CN104266894B - A kind of mine microquake signal preliminary wave moment extracting method based on correlation analysis - Google Patents

Abstract

A kind of mine microquake signal preliminary wave moment extracting method based on correlation analysis, belongs to the analysis and processing method of mine microquake signal.Its function is that Intelligent Recognition two signal time is poor, realizes automatic shift alignment and judge mine microquake signal correlation, and the method is broadly divided into five steps: one is to read mine microquake signal；Two correlation functions being to solve for two mine microquake signals；Three is according to the correlation function solved, and obtains time difference during maximal correlation；Four is to shift one of them signal according to the time difference calculated；Five is that two mine microquake signals after alignment are solved its correlation coefficient, it is judged that its dependency.The present invention preferably resolves Intelligent Recognition mine microquake signal arrival time difference and judges that mine microquake signal is whether from the problem of same focus.

Description

A kind of mine microquake signal preliminary wave moment extracting method based on correlation analysis

Technical field

The present invention relates to the analysis and processing method of mine microquake signal, micro-in particular to a kind of mine based on correlation analysis Shake signal preliminary wave moment extracting method.

Background technology

In the shock event such as natural earthquake, engineering explosion, blast, wherein some energy is necessarily converted into shock wave Form, it can be propagated to around centered by focus.Mine microquake signal first break picking be vibrations research in the most crucial and Important problem.In seismic source location, the moment of the pickup of accurate quick to preliminary wave is by accurate seismic source location or signal The basis analyzed.The mine microquake signal collected is analyzed and processes, and determines mine microquake signal first arrival with this The technology of ripple due in is referred to as first break picking technology.First break picking technology is in military, civilian and industrial engineering field all There is applied research widely.

And ore deposit shake is the dynamic phenomenon of the mine rock mass sudden destroying that mining activity induces, shaft production in its serious threat With the life security of miner, it is monitored in real time and early warning has important theory and realistic meaning.Mine's shock signal is relative For seismic signal, the feature such as have that earthquake magnitude is little, focus is shallow and coverage is limited, it is possible to referred to as mine microquake Signal.Owing to can substantially mark out danger zone by the characteristic distributions and occurrence frequency of analyzing mine microquake signal, it is achieved Effective monitoring and early warning, and because the release of mine microquake signal energy own is less, seismic wave transmission range is limited, thus it requires It is higher that the positioning precision of mine microquake signal compares common seismic signal.And the raising of positioning precision needs the preliminary wave moment more For extracting fast and accurately.

The most general mine microquake signal preliminary wave automatic Picking technology used there is no the characteristic being specifically designed for mine microquake signal With require.When preliminary wave automatic Picking purpose is intended to determine the boundary in its signal between pure noise signal and useful signal Carve, be the most all that the change of the amplitude according to mine microquake signal, frequency and phase place determines this moment.Traditional first arrival Pick-up method is broadly divided into two big classes: a class is method based on earthquake record temporal characteristics, as extremum method (peakvalue's checking), Calculus of finite differences；This kind of method is more sensitive to noise ratio, when the noise of earthquake record is more serious, it is difficult to accurately first break picking.Separately One class method is method based on earthquake record global feature, such as method of correlation；Although this kind of method has preferably suppression to noise Effect, but affected by factors such as similaritys between seismic channel, for complicated earthquake record, the precision of first break pickup also can It is affected.

Up to the present, it has been proposed that the method for many first break pickup, as manually pick up method, method of correlation, energy ratio function, Peak amplitude method, FRACTAL DIMENSION method and neural network etc..

Artificial pickup method is simple, but is affected relatively big by anthropic factor and subjective factors, is readily incorporated personal error, can be straight Connect and cause resultant error to increase.

Gelchinsky and Shtivelman proposes a kind of method that neighboring track carries out cross-correlation, it is assumed that the arteries and veins in each road Rush shape not change.Method of correlation is continued to affecting relatively big by first arrival, and very close with the choice relation of wavelet, with Window scope when Shi Yaoqiu selects suitable, these have certain difficulty in Practical Project.

Hatherrly proposes Linear Least Square Predicting Technique and corrects, with flex point, the method combined, and he proposes first to identify First peak value and flex point, then estimate the statistics difference of the two.The grade of Huang Cheng uses statistical method by seismic first break record It is divided into signal and two parts of noise, and makes the difference between this two parts statistical nature for maximum.Statistical nature method receives ground Shake waveform similarity impact is relatively big, and precision can cause certain impact.

Signal energy in the ratio of energy use cycle and the ratio of window energy time total, more sensitive to first arrival, later arrivals decline Subtract ratio comparatively fast, so the maximum of points of ratio as the approximation of first arrival and is made suitable time shift, be initial time. Coppens proposes the method carrying out energy comparison when different size of in window.Jiang Yule etc. propose same polarity energy ratio Method, the energy ratio function i.e. improved.Owing to the capacity of resisting disturbance of the ratio of energy is the best, so occurring for first arrival waveform The initial time of the area pickup of significant change is not accurate enough.

The process of time domain fractal dimension method first break picking must interpolation, and the accuracy of the strong dependence interpolation of result.Fabio Boschetti etc. propose a kind of first arrival detection algorithm based on FRACTAL DIMENSION, and the method is along with it occurs in signal based on seismic channel The feature that fractal dimension changes is to determine seismic channel first arrival.But its to time window and step-length choose the most sensitive, the most not Careful will be serious affect its result.

Neural network utilizes multiparameter feature to carry out pattern recognition, and the temporal characteristics and the entirety that make full use of earthquake record are special Levy.Not only there is due to neutral net parallel processing, self-organizing self-learning capability, and there is height robustness, fault-tolerance Mapping, calculating and classification capacity with height.The East Sea, the village etc. use regards earthquake record first break pickup as a pattern recognition Journey, makes full use of temporal characteristics and the global feature of earthquake record, carries out earthquake record first arrival by Artificial Neural Network and picks up Taking, it can obtain preferable actual effect.But it is the highest that its major defect is algorithm complexity, search needs certain time.

Seismic signal passes through Wavelet Multiresolution Decomposition, can efficiently separate, eliminate noise, beneficially FRACTAL DIMENSION and nerve net Network method improves the precision of first break picking.Luo Guang proposes modified model first break picking based on wavelet transformation method, and Yang Junfeng carries Three-component seismic phase method of identification based on wavelet transformation, energy factors method are gone out.But it remains a need for manually choosing data segment sometimes It is picked up reducing the pick-up time.

In addition, certain methods is had to also rely on the contrast between this road and its shortcut, although this class method has one to noise Fixed suppression, but affected by factors such as similaritys between seismic channel, for complicated earthquake record, first break pickup Precision also can be affected.

In sum, seismic signal is an one-dimensional time series, due to this sequence merely illustrate time and amplitude it Between relation, and have the interference of noise signal, the pickup to directly carrying out seismic first breaks causes sizable difficulty.Solve Certainly problems can use intelligent algorithm as described above, but generally have higher algorithm complex due to intelligent algorithm, Be not suitable for needing quickly and having the mine microquake environment of higher positioning accuracy, and the present invention preferably resolves first arrival This problem of the pickup of ripple time, and there is relatively low algorithm complex.

Summary of the invention

It is an object of the invention to for problem present in prior art, it is provided that a kind of mine microquake based on correlation analysis is believed Number preliminary wave moment extracting method, solves preliminary wave time intelligence identification, mine microquake signal pair in mine microquake signal processing Neat and judge that mine microquake signal is whether from the problem of same focus.

Realize the technical scheme of the object of the invention: the mine microquake signal preliminary wave moment extracting method of the present invention includes five steps Rapid: one is to read mine microquake signal；Two correlation functions being to solve for two mine microquake signals；Three is according to the phase solved Close function, obtain time difference during maximal correlation；Four is to shift one of them signal according to the time difference calculated； Five is that two mine microquake signals after alignment are solved its correlation coefficient, it is judged that its dependency；Concrete grammar step is as follows:

(1) read mine microquake signal: two mine microquake signal-obtainings are entered system, define x_nIt it is a mine microquake signal Sequence, y_nFor another mine microquake signal sequence, it is desirable to two sequences must be isometric, set its sequence length as L, and two letters Number sample frequency is f_s；

(2) correlation function of two mine microquake signals is solved: solve mine microquake signal correction function according to following formula, in phase When closing function acquirement maximum, calculate the time difference between mine microquake signal according to sampled point and sample rate； Correlation function formula according to discrete signal sequence calculates correlation function R_xyFor:

$R_{\mathrm{xy}}\left(m\right)=\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}x\left(n\right)y(n+m)$

Wherein variable m span is 0 to L；

(3) according to the correlation function solved, time difference during maximal correlation is obtained: according to the time difference calculated to ore deposit Mountain microseismic signals is alignd；Calculate time difference during maximal correlation according to correlation function, sampled point with sample rate, obtain phase Close function maxima R_maxFor:

R_max=max [| R_xy(m)|]

(4) according to the time difference calculated, one of them signal is shifted: the mine microquake signal after alignment is carried out Correlation analysis, and set threshold value to judge whether mine microquake signal comes from same focus； Acquirement initial time, to the total number of sample points in correlation function maximum absolute value moment, is denoted as N, then time deviation T_offset:

T_offset=N/f_s

(5) two mine microquake signals after alignment are solved its correlation coefficient, it is judged that its dependency:

The correlation matrix R of two mine microquake signals_corrFor:

$R_{\mathrm{corr}}=\left[\begin{array}{cc}{\mathrm{\ρ}}_{11}& {\mathrm{\ρ}}_{12}\\ {\mathrm{\ρ}}_{21}& {\mathrm{\ρ}}_{22}\end{array}\right]$

Wherein ρ_ij=E ((X_i-E(X_i))·(Y_j-E(Y_j))), wherein variable i, variable j span are 1 to 2, and X is x_n Sequence overall, Y is y_nSequence overall, E is mathematic expectaion；The correlation matrix leading diagonal obtained represents from phase Guan Xing, it is always 1；Counter-diagonal is the dependency of two signals, and absolute value is the biggest, and dependency is the best, its span For [0,1]；Cut off value can be any value between 0 to 1 according to practical situation different set；Whether final determining program needs Terminate, if need to terminate, ending method, if need to continue to run with, return and read data step, read new data and shift Alignment and dependency compare.

Beneficial effect, owing to have employed such scheme, Intelligent Recognition two signal time is poor, realize automatic shift alignment and judgement Mine microquake signal correlation, it is achieved Intelligent Recognition mine microquake signal arrival time difference；Realize the intelligence of mine microquake signal Alignment, in order to store, analyze and process；Compare whether two mine microquake signals come from same focus by dependency； Solve Intelligent Recognition mine microquake signal arrival time difference and judge that mine microquake signal is whether from the problem of same focus.

Described mine microquake signal preliminary wave moment extracting method is due to the method using correlation analysis, and algorithm is easy, compares The method tools such as fuzzy recognition method in background technology, time domain fractal dimension method, neural net method, small wave converting method There is less program complexity, be more suitably applied to the use that there is a large amount of vibration data, promptness requires high field close, The methods such as artificial cognition method, the ratio of energy of comparing decrease the error that subjective factors produces.

Accompanying drawing explanation

Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, the present invention's Schematic description and description is used for explaining the present invention, is not intended that inappropriate limitation of the present invention.

Fig. 1 is the flow chart of the inventive method.

Fig. 2 is the system structure schematic diagram of the embodiment of the present invention.

Fig. 3 is a, b, c, d signal waveforms of the embodiment of the present invention.

Fig. 4 is a signal and b signal correction function and displacement alignment effect analogous diagram in the inventive method embodiment.

Fig. 5 is a signal and d signal correction function and displacement alignment effect analogous diagram in the inventive method embodiment.

Detailed description of the invention

The invention will be further described for middle embodiment:

Mine microquake signal preliminary wave moment extracting method includes five steps: one is to read mine microquake signal；Two are to solve for The correlation function of two mine microquake signals；Three is according to the correlation function solved, and obtains time difference during maximal correlation；Four It is according to the time difference calculated, one of them signal to be shifted；Five is that two mine microquake signals after alignment are solved it Correlation coefficient, it is judged that its dependency；Concrete grammar step is as follows:

(1) read mine microquake signal: two mine microquake signal-obtainings are entered system, define x_nIt it is a mine microquake signal Sequence, y_nFor another mine microquake signal sequence, it is desirable to two sequences must be isometric, set its sequence length as L, and two letters Number sample frequency is f_s；

(2) correlation function of two mine microquake signals is solved: solve mine microquake signal correction function according to following formula, in phase When closing function acquirement maximum, calculate the time difference between mine microquake signal according to sampled point and sample rate； Correlation function formula according to discrete signal sequence calculates correlation function R_xyFor:

$R_{\mathrm{xy}}\left(m\right)=\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}x\left(n\right)y(n+m)$

Wherein variable m span is 0 to L；

(3) according to the correlation function solved, time difference during maximal correlation is obtained: according to the time difference calculated to ore deposit Mountain microseismic signals is alignd；Calculate time difference during maximal correlation according to correlation function, sampled point with sample rate, obtain phase Close function maxima R_maxFor:

R_max=max [| R_xy(m)|]

(4) according to the time difference calculated, one of them signal is shifted: the mine microquake signal after alignment is carried out Correlation analysis, and set threshold value to judge whether mine microquake signal comes from same focus； Acquirement initial time, to the total number of sample points in correlation function maximum absolute value moment, is denoted as N, then time deviation T_offset:

T_offset=N/f_s

(5) two mine microquake signals after alignment are solved its correlation coefficient, it is judged that its dependency:

The correlation matrix R of two mine microquake signals_corrFor:

$R_{\mathrm{corr}}=\left[\begin{array}{cc}{\mathrm{\ρ}}_{11}& {\mathrm{\ρ}}_{12}\\ {\mathrm{\ρ}}_{21}& {\mathrm{\ρ}}_{22}\end{array}\right]$

Wherein ρ_ij=E ((X_i-E(X_i))·(Y_j-E(Y_j))), wherein variable i, variable j span are 1 to 2, and X For x_nSequence overall, Y is y_nSequence overall, E is mathematic expectaion；The correlation matrix leading diagonal obtained represents Autocorrelation, it is always 1；Counter-diagonal is the dependency of two signals, and absolute value is the biggest, and dependency is the best, its value Scope is [0,1]；Cut off value can be any value between 0 to 1 according to practical situation different set.Final determining program is No needs is terminated, if need to terminate, ending method, if need to continue to run with, returns and reads data step, reads new data and carries out Displacement alignment and dependency compare.

The present invention utilizes correlation function that signal carries out correlation analysis, thus obtain the preliminary wave moment, to carry out signal the most right Neat and judge that whether mine microquake signal is from same focus.

Accompanying drawing 1 is the flow chart of the inventive method, and the described mine microquake signal preliminary wave moment based on correlation analysis extracts Method, according to the flow process of accompanying drawing 1 complete the moment extract, signal alignment and judge signal whether from same focus, it is basic Step is as follows:

Read mine microquake signal x_n, mine microquake signal y_n, its sample frequency is f_s；

Correlation function formula calculating correlation function according to discrete signal sequence:

$R_{\mathrm{xy}}\left(m\right)=\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}x\left(n\right)y(n+m)---\left(1\right)$

Time difference during maximal correlation is calculated with sample rate according to correlation function, sampled point

Acquirement correlation function maximum:

R_max=max [| R_xy(m)|] (2)

Acquirement initial time, to the total number of sample points in correlation function maximum absolute value moment, is denoted as N

Then time deviation:

T_offset=N/f_s (3)

The T calculated according to previous step_offsetBy two signal displacement alignment, in order to next step calculates correlation coefficient；

The correlation matrix of two signals is sought according to formula:

$R_{\mathrm{corr}}=\left[\begin{array}{cc}{\mathrm{\ρ}}_{11}& {\mathrm{\ρ}}_{12}\\ {\mathrm{\ρ}}_{21}& {\mathrm{\ρ}}_{22}\end{array}\right]---\left(4\right)$

Wherein ρ_ij=E ((X_i-E(X_i))·(Y_j-E(Y_j))), and X is x_nSequence overall, Y is y_nSequence overall, E For mathematic expectaion.The correlation matrix leading diagonal obtained represents autocorrelation, and it is always 1；Counter-diagonal is two letters Number dependency, absolute value is the biggest, and dependency is the best.Generally two signal correction absolute coefficient are between 0-0.3 Close for microfacies；Close for reality between 0.3-0.5；It it is significant correlation between 0.5-0.8；It is highly correlated between 0.8-1.0.Should By setting cut off value, method judges whether two signals have a high correlation, thus judge its whether from same focus, This cut off value can be modified according to actual needs.

Described mine microquake signal preliminary wave moment extracting method based on correlation analysis, can be applicable to the distributed ore deposit of networking In the seismic source location system of mountain, accompanying drawing 2 is this system structure schematic diagram, 2 is embodied as described method below in conjunction with the accompanying drawings Mode is described in detail.In accompanying drawing 2, round dot is sensor node, comprises microseismic sensors, can monitor mine microquake Signal.When focus occurs vibrations, the sensor node closed on can collect mine microquake data.Accompanying drawing 3 is for receiving ore deposit The waveform of four different shaking sensor records of mountain microseismic signals, wherein tri-signals of a, b, c come from same focus, D comes from another focus being different from other three signals.In the present embodiment, set according to great amount of samples and correlation experience Determine whether from the cut off value of same focus be 0.5.

First a Yu b signal is carried out correlation analysis.Accompanying drawing 4 is a signal and b signal phase in the inventive method embodiment Close function and displacement alignment effect analogous diagram.Calculate the correlation function of a with b signal according to formula (1) and map, the phase tried to achieve Pass function is specifically distributed sees accompanying drawing 4.Correlation function maximum R is obtained according to formula (2)_max=250.8038.Calculate according to formula (3) A Yu b signal two signal time is poor, T_offset=-0.0393 (s).Two signal correction coefficient matrixes are calculated according to formula (4).? In correlation matrix, leading diagonal is always 1, and counter-diagonal that is two signal correlation, the biggest then dependency of absolute value is the best. A with the b signal correction coefficient matrix tried to achieve is:

$R_{\mathrm{ab}}=\left[\begin{array}{cc}1.0000& -0.5462\\ -0.5462& 1.0000\end{array}\right]$

Being taken absolute value by its correlation matrix counter-diagonal, obtaining a signal with b signal correlation is 0.5462, more than setting Fixed cut off value 0.4, the most described method judge a signal from b signal from different focus.

Secondly a Yu d signal is carried out correlation analysis.Accompanying drawing 5 is a signal and d signal phase in the inventive method embodiment Close function and displacement alignment effect analogous diagram.Calculate the correlation function of a with d signal according to formula (1) and map, the phase tried to achieve Pass function is specifically distributed sees accompanying drawing 5.Correlation function maximum R is obtained according to formula (2)_max=31.5554.Calculate according to formula (3) A Yu b signal two signal time is poor, T_offset=0.0533 (s).Two signal correction coefficient matrixes are calculated according to formula (4).Try to achieve A with d signal correction coefficient matrix be:

$R_{\mathrm{ad}}=\left[\begin{array}{cc}1.0000& -0.0566\\ -0.0566& 1.0000\end{array}\right]$

Being taken absolute value by its correlation matrix counter-diagonal, obtaining a signal with d signal correlation is 0.0566, less than setting Fixed cut off value 0.4, the most described method judge a signal from d signal from different focus.

Above in association with accompanying drawing, the detailed description of the invention of the present invention is described, but these explanations can not be considered as limiting this The scope of invention, protection scope of the present invention is limited by appended claims, any on the basis of the claims in the present invention Change be all protection scope of the present invention.

Claims (1)

1. a mine microquake signal preliminary wave moment extracting method based on correlation analysis, is characterized in that: mine microquake Signal preliminary wave moment extracting method includes five steps: one is to read mine microquake signal；Two are to solve for two mine microquake letters Number correlation function；Three is according to the correlation function solved, and obtains time difference during maximal correlation；Four is according to calculating Time difference one of them signal is shifted；Five is that two mine microquake signals after alignment are solved its correlation coefficient, sentences Its dependency disconnected；Concrete grammar step is as follows:

(1) read mine microquake signal: two mine microquake signal-obtainings are entered system, define x_nIt it is a mine microquake signal Sequence, y_nFor another mine microquake signal sequence, it is desirable to two sequences must be isometric, set its sequence length as L, and two letters Number sample frequency is f_s；

(2) correlation function of two mine microquake signals is solved: solve mine microquake signal correction function according to following formula, in phase When closing function acquirement maximum, calculate the time difference between mine microquake signal according to sampled point and sample rate；

Correlation function formula according to discrete signal sequence calculates correlation function R_xyFor:

$R_{xy}\left(m\right)=\underset{n=-\mathrm{\∞}}{\overset{\mathrm{\∞}}{\Σ}}x\left(n\right)y(n+m)$

Wherein variable m span is 0 to L；

(3) according to the correlation function solved, time difference during maximal correlation is obtained: according to the time difference calculated to ore deposit Mountain microseismic signals is alignd；Calculate time difference during maximal correlation according to correlation function, sampled point with sample rate, obtain phase Close function maxima R_maxFor:

R_max=max [| R_xy(m)|]

(4) according to the time difference calculated, one of them signal is shifted: the mine microquake signal after alignment is carried out Correlation analysis, and set threshold value to judge whether mine microquake signal comes from same focus；

Acquirement initial time, to the total number of sample points in correlation function maximum absolute value moment, is denoted as N, then time deviation T_offset:

T_offset=N/f_s

(5) two mine microquake signals after alignment are solved its correlation coefficient, it is judged that its dependency:

The correlation matrix R of two mine microquake signals_corrFor:

$R_{corr}=\left[\begin{array}{cc}{\mathrm{\ρ}}_{11}& {\mathrm{\ρ}}_{12}\\ {\mathrm{\ρ}}_{21}& {\mathrm{\ρ}}_{\mathrm{22}}\end{array}\right]$

Wherein ρⁱ _j=E ((X_i-E(X_i))·(Y_j-E(Y_j))), wherein variable i, variable j span are 1 to 2, and X For x_nSequence overall, Y is y_nSequence overall, E is mathematic expectaion；The correlation matrix leading diagonal table obtained Showing autocorrelation, it is always 1；Counter-diagonal is the dependency of two signals, and absolute value is the biggest, and dependency is the best, and it takes Value scope is [0,1]；Cut off value is according to any value that practical situation different set is between 0 to 1；Final determination methods is No needs is terminated, if need to terminate, ending method, if need to continue to run with, returns and reads data step, reads new data and carries out Displacement alignment and dependency compare.