CN107356962B - Micro-seismic Signals localization method and device based on fibre optical sensor - Google Patents
Micro-seismic Signals localization method and device based on fibre optical sensor Download PDFInfo
- Publication number
- CN107356962B CN107356962B CN201710576872.6A CN201710576872A CN107356962B CN 107356962 B CN107356962 B CN 107356962B CN 201710576872 A CN201710576872 A CN 201710576872A CN 107356962 B CN107356962 B CN 107356962B
- Authority
- CN
- China
- Prior art keywords
- feature
- micro
- spectrum envelope
- seismic signals
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/22—Transmitting seismic signals to recording or processing apparatus
- G01V1/226—Optoseismic systems
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V2210/00—Details of seismic processing or analysis
- G01V2210/40—Transforming data representation
- G01V2210/48—Other transforms
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The application is the Micro-seismic Signals localization method and device about a kind of based on fibre optical sensor.This method comprises: obtaining Micro-seismic Signals and its location information in preset time;The Micro-seismic Signals include K class acoustic vibration signal;One kind acoustic vibration signal every in the Micro-seismic Signals is pre-processed;Extract the Fourier transformation feature and its spectrum envelope feature and bi-orthogonal wavelet transformation feature and its spectrum envelope feature of pretreated Micro-seismic Signals;By the Fourier transformation feature and its spectrum envelope feature and the training of the bi-orthogonal wavelet transformation feature and its spectrum envelope feature and test target locating and tracking training pattern;The i.e. fibre optical sensor that outputs test result obtains the corresponding ground target position of vibration signal.The accuracy and robustness of vibration signal locating and tracking can be improved in the application.
Description
Technical field
This application involves technical field of optical fiber sensing more particularly to a kind of Micro-seismic Signals positioning based on fibre optical sensor
Method and device.
Background technique
Fibre optical sensor using light wave as the carrier of signal, based on light modulation effect outside environmental elements (such as humidity,
The factors such as pressure, electric field) physical features parameter (such as intensity, wavelength, frequency, the phase of corresponding adjustment light wave when changing
Deng);Then the transmission medium of light wave is used optical fibers as;Finally using photodetector receive light wave, determine phase of light wave and
The variation of light intensity, to obtain the situation of change of detected outside environmental elements.
Since optical fiber itself is not charged, electromagnetism interference, good insulating, high sensitivity, small in size, light-weight, high temperature resistant
The advantages that, so that fibre optical sensor is suitable for (high temperature, inflammable and explosive, strong electromagnetic etc. under some rugged environments of long term monitoring
Place) external environment information, for example, the fields such as military affairs, aerospace, electric power, railway, oil well have huge development latent
Power.
Fibre optical sensor can be used for detecting ground target, be used as seismic sensor at this time.Ground target (pedestrian,
Vehicle) movement when excitation is carried out to generating seismic wave to ground, which can propagate along earth's surface medium to four sides.
For example, seismic wave is mainly regular impact signal when ground pedestrian movement (start, run, jumping).Vehicle movement
When, seismic wave specifically includes that first is that impact of the Vehicular vibration caused by vehicle motor to ground;Second is that the suspension of vehicle itself
Impact of the self-vibration to ground;Third is that the tire or crawler belt of vehicle are periodically clapped caused by ground during the motion
It beats.Then above-mentioned seismic signal is acquired using seismic sensor, above-mentioned signal is analyzed again later, may be implemented pair
Ground target detected, is classified and locating and tracking etc..
During realizing application scheme, inventor's discovery: light sensor locating and tracking ground target is being utilized
Seismic wave when, since Micro-seismic Signals are generally fainter, and external environment it is complicated and changeable cause detect signal it is more similar,
Accurate positionin can not be made to Micro-seismic Signals in time.
Summary of the invention
To overcome the problems in correlation technique, the embodiment of the present application provides a kind of earthquake motion based on fibre optical sensor
Signal framing method and device, the technical issues of to solve in the related technology.
According to the embodiment of the present application in a first aspect, providing a kind of Micro-seismic Signals positioning side based on fibre optical sensor
Method, which comprises
Obtain Micro-seismic Signals and its location information in preset time;The Micro-seismic Signals include K class acoustic vibration signal,
K is positive integer;
One kind acoustic vibration signal every in the Micro-seismic Signals is pre-processed;
The Fourier transformation feature and its spectrum envelope feature and biorthogonal wavelet for extracting pretreated Micro-seismic Signals become
Change feature and its spectrum envelope feature;
By the Fourier transformation feature and its spectrum envelope feature and the bi-orthogonal wavelet transformation feature and its spectrum envelope
Feature training and test target locating and tracking training pattern;
The i.e. fibre optical sensor that outputs test result obtains the corresponding ground target position of vibration signal.
Optionally, Micro-seismic Signals are obtained in preset time and its step of location information, comprising:
Each moment within a preset time records the L signal data and a position data of every a kind of signal;L is
Positive integer.
Optionally, pretreated step is carried out to one kind acoustic vibration signal every in the Micro-seismic Signals, comprising:
Calculate the average value of T*L signal data of Micro-seismic Signals;T is the quantity at the acquisition moment in preset time period
And be positive integer, L is positive integer;
For every a kind of acoustic vibration signal, the ratio pair of each signal data and the average value of the acoustic vibration signal is calculated
Such acoustic vibration signal is normalized.
Optionally, the Fourier transformation feature and its spectrum envelope feature and biorthogonal of pretreated Micro-seismic Signals are extracted
The step of Wavelet Transform Feature and its spectrum envelope feature, comprising:
Fourier transformation and bi-orthogonal wavelet transformation are carried out to pretreated Micro-seismic Signals respectively, obtain Fourier's change
Change feature and bi-orthogonal wavelet transformation feature.
Optionally, the Fourier transformation feature and its spectrum envelope feature and biorthogonal of pretreated Micro-seismic Signals are extracted
The step of Wavelet Transform Feature and its spectrum envelope feature, comprising:
The pretreated Micro-seismic Signals are decomposed using empirical mode decomposition method, obtain multiple essential moulds
State function component;
Obtain the spectrum envelope feature of each essential mode function component.
Optionally, by the Fourier transformation feature and its spectrum envelope feature and the bi-orthogonal wavelet transformation feature and its
The step of training of spectrum envelope feature and test target locating and tracking training pattern, comprising:
The Fourier transformation feature at [1, T/2] moment and its spectrum envelope feature F [1, T/2] and bi-orthogonal wavelet transformation is special
Sign and its spectrum envelope feature W [1, T/2] and the corresponding position signal of features described above are defined as training set;It will [T/2, the T] moment
Fourier transformation feature and its spectrum envelope feature F [T/2, T] and bi-orthogonal wavelet transformation feature and its spectrum envelope feature W [T/
2, T] and the corresponding position signal of features described above is defined as test set;T be preset time period in acquisition the moment quantity and
For positive integer;
The target locating training pattern is trained using the training set;
The target locating training pattern is tested using the test set.
Optionally, the target locating training pattern includes: neural network and projection length with time delay
Phase Memory Neural Networks.
According to the second aspect of the embodiment of the present application, a kind of Micro-seismic Signals positioning dress based on fibre optical sensor is provided
It sets, described device includes:
Signal acquisition module, for obtaining Micro-seismic Signals and its location information in preset time;The Micro-seismic Signals
Including K class acoustic vibration signal, K is positive integer;
Preprocessing module, for being pre-processed to one kind acoustic vibration signal every in the Micro-seismic Signals;
Feature obtains module, and the Fourier transformation feature and its spectrum envelope for extracting pretreated Micro-seismic Signals are special
Bi-orthogonal wavelet transformation feature of seeking peace and its spectrum envelope feature;
Training test module, for becoming the Fourier transformation feature and its spectrum envelope feature and the biorthogonal wavelet
Change feature and its training of spectrum envelope feature and test target locating and tracking training pattern;
As a result output module, for outputing test result, i.e. fibre optical sensor obtains the corresponding ground target institute of vibration signal
In position.
Optionally, it includes transform characteristics acquiring unit and spectrum envelope acquiring unit that the feature, which obtains module,;
The transform characteristics acquiring unit, for pretreated Micro-seismic Signals to be carried out with Fourier transformation and double respectively
Orthogonal wavelet transformation obtains Fourier transformation feature and bi-orthogonal wavelet transformation feature;
The spectrum envelope acquiring unit includes:
Component acquisition submodule, for being carried out using empirical mode decomposition method to the pretreated Micro-seismic Signals
It decomposes, obtains multiple essential mode function components;
Spectrum envelope acquisition submodule, for obtaining the spectrum envelope feature of each essential mode function component.
Optionally, the trained test module includes:
Definition unit, for will [1, the T/2] moment Fourier transformation feature and its spectrum envelope feature F [1, T/2] and pair
Orthogonal wavelet transformation feature and its spectrum envelope feature W [1, T/2] and the corresponding position signal of features described above are defined as training
Collection;Will [T/2, the T] moment Fourier transformation feature and its spectrum envelope feature F [T/2, T] and bi-orthogonal wavelet transformation feature and
Its spectrum envelope feature W [T/2, T] and the corresponding position signal of features described above are defined as test set;T is in preset time period
It acquires the quantity at moment and is positive integer;
Training unit, for being trained using the training set to the target locating training pattern;
Test cell, for being tested using the test set the target locating training pattern.
Optionally, the target locating training pattern includes: neural network unit and projection with time delay
Shot and long term Memory Neural Networks unit.
The technical solution that embodiments herein provides can include the following benefits:
The above method provided by the embodiments of the present application, by obtaining Micro-seismic Signals and its location information in preset time;
Then one kind acoustic vibration signal every in Micro-seismic Signals is pre-processed;Fu of pretreated Micro-seismic Signals is extracted later
In leaf transformation feature and its spectrum envelope feature and bi-orthogonal wavelet transformation feature and its spectrum envelope feature;Furthermore by above-mentioned Fourier
Simultaneously test target is fixed for transform characteristics and its spectrum envelope feature and the training of above-mentioned bi-orthogonal wavelet transformation feature and its spectrum envelope feature
Position track training model;During the test, above-mentioned target locating training pattern outputs test result i.e. fibre optical sensor
Obtain the corresponding ground target position of vibration signal.As it can be seen that by obtain Micro-seismic Signals Fourier transformation feature and
Its spectrum envelope feature and bi-orthogonal wavelet transformation feature and its spectrum include feature, after practical application is tested, with correlated characteristic
Extracting method compares, and the application is capable of providing the feature with more discrimination;Also, target locating is used in the application
Training pattern improves the accuracy and robustness of vibration signal locating and tracking.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
The application can be limited.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the application
Example, and together with specification it is used to explain the principle of the application.
Fig. 1 is the process signal for the Micro-seismic Signals localization method based on fibre optical sensor that one embodiment of the application provides
Figure;
Fig. 2 is that the process for the Micro-seismic Signals localization method based on fibre optical sensor that another embodiment of the application provides is shown
It is intended to;
Fig. 3 is that the process for the Micro-seismic Signals localization method based on fibre optical sensor that the another embodiment of the application provides is shown
It is intended to;
Fig. 4 is the neural network block diagram that one embodiment of the application provides;
Fig. 5 is time-delay neural network block diagram provided by the embodiments of the present application;
Fig. 6 is projection shot and long term Memory Neural Networks structure chart provided by the embodiments of the present application;
Fig. 7 is the block diagram for the Micro-seismic Signals positioning device based on fibre optical sensor that one embodiment of the application provides;
Fig. 8 is the block diagram for the Micro-seismic Signals positioning device based on fibre optical sensor that another embodiment of the application provides;
Fig. 9 is the block diagram for the Micro-seismic Signals positioning device based on fibre optical sensor that the application another embodiment provides;
Figure 10 is the block diagram for the Micro-seismic Signals positioning device based on fibre optical sensor that the another embodiment of the application provides.
Specific embodiment
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all embodiments consistent with the application.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects be described in detail in claims, the application.
Fig. 1 is the process of the Micro-seismic Signals localization method shown according to an exemplary embodiment based on fibre optical sensor
Schematic diagram.As shown in Figure 1, Micro-seismic Signals localization method includes:
Step 101, Micro-seismic Signals and its location information in preset time are obtained;The Micro-seismic Signals include K class sound
Vibration signal, K are positive integer;
Step 102, one kind acoustic vibration signal every in the Micro-seismic Signals is pre-processed;
Step 103, extract pretreated Micro-seismic Signals Fourier transformation feature and its spectrum envelope feature and it is double just
Hand over Wavelet Transform Feature and its spectrum envelope feature;
Step 104, by the Fourier transformation feature and its spectrum envelope feature and the bi-orthogonal wavelet transformation feature and
The training of its spectrum envelope feature and test target locating and tracking training pattern;
Step 105, the i.e. fibre optical sensor that outputs test result obtains the corresponding ground target position of vibration signal.
The above method provided by the embodiments of the present application, by obtaining Micro-seismic Signals and its location information in preset time;
Then one kind acoustic vibration signal every in Micro-seismic Signals is pre-processed;Fu of pretreated Micro-seismic Signals is extracted later
In leaf transformation feature and its spectrum envelope feature and bi-orthogonal wavelet transformation feature and its spectrum envelope feature;Furthermore by above-mentioned Fourier
Simultaneously test target is fixed for transform characteristics and its spectrum envelope feature and the training of above-mentioned bi-orthogonal wavelet transformation feature and its spectrum envelope feature
Position track training model;During the test, above-mentioned target locating training pattern outputs test result i.e. fibre optical sensor
Obtain the corresponding ground target position of vibration signal.As it can be seen that by obtain Micro-seismic Signals Fourier transformation feature and
Its spectrum envelope feature and bi-orthogonal wavelet transformation feature and its spectrum include feature, after practical application is tested, and only extract it
A kind of method of middle feature compares, and the application is capable of providing the feature with more discrimination;Also, target is used in the application
Locating and tracking training pattern improves the accuracy and robustness of vibration signal locating and tracking.
In a step 101, fibre optical sensor acquisition Micro-seismic Signals and its location information are utilized in the application.For example, with
Distributed fiberoptic sensor acquires one earthquake as acquisition equipment, every predetermined period (can be configured according to default scene)
Dynamic signal, then in predetermined period at acquisition T time, recording [1, T] moment whole K class acoustic vibration signal, (such as personnel walk step
Sound, personnel are run footsteps, running car sound, truck traveling sound etc.) and its sound source location information.It is each at each moment
Class acoustic vibration signal all records L signal data and a location information.T and L is all positive integer.
In a step 102, the application pre-processes every a kind of signal data.Such as pretreatment can be at normalization
Reason, comprising:
Step 201, the average value of T*L signal data of Micro-seismic Signals is calculated.
Step 202, for every a kind of acoustic vibration signal, the average value of each signal data He the acoustic vibration signal is calculated
Ratio such acoustic vibration signal is normalized.
In one embodiment, calculate T*L signal data average value, then calculate each signal data with it is above-mentioned average
The ratio of value is normalized each signal data.By pre-processing to signal data in the application, difference can be made
The signal data of amplitude is in the same order of magnitude, is convenient for comparative analysis, and accelerate Data Convergence speed using subsequent processes.Separately
Outside, faintly vibration signal and the strongly difference between vibration signal can suitably be reduced by normalization, improves and calculates essence
Degree.
In step 103, on the basis of Micro-seismic Signals after the pre-treatment, Fourier transformation feature and biorthogonal are obtained
Wavelet Transform Feature and their corresponding spectrum envelope features.
The application carries out Fourier transformation and bi-orthogonal wavelet transformation to pretreated Micro-seismic Signals respectively, obtains Fu
In leaf transformation feature and bi-orthogonal wavelet transformation feature.
In one embodiment, the signal time-domain value that S (t) is t moment is defined, is obtained after discrete Fourier transform:
Wherein, k=0,1 ... ..., N-1, X (k) are the discrete Fourier transform value of S (t), and N is the length for converting spectrum sequence
Degree.
Fourier transformation is carried out to vibration signal data in the embodiment of the present application, which is decomposed
For a series of unlimited superposition of the sine wave signal of different frequencies, i.e., reluctant time-domain signal is converted to and is easy to analyze
Frequency-region signal.But since the vibration signal data are non-stationary signal, cause Fourier transformation that data can not be effectively treated
Time domain minutia, therefore increase wavelet transformation in the embodiment of the present application, to obtain the non-stationary signal i.e. wink of vibration signal
Between change.The wavelet transformation can division time frequency space heterogeneous can more preferably be reflected by automatic adjusument time-frequency window
The instantaneous variation of non-stationary signal.
In another embodiment, the application uses biorthogonal wavelet, and one of function decomposes signal data, another
Signal is reconstructed in a function, can take into account the accuracy and symmetry of pretreated Micro-seismic Signals in this way.
For example, choose two sequences hs (n) andAs there is limit for length's unit impulse response (Finite Impulse
Response, FIR) filter impulse response, two groups of scaling function phis (t) andGenerated subspace V respectivelykWithAnd
Have:
<φ(2-kT), (2 φ-kT-k) >=2kδ(k) (5)
Wherein, n represents time series, is positive integer;K represents the displacement in time series, is integer.
It will be appreciated that the basic function that orthogonal wavelet constitutes the flexible and translation of wavelet function is completely orthogonal, and this Shen
Please in using biorthogonal wavelet to there is orthogonality between wavelet function of the different scale under flexible, and with passing through translation between scale
Obtain there is no orthogonality between wavelet function system, can make in this way φ (t) andWith more symmetry, and reduce wavelet transformation
Caused phase distortion in treatment process.
In one embodiment of the application, after extracting Fourier transformation feature and bi-orthogonal wavelet transformation feature, such as Fig. 3 institute
Show, comprising:
Step 301, the pretreated Micro-seismic Signals are decomposed using empirical mode decomposition method, is obtained more
A essence mode function component.For example, empirical mode decomposition method is the time scale feature according to vibration signal data itself
Signal decomposition is carried out, without presetting any basic function.Decomposable process is: it is all to find out vibration signal data sequence X (t)
Maximum point, and the coenvelope line to form vibration signal data sequence is fitted with cubic spline functions.Equally, institute is found out
Some minimum points, and all minimum points are fitted to the lower envelope line to form data by cubic spline functions.On
The mean value of envelope and lower envelope line is denoted as ml, and subtracts average envelope ml with vibration signal data sequence X (t), obtains one
A new data sequence h:X (t)-ml=hl.If there is also negative local maximums and positive local minimum by new data sequence h
Value, illustrating this also is not an intrinsic mode functions, needs to continue " to screen ", until obtaining multiple essential mode functions point
Amount.
Step 302, the spectrum envelope feature of each essential mode function component is obtained.
By the above process, step 103 final output Fourier transformation feature and its spectrum envelope feature, and, biorthogonal is small
Wave conversion feature and its spectrum envelope feature.
At step 104, according to above-mentioned Fourier transformation feature and its spectrum envelope feature, and, bi-orthogonal wavelet transformation is special
Sign and its spectrum envelope feature carry out ground target locating and tracking.
In one embodiment, the application uses target locating model,
As shown in figure 4, the model framework of the application includes neural network and projection shot and long term memory with time delay
Neural network.Wherein, the neural network for the characteristic elder generation input tape having time delay that step 103 obtains, then sometimes by band
Between postpone neural network output as project shot and long term Memory Neural Networks input.As it can be seen that the application uses two kinds of minds
The mode combined through network, the neural network that not only can use time delay includes more time-based context relations
The advantages of, can also include using shot and long term neural network more context relation based on content the advantages of.That is, the application is real
It applies example and constructs a set of deep neural network model all more complicated on time and content, to have to vibration signal feature
Higher differentiation precision.
As shown in figure 5, having neural network feedforward neural network (the Feedforward Neural of time delay
Network on the basis of), the time delay depth that different front and back frame informations can be added between each hidden layer has been further introduced into it
Neural network (Time-Delay Neural Network, TDNN), it is traditional in this architecture, pass through narrow sense
Context realizes the study of initialization, further carries out hidden layer training from extensive context with deeper layer, can be effective
Ground learns the characterization from Short-term characteristic.Each layer in TDNNs is all run with different temporal resolutions, more with network
High-level increase, higher layer will be able to learn more time-based context relations.In addition, in the backpropagation phase
Between, the lower level of network is updated by the gradient accumulated in all time steps of Input context.Therefore, network compared with
Low layer is forced study conversion invariant features transformation.The common each hidden layer of DNN only receives the current output of previous hidden layer, with
DNN is compared, and context extension has also been made to hidden layer in TDNNs, i.e. TDNNs can be by the current output of hidden layer and its front and back several moment
Output be stitched together, the input as next hidden layer.
As shown in fig. 6, projection shot and long term Memory Neural Networks are the modified versions of the shot and long term Memory Neural Networks of standard,
Y_t in shot and long term Memory Neural Networks is done into primary projection dimensionality reduction, is re-used as exporting and for recycling.Y_t is reduced in this way
With the dimension of y_ (t-1), it is assumed that halve dimension when projection, then most of parameter amount can be made to halve.Project shot and long term note
Neural network structure is recalled as shown in figure 3, its formula is as follows:
gt=tanh (Wxgxt+Wrgrt-1+bg); (6)
it=σ (Wxixt+Wrirt-1+Wcict-1+bi); (7)
ft=σ (Wxfxt+Wrfrt-1+Wcfct-1+bf); (8)
ct=it⊙gt+ft⊙ct-1; (9)
ot=σ (Wxoxt+Wrort-1+Wcoct-1+bo); (10)
mt=ot⊙tanh(ct); (11)
yt=Wrpmt; (12)
rt=yt(1:nr); (13)
Wherein, xt、ytIt is currently to output and input;gtIt is that treated to input, merges current input and previous moment is defeated
Information out;ctIt is the state value of first cell CELL, for saving historical information;it、ftAnd otIt is input gate respectively, forgets door
And out gate, control selections are how many respectively or which treated input, forget how many or which previous moment state value with
And select how many or which state activation value as output;σ is Sigmoid function (also referred to as S sigmoid growth curve);W indicates to correspond to
Weight (such as WxgIndicate current input xtWith treated input gtBetween weight), b is bias vector, from CELL to each
A matrix is diagonal matrix;⊙ indicates that the element of two vectors is corresponding and is multiplied.I in formulat、ft、ot、gt、ctAnd ytDimension
It is identical.WrpFor projection matrix, rt=yt(1:nr) indicate rtIt is ytPreceding nrA element, i in formulat、ft、ot、gt、ctAnd mtDimension
It is identical.
In practical applications, general projection matrix can be ytDimension become mtThe half of dimension, and rtGenerally take ytBefore
Half part.
Another embodiment, the training method to the target locating model include: training set/test set select, model
Training, model measurement.
Wherein, training set/test set, which is selected, includes:
The Fourier transformation feature at [1, T/2] moment and its spectrum envelope feature F [1, T/2] and bi-orthogonal wavelet transformation is special
Sign and its spectrum envelope feature W [1, T/2] and the corresponding position signal of features described above are defined as training set.That is preset time
First half is as training set.
The Fourier transformation feature at [T/2, T] moment and its spectrum envelope feature F [T/2, T] and bi-orthogonal wavelet transformation is special
Sign and its spectrum envelope feature W [T/2, T] and the corresponding position signal of features described above are defined as test set.That is preset time
Latter half is as test set.
Target locating model is trained using training set in the embodiment of the present application.Then using test set to instruction
Target locating model after white silk is tested, and the test result of target locating model output is obtained.
It will be appreciated that test result, that is, fibre optical sensor of above-mentioned target locating model output obtains vibration signal
Corresponding ground target position.
According to the second aspect of the embodiment of the present application, a kind of Micro-seismic Signals positioning dress based on fibre optical sensor is provided
It sets, as shown in fig. 7, described device includes:
Signal acquisition module 701, for obtaining Micro-seismic Signals and its location information in preset time;The earthquake motion letter
Number include K class acoustic vibration signal, K is positive integer;
Preprocessing module 702, for being pre-processed to one kind acoustic vibration signal every in the Micro-seismic Signals;
Feature obtains module 703, for extracting the Fourier transformation feature and its spectrum packet of pretreated Micro-seismic Signals
Network feature and bi-orthogonal wavelet transformation feature and its spectrum envelope feature;
Training test module 704, for the Fourier transformation feature and its spectrum envelope feature and the biorthogonal is small
Wave conversion feature and its training of spectrum envelope feature and test target locating and tracking training pattern;
As a result output module 705, for outputing test result, i.e. fibre optical sensor obtains vibration signal corresponding ground appearance
Mark position.
Optionally, as shown in figure 8, it includes that transform characteristics acquiring unit 801 and spectrum envelope obtain that features described above, which obtains module 703,
Take unit 802;
Transform characteristics acquiring unit 801, for pretreated Micro-seismic Signals to be carried out with Fourier transformation and double respectively
Orthogonal wavelet transformation obtains Fourier transformation feature and bi-orthogonal wavelet transformation feature;
As shown in figure 9, spectrum envelope acquiring unit 802 includes:
Component acquisition submodule 901, for utilizing empirical mode decomposition method to the pretreated Micro-seismic Signals
It is decomposed, obtains multiple essential mode function components;
Spectrum envelope acquisition submodule 902, for obtaining the spectrum envelope feature of each essential mode function component.
Optionally, as shown in Figure 10, above-mentioned trained test module 704 includes:
Definition unit 1001 is used for the Fourier transformation feature and its spectrum envelope feature F [1, T/2] at [1, T/2] moment
It is defined as instructing with bi-orthogonal wavelet transformation feature and its spectrum envelope feature W [1, T/2] and the corresponding position signal of features described above
Practice collection;By the Fourier transformation feature at [T/2, T] moment and its spectrum envelope feature F [T/2, T] and bi-orthogonal wavelet transformation feature
And its spectrum envelope feature W [T/2, T] and the corresponding position signal of features described above are defined as test set;T is in preset time period
Acquisition the moment quantity and be positive integer;
Training unit 1002, for being trained using the training set to the target locating training pattern;
Test cell 1003, for being tested using the test set the target locating training pattern.
Optionally, above-mentioned target locating training pattern includes: neural network unit and projection with time delay
Shot and long term Memory Neural Networks unit.
About the device in above-described embodiment, the concrete mode that wherein each unit or module execute operation is having
It closes and is described in detail in the embodiment of this method, no detailed explanation will be given here.
Those skilled in the art will readily occur to its of the application after considering specification and practicing disclosure disclosed herein
Its embodiment.This application is intended to cover any variations, uses, or adaptations of the application, these modifications, purposes or
Person's adaptive change follows the general principle of the application and including the undocumented common knowledge in the art of the application
Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the application are by following
Claim is pointed out.
It should be understood that the application is not limited to the precise structure that has been described above and shown in the drawings, and
And various modifications and changes may be made without departing from the scope thereof.Scope of the present application is only limited by the accompanying claims.
Claims (9)
1. a kind of Micro-seismic Signals localization method based on fibre optical sensor, which is characterized in that the described method includes:
Obtain Micro-seismic Signals and its location information in preset time;The Micro-seismic Signals include K class acoustic vibration signal, and K is
Positive integer;
One kind acoustic vibration signal every in the Micro-seismic Signals is pre-processed;
Fourier transformation feature and its spectrum envelope feature and the bi-orthogonal wavelet transformation for extracting pretreated Micro-seismic Signals are special
Sign and its spectrum envelope feature;
By the Fourier transformation feature and its spectrum envelope feature and the bi-orthogonal wavelet transformation feature and its spectrum envelope feature
Training and test target locating and tracking training pattern;
The i.e. fibre optical sensor that outputs test result obtains the corresponding ground target position of vibration signal;
Wherein, by the Fourier transformation feature and its spectrum envelope feature and the bi-orthogonal wavelet transformation feature and its spectrum envelope
The step of feature training and test target locating and tracking training pattern, comprising:
Will [1, the T/2] moment Fourier transformation feature and its spectrum envelope feature F [1, T/2] and bi-orthogonal wavelet transformation feature and
Its spectrum envelope feature W [1, T/2] and the corresponding position signal of features described above are defined as training set;By Fu at [T/2, T] moment
In leaf transformation feature and its spectrum envelope feature F [T/2, T] and bi-orthogonal wavelet transformation feature and its spectrum envelope feature W [T/2, T],
And the corresponding position signal of features described above is defined as test set;T is the quantity at the acquisition moment in preset time period and is positive
Integer;
The target locating training pattern is trained using the training set;
The target locating training pattern is tested using the test set.
2. Micro-seismic Signals localization method according to claim 1, which is characterized in that obtain earthquake motion letter in preset time
Number and its step of location information, comprising:
Each moment within a preset time records the L signal data and a position data of every a kind of signal;L is positive whole
Number.
3. Micro-seismic Signals localization method according to claim 1, which is characterized in that each in the Micro-seismic Signals
Class acoustic vibration signal carries out pretreated step, comprising:
Calculate the average value of T*L signal data of Micro-seismic Signals;T is the quantity at the acquisition moment in preset time period and is
Positive integer, L are positive integer;
For every a kind of acoustic vibration signal, the ratio of each signal data and the average value of the acoustic vibration signal is calculated to such
Acoustic vibration signal is normalized.
4. Micro-seismic Signals localization method according to claim 1, which is characterized in that extract pretreated earthquake motion letter
Number Fourier transformation feature and its spectrum envelope feature and bi-orthogonal wavelet transformation feature and its step of spectrum envelope feature, packet
It includes:
Fourier transformation and bi-orthogonal wavelet transformation are carried out to pretreated Micro-seismic Signals respectively, obtain Fourier transformation spy
It seeks peace bi-orthogonal wavelet transformation feature.
5. Micro-seismic Signals localization method according to claim 1, which is characterized in that extract pretreated earthquake motion letter
Number Fourier transformation feature and its spectrum envelope feature and bi-orthogonal wavelet transformation feature and its step of spectrum envelope feature, packet
It includes:
The pretreated Micro-seismic Signals are decomposed using empirical mode decomposition method, obtain multiple essential mode letters
Number component;
Obtain the spectrum envelope feature of each essential mode function component.
6. Micro-seismic Signals localization method according to claim 1, which is characterized in that the target locating training mould
Type includes: neural network and projection shot and long term Memory Neural Networks with time delay.
7. a kind of Micro-seismic Signals positioning device based on fibre optical sensor, which is characterized in that described device includes:
Signal acquisition module, for obtaining Micro-seismic Signals and its location information in preset time;The Micro-seismic Signals include K
Class acoustic vibration signal, K are positive integer;
Preprocessing module, for being pre-processed to one kind acoustic vibration signal every in the Micro-seismic Signals;
Feature obtain module, for extract pretreated Micro-seismic Signals Fourier transformation feature and its spectrum envelope feature and
Bi-orthogonal wavelet transformation feature and its spectrum envelope feature;
Training test module, for the Fourier transformation feature and its spectrum envelope feature and the bi-orthogonal wavelet transformation is special
Sign and its training of spectrum envelope feature and test target locating and tracking training pattern;
As a result it is in place to obtain the corresponding ground target institute of vibration signal for the i.e. fibre optical sensor that outputs test result for output module
It sets;
The trained test module includes:
Definition unit is used for the Fourier transformation feature at [1, T/2] moment and its spectrum envelope feature F [1, T/2] and biorthogonal
Wavelet Transform Feature and its spectrum envelope feature W [1, T/2] and the corresponding position signal of features described above are defined as training set;It will
The Fourier transformation feature and its spectrum envelope feature F [T/2, T] and bi-orthogonal wavelet transformation feature and its spectrum packet at [T/2, T] moment
Network feature W [T/2, T] and the corresponding position signal of features described above are defined as test set;When T is the acquisition in preset time period
The quantity at quarter and be positive integer;
Training unit, for being trained using the training set to the target locating training pattern;
Test cell, for being tested using the test set the target locating training pattern.
8. Micro-seismic Signals positioning device according to claim 7, which is characterized in that it includes becoming that the feature, which obtains module,
Change feature acquiring unit and spectrum envelope acquiring unit;
The transform characteristics acquiring unit, for carrying out Fourier transformation and biorthogonal respectively to pretreated Micro-seismic Signals
Wavelet transformation obtains Fourier transformation feature and bi-orthogonal wavelet transformation feature;
The spectrum envelope acquiring unit includes:
Component acquisition submodule, for being divided using empirical mode decomposition method the pretreated Micro-seismic Signals
Solution obtains multiple essential mode function components;
Spectrum envelope acquisition submodule, for obtaining the spectrum envelope feature of each essential mode function component.
9. Micro-seismic Signals positioning device according to claim 7, which is characterized in that the target locating training mould
Type includes: neural network unit and projection shot and long term Memory Neural Networks unit with time delay.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710576872.6A CN107356962B (en) | 2017-07-14 | 2017-07-14 | Micro-seismic Signals localization method and device based on fibre optical sensor |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710576872.6A CN107356962B (en) | 2017-07-14 | 2017-07-14 | Micro-seismic Signals localization method and device based on fibre optical sensor |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107356962A CN107356962A (en) | 2017-11-17 |
CN107356962B true CN107356962B (en) | 2019-03-08 |
Family
ID=60292702
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710576872.6A Active CN107356962B (en) | 2017-07-14 | 2017-07-14 | Micro-seismic Signals localization method and device based on fibre optical sensor |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107356962B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108387925A (en) * | 2017-05-08 | 2018-08-10 | 武汉普惠海洋光电技术有限公司 | Based on optical fiber the localization method of acoustic detector intruder detection system |
CN112047212B (en) * | 2020-08-31 | 2022-08-12 | 日立楼宇技术(广州)有限公司 | Elevator operation control method, device, computer equipment and storage medium |
CN116299705A (en) * | 2023-03-21 | 2023-06-23 | 北京航天极峰科技有限公司 | Vibration detection method and system |
CN117875134A (en) * | 2024-02-29 | 2024-04-12 | 重庆交通大学 | Large-span arch bridge shock resistance probability assessment method based on time-frequency hybrid calculation |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1164886C (en) * | 2002-12-10 | 2004-09-01 | 西安交通大学 | Oil gas pipeline leak intelligent on line monitoring method based on distribution type optical fibre sensor |
CN101556724B (en) * | 2008-04-09 | 2011-06-01 | 中国科学院半导体研究所 | Safety management system of optical fiber perimeter and pattern recognition method thereof |
CN103033842B (en) * | 2012-12-18 | 2016-01-20 | 电子科技大学 | Time division multiplexing array formula optical fiber grating earthquake ripple real-time monitoring system |
CN103090961B (en) * | 2013-02-01 | 2014-11-26 | 华中科技大学 | Disturbance source positioning method of distributed type optical fiber sensing system |
CN103292889B (en) * | 2013-05-23 | 2014-12-10 | 无锡聚为传感科技有限公司 | Distributed optical fiber vibrating sensor vibration source locating method |
CN103605151B (en) * | 2013-11-20 | 2016-03-02 | 中北大学 | Based on the Distributed Cluster ripple shallow-layer microseism localization method of phase measurement |
CN106646587B (en) * | 2016-12-29 | 2018-08-21 | 北京知觉科技有限公司 | Object detection and recognition method and system based on acoustic vibration signal |
-
2017
- 2017-07-14 CN CN201710576872.6A patent/CN107356962B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN107356962A (en) | 2017-11-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107356962B (en) | Micro-seismic Signals localization method and device based on fibre optical sensor | |
Shao et al. | Data-driven distributed optical vibration sensors: a review | |
Wu et al. | One-dimensional CNN-based intelligent recognition of vibrations in pipeline monitoring with DAS | |
William et al. | Classification of military ground vehicles using time domain harmonics' amplitudes | |
Liu et al. | Deep learning and recognition of radar jamming based on CNN | |
CN109474890A (en) | A kind of indoor objects passive tracking method based on WiFi Doppler frequency shift | |
US10914573B2 (en) | Shearography detection and classification | |
CN110647788B (en) | Human daily behavior classification method based on micro-Doppler characteristics | |
Li et al. | A novel method for frequency feature extraction of ship radiated noise based on variational mode decomposition, double coupled Duffing chaotic oscillator and multivariate multiscale dispersion entropy | |
Wang et al. | Rapid response DAS denoising method based on deep learning | |
CN112415583A (en) | Seismic data reconstruction method and device, electronic equipment and readable storage medium | |
Xu et al. | Seismic target recognition based on parallel recurrent neural network for unattended ground sensor systems | |
Ristea et al. | Estimating the magnitude and phase of automotive radar signals under multiple interference sources with fully convolutional networks | |
Kimiaefar et al. | Seismic random noise attenuation using artificial neural network and wavelet packet analysis | |
Vos et al. | A generative machine learning approach to RFI mitigation for radio astronomy | |
Van Komen et al. | A convolutional neural network for source range and ocean seabed classification using pressure time-series | |
Martisius et al. | EEG dataset reduction and classification using wave atom transform | |
Ghavidel et al. | Sonar data classification by using few-shot learning and concept extraction | |
Lan et al. | Recognition of moving ground targets by measuring and processing seismic signal | |
CN112652326B (en) | Ambient sound identification method based on pseudo-color time-frequency image and convolution network | |
Esmaeilpour et al. | From environmental sound representation to robustness of 2D CNN models against adversarial attacks | |
Yoshimura et al. | Preliminary investigation of visualizing human activity recognition neural network | |
Wang et al. | The methodological pitfall of dataset-driven research on deep learning: An IoT example | |
CN116091897A (en) | Distributed optical fiber sensing event identification method and system based on light weight | |
Huang et al. | Fully modelling based intrusion discrimination in optical fiber perimeter security system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20220825 Address after: Room 607, Building A, West District, Shangdi International Pioneer Park, No. 1 Information Road, Haidian District, Beijing 100085 Patentee after: Beijing Fibo Optoelectronics Technology Co.,Ltd. Address before: D312, floor 2, block D, No. 9, Shangdi Third Street, Haidian District, Beijing 100085 Patentee before: BEIJING PERCEPTION TECHNOLOGY Co.,Ltd. |
|
TR01 | Transfer of patent right |