-
本文中的数据信号来源于实际分布式光纤安防系统。该系统采用的是马赫-曾德尔干涉结构, 如图 1所示。激光器输出的激光通过3dB耦合器1分为功率相等的两束, 分别进入马赫-曾德尔干涉结构的两臂, 经过一段距离传输后再通过一个3dB耦合器2进行干涉, 干涉光强被光电探测器检测处理后成为待分析的信号。
待分析信号之所以能够携带光纤受到的扰动信息, 是因为上述马赫-曾德尔干涉结构的两臂采用了不同的处理, 参考臂L2使用的是应力不敏感光纤, 而传感臂L1则采用了应力敏感光纤。在应力信号作用下L1中光信号会由于光纤的纵向应变、横向应变以及光弹效应的影响[5-6]而产生一个附加相位Δφ, 由于L2对应力不敏感, 所以在应力作用下光信号相位不会改变。探测器接收到的干涉光强可表示为[7]:
$ I = {I_1} + {I_2} + 2\sqrt {{I_1}{I_2}} \cos \left( {\Delta \varphi + \Delta \psi } \right) $
(1) 式中, I1和I2分别代表L1和L2中的光强, Δψ是由系统制作工艺等因素引入的初始相位差, 系统参量确定后不会再改变。因此, 对信号分析的主要工作落在了如何通过最终的强度信号来反演由应力导致的附加相位变化上。
-
希尔伯特-黄变换(HHT)中希尔伯特变换是基础, 由HUANG提出的EMD方法是核心。通过EMD得到的所有IMF有两个共同的特征:(1)过零点和极值点个数相同或相差为1;(2)通过极大值点拟合得到的上包络曲线和由极小值得到的下包络曲线的局部均值等于0[8-11]。
采用希尔伯特-黄变换处理信号的基本步骤是:首先利用EMD方法对信号进行分解, 得到信号的IMF表示, 然后对每一个IMF进行希尔伯特变换。下面以光纤振动信号为例简要介绍EMD方法的实现流程, 如图 2所示。首先确定出待分析信号x(t)(即本文中的光纤振动信号)的所有极值点, 然后由极大值点和极小值点分别可确定一条拟合曲线(即max和min曲线), 作为x(t)的上下包络线。记上下包络线均值为s1, x(t)与s1的差定义为:
$ {g_{1,1}}\left( t \right) = x\left( t \right) - {s_1} $
(2) 对不同的初始信号, g1, 1(t)可能是一个IMF分量, 也可能不是。若是, g1, 1(t)就作为x(t)的第1阶IMF, 记为I(1)=g1, 1(t); 否则继续对g1, 1(t)进行以上操作。假设经过m次操作后得到了g1, m(t), g1, m(t)是否是一个IMF, 必须有一个可度量的参照。经研究发现, 它可用连续两个处理结果之间的标准差S来评判:
$ S = \sum\limits_{t = 0}^T {\left| {\frac{{{{\left| {{g_{1,\left( {m - 1} \right)}}\left( t \right) - {g_{1,m}}\left( t \right)} \right|}^2}}}{{{{\left[ {{g_{1,\left( {m - 1} \right)}}\left( t \right)} \right]}^2}}}} \right|} $
(3) 大量经验研究表明, S取值在0.2~0.3较为合适, 既可保证IMF的稳定性, 又可以使IMF具有相应的物理意义[12]。
当g1, m(t)满足S条件时, 记I(1)= g1, m(t)。然后将信号x(t)与其第1阶IMF的差值——r1(t)=x(t)-I(1)作为新的数据重复上述分解过程。假设经过n次分解后所得I(n)或者残余量r(t)小于某一阈值或者残余量已经成为单调函数, 则结束整个EMD分解过程。至此, 原始信号x(t)可表示为n阶IMF分量和残余量r(t)的和[8-10], 即:
$ x\left( t \right) = r\left( t \right) + \sum\limits_{k = 1}^n {I\left( k \right)} $
(4) 任一时域信号x(t)的希尔伯特变换y(t)定义为x(t)与$\frac{{\rm{1}}}{{{\rm{ \mathit{ π} }}t}}$的卷积[13-15], 即:
$ y\left( t \right) = x\left( t \right) \otimes \frac{1}{{{\rm{ \mathsf{ π} }}t}} = \frac{1}{{\rm{ \mathsf{ π} }}}p\int {\frac{{x\left( \tau \right)}}{{t - \tau }}{\rm{d}}\tau } $
(5) 式中, p代表柯西主值, 由x(t)及其希尔伯特变换可确定对应的解析信号z(t)为:
$ z\left( t \right) = x\left( t \right) + {\rm{i}}y\left( t \right) = a\left( t \right){{\rm{e}}^{{\rm{i}}\theta \left( t \right)}} $
(6) 式中, a(t)称为时域信号x(t)的瞬时幅值, θ(t)为其瞬时相位, 两者各由下式确定:
$ \left\{ \begin{array}{l} a\left( t \right) = \sqrt {{x^2}\left( t \right) + {y^2}\left( t \right)} \\ \theta \left( t \right) = \arctan \frac{{y\left( t \right)}}{{x\left( t \right)}} \end{array} \right. $
(7) 由相位函数求导可得信号的瞬时频率函数:
$ f\left( t \right) = \frac{1}{{2{\rm{ \mathsf{ π} }}}}\frac{{{\rm{d}}\theta \left( t \right)}}{{{\rm{d}}t}} $
(8) 对由原始信号x(t)通过EMD分解得到的每一阶IMF进行希尔伯特变换, 就可得到x(t)的一组由幅值和相位都随时间变化的组分表示:
$ x\left( t \right) = {\mathop{\rm Re}\nolimits} \sum\limits_{k = 1}^n {{a_k}\left( t \right){{\rm{e}}^{{\rm{i}}{\theta _k}\left( t \right)}}} $
(9) 上式中省去了残余量r(t), Re表示取所得结果的实数部分。更进一步可将相位函数表示为瞬时频率函数对时间的积分, 即:
$ H\left( {f,t} \right) = x\left( t \right) = {\mathop{\rm Re}\nolimits} \sum\limits_{k = 1}^n {{a_k}\left( t \right)\exp \left[ {{\rm{i}}2{\rm{ \mathsf{ π} }}\int {{f_k}\left( t \right){\rm{d}}t} } \right]} $
(10) 式中, H(f, t)是随时间和频率变化的信号幅度。这种由瞬时频率、瞬时幅值来表示原始光纤振动信号的方式叫做光纤振动信号的Hilbert谱表示法, 即光纤振动信号的时频谱。
HHT和CWT用于光纤振动信号分析的对比研究
Comparison study of optical fiber vibration signals using HHT and CWT
-
摘要: 为了提高光纤安防系统的准确预警能力,采用希尔伯特-黄变换法对光纤振动传感器信号产生原理和典型的光纤振动信号进行了分析,获得了传感器不同状态下信号的时频谱,并与通过连续小波变换得到的时频谱进行了对比分析。结果表明,希尔伯特-黄变换在光纤安防系统的时频分析中具有重要价值。Abstract: In order to improve the ability of accurate early warning of optical fiber security system, Hilbert-Huang transform (HHT) technology was adopted to analyze the principle of optical fiber vibration sensor's signal. Time-frequency spectrums of different signals were obtained. The comparison of time-frequency spectrums of optical fiber vibration signals by HHT and continuous wavelet transform (CWT) was analyzed. The results show that HHT has an important value in time-frequency analysis of optical fiber security system.
-
-
[1] ZHANG L N, LI F Ch, PENGB Q, et al. The basic principle of HHT method and application discussion[J]. Journal of East China Institute of Technology, 2013, 36(4):419-423(in Chinese). [2] ZHONG Y M, QIN Sh R, TANG B P. Study on the marginal spectrum in Hilbert-Huang transform[J]. Systems Engineering and Electronics, 2004, 26(9):1323-1326(in Chinese). [3] LIU Sh G, LI Z R, LIU Q. Study of self-mixing interference signal processing based on time-frequency analysis[J].Laser Technology, 2009, 33(6):626-629(in Chinese). [4] JIANG Sh Ch, CAO Y Q, SUI Q M, et al. Study on improved Hilbert-Huang transform-based ultra-narrow-bandwidth laser gas detection noise elimination[J]. Acta Optica Sinica, 2014, 34(1):1-6(in Chinese). [5] CAO L D, LIU X, QIAO F F, et al. Testing of pressure and temperature sensor based on all fiber Mach-Zehnder interferometer[J].Optics & Optoelectronics Technology, 2015, 13(3):34-37(in Chinese). [6] YU X M, LUO G M, ZHU Zh M, et al. The multi target recognition of intrusion signal of perimeter security with distributed fiber-optic sensor[J]. Opto-Electronic Engineering, 2014, 41(1):36-41(in Chinese). [7] FAN L Y, ZHAO R F, JIANG W W, et al. M-Z interferometer using a twin-core fiber for strain sensing[J]. Journal of Optoelectronics·Laser, 2010, 21(10):1488-1491(in Chinese). [8] OMID B, SOHEIL R. Enhanced Hilbert-Huang transform and itsapplication to modal identification[J]. The Structural Design of Tall and Special Buildings, 2014, 23(4):239-253. doi: 10.1002/tal.v23.4 [9] CHANG C C, HSIAO T C, HSU H Y. Frequency range extension of spectral analysis of pulse rate variability based on Hilbert-Huang transform[J]. Medical & Biological Engineering & Computing, 2014, 52(4):343-351. [10] PU L, YE Y T, WU Y F, et al. Laser interferometer signal processing based on empirical mode decomposition[J].Laser Technology, 2011, 35(3):299-301(in Chinese). [11] JI Ch P, ZHAO L N. Research of wave filter method of human sphygmus signal[J].Laser Technology, 2016, 40(1):42-46(in Chinese). [12] ZHANG Y P. HHT analysis and application of blasting vibration signal[M].Beijing:Metallurgical Industry Press, 2008:32-81(in Chinese). [13] HOU R N. Application of Hilbert transformin reconstruction of in-line digital holography[J].Laser Technology, 2013, 37(3):362-364(in Chinese). [14] GAO Y F, GUO H J, SONG X F, et al. Digital holographic phase reconstruction technology based on Hilbert transform[J].Laser Technology, 2015, 39(2):266-269(in Chinese). [15] LI X Y, HUANG Zh H, ZHANG H, et al. Phase analysis method based on Hilbert transform in dynamic speckle pattern interferometry[J]. Journal of Optoelectronics·Laser, 2015, 26(6):1111-1117(in Chinese).