threshold method[J]. Laser Technology, 2019, 43(1): 63-68(in Chinese).
    [8]

    LI Sh Y. Improved wavelet threshold denoising method and its simulation using MATLAB[J]. Noise and Vibration Control, 2010, 30(2): 121-124(in Chinese).
    [9]

    LI J, ZENG L D, YUAN Sh Zh, et al. Research on the application of laser Doppler velocimeter in aircraft navigation[J]. Applied Laser, 2017, 37(6): 870-880(in Chinese).
    [10]

    BAI T, WU J, LI M L, et al. Application of DRNN in voice mea-surement system of laser Doppler vibrometer[J]. Laser Technology, 2019, 43(1): 109-114(in Chinese).
    [11]

    WANG H. Research on data processing and application of laser Doppler vibration measurement[D]. Hangzhou: Zhejiang University, 2018: 9-69(in Chinese).
    [12]

    XIANG B P, ZHOU J, NI L, et al. Research on improved wavelet packet threshold denoising algorithm based on sample entropy[J]. Journal of Vibration, Measurement & Diagnosis, 2019, 39(2): 410-415(in Chinese).
    [13]

    LIU M S, SUN Zh Y. Application of improved wavelet denoising method in low-frequency oscillation analysis of power system[J]. Journal of Physics Conference Series, 2020, 1633(1): 95-103. doi: 10.1088/1742-6596/1633/1/012115/pdf
    [14]

    HAO J J, LIU Y G, LIAO G, et al. A signal de-noising with improved wavelet threshold function[J]. Journal of Chongqing University of Technology(Natural Science Edition), 2019, 33(4): 93-97(in Chinese).
    [15]

    ZHAO H B, ZHANG D, YANG J K, et al. Application of wavelet layered method for laser Doppler velocimetry signal[J]. Laser Technology, 2019, 43(1): 103-108(in Chinese).
    [16] CAO J J, HU L L, ZHAO R. Improved threshold de-noising method of fiber bragg grating sensor signal based on wavelet transform[J]. Chinese Journal of Sensors and Actuators, 2015, 28(4): 521-525(in Chinese).

    [17]

    WU G W, WANG Ch M, BAO J D, et al. A wavelet threshold de-noising algorithm based on adaptive threshold function[J]. Journal of Electronics & Information Technology, 2014, 36(6): 1340-1347(in Chinese).
    [18]

    AMIR A S, SEYYED M S. Decision tree-based method for optimum decomposition level determination in wavelet transform for noise reduction of partial discharge signals[J]. IET Science, Measurement & Technology, 2020, 14(1): 9-16.
    [19]

    ZHU J J, ZHANG Zh T, KUANG C L, et al. A reliable evaluation indicator of wavelet de-noising[J]. Geomatics and Information Science of Wuhan University, 2015, 40(5): 688-694(in Chinese).
    [20]

    TAO K, ZHU J J. A hybrid indicator for determining the best decomposition scale of wavelet denoising[J]. Acta Geodaetica et Cartographica Sinica, 2012, 41(5): 749-755(in Chinese).
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(11) / Tables(2)

Article views(6761) PDF downloads(25) Cited by()

Proportional views

Laser Doppler vibration signal processing based on wavelet denoising

    Corresponding author: GAN Xuehui, xuehuig@dhu.edu.cn
  • 1. Shanghai Collaborative Innovation Center for High Performance Fiber Composites, Shanghai 201620, China
  • 2. College of Mechanical Engineering, Donghua University, Shanghai 201620, China

Abstract: In order to reduce the interference of noise in laser Doppler vibration signal, a laser Doppler vibration signal processing method based on improved wavelet de-noising was proposed. The scale was introduced into the threshold function, and a new evaluation index was established to select the optimal decomposition level. The improved algorithm was used to process the vibration signal. The improved algorithm and the original algorithm were adopted for processing the vibration signal. The simulation analysis and experimental verification were then carried out, and the vibration data before and after processing were obtained. The results indicate that the signal-to-noise ratio of simulated signals processed by the improved algorithm is 19.4% higher than that of the original soft and hard threshold algorithm. The measured tuning fork vibration frequency is 515Hz, which is consistent with the actual tuning fork frequency. This result is helpful to reduce the influence of noise in the laser Doppler vibration signal and obtain the vibration state.

引言
  • 激光多普勒测振技术是基于多普勒效应通过解析被测物体的振动所产生的多普勒频移量,来得到物体的振动状态[1-2]。由于其响应速度快、检测精度高,在运动状态检测、模态分析、结构损伤识别等领域应用越来越广泛[3-6]。然而由于外界环境的干扰,采集的激光干涉信号不仅携带被测物体的振动信号,而且携带更多的是外界环境噪声;甚至在大多数情况下, 振动信号会被噪声完全掩盖。因此, 如何实现强噪声环境下提取目标振动信号成为首要任务和难点[7]

    小波变换能同时对信号时域和频域进行多尺度细化分析,较好地区分信号中的突变部分和噪声,进而对信号进行去噪[8]。其中小波阈值去噪法被广泛应用于激光多普勒振动信号处理[9-11], 但是由于其去噪会产生固定偏差, 许多学者也对小波阈值去噪算法做了改进。XIANG等人[12]将样本熵表征噪声强度引入阈值函数,使阈值函数更适应小波系数的分布,但是用表征信号强度的熵表征噪声强度会造成较大偏差。LIU等人[13]构造了随尺度变化的局部阈值,同时引入正弦函数改进阈值函数,解决了软硬阈值函数存在的不足,但是仅将信噪比作为指标会导致无法找到最优分解层数。HAO等人[14]将均方根误差和信噪比作为评价指标选择最优分解层数,解决了单一评价指标存在的不足,但是单纯用现有指标线性叠加仍会产生偏差。上述学者在阈值函数和分解层数评价指标方面对算法做了改进,去噪效果有一定提升,但仍存在问题:难以寻找最优分解层数、忽略有用信号小波系数随尺度的分布。

    本文中将尺度引入阈值函数模型、选用局部阈值、将均方根误差变化量、平滑度变化量处理后线性叠加作为新的分解层数评价指标来改进小波阈值去噪算法,将改进后算法用于激光多普勒测振信号处理,并且通过仿真和音叉去噪实验验证了改进算法的有效性。

1.   小波阈值去噪原理及模型改进
  • 小波阈值去噪原理是通过估计噪声强度生成阈值,利用阈值函数处理小波系数去除部分噪声相关的分量,从而提高有用信号的比重。主要分为以下3个步骤:(1)通过小波变换将含噪信号分解成若干层的小波系数; (2)利用阈值函数处理各层小波系数去除与噪声有关的分量; (3)通过小波逆变换重构信号[15]

  • 阈值是影响去噪效果的一个重要因素,低尺度的小波系数是大于高尺度的,包含的噪声系数也越大,对不同尺度利用相同阈值处理,会影响去噪效果。因此本文中使用局部阈值。局部阈值λj的公式为[16]

    式中,σj是第j层小波系数的噪声标准差,wj是第j层小波系数数列,λ是根据阈值规则计算得出的阈值。(1)式可以满足小尺度下取较高阈值,避免全局阈值产生恒定偏差。median为中值函数,表示求数列中值。

    小波阈值去噪关键在于阈值函数的选取[17],针对软阈值和硬阈值的缺点,提出了基于局部阈值的改进阈值函数。改进后阈值函数g(x) 形式为:

    式中,x是小波系数数列中的数。

    改进后的阈值函数在x=λj处连续,克服了硬阈值的缺点。

    x>0时:

    x < 0时:

    由(4)式~(9)式可知, g(x)=x是阈值函数的一条渐近线,随着x变大处理后的小波系数向原小波系数不断逼近,同时随着尺度增加,逼近程度不断增加,符合小波系数的组成随尺度的变化情况,解决了软阈值函数产生较大偏差的问题。

    采集的激光多普勒振动信号中噪声并不是稳定的,某一时刻外界噪声可能急剧变大,每次采样都需要寻找最优分解层数,一般来说分解层数为1~8[18]。参考现有融合评价指标,作者提出了自适应分解层数[19-20],选取均方误差和平滑度融合得到评价指标T,来选择最优分解层数。具体计算方法如下。

    由于有用信号未知,无法通过计算有用信号比重来评价去噪后信号质量,所以借助变化率特征构建2个描述指标:均方根误差(root mean square error, RMSE)变化量Δe、平滑度变化量Δr[20]。具体表达如下:

    式中,e(m)表示均方根误差;r(m)表示平滑度,Δe(m)表示m尺度下均方根误差变化量;Δr(m)表示m尺度下平滑度变化量。

    将上述两个指标归一化处理:

    计算两个指标的权重系数:

    式中,$ \overline{\Delta e^{\prime}}$表示归一化均方根误差变化量的均值,$ \overline{\Delta r^{\prime}} $表示归一化平滑度变化量的均值,We表示归一化均方根误差变化量的权重系数,Wr表示归一化平滑度变化量的权重系数。

    最后通过线性组合叠加得到评价指标T

2.   去噪仿真实验
  • 为验证改进小波阈值去噪算法的效果,通过MATLAB创建信噪比为5dB,有用信号为S=3sin(t+π/4)+4cos(t+π/3)的含噪信号, 其中S表示生成的模拟信号,t是时间。利用自己改进的算法和现有软阈值、硬阈值、参考文献[14]中改进算法对生成的模拟信号去噪。实验中采用db4基小波,无偏阈值规则,先利用本文中提出的改进算法去噪可以得到一个信噪比,在保持阈值规则、小波基不变情况下采用软阈值、硬阈值和参考文献[14]中去噪算法,比较去噪后信噪比的大小。图 1为含噪信号,图 2为参考文献[14]中算法去噪后信号,图 3为软阈值法去噪后信号,图 4为硬阈值法去噪后信号,图 5为本文中提出的改进算法去噪后信号。

    Figure 1.  Noisy signal

    Figure 2.  Signal processed by validate algorithm

    Figure 3.  Signal processed by soft threshold algorithm

    Figure 4.  Signal processed by hard threshold algorithm

    Figure 5.  Signal processed by the improved algorithm

    从上述图中可以看出, 由于模拟信号复杂程度较低,所以采用上述4种方法去噪效果都较好,不存在突出尖峰;但是软硬阈值法去噪后峰值相差较大,并且曲线不光滑;参考文献[14]中算法和软硬阈值法相比, 去噪效果有了一定改进,但是仍然存在局部的不光滑曲线; 本文中提出的改进去噪算法峰值平齐,还接近实际峰值,同时去噪后曲线平滑程度高。为进一步突出表现去噪效果,分别选择db4, sym6, coif3小波作为小波基比较去噪后信噪比(signal-to-noise ratio, SNR)。表 1中为各种算法去噪后信噪比,表 2中为改进算法处理仿真信号信噪比比原有算法处理效果提升百分比。

    the improved algorithm/dB hard threshold algorithm/dB soft threshold algorithm/dB validate algorithm/dB
    db4 25.28 16.96 16.97 22.96
    sym6 26.44 22.14 22.14 23.94
    coif3 25.05 19.75 20.14 23.26

    Table 1.  SNR of each algorithm after denoising

    compare with the original threshold algorithm/% compare with the validate algorithm/%
    db4 50.0 10.1
    sym6 19.4 10.4
    coif3 24.4 7.7

    Table 2.  Improvement percentage of SNR

    表 1表 2中的数据可以看出:采用改进的算法去噪效果都比原有方法好,用不同小波基时改进效果也会发生改变,改进的算法比原有算法信噪比至少提升19.4%,比参考文献[14]中算法信噪比仍提升约7.7%,表明改进的算法具有更好的去噪效果。

3.   信号去噪实验
  • 实际采集的激光多普勒振动信号较复杂,仿真实验并不能体现对实际信号的去噪效果,因此需要对实际的采集信号处理验证改进小波去噪算法的效果。本文中基于多普勒效应搭建如图 6所示的激光多普勒测振实验平台。光源选择波长为632.8nm、功率为5mW的He-Ne激光器,光电探测器选择卓立汉光公司的DSi300硅光电探测器,声光频移器选择中国电子科技SGT40-633-2PA一体化声光调制器,载波频率为40MHz。以频率为512Hz的音叉为振动物体,调整光路得到干涉牛顿环后敲击音叉,通过光电探测器和采集卡传输振动信号, 基于LabVIEW和MATLAB联合编写振动信号处理程序。

    Figure 6.  Laser Doppler experimental device

    通过搭建的实验平台和软件系统,设置采样频率为5000Hz,采样数为2000,分别利用本文中提出的改进去噪算法、参考文献[14]中算法以及现有的软硬阈值函数去噪算法对信号处理,观察时域和频域信号比较去噪效果。图 7是测量得到的原始信号(包括时域和频域信号),图 8是软阈值函数算法处理后频域信号,图 9是硬阈值函数算法处理后频域信号,图 10是参考文献[14]中算法处理后频域信号,图 11是改进去噪算法处理后频域信号。

    Figure 7.  Raw signal

    Figure 8.  Frequency-domain signal processed by soft threshold algorithm

    Figure 9.  Frequency-domain signal processed by hard threshold algorithm

    Figure 10.  Frequency-domain signal processed by validate algorithm

    Figure 11.  Frequency-domain signal processed by improved algorithm

  • 图 7所示, 采集到的音叉振动信号包含许多噪声干扰难以看出音叉的振动频率,同时具有很强的低频外部噪声,所以在利用去噪算法处理前采用一个低通滤波器滤除低频噪声,分别采用不同的去噪处理算法处理采集的振动信号。考虑到在音叉上粘贴了一个增强反光效率的小镜片,所以采集到的频率会与音叉固有频率(512Hz)有一定偏差。由于采样点数过多,曲线较密,时域信号较难看出去噪效果,所以在频域图里观察频率组成来检验去噪效果,由图 8图 9可以看出, 现有的软硬阈值函数去噪算法由于固定的分解层数限制,保留的噪声仍然过多,无法实现振动频率提取。由图 10可以看出, 参考文献[14]中去噪算法相比现有算法有了改进,可以提取到振动频率,但是低频和高频噪声仍然有较高幅值。观察图 11可以发现, 本文中提出的改进去噪算法低频和高频段振动幅值较小,因此该改进算法去噪效果更优,对高低频噪声都有较好抑制作用。

4.   结论
  • 利用改进去噪算法处理信噪比为5dB的仿真信号,信噪比变为25dB,改进的算法比原有算法信噪比大约提升19.4%,提升了信号的去噪效果。随后利用改进算法处理频率为512Hz的音叉振动信号,可以得到音叉振动频率为515Hz,由于在音叉上粘贴一个增强反光效率的小镜片会造成固有频率发生细小改变,所以, 该算法在保留有用信号基础上对低频和高频噪声都有较好滤除效果。结合仿真和实验结果分析,去噪后信号信噪比提高,噪声得到去除,有用信号得以保留,验证了利用改进算法处理激光多普勒振动信号,获取振动状态是可行的。

Reference (20)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return