CN103134580A - Signal processing method for distributed optical fiber vibration measurement system based on wavelet analysis - Google Patents

Abstract

The invention discloses a signal processing method for a distributed optical fiber vibration measurement system based on wavelet analysis. The time domain and the frequency domain of signals are simultaneously analyzed by means of wavelet analysis, the signals are multi-resolution, and each moment of the signals can be analyzed, so that original signals are effectively extracted from the signals mixed with strong noise. The signal processing method has the advantages that on one hand, the noise and background signals can be effectively removed, real vibration signals are effectively restored, on the other hand, multi-resolution analysis is performed for the vibration signals, misinformation vibration signals and the real intrusion vibration signals are distinguished by the aid of experience wavelet coefficient threshold values, and accordingly, the misinformation rate and the missing report rate of the system are reduced.

Description

A kind of signal processing method of the distributed optical fiber vibration measuring system based on wavelet analysis

Technical field

The present invention relates to the signal processing technology field of distributed optical fiber vibration measuring system, particularly a kind of signal processing method of the distributed optical fiber vibration measuring system based on wavelet analysis.

Background technology

The distributed optical fiber vibration measuring system is a kind of orientable distributed Vibration-Measuring System of the long distance based on Principle of fiber optical sensing technique, and this type systematic is used as security intrusion system and is widely used in the field of the long distance monitoring of the needs such as national defence boundary line, petroleum delivery pipeline, high-speed railway guard rail.This system synthesis has adopted optical time domain reflection technology (OTDR) and fiber optic interferometric technology, therefore have advantages of simultaneously that OTDR technological orientation precision is high and interference of light technology detection sensitivity is good, therefore be fit to very much be applied to the field of the long-distance distributed monitoring of needs such as national defence boundary line, oil pipeline and high ferro guard rail.

The distributed optical fiber vibration measuring system is based on the full optical fiber interference principle, adopt optical cable as vibration transducer, toward the optical fiber transmission laser of optical cable the inside, and the reflection wave of reception laser, by signal processing method exploring laser light echoing characteristicss such as opto-electronic conversion, data acquisitions, demodulate vibration signal.In actual application, when having the vibration events such as personnel activity or mechanical dredge operation to occur when the sensing optic cable periphery, the vibration signal that vibration event produces can cause optical cable generation strain, causes the phase place of laser in optical cable and polarization state to change.System analyzes the vibration signal that collects, and judges whether to consist of intrusion event, and if it is produce and report to the police, and the position of location intrusion event generation.

The distributed optical fiber vibration measuring system is based on the full optical fiber interference principle, and because interference light intensity is very weak, so the electric signal that produces after opto-electronic conversion (generally adopting APD as photoelectric commutator) is also very faint, and signal almost all is submerged in noise.Because system sensitivity is very high, the slight vibration of optical cable also can cause the wrong report of system, has limited such range of application simultaneously.At present, the distributed optical fiber vibration measuring system mainly is based on time domain approach to the disposal route of signal, and the method adopts the data and curves that does not collect in the same time to subtract each other.If there is no generation of vibration, its interference signal does not change, and the result of subtracting each other should be zero straight line for a numerical value in theory; If generation of vibration is arranged, its interference signal changes, and the result of subtracting each other is exactly the reflection vibration signal curve of (interfere and change) in theory, has judged whether thus generation of vibration.Yet, because noise ratio is large and system sensitivity is very high, the slight vibration of optical cable also can cause the variation of interference signal, be difficult to find a threshold value, can distinguish exactly which class vibration event is wrong report, and which class event is to invade and the vibration event of generation really, namely based on the signal procesing in time domain method, the misreport of system rate is high, and rate of failing to report is also high, has limited the application of system.How reducing misreport of system rate and rate of failing to report, is the difficult problem of current distributed optical fiber vibration measuring system development.

Summary of the invention

In order to solve problems of the prior art, the present invention proposes a kind of signal processing method of the distributed optical fiber vibration measuring system based on wavelet analysis, this method was analyzed signal from time and frequency two aspects, filtering appts noise well, can also carry out multiresolution analysis to vibration signal, distinguish wrong report vibration signal and true invasion vibration signal with experience wavelet coefficient threshold values, thereby reduce misreport of system rate and system's rate of failing to report.

In order to achieve the above object, the present invention adopts a kind of signal processing method of the distributed optical fiber vibration measuring system based on wavelet analysis.

Wavelet analysis is a kind of signal analysis method, and it can carry out time and frequency domain analysis simultaneously to signal, has the characteristics of multiresolution, can extract original signal preferably from the signal that is mixed with very noisy, is described as the microscope of analytic signal.Owing to having multi-resolution characteristics, wavelet analysis can be analyzed constantly to each of signal, and this is very effective for the analysis that has the mutability signal as vibration signal.

Usually the signal processing method based on wavelet analysis comprises wavelet transformation (also being wavelet decomposition) and wavelet reconstruction, the principle of the wavelet transform of often using in digital signal processing as shown in Figure 1:

The principle of wavelet transformation is that original signal s is transformed into wavelet coefficient w, w=[a wherein, d], comprise approximate (approximation) coefficient a and details (detail) coefficient d, that is:

Approximation coefficient a has reflected signal averaging composition (low frequency), as the profile of signal;

Detail coefficients d has reflected signal intensity composition (high frequency), as the sign mutation part;

Like this, original signal s can resolve into the approximate a of small echo and wavelet details d sum, namely

s＝a+d，

Wavelet coefficient w=[a, d] component, multiply by basis function, form wavelet decomposition:

Wavelet approximation coefficients a * basis function g=approximate factorization a, the i.e. outline portion of signal (approximate part);

Wavelet details coefficient d * basis function h=details is decomposed d, the i.e. detail section of signal;

And pairing approximation decomposition part a can also continue to decompose again, obtains the wavelet coefficient that the second layer decomposes; The rest may be inferred, can carry out the K layer to original signal s and decompose, thereby obtain ground floor wavelet details coefficient, second layer wavelet details coefficient, the 3rd layer of wavelet details coefficient ..., K layer wavelet details coefficient and K layer wavelet approximation coefficients.

Wavelet reconstruction (be signal recover) is that the coefficient in above-mentioned little broken conversion is brought into the calculating of wavelet reconstruction formula and obtained, the principle of wavelet reconstruction as shown in Figure 2:

The signal characteristic of distributed optical fiber vibration measuring system is: when vibration did not occur, background signal changed slowly; When generation of vibration was arranged, vibration signal produced sudden change, the conversion jump signal faster that namely superposeed on background signal slowly, but with respect to noise, this jump signal is still low frequency signal; And be superimposed upon noise in signal, be high-frequency signal with respect to vibration signal; That is:

The signal of distributed optical fiber vibration measuring system=ultralow frequency background signal+low frequency jump signal (vibration)+HF noise signal.

Specific implementation based on the signal processing method of the distributed optical fiber vibration measuring system of wavelet analysis is as follows:

The first step: signal is carried out K (K＞1) layer wavelet decomposition, obtain the ground floor wavelet details coefficient of signal after decomposition, second layer wavelet details coefficient, the 3rd layer of wavelet details coefficient ..., K layer wavelet details coefficient and K layer wavelet approximation coefficients;

Second step: ground floor wavelet details coefficient is done threshold values process: the high fdrequency component of this coefficient representation signal, namely noise component, can process layer detail coefficients according to threshold values, for example whole zero clearings of detail coefficients;

The 3rd step: K layer wavelet approximation coefficients done threshold values process: the ultralow frequency component of this coefficient representation signal, namely represent background signal, can process according to threshold values for example whole zero clearings of approximation coefficient to K layer approximation coefficient;

The 4th step: other coefficients except this above-mentioned ground floor wavelet details coefficient and these two coefficients of K layer wavelet approximation coefficients are done threshold values process: the jump signal (part that will detect) that during largely representing generation of these coefficients vibration event, system produces, to this part signal, can be according to the size of coefficient, duration etc. are analyzed, process with threshold value, the filtering invalid signals (as the wrong report etc.);

The 5th step: the wavelet coefficient after above-mentioned the second～the 4th step is processed through threshold values carries out wavelet reconstruction, obtains denoising, goes background and removes the cleaner time domain vibration signal that has that disturbs after vibration (belonging to wrong report vibrates);

The 6th step: the time domain vibration signal after the 5th reconstruct that obtains of step is carried out threshold values process, the size of threshold values is rule of thumb obtained; For example put 1 for the data that surpass threshold values, representative has true intrusion event to occur, and sets to 0 less than or equal to the data of threshold values, and representative does not have intrusion event to occur;

The 7th step: the vibration event that has intrusion event to occur is positioned and produce warning message.

Especially, the wavelet decomposition number of plies is larger, and the processing time is also longer, in practical application, should decompose as required the suitable number of plies.

Beneficial effect of the present invention is: the method is analyzed signal from time domain and frequency domain two aspects, can effectively remove noise and background signal on the one hand, effectively restores true vibration signal; On the other hand, due to vibration signal is carried out multiresolution analysis, distinguish wrong report vibration signal and true invasion vibration signal with experience wavelet coefficient threshold values, thereby reduced misreport of system rate and system's rate of failing to report.

Fig. 1 is the wavelet transform schematic diagram.

Fig. 2 is discrete wavelet reconfiguration principle figure.

Embodiment:

The below is decomposed into example so that signal is carried out 3 layers, the signal of distributed optical fiber vibration measuring system is carried out 3 layers of test findings after decomposition to be shown: after wavelet transformation, ground floor wavelet details coefficient mainly reflects the feature of noise, the 3rd layer of wavelet approximation coefficients mainly reflected the feature of background signal, and remaining second layer wavelet details coefficient and the 3rd layer of wavelet details coefficient have mainly reflected the feature of vibration signal.Choosing of threshold value is also empirical value according to test findings.Concrete performing step is:

The first step is carried out 3 layers of wavelet decomposition to the data S (i.e. digital signal after analog to digital conversion) that collects, and obtains ground floor wavelet details coefficient D ₁, second layer wavelet details coefficient D ₂The 3rd layer of wavelet details coefficient D ₃With the 3rd layer of wavelet approximation coefficients A ₃Wherein

$D_1=[D_1\left(0\right),D_1\left(1\right)...D_1(\frac{N}{2}-1)],$

${\mathrm{<figure-callout\; id="2"\; label="D"\; filenames="HSB00000791058100011.png"\; state="\{\{state\}\}">D</figure-callout>}}_2=[D_2\left(0\right),D_2\left(1\right)...D_2(\frac{N}{4}-1)]$

$D_3=[D_3\left(0\right),D_3\left(1\right)...D_3(\frac{N}{8}-1)]$

$A_3=[A_3\left(0\right),A_3\left(1\right)...A_3(\frac{N}{8}-1)],$

Wherein N is the length of data S.

Second step is to ground floor wavelet details coefficient D ₁Do following processing:

D′ ₁(n)＝0；

Wherein $n=\mathrm{0,1,2}...,\frac{N}{2}-1;$

The 3rd step is to second layer wavelet details coefficient D ₂, do following processing:

${D^{\&prime;}}_2\left(n\right)=\left\{\begin{array}{c}{D}_2\left(n\right),\left(\right|D_2\left(n\right)|\&GreaterEqual;\frac{{\mathrm{\&delta;}}_1}{n+1})\\ 0,\left(\right|D_2\left(n\right)|<\frac{{\mathrm{\&delta;}}_1}{n+1})\end{array}\right.,$

Wherein $n=\mathrm{0,1,2}...,\frac{N}{4}-1;$

Further, δ rule of thumb ₁Value is 0.001139;

The 4th step is to the 3rd layer of wavelet details coefficient D ₃, do following processing:

${D^{\&prime;}}_3\left(n\right)=\left\{\begin{array}{c}{D}_3\left(n\right),\left(\right|D_3\left(n\right)|\&GreaterEqual;\frac{{\mathrm{\&delta;}}_2}{n+1})\\ 0,\left(\right|D_3\left(n\right)|<\frac{{\mathrm{\&delta;}}_2}{n+1})\end{array}\right.,$

Wherein $n=\mathrm{0,1,2}...,\frac{N}{8}-1,$

Further, δ rule of thumb ₂Value is 0.07625

The 5th step is to the 3rd layer of wavelet approximation coefficients A ₃Do following processing:

A′ ₃(n)＝0；

Wherein $n=\mathrm{0,1,2}...,\frac{N}{8}-1;$

The 6th goes on foot, and above-mentioned second step～5th is gone on foot the wavelet coefficient D ' of the processing that obtains ₁, D ' ₂, D ' ₃, A ' ₃Bring the inverse wavelet transform formula into and carry out wavelet reconstruction (be signal recover), obtain denoising, go background and remove the cleaner time domain vibration signal S ' that has that disturbs after vibration (belonging to the wrong report vibration).

In the 7th step, the time-domain signal S ' that the 6th step was obtained carries out the threshold values processing:

$Y\left(n\right)=\left\{\begin{array}{c}1,\left(\right|S^{\&prime;}\left(n\right)|\&GreaterEqual;\frac{\mathrm{\&gamma;}}{n+1})\\ 0,\left(\right|S^{\&prime;}\left(n\right)|<\frac{\mathrm{\&gamma;}}{n+1})\end{array}\right.;$

Wherein, n=0,1,2,3...N-1

Further, rule of thumb the γ value is 2.5213

The 8th step, Y is scanned, when Y (n) is 1, think that the position at optical cable corresponding with n place invasion vibration event (not belonging to wrong report) has occured occured, system produces warning, and orients the event occurrence positions.Further, if find that a plurality of data are at 1 o'clock, thinking has a plurality of positions that true invasion vibration event occurs.

Claims (1)

1. the signal processing method based on the distributed optical fiber vibration measuring system of wavelet analysis, is characterized in that, comprises the steps:

The first step: signal is carried out K (K＞1) layer wavelet decomposition, obtain the ground floor wavelet details coefficient of signal after decomposition, second layer wavelet details coefficient, the 3rd layer of wavelet details coefficient ..., K layer wavelet details coefficient and K layer wavelet approximation coefficients;

Second step: ground floor wavelet details coefficient is done threshold values process: the high fdrequency component of this coefficient representation signal, namely noise component, can process layer detail coefficients according to threshold values;

The 3rd step: K layer wavelet approximation coefficients done threshold values process: the ultralow frequency component of this coefficient representation signal, namely represent background signal, can process K layer approximation coefficient according to threshold values;

The 4th step: other coefficients except this above-mentioned ground floor wavelet details coefficient and these two coefficients of K layer wavelet approximation coefficients are done threshold values process: the jump signal (part that will detect) that during largely representing generation of these coefficients vibration event, system produces, to this part signal, can be according to the size of coefficient, duration etc. are analyzed, process the filtering invalid signals with threshold value;

The 5th step: the wavelet coefficient after above-mentioned the second～the 4th step is processed through threshold values carries out wavelet reconstruction, obtains denoising, goes background and removes the cleaner time domain vibration signal that has that disturbs after vibrating;

The 6th step: the time domain vibration signal after the 5th reconstruct that obtains of step is carried out threshold values process, the size of threshold values is rule of thumb obtained;

The 7th step: the vibration event that has intrusion event to occur is positioned and produce warning message.

Images