CN106301575A - The sorting technique of a kind of fiber-optic vibration signal and device and optical fiber sensing system - Google Patents

Abstract

The invention discloses sorting technique, device and the optical fiber sensing system of a kind of fiber-optic vibration signal.Wherein, the method includes: processing terminal obtains fiber-optic vibration signal, and wherein, described fiber-optic vibration signal is converted to by the optical signal of fiber reflection；Described fiber-optic vibration signal is divided into multiple subsignal；Utilize SVM algorithm, the plurality of subsignal is divided at least one preset signals apoplexy due to endogenous wind, wherein, the feature of the fiber-optic vibration signal that preset signals class described in produces corresponding to a vibration source.By the way, the present invention is capable of fiber-optic vibration signal correspondence vibration source and classifies, and improves and identifies accuracy.

Description

Optical fiber vibration signal classification method and device and optical fiber sensing system

Technical Field

The invention relates to the field of optical fiber communication, in particular to a method and a device for classifying optical fiber vibration signals and an optical fiber sensing system.

Background

With the development of optical fiber processing technology, optical fibers have been used as sensors in various detection fields for measuring ambient environmental conditions, such as whether an object passes through, and what the object is. When the optical fiber sensor is affected by external interference, partial characteristics of transmission light in the optical fiber are changed, the optical fiber sensing system receives optical signals of the optical fiber, performs photoelectric conversion, analyzes characteristics of the converted electrical signals to judge the change of the optical characteristics, and further determines the environmental conditions of the corresponding position of the optical fiber, such as vehicle intrusion, and the like, so that the monitoring of the environment can be realized. .

The environment around the optical fiber is complex, and there are usually two vibration sources occurring simultaneously. At this time, the collected optical fiber vibration signal includes a mixed signal of the vibration signals of the two vibration sources, and if the vibration source identification is directly performed on the mixed signal, since the mixed signal includes the vibration signal characteristics of the two vibration sources, an identification error is easily caused, so that the mixed signal needs to be classified before the identification.

Disclosure of Invention

The invention mainly solves the technical problem of providing a method and a device for classifying optical fiber vibration signals and an optical fiber sensing system, which can realize classification of the optical fiber vibration signals corresponding to vibration sources and improve identification accuracy.

In order to solve the technical problems, the invention adopts a technical scheme that: the classification method of the optical fiber vibration signal comprises the following steps: the method comprises the steps that a processing terminal obtains an optical fiber vibration signal, wherein the optical fiber vibration signal is obtained by converting an optical signal reflected by an optical fiber; dividing the optical fiber vibration signal into a plurality of sub-signals; and dividing the sub-signals into at least one preset signal class by utilizing a Support Vector Machine (SVM) algorithm, wherein one preset signal class corresponds to the characteristics of the optical fiber vibration signal generated by a vibration source.

Wherein the step of dividing the plurality of sub-signals into at least one preset signal class using a Support Vector Machine (SVM) algorithm comprises: when the number of the locally stored preset signal classes exceeds 2, selecting two preset signal classes from the locally stored preset signal classes, forming a two-class classifier by utilizing an SVM algorithm, and dividing each sub-signal into one of the two selected preset signal classes by utilizing the two-class classifier; reselecting the divided preset signal classes and the unselected one of the preset signal classes, forming a two-class classifier by using an SVM algorithm, dividing each sub-signal into one of the two reselected preset signal classes by using the newly formed two-class classifier, and repeating the step until each preset signal class in the preset signal classes is selected; and respectively taking the preset signal classes into which the plurality of sub-signals are finally divided as the preset signal classes to which the plurality of sub-signals belong.

And the selection sequence of the preset signal classes is arranged from large to small according to the difference degree between the characteristic data in the preset signal classes.

Wherein after the step of dividing the plurality of sub-signals into at least one preset signal class, the method further comprises: matching sub-signals belonging to the same preset signal class with a preset vibration source model; and if the matching is successful, determining the vibration source type of the sub-signals belonging to the same preset signal class as the vibration source type corresponding to the preset vibration model.

Wherein the step of acquiring the optical fiber vibration signal comprises: acquiring an optical fiber vibration signal generated in a certain position of an optical fiber within a period of time; the step of dividing the fiber vibration signal into a plurality of sub-signals comprises: and sampling the optical fiber vibration signal for multiple times to obtain a plurality of sub-signals.

Wherein, the step of sampling the optical fiber vibration signal for multiple times to obtain a plurality of sub-signals comprises: and acquiring the optical fiber vibration signals by using a high-speed data acquisition card FPGA to obtain the plurality of sub signals.

Wherein after the step of splitting the fiber vibration signal into a plurality of sub-signals, the method further comprises: carrying out singular value decomposition on a matrix formed by the plurality of sub-signals by using a Singular Value Decomposition (SVD) algorithm to obtain a plurality of sub-matrices; acquiring the submatrix with the maximum singular value, and extracting each row of elements of the submatrix with the maximum singular value to assemble a one-dimensional signal; and performing wavelet de-noising on the one-dimensional signals to obtain a plurality of sub-signals generated at a certain position of the optical fiber.

In order to solve the technical problem, the invention adopts another technical scheme that: provided is an optical fiber vibration signal classification device including: the optical fiber vibration detection device comprises an acquisition module, a detection module and a control module, wherein the acquisition module is used for acquiring an optical fiber vibration signal, and the optical fiber vibration signal is obtained by converting an optical signal reflected by an optical fiber; the dividing module is used for dividing the optical fiber vibration signal into a plurality of sub-signals; the classification module is used for classifying the sub-signals into at least one preset signal class by utilizing a Support Vector Machine (SVM) algorithm, wherein one preset signal class corresponds to the characteristics of the optical fiber vibration signal generated by a vibration source.

Wherein the classification module comprises: the selection unit is used for selecting two preset signal classes from the plurality of locally stored preset signal classes when the number of the locally stored preset signal classes exceeds 2, forming a two-class classifier by utilizing an SVM algorithm, and dividing each sub-signal into one of the two selected preset signal classes by utilizing the two-class classifier; and the classifying unit is used for reselecting the preset signal class obtained by the division and an unselected preset signal class in the preset signal classes, forming a two-class classifier by using an SVM algorithm, dividing each sub-signal into one of the two reselected preset signal classes by using the newly formed two-class classifier, repeating the step until each preset signal class in the preset signal classes is selected, and taking the preset signal class obtained by the last division of the sub-signals as the preset signal class to which the sub-signals belong.

In order to solve the technical problems, the invention adopts another technical scheme that: the optical fiber sensing system comprises an optical fiber sensor and a processing terminal; the optical fiber sensor is used for sending a first optical signal at one end and receiving a second optical signal obtained by reflecting the first optical signal from the one end; the processing terminal is used for classifying the optical fiber vibration signals corresponding to the second optical signals, and the processing terminal comprises the optical fiber vibration signal classification device so as to classify the optical fiber vibration signals.

According to the scheme, the processing terminal classifies a plurality of sub-signals of the optical fiber vibration signal by adopting the SVM algorithm, and the SVM algorithm can classify linear indivisible data, so that the sub-signals belonging to different vibration sources in the optical fiber vibration signal can be divided, the effective classification of the optical fiber vibration signal corresponding to the vibration sources is realized, and the accuracy of vibration source identification of the follow-up sub-signals can be improved after the sub-signals of the different vibration sources are divided.

Drawings

FIG. 1 is a flow chart of an embodiment of a method for classifying optical fiber vibration signals according to the present invention;

FIG. 2 is a schematic structural diagram of an embodiment of an optical fiber sensing system of the present invention;

FIG. 3 is a flowchart of the sub-steps involved in step S13 shown in FIG. 1;

FIG. 4 is a schematic flow chart of constructing a classifier according to an embodiment of the method for classifying optical fiber vibration signals of the present invention;

FIG. 5 is a flow chart of another embodiment of the method for classifying optical fiber vibration signals according to the present invention;

FIG. 6 is a schematic structural diagram of an embodiment of an optical fiber vibration signal classification apparatus according to the present invention;

fig. 7 is a schematic structural diagram of another embodiment of the optical fiber vibration signal classification apparatus according to the present invention.

Detailed Description

In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular system structures, interfaces, techniques, etc. in order to provide a thorough understanding of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

Referring to fig. 1, a flowchart of an embodiment of a method for classifying optical fiber vibration signals according to the present invention includes:

s11: and the processing terminal acquires the optical fiber vibration signal.

The optical fiber vibration signal is an optical fiber electrical signal which is obtained by converting an optical signal reflected by an optical fiber. For example, the optical fiber vibration signal is an electrical signal converted from an optical signal reflected by a certain position of the optical fiber within a period of time.

Fig. 2 shows an optical fiber sensing system, which can detect multiple concurrent vibration sources simultaneously by detecting the reflected light interference intensity variation caused by the phase variation of the backscattered signal in an optical pulse modulation manner, so as to realize early warning and positioning of the vibration sources, for example, in conjunction with fig. 2. The optical fiber sensing system comprises an optical fiber sensor 21, an optical system 23, a photoelectric conversion circuit 24 and a processing terminal 22 which are connected in sequence.

The fiber optic sensor 21 is disposed in an environment to be monitored, such as the ground, to monitor the environmental condition. The optical fiber sensor 21 can use a free fiber core in a common communication optical cable as a sensing unit to perform distributed multi-point vibration measurement. The basic principle is that when external vibration acts on the communication optical cable, the fiber core in the optical cable is deformed, so that the length and the refractive index of the fiber core are changed, and the phase of light in the optical cable is changed. When light is transmitted in the optical cable, Rayleigh scattered light is continuously transmitted backwards due to the action of photons and fiber core lattices. When vibration occurs outside, the phase of the back rayleigh scattering light changes, and the signal light carrying the outside vibration information is processed by the optical system 23 to convert the weak phase change into light intensity change, and then is subjected to photoelectric conversion by the photoelectric conversion circuit 24 and corresponding signal processing, and then enters the human processing terminal 22 for data analysis. The processing terminal 22 judges the occurrence of a vibration event based on the result of the analysis, and confirms the vibration location.

Specifically, the optical fiber sensor 21 periodically emits a first optical signal from one end, the first optical signal may be a pulse signal, such as laser with a pulse width of 10ns, the first optical signal is subjected to rayleigh scattering at various positions in the optical fiber cable to form a second optical signal, and the second optical signal is reflected back to one end of the optical fiber sensor 21. The optical fiber sensor 21 outputs the second optical signal from the one end. The optical system 23 samples the second optical signal to obtain a plurality of optical signals corresponding to different optical fiber positions. The sampling interval may collect optical signals transmitted by the optical fiber at a set distance, for example, the first sampling optical signal corresponds to an optical signal reflected at a position 1 meter away from one end of the optical fiber, the second sampling optical signal corresponds to an optical signal reflected at a position 2 meters away from one end of the optical fiber, and so on.

The optical system 23 converts the sampled optical signals into corresponding electrical signals through the photoelectric conversion circuit 24 for signal processing. Here, the analog signal may be obtained by conversion by a general photoelectric conversion circuit 24 such as APD, and the analog signal may be converted into a digital signal by an analog-to-digital converter and transmitted to the processing terminal 22. The converted optical fiber digital signal can be directly used as the optical fiber vibration signal, or the optical fiber digital signal with the fluctuation exceeding a set degree can be used as the optical fiber vibration signal.

In other embodiments, at least one of the partial steps of the optical system, the photoelectric conversion step and the analog-to-digital conversion step may be executed by the processing terminal 22, for example, the processing terminal 22 receives a plurality of sampled optical signals collected from the second optical signal reflected by the optical fiber sensor 21, converts the sampled optical signals into a plurality of sampled electrical signals, and performs analog-to-digital conversion on the sampled electrical signals to obtain the optical fiber vibration signals; or after the analog-to-digital conversion is carried out on the sampling electric signal, the stable characteristics of the sampling electric signal in a time domain and a space domain are analyzed; detecting the sampled electric signal by adopting different detection modes corresponding to the stable characteristics of different time domains and space domains of the sampled electric signal; and if the detection is passed, confirming the sampling electric signal as an optical fiber vibration signal.

S12: and dividing the optical fiber vibration signal into a plurality of sub-signals.

For example, the optical fiber vibration signal is generated at a certain position of the optical fiber for a period of time, and the processing terminal 22 performs multiple sampling on the optical fiber vibration signal to obtain multiple sub-signals. The sampling interval may be adjusted according to actual conditions, generally, the sampling interval is between 0 μ s and 100 μ s, such as 40 μ s, 10 μ s, 100 μ s, and the like, and in an application, the acquisition interval is 0, that is, the processing terminal acquires m sub-signals continuously, where m is an integer greater than 1.

In this embodiment, the Processing terminal 22 may further include a high-speed data acquisition card (FPGA) module and a Digital Signal Processing (DSP) module, where the FPGA module is configured to acquire the optical fiber vibration Signal to obtain a plurality of sub signals. The FPGA module buffers the collected optical fiber sampling signal in an FIFO buffer in the FPGA module, a half-full signal line of the FIFO is connected with the DSP module, and when the FIFO is half full, EDMA transaction of the DSP is triggered to transfer the related data of the optical fiber sampling signal from the FIFO to a memory of the DSP, such as SDRAM. And when the data length in the memory reaches a system set value, processing the optical fiber sampling signal in the memory, such as signal classification, vibration source identification and the like. Furthermore, the following steps of singular value decomposition and noise cancellation may be performed before storing the signal into the memory of the DSP, or the collected optical fiber sampling signal may be directly stored into the memory of the DSP, and the following steps of singular value decomposition and noise cancellation may be performed after the data length of the memory reaches a set value.

S13: and dividing the plurality of sub-signals into at least one preset signal class by utilizing a support vector machine algorithm.

Wherein one of the predetermined signal classes corresponds to a characteristic of an optical fiber vibration signal generated by a vibration source. That is, the characteristic data of the optical fiber vibration signal generated by a vibration source is classified as the attribute or characteristic of a preset signal class. The characteristic data may be vibration intensity, vibration frequency, vibration duration, etc. of the optical fiber vibration signal.

The optical fiber vibration signals mainly acquired by the invention may include vibration signals generated by a plurality of vibration sources, for example, the sub-signals acquired for the first time may be generated mainly by passing of pedestrians, and the sub-signals acquired for the fifth time may be generated by passing of vehicles, so that the sub-signals need to be classified according to different vibration sources.

Because some characteristics of the optical fiber vibration signals generated by different vibration sources are not greatly different, the optical fiber vibration signals are not linearly separable. In this embodiment, a Support Vector Machine (SVM) algorithm is adopted to classify the linear indivisible data.

Specifically, the processing terminal acquires vibration signals of different vibration sources in advance, extracts and learns vibration signal characteristics of the different vibration sources, and obtains a plurality of preset signal classes representing the vibration signal characteristics of the different vibration sources. A classifier for classifying the plurality of predetermined signal classes is obtained according to the following formula 1:

$\left\{\begin{array}{c}\min \frac{1}{2}\left|\right|w|{|}^2\\ S.T.y_i\[(w\·x_i)+b\]\≥1-{\mathrm{\ζ}}_i,(i=1,2,\mathrm{...},n)\\ {\mathrm{\ζ}}_i\≥0\end{array}\right.---\left(1\right)$

wherein the | | w | | non-woven cells²For the objective function, 1/| w | |, referred to as the geometric interval, is the Euclidean distance between the sub-signal and the classification hyperplane, and S.T represents the following constraint condition for the objective function, and ζ is_iAs a relaxation variable, said x_iThe vibration source signal is an ith vibration signal, y is an identification value of an ith preset signal class, n is the number of the obtained vibration source signals, and b is a set constant. The identification value of the preset signal class can be represented by-1, and the like, and the b can be set according to different optical fiber application environments or optical fiber requirements. The relaxation variables mentioned above can be determined in advance from the program test, finding suitable values by a number of trials.

After conversion to the form of equation 1 above, the problem becomes a convex optimization problem, or more specifically, a convex quadratic programming problem because the objective function is now quadratic and the constraints are linear. This problem can be solved using any off-the-shelf optimization package of qp (predictive programming), which ends up in a sentence: under certain constraint conditions, the target is optimal, and the loss is minimum.

Although this problem is indeed a standard QP problem, it also has its special structure, and after transforming the lagrangian dual to dual variable optimization problem, it can find a more efficient method to solve, and usually this method is much more efficient than directly using the general QP optimization package for optimization.

That is, besides the conventional method for solving the QP problem, an optimal solution can be obtained by solving the dual problem, which is the dual algorithm of the support vector machine under the linear separable condition, and this has the advantages that: one dual problem tends to be easier to solve; the two can naturally introduce the kernel function, and further popularize to the nonlinear classification problem.

After obtaining the classifier, the classifier is used to classify the sub-signals obtained in S12 to classify each sub-signal into the preset signal class.

Referring to fig. 3, in an embodiment, the step S13 includes the following sub-steps:

s131: when the number of the locally stored preset signal classes exceeds 2, two preset signal classes are selected from the locally stored preset signal classes, an SVM algorithm is utilized to form a two-class classifier, and each sub-signal is divided into one of the two selected preset signal classes by the two-class classifier.

S132: and reselecting the divided preset signal class and the unselected one of the plurality of preset signal classes, forming a two-class classifier by using an SVM algorithm, and dividing each sub-signal into one of the two reselected preset signal classes by using the newly formed two-class classifier.

Step S132 is repeated until each of the plurality of preset signal classes has been selected.

S133: and respectively taking the preset signal classes into which the plurality of sub-signals are finally divided as the preset signal classes to which the plurality of sub-signals belong.

In S131 and S132, the two classes of classifiers formed by the SVM algorithm can be obtained according to the method 1.

For the convenience of understanding, the classification of the sub-signals is described with reference to fig. 4, and there are 5 preset signal classes including 1 st to 5 th preset signal classes of the processing terminal. The processing terminal extracts a sub-signal, selects the 1 st and 5 th preset signal classes from the preset signal classes to form a second class classifier, and inputs the sub-signal into the second class classifier; if the classification result is that the sub-signal belongs to the 1 st preset signal class, selecting a 4 th preset signal class and the 1 st preset signal class from the remaining 3 preset signal classes to form a second class separator, and inputting the sub-signal into the second class separator; if the classification result is that the sub-signal belongs to the 1 st preset signal class, selecting the 3 rd preset signal class and the 1 st preset signal class from the remaining two preset signal classes to form a second class separator, and inputting the sub-signal into the second class separator; and if the obtained classification result is that the sub-signal belongs to the 1 st preset signal class, forming a second class separator from the remaining 2 nd preset signal class and the 1 st preset signal class, inputting the sub-signal into the second class separator, and if the obtained classification result is that the sub-signal belongs to the 1 st preset signal class, dividing the sub-signal into the first preset signal class. The other sub-signals are classified using the same principle as described above.

Of course, the selection order of the preset signal classes is arranged from large to small according to the difference degree between the feature data in the preset signal classes, that is, the difference degree between the feature data of the vibration signals included in the two preset signal classes selected for the first time is greater than the difference degree between the feature data of the vibration signals included in the two preset signal classes selected for the second time … …, and so on. Of course, the selection order of the preset signal classes may also be arranged in sequence from small to large according to the difference degree between the feature data in the plurality of preset signal classes, and is not limited herein.

In this embodiment, the processing terminal classifies a plurality of sub-signals of the optical fiber vibration signal by using the SVM algorithm, and since the SVM algorithm can classify linearly-indistinguishable data, sub-signals belonging to different vibration sources in the optical fiber vibration signal can be classified, so that effective classification of the optical fiber vibration signal corresponding to the vibration sources is realized, and after the sub-signals of different vibration sources are classified, accuracy of vibration source identification of subsequent sub-signals can be improved.

Referring to fig. 5, fig. 5 is a flowchart illustrating a method for classifying optical fiber vibration signals according to still another embodiment of the present invention. The method of the embodiment is executed by a processing terminal, 52: the method comprises the following steps:

s51: and the processing terminal acquires an optical fiber vibration signal, wherein the optical fiber vibration signal is obtained by converting an optical signal reflected by an optical fiber.

S52: and dividing the optical fiber vibration signal into a plurality of sub-signals.

The above steps of S51 and S52 are as described in S11 and S12 of the above embodiment, please refer to the above related description.

S53: and carrying out singular value decomposition on the matrix formed by the plurality of sub-signals by using a singular value decomposition algorithm to obtain a plurality of sub-matrices.

In this embodiment, the processing terminal processes the acquired multiple sub-signals by using a Singular Value Decomposition (SVD) algorithm to obtain multiple sub-signals with high signal-to-noise ratio and uniform noise.

Specifically, the step S53 may specifically include the following sub-steps:

substep S531: and forming the plurality of subsignals { x (1), x (2), …, x (n) } into a k × l order matrix H.

The plurality of optical fiber sampling signals form a one-dimensional signal sequence X ═ { X (1), X (2), …, X (n) }, and the one-dimensional signal sequence X forms a k × l order matrix H.

Wherein,k ═ l ═ n, and k and l are positive integers.

Substep S532: performing singular value decomposition on the matrix H by using the following formula 11 to obtain a plurality of sub-matrices

Wherein, the matrix H is a k × l unitary matrix; the matrix H has a rank m, and the k-l order matrix H can be represented as the sum of m k-l order sub-matrices by SVD.

U is an orthogonal matrix of order k × k, and V^TA conjugate matrix representing a matrix V, said V being an orthogonal matrix of order l × l, saidIs a diagonal matrix, theIs the i-th singular value of the matrix H, andsaid u is_iIs the ith column vector of the matrix U, said v_iIs the ith column vector of matrix V, the H_iIs composed of u_iAnd v_iThe sub-matrix of (2).

If matrix H represents time frequency information, corresponding uⁱAnd vⁱConsidered as a frequency vector and a time vector. Thus, the time-frequency information in H is decomposed into a series of units of uⁱAnd vⁱIn the constructed time-frequency subspace. Therefore, a sub-space is selected for reconstruction, and a signal with a specific component can be extracted, for example, a sub-matrix with the largest singular value is selected, so that main data features contained in the sub-signals can be extracted and obtained.

S54: and acquiring the submatrix with the maximum singular value, and extracting each row of elements of the submatrix with the maximum singular value to assemble a one-dimensional signal.

For example, a plurality of decomposed sub-matrices are obtainedSubmatrix with maximum medium singular valueAnd the submatrix with the maximum singular valueAs one-dimensional signals S respectively_jForming said one-dimensional signal S_j＝{S_j,1,S_j，2,…,S_j,mWherein, the S_j,mRepresenting a sub-matrixThe m-th row vector of (2).

S55: and performing wavelet de-noising on the one-dimensional signals to obtain a plurality of sub-signals generated at a certain position of the optical fiber.

In this embodiment, the processing terminal performs wavelet threshold denoising on the one-dimensional signal obtained in step S54, where the denoising manner may include a hard threshold and a soft threshold.

The hard threshold is specifically defined as a one-dimensional signal S obtained by performing the step S54 according to the following formula 12_jEach element of (1S)_j,1,S_j，2,…,S_j,mIs processed by each processed element S_j,1',S_j，2',…,S_j,m' } forming a fiber signal S generated in the period of time at a certain position of the optical fiber_j'，

${S_{j,z}}^{\′}=\left\{\begin{array}{cc}{S}_{j,z}& \left|S_{j,z}\right|>T\\ 0& \left|S_{j,z}\right|\≤T\end{array}\right.---\left(12\right)$

Wherein, the S_j,zIs said one-dimensional signal S_jThe z-th element of (1), the S_j,z' is the optical fiber vibration signal S_j' the z-th sub-signal (element), the T is a set value.

The soft threshold is specifically determined by the following formula 13 for the one-dimensional signal S obtained in the step S54_jEach element of (1S)_j,1,S_j，2,…,S_j,mAnd by each processed element { S }_j,1',S_j，2',…,S_j,m' } forming a fiber signal S generated in the period of time at a certain position of the optical fiber_j'，

${S_{j,z}}^{\′}=\left\{\begin{array}{cc}sgn\left(S_{j,z}\right)\left(\right|S_{j,z}|-\mathrm{\α}T)& \left|S_{j,z}\right|>T\\ 0& \left|S_{j,z}\right|\≤T\end{array}\right.---\left(13\right)$

Wherein, the S_j,zIs said one-dimensional signal S_jThe z-th element of (1), the S_j,z' is the optical fiber vibration signal S_j' the z-th sub-signal (element) in (α) is a scaling factor, 0 ≦ α ≦ 1, sgn (x) is a sign function, and T is a set value.

In a specific application, the T may be input by a user according to an actual situation, or set correspondingly after judging and learning the actual situation.

After the wavelet denoising, the processing terminal can classify, vibrate the source identification or perform other processing on the denoised sub-signals.

S56: and dividing the sub-signals into at least one preset signal class by utilizing an SVM algorithm, wherein one preset signal class corresponds to the characteristics of the optical fiber vibration signal generated by a vibration source.

Please refer to the above description of S13, which is not repeated herein.

S57: and matching the sub-signals belonging to the same preset signal class with the preset vibration source model, and executing S58 if the matching is successful. If the matching is unsuccessful, the processing terminal can extract the characteristics of the sub-signals and form a new preset signal class according to the characteristics of the sub-signals.

S58: and determining the vibration source type of the sub-signals belonging to the same preset signal class as the vibration source type corresponding to the preset vibration model.

For example, if the sub-signal belonging to the class 1 preset signal class matches with the feature of the first preset vibration model in the above example, the sub-signal is a vibration signal, and the vibration source is determined to be a pedestrian passing through. The processing terminal calculates the optical fiber position corresponding to the optical fiber signal and can also send information to related equipment so as to inform the optical fiber position that pedestrians pass through.

Specifically, the method for the processing terminal to determine whether the sub-signal matches the preset vibration model may include the following steps:

s571: the processing terminal acquires J frame sub-signals belonging to the same preset signal class, wherein J is more than or equal to 1.

S572: and extracting feature vectors of each frame of sub-signals to form an array T [ J ] ═ { T (0), T (J), … and T (J-1) }, and acquiring an array R [ I ] ═ R (0), R (I), … and R (I-1) } formed by the feature vectors of the preset signal model.

And the extraction mode of the feature vector of the sub-signal is consistent with that of the feature vector of the preset signal model.

For example, the processing terminal stores at least one preset signal model, each preset signal model corresponds to a plurality of feature vectors R (0), R (I), …, R (I-1) of the optical fiber vibration signal including a vibration source, wherein I is a timing index of a signal frame of the preset signal model, I-0 is a start sub-signal frame of the preset signal model, I-1 is an end sub-signal frame of the preset signal model, so I is a total number of sub-signal frames included in the preset signal model, and R (I) is a feature vector of a sub-signal of an ith frame of the preset signal model. The processing terminal extracts feature vectors of the sub-signals of the 1 st frame to the J th frame, which are sequentially corresponding to T (0), T (J), …, and T (J-1), where J is a timing index of a signal frame of the optical fiber signal, J is 0 which is a start sub-signal frame of the optical fiber signal, J is J-1 which is an end sub-signal frame of the optical fiber signal, so J is a total number of frames of the sub-signals included in the optical fiber signal, and T (i) is a feature vector of a sub-signal of the J th frame of the optical fiber signal. The above I and J are both greater than 1, and may be equal or unequal, and are not limited herein.

It should be noted that the way of extracting the feature vector of the sub-signal by the processing terminal is consistent with the way of extracting the feature vector in the preset signal model, so as to ensure the accurate comparison between the following two. That is, the preset signal model and the optical fiber signal use the same type of feature vector.

The extraction method may be various, for example, parameters obtained by Linear Prediction Coding (LPC) and capable of representing the characteristics of the sub-signal, such as LPC coefficients or cepstral coefficients. In another embodiment, the step of extracting the feature vector of each frame signal comprises: and carrying out LPC analysis on each frame of sub-signal to obtain corresponding cepstrum coefficient, and taking the cepstrum coefficient of each frame of sub-signal as a feature vector of the cepstrum coefficient.

S573: the distance g (R (0), T (0)) between the feature vector T (0) and the feature vector R (0) and the parameter M are determined.

Wherein said M is positively correlated with the difference between said I and J. For example, M + I-J, M being a set constant. In one application, m may be set to one tenth to one thirty times I or J and less than 10.

In this embodiment, the processing terminal calculates the distance g (R (0), T (0)) between the feature vector T (0) and the feature vector R (0) using formula 14.

g(R(0),T(0))＝2d(T(0),R(0)) (14)

Please refer to formula 16 of step S44 and the related description thereof for the definition of d.

S574: and the processing terminal calculates the distances g (R (I), T (J)) between each feature vector T (J) of the array T [ J ] and at least part of feature vectors R (I) of the array R [ I ] in sequence according to the distances g (R (0), T (0)) until the distances g (R (I-1), T (J-1)) between the feature vectors T (J-1) and R (I-1) are calculated.

Wherein g (R (i), T (j)) is calculated from g (R (i-1), T (j)), g (R (i-1), T (j-1)), or g (R (i), T (j-1)). For example, the processing terminal calculates g (r (i), t (j)) by using formula 15 and formula 16;

$g\left(R\right(i),T(j\left)\right)=min\left\{\begin{array}{c}g(R(i-1),T\left(j\right))+d(R\left(i\right),T\left(j\right))\\ g(R(i-1),T(j-1))+2d(R\left(i\right),T\left(j\right))\\ g(R\left(i\right),T(j-1))+d(R\left(i\right),T\left(j\right))\end{array}\right\}---\left(15\right)$

$d=\sqrt{\frac{{(y_1-x_1)}^2+{(y_2-x_2)}^2+\mathrm{......}+{(y_n-x_n)}^2}{n}}---\left(16\right)$

wherein the feature vector T (j) is represented as (y)₁，…，y_n) Said feature vector R (i) is represented by (x)₁，…，x_n). Of course, in other embodiments, the distance function d may also use Euclidean distance, which is

Wherein the partial feature vector R (I) corresponding to each feature vector t (j) includes feature vectors R (max (j-M,0)) to R (min (j + M, I-1)) in the array R [ I ].

The above sequential calculation can be expressed as: the distance between each feature vector T (J) and the same feature vector R (I) is calculated according to the element sequence of the array T [ J ], and the distance between the feature vector R (I) and the same feature vector T (J) is calculated according to the element sequence of the array R [ I ]. As shown in the above formula 12, the distance between each feature vector t (j) and the feature vector r (i) depends on the distance between the previous feature vectors, and thus needs to be calculated according to the array order.

Wherein, the S574 may specifically include the following sub-steps:

sequentially calculating the distance g (R (0), T (J)) between each feature vector T (J) of the array T [ J ] and the feature vector R (0) according to the distance g (R (0), T (0));

sequentially calculating the distance g (R (I), T (J)) between each feature vector T (J) of the array T [ J ] and at least part of the feature vectors R (I) of the array R [ I ], respectively.

Wherein, when j is equal to 0, the partial feature vector R (I) corresponding to the feature vector T (0) includes all feature vectors in the array R [ I ], and when j is equal to 0, the partial feature vector R (I) corresponding to the feature vector T (j) includes feature vectors R (max (j-M,1)) to R (min (j + M, I-1)) in the array R [ I ].

S575: and calculating the ratio of the distance g (R (I-1), T (J-1)) to the sum of I and J to serve as the similar distance of the sub-signal and the preset signal model.

For example, after the processing terminal obtains the distance g (R (I-1), T (J-1)) matched to the feature vector R (I-1) and the feature vector T (J-1), the similar distance s between the sub-signal and the preset signal model is calculated according to the formula 17;

$s=\frac{g(R(I-1),T(J-1))}{I+J}---\left(17\right).$

s576: and if the similar distance meets the set condition, the processing terminal determines the vibration source type of the sub-signal as the vibration source type corresponding to the preset signal model.

The set condition is, for example, less than the set similarity distance, or the minimum similarity distance among all the preset signal models. For example, the processing terminal stores a plurality of preset signal models, the processing terminal executes the steps S573 to S575 for a plurality of times to obtain the similar distance between each preset signal model and the optical fiber vibration model, and the processing terminal classifies the vibration source type of the optical fiber vibration model into the vibration source type corresponding to the preset signal model with the smallest similar distance. Of course, the setting condition may be other conditions according to different requirements of specific applications, and is not limited specifically herein.

The matching mode determines the vibration source type of the sub-signals according to the similarity, vibration source classification of the sub-signals is achieved, the classification mode can accurately classify the vibration sources, accuracy of vibration source identification is improved, the processing terminal only calculates the distance between each feature vector T (J) of the array T [ J ] and part of feature vectors R (I) of the array R [ I ] according to a set rule, operation amount is reduced, identification speed and efficiency are improved, and processing resources of the processing terminal are saved.

Referring to fig. 6, a schematic structural diagram of an embodiment of an optical fiber vibration signal classification apparatus according to the present invention includes:

the obtaining module 61 is configured to obtain an optical fiber vibration signal, where the optical fiber vibration signal is obtained by converting an optical signal reflected by an optical fiber.

And a dividing module 62, configured to divide the optical fiber vibration signal into a plurality of sub-signals.

The classification module 63 is configured to classify the plurality of sub-signals into at least one preset signal class by using a support vector machine SVM algorithm, where one preset signal class corresponds to a feature of an optical fiber vibration signal generated by a vibration source.

Optionally, the classification module 62 includes:

the selection unit is used for selecting two preset signal classes from the plurality of locally stored preset signal classes when the number of the locally stored preset signal classes exceeds 2, forming a two-class classifier by utilizing an SVM algorithm, and dividing each sub-signal into one of the two selected preset signal classes by utilizing the two-class classifier;

and the classifying unit is used for reselecting the preset signal class obtained by the division and an unselected preset signal class in the preset signal classes, forming a two-class classifier by using an SVM algorithm, dividing each sub-signal into one of the two reselected preset signal classes by using the newly formed two-class classifier, repeating the step until each preset signal class in the preset signal classes is selected, and taking the preset signal class obtained by the last division of the sub-signals as the preset signal class to which the sub-signals belong.

Optionally, the selection order of the preset signal classes is arranged in sequence from large to small according to the difference degree between the feature data in the plurality of preset signal classes.

Optionally, the apparatus further comprises an identification module configured to: matching sub-signals belonging to the same preset signal class with a preset vibration source model; and if the matching is successful, determining the vibration source type of the sub-signals belonging to the same preset signal class as the vibration source type corresponding to the preset vibration model.

Optionally, the obtaining module 61 is specifically configured to obtain an optical fiber vibration signal generated by a certain position of the optical fiber within a period of time; the dividing module 62 is specifically configured to perform multiple sampling on the optical fiber vibration signal to obtain multiple sub-signals.

Optionally, the dividing module 62 is specifically configured to acquire the optical fiber vibration signal by using an FPGA to obtain the plurality of sub signals.

Optionally, the apparatus further includes a denoising module, configured to perform singular value decomposition on a matrix formed by the plurality of sub-signals by using an SVD algorithm to obtain a plurality of sub-matrices; acquiring the submatrix with the maximum singular value, and extracting each row of elements of the submatrix with the maximum singular value to assemble a one-dimensional signal; and performing wavelet de-noising on the one-dimensional signals to obtain a plurality of sub-signals generated at a certain position of the optical fiber. Optionally, the sampling module 62 is specifically configured to determine an envelope of the optical fiber signal from a peak value of the optical fiber signal; and sampling the envelope curve of the optical fiber signal to obtain a plurality of sampling signals with set time length.

The modules of the processing terminal are respectively configured to execute corresponding steps in the method embodiments, and the specific execution process is as described in the above method embodiments and is not described herein again.

Referring to fig. 7, a schematic structural diagram of another embodiment of the optical fiber vibration signal classification apparatus according to the present invention, the apparatus 70 includes a processor 71, a memory 72, a receiver 73 and a bus 74. The processor 71, the memory 72, and the receiver 73 may be one or more, and only one is illustrated in fig. 7.

The receiver 73 is used for receiving information transmitted from an external device. For example, the optical fiber signal detected by the optical fiber sensor is received.

The memory 72 is used for storing and providing the computer program to the processor 71 and may store data processed by the processor 71, such as the fiber signal light received by the receiver 73. The memory 72 may include at least one of read-only memory, random access memory, and non-volatile random access memory (NVRAM), among others.

The memory 72 stores a computer program comprising elements, executable modules or data structures, or a subset thereof, or an expanded set thereof:

and (3) operating instructions: including various operational instructions for performing various operations.

Operating the system: including various system programs for implementing various basic services and for handling hardware-based tasks.

In an embodiment of the present invention, processor 71 performs the following operations by calling the operation instructions stored in memory 72 (which may be stored in the operating system).

In particular, processor 71 may execute computer programs in memory 72 for:

acquiring an optical fiber vibration signal by a receiver 73, wherein the optical fiber vibration signal is obtained by converting an optical signal reflected by an optical fiber;

dividing the optical fiber vibration signal into a plurality of sub-signals;

and dividing the sub-signals into at least one preset signal class by utilizing an SVM algorithm, wherein one preset signal class corresponds to the characteristics of the optical fiber vibration signal generated by a vibration source.

Optionally, the step of dividing the sub-signals into at least one preset signal class by the processor 71 using the SVM algorithm includes:

when the number of the locally stored preset signal classes exceeds 2, selecting two preset signal classes from the locally stored preset signal classes, forming a two-class classifier by utilizing an SVM algorithm, and dividing each sub-signal into one of the two selected preset signal classes by utilizing the two-class classifier;

reselecting the divided preset signal classes and the unselected one of the preset signal classes, forming a two-class classifier by using an SVM algorithm, dividing each sub-signal into one of the two reselected preset signal classes by using the newly formed two-class classifier, and repeating the step until each preset signal class in the preset signal classes is selected;

and respectively taking the preset signal classes into which the plurality of sub-signals are finally divided as the preset signal classes to which the plurality of sub-signals belong.

Optionally, the selection order of the preset signal classes is arranged in sequence from large to small according to the difference degree between the feature data in the plurality of preset signal classes.

Optionally, the processor 71 is further configured to: matching sub-signals belonging to the same preset signal class with a preset vibration source model; and if the matching is successful, determining the vibration source type of the sub-signals belonging to the same preset signal class as the vibration source type corresponding to the preset vibration model.

Optionally, the step of acquiring the fiber vibration signal performed by the processor 71 includes: acquiring an optical fiber vibration signal generated in a certain position of an optical fiber within a period of time; the processor 71 performs the step of dividing the fiber vibration signal into a plurality of sub-signals including: and sampling the optical fiber vibration signal for multiple times to obtain a plurality of sub-signals.

Optionally, the step of performing, by the processor 71, multiple sampling on the fiber vibration signal to obtain multiple sub-signals includes: and acquiring the optical fiber vibration signals by using the FPGA to obtain the plurality of sub-signals.

Optionally, the processor 71 is further configured to: carrying out singular value decomposition on a matrix formed by the plurality of sub-signals by utilizing an SVD algorithm to obtain a plurality of sub-matrices; acquiring the submatrix with the maximum singular value, and extracting each row of elements of the submatrix with the maximum singular value to assemble a one-dimensional signal; and performing wavelet de-noising on the one-dimensional signals to obtain a plurality of sub-signals generated at a certain position of the optical fiber.

The processor 71 may also be referred to as a CPU (Central Processing Unit). In a particular application, the various components of the terminal are coupled together by a bus 74, where the bus 74 may include a power bus, a control bus, a status signal bus, and the like, in addition to a data bus. But for clarity of illustration the various buses are labeled as bus 74 in the figures.

The method disclosed in the above embodiments of the present invention may also be applied in the processor 71, or implemented by the processor 71. The processor 71 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware or instructions in the form of software in the processor 71. The processor 71 may be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components. The various methods, steps and logic blocks disclosed in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in a memory 72, and the processor 71 reads information in the corresponding memory and performs the steps of the above method in combination with hardware thereof.

In the scheme, the processing terminal classifies a plurality of sub-signals of the optical fiber vibration signal by adopting the SVM algorithm, and the SVM algorithm can classify linear indivisible data, so that the sub-signals belonging to different vibration sources in the optical fiber vibration signal can be divided, the effective classification of the optical fiber vibration signal corresponding to the vibration sources is realized, and the accuracy of vibration source identification of the follow-up sub-signals can be improved after the sub-signals of the different vibration sources are divided.

In the embodiments provided in the present invention, it should be understood that the disclosed method and apparatus can be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules or units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated units in the other embodiments described above may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, a network device, or the like) or a processor (processor) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes performed by the present specification and drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims (10)

1. A method for classifying a fiber vibration signal, comprising:

the method comprises the steps that a processing terminal obtains an optical fiber vibration signal, wherein the optical fiber vibration signal is obtained by converting an optical signal reflected by an optical fiber;

dividing the optical fiber vibration signal into a plurality of sub-signals;

and dividing the sub-signals into at least one preset signal class by utilizing a Support Vector Machine (SVM) algorithm, wherein one preset signal class corresponds to the characteristics of the optical fiber vibration signal generated by a vibration source.

2. The classification method according to claim 1, wherein the step of classifying the plurality of sub-signals into at least one preset signal class using a Support Vector Machine (SVM) algorithm comprises:

when the number of the locally stored preset signal classes exceeds 2, selecting two preset signal classes from the locally stored preset signal classes, forming a two-class classifier by utilizing an SVM algorithm, and dividing each sub-signal into one of the two selected preset signal classes by utilizing the two-class classifier;

reselecting the divided preset signal classes and the unselected one of the preset signal classes, forming a two-class classifier by using an SVM algorithm, dividing each sub-signal into one of the two reselected preset signal classes by using the newly formed two-class classifier, and repeating the step until each preset signal class in the preset signal classes is selected;

and respectively taking the preset signal classes into which the plurality of sub-signals are finally divided as the preset signal classes to which the plurality of sub-signals belong.

3. The classification method according to claim 2, wherein the selection order of the preset signal classes is arranged in order from large to small according to the degree of difference between the feature data in the plurality of preset signal classes.

4. The classification method according to claim 1, wherein after the step of classifying the plurality of sub-signals into at least one preset signal class, the method further comprises:

matching sub-signals belonging to the same preset signal class with a preset vibration source model;

and if the matching is successful, determining the vibration source type of the sub-signals belonging to the same preset signal class as the vibration source type corresponding to the preset vibration model.

5. The method of classification of claim 1, wherein the step of acquiring a fiber vibration signal comprises:

acquiring an optical fiber vibration signal generated in a certain position of an optical fiber within a period of time;

the step of dividing the fiber vibration signal into a plurality of sub-signals comprises:

and sampling the optical fiber vibration signal for multiple times to obtain a plurality of sub-signals.

6. The method of classifying according to claim 5, wherein the step of sampling the fiber vibration signal a plurality of times to obtain a plurality of sub-signals comprises:

and acquiring the optical fiber vibration signals by using a high-speed data acquisition card FPGA to obtain the plurality of sub signals.

7. The method of sorting according to claim 1, wherein after the step of splitting the fiber vibration signal into a plurality of sub-signals, the method further comprises:

carrying out singular value decomposition on a matrix formed by the plurality of sub-signals by using a Singular Value Decomposition (SVD) algorithm to obtain a plurality of sub-matrices;

acquiring the submatrix with the maximum singular value, and extracting each row of elements of the submatrix with the maximum singular value to assemble a one-dimensional signal;

and performing wavelet de-noising on the one-dimensional signals to obtain a plurality of sub-signals generated at a certain position of the optical fiber.

8. An optical fiber vibration signal classification apparatus, comprising:

the optical fiber vibration detection device comprises an acquisition module, a detection module and a control module, wherein the acquisition module is used for acquiring an optical fiber vibration signal, and the optical fiber vibration signal is obtained by converting an optical signal reflected by an optical fiber;

the dividing module is used for dividing the optical fiber vibration signal into a plurality of sub-signals;

the classification module is used for classifying the sub-signals into at least one preset signal class by utilizing a Support Vector Machine (SVM) algorithm, wherein one preset signal class corresponds to the characteristics of the optical fiber vibration signal generated by a vibration source.

9. The classification apparatus of claim 8, wherein the classification module comprises:

the selection unit is used for selecting two preset signal classes from the plurality of locally stored preset signal classes when the number of the locally stored preset signal classes exceeds 2, forming a two-class classifier by utilizing an SVM algorithm, and dividing each sub-signal into one of the two selected preset signal classes by utilizing the two-class classifier;

and the classifying unit is used for reselecting the preset signal class obtained by the division and an unselected preset signal class in the preset signal classes, forming a two-class classifier by using an SVM algorithm, dividing each sub-signal into one of the two reselected preset signal classes by using the newly formed two-class classifier, repeating the step until each preset signal class in the preset signal classes is selected, and taking the preset signal class obtained by the last division of the sub-signals as the preset signal class to which the sub-signals belong.

10. An optical fiber sensing system is characterized by comprising an optical fiber sensor and a processing terminal;

the optical fiber sensor is used for sending a first optical signal at one end and receiving a second optical signal obtained by reflecting the first optical signal from the one end;

the processing terminal is configured to classify the optical fiber vibration signal corresponding to the second optical signal, wherein the processing terminal includes the optical fiber vibration signal classification apparatus according to claim 8 or 9, so as to classify the optical fiber vibration signal.