CN104266894A - Mine microearthquake signal preliminary wave moment extracting method based on correlation analysis - Google Patents

Abstract

The invention discloses a mine microearthquake signal preliminary wave moment extracting method based on correlation analysis, belonging to analytic processing methods of mine microearthquake signals. The mine microearthquake signal preliminary wave moment extracting method has the functions of intelligently identifying a time difference between two signals, realizing automatic displacement and alignment and judging the correlation of the mine microearthquake signals. The mine microearthquake signal preliminary wave moment extracting method mainly comprises the following five steps: firstly, reading the mine microearthquake signals; secondly, solving a correlation function of the two mine microearthquake signals; thirdly, solving a time difference during the maximum correlation according to the solved correlation function; fourthly, displaying one signal according to the calculated time difference; and fifthly, solving a correlation coefficient of the two aligned mine microearthquake signals, and judging the correlation. According to the mine microearthquake signal preliminary wave moment extracting method, the problems of intelligently identifying the time difference of arrival of the mine microearthquake signal and judging whether the mine microearthquake signals come from the same focus are preferably solved.

Description

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

Technical field

The present invention relates to the analysis and processing method of mine microquake signal, particularly in a kind of mine microquake signal primary wave moment extracting method based on correlation analysis.

Background technology

In the shock event such as natural earthquake, engineering explosion, blast, wherein some energy must be converted into the form of shock wave, and it can propagate towards periphery centered by focus.The first break picking of mine microquake signal is problem very crucial and important in vibrations research.In seismic source location, the pickup of accurate quick is the basis of carrying out accurate seismic source location or signal analysis to the moment of primary wave.Treatment and analysis is carried out to the mine microquake signal collected, and determines that the technology of mine microquake signal primary wave due in is called first break picking technology with this.First break picking technology all has applied research widely in military, civilian and Industrial Engineering field.

And ore deposit shake is the dynamic phenomenon of the mine rock mass sudden destroying that mining activity brings out, the life security of shaft production and miner in its serious threat, carries out Real-Time Monitoring and early warning has important theory and realistic meaning to it.Mine's shock signal, for seismic signal, has that earthquake magnitude is little, focus is shallow and the feature such as coverage is limited, so can be referred to as mine microquake signal.Due to roughly danger zone can be marked out by the characteristic distributions and occurrence frequency analyzing mine microquake signal, realize effective monitoring index system, and because of the fault offset of mine microquake signal own less, seismic wave transmission range is limited, thus it is higher to require the positioning precision of mine microquake signal to compare common seismic signal.And the raising of positioning precision needs the primary wave moment to extract more fast and accurately.

At present the general mine microquake signal primary wave automatic Picking technology adopted there is no specially for the characteristic of mine microquake signal and requirement.Primary wave automatic Picking object is the boundary moment will determined in its signal between pure noise signal and useful signal, is all that this moment is determined in the change of amplitude, frequency and the phase place according to mine microquake signal usually.Traditional first break pickup method is mainly divided into two large classes: a class is the method based on seismologic record temporal characteristics, as extremum method (peakvalue's checking), method of difference; These class methods are more responsive to noise ratio, when the noise of seismologic record is more serious, are difficult to accurate first break picking.Another kind of method is the method based on seismologic record global feature, as correlation method; Although these class methods have good inhibiting effect to noise, be subject to the impact of the factor such as similarity between seismic trace, for complicated seismologic record, the precision of first break pickup also can be affected.

Up to the present, proposed the method for many first break pickup, as manually picked up method, correlation method, energy ratio function, peak amplitude method, FRACTAL DIMENSION method and neural network etc.

Artificial pickup method is simple, but affects comparatively large by human factor and subjective factor, easily introduces personal error, can directly cause resultant error to increase.

Gelchinsky and Shtivelman proposes a kind of method that neighboring track carries out cross-correlation, and it supposes that the pulse shape in each road does not change.Correlation method is comparatively large by continuing to affecting of first arrival, and very close with the choice relation of wavelet, and window scope when simultaneously requiring to select suitable, these have certain difficulty in Practical Project.

Hatherrly proposes Linear Least Square forecasting techniques and corrects with flex point the method combined, and he proposes first to identify first peak value and flex point, the statistics difference then both estimation.Huang Cheng etc. adopt statistical method seismic first break record to be divided into signal and noise two parts, and make the difference between this two parts statistical nature be maximum.It is comparatively large that statistical nature method receives the impact of seismic waveshape similarity, can cause certain influence to precision.

Signal energy in the ratio of energy life cycle and total time window energy ratio, more responsive to first arrival, subsequent event attenuation ratio is very fast, so the maximum of points of ratio is done suitable time shift as the approximate value of first arrival, is initial time.Coppens propose different size time window in carry out the method for energy comparison.Jiang Yule etc. propose same polarity the ratio of energy, the energy ratio function namely improved.Because the antijamming capability of the ratio of energy is good not enough, so the initial time picked up for the area of first arrival waveform generation 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. proposes a kind of first arrival detection algorithm based on FRACTAL DIMENSION, and the method is along with signal occurs that the feature that its fractal dimension changes determines seismic trace first arrival based on seismic trace.But when its pair window and step-length choose very responsive, careless slightly will be serious affect its result.

Neural network utilizes multiparameter feature to carry out pattern-recognition, makes full use of temporal characteristics and the global feature of seismologic record.Because neural network not only has parallel processing, self-organization self-learning capability, and there is height robustness, fault-tolerance and mapping highly, calculating and classification capacity.The East Sea, the village etc. adopt regards seismologic record first break pickup as a mode identification procedure, and make full use of temporal characteristics and the global feature of seismologic record, carry out seismologic record first break pickup by Artificial Neural Network, it can obtain good actual effect.But it is too high that its major defect is algorithm complexity, search needs certain hour.

Seismic signal by Wavelet Multiresolution Decomposition, can be separated, stress release treatment effectively, is conducive to the precision that FRACTAL DIMENSION and neural network improve first break picking.Luo Guang proposes the first break picking method of modified based on wavelet transformation, and Yang Junfeng proposes three-component seismic phase method of identification, energy factors method based on wavelet transformation.But it still needs manually to choose data segment to carry out to pick up to reduce the pick-up time sometimes.

In addition, there is certain methods also to rely on contrast between this road and its shortcut, although these class methods have certain suppression to noise, be subject to the impact of the factor such as similarity between seismic trace, for complicated seismologic record, the precision of first break pickup also can be affected.

In sum, seismic signal is the time series of an one dimension, because this sequence merely illustrates the relation between time and amplitude, and has the interference of noise signal, causes sizable difficulty to the pickup directly carrying out seismic first breaks.Solution problems can adopt intelligent algorithm as described above, but because intelligent algorithm generally has higher algorithm complex, be not suitable for needing fast and have the mine microquake environment of higher positioning accuracy, and the present invention preferably resolves this problem of pickup to the primary wave time, and there is lower algorithm complex.

Summary of the invention

The object of the invention is for Problems existing in prior art, a kind of mine microquake signal primary wave moment extracting method based on correlation analysis is provided, solves primary wave time intelligence identification in mine microquake signal transacting, mine microquake signal alignment and judge mine microquake signal whether from the problem of same focus.

Realize the technical scheme of the object of the invention: mine microquake signal primary wave moment extracting method of the present invention comprises five steps: one is read mine microquake signal; Two is the related functions solving two mine microquake signals; Three is the related functions according to solving, and obtains mistiming during maximal correlation; Four is be shifted to one of them signal according to the mistiming calculated; Five be to alignment after two mine microquake signals solve its related coefficient, judge its correlativity; Concrete grammar step is as follows:

(1) mine microquake signal is read: two mine microquake signal-obtainings are entered system, definition x _nbe a mine microquake burst, y _nfor another mine microquake burst, require that two sequences must be isometric, setting its sequence length is L, and two signal sampling frequencies are f _s;

(2) solve the related function of two mine microquake signals: solve mine microquake signal correction function according to following formula, when related function obtains maximal value, calculate the mistiming between mine microquake signal according to sampled point and sampling rate; According to the related function formulae discovery related function R of discrete signal sequence _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 related function solved, mistiming during maximal correlation is obtained: according to the mistiming calculated, mine microquake signal is alignd; Mistiming when calculating maximal correlation according to related function, sampled point and sampling rate, obtain related function maximal value R _maxfor:

R _max＝max[|R _xy(m)|]

(4) according to the mistiming calculated, one of them signal is shifted: correlation analysis is carried out to the mine microquake signal after alignment, and sets threshold value to judge whether mine microquake signal comes from same focus; Obtain the total number of sample points of initial time to the related function maximum absolute value moment, be denoted as N, then time deviation T _offset:

T _offset＝N/f _s

(5) its related coefficient is solved to two mine microquake signals after alignment, judges its correlativity:

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 _nsequence overall, Y is y _nsequence overall, E is mathematical expectation; The correlation matrix principal diagonal obtained represents autocorrelation, and it is always 1; Counter-diagonal is the correlativity of two signals, and absolute value is larger, and correlativity is better, and its span is [0,1]; Cut off value can be any value between 0 to 1 according to actual conditions different set; Final determining program is the need of end, if need terminate, ending method, run, return reading data step if need continue, reading new data carries out displacement alignment and correlativity compares.

Beneficial effect, owing to have employed such scheme, Intelligent Recognition two signal time is poor, realize automatic shift alignment and judge mine microquake signal correlation, realizes Intelligent Recognition mine microquake signal arrival time difference; Realize mine microquake signal intelligence alignment so that store, treatment and analysis; Compare two mine microquake signals by correlativity and whether come from same focus; Solve Intelligent Recognition mine microquake signal arrival time difference and judge mine microquake signal whether from the problem of same focus.

Described mine microquake signal primary wave moment extracting method is owing to adopting the method for correlation analysis, algorithm is easy, the methods such as the fuzzy recognition method in background technology, time domain fractal dimension method, neural net method, small wave converting method of comparing have less program complexity, be more suitable for being applied to the use that there is a large amount of vibration data, promptness required to high field conjunction, the methods such as artificial cognition method, the ratio of energy of comparing decrease the error that subjective factor produces.

Accompanying drawing explanation

Accompanying drawing described herein is used to provide a further understanding of the present invention, and form a application's part, schematic description and description of the present invention, for explaining the present invention, does not form inappropriate limitation of the present invention.

Fig. 1 is the process flow diagram of the inventive method.

Fig. 2 is the system architecture 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 the alignment effect analogous diagram that is shifted in the inventive method embodiment.

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

Embodiment

Below in conjunction with embodiment in accompanying drawing, the invention will be further described:

Mine microquake signal primary wave moment extracting method comprises five steps: one is read mine microquake signal; Two is the related functions solving two mine microquake signals; Three is the related functions according to solving, and obtains mistiming during maximal correlation; Four is be shifted to one of them signal according to the mistiming calculated; Five be to alignment after two mine microquake signals solve its related coefficient, judge its correlativity; Concrete grammar step is as follows:

(1) mine microquake signal is read: two mine microquake signal-obtainings are entered system, definition x _nbe a mine microquake burst, y _nfor another mine microquake burst, require that two sequences must be isometric, setting its sequence length is L, and two signal sampling frequencies are f _s;

(2) solve the related function of two mine microquake signals: solve mine microquake signal correction function according to following formula, when related function obtains maximal value, calculate the mistiming between mine microquake signal according to sampled point and sampling rate; According to the related function formulae discovery related function R of discrete signal sequence _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 related function solved, mistiming during maximal correlation is obtained: according to the mistiming calculated, mine microquake signal is alignd; Mistiming when calculating maximal correlation according to related function, sampled point and sampling rate, obtain related function maximal value R _maxfor:

R _max＝max[|R _xy(m)|]

(4) according to the mistiming calculated, one of them signal is shifted: correlation analysis is carried out to the mine microquake signal after alignment, and sets threshold value to judge whether mine microquake signal comes from same focus; Obtain the total number of sample points of initial time to the related function maximum absolute value moment, be denoted as N, then time deviation T _offset:

T _offset＝N/f _s

(5) its related coefficient is solved to two mine microquake signals after alignment, judges its correlativity:

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 _nsequence overall, Y is y _nsequence overall, E is mathematical expectation; The correlation matrix principal diagonal obtained represents autocorrelation, and it is always 1; Counter-diagonal is the correlativity of two signals, and absolute value is larger, and correlativity is better, and its span is [0,1]; Cut off value can be any value between 0 to 1 according to actual conditions different set.Final determining program is the need of end, if need terminate, ending method, run, return reading data step if need continue, reading new data carries out displacement alignment and correlativity compares.

The present invention utilizes related function to carry out correlation analysis to signal, thus obtains the primary wave moment, carries out signal automatic aligning and judge that whether mine microquake signal is from same focus.

Accompanying drawing 1 is the process flow diagram of the inventive method, the described mine microquake signal primary wave moment extracting method based on correlation analysis, and complete moment extraction, signal alignment according to the flow process of accompanying drawing 1 and judge that whether signal is from same focus, its basic step is as follows:

Read mine microquake signal x _n, mine microquake signal y _n, its sample frequency is f _s;

Related function formulae discovery related function according to discrete signal sequence:

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

Mistiming when calculating maximal correlation according to related function, sampled point and sampling rate

Obtain related function maximal value:

R _max＝max[|R _xy(m)|] (2)

Obtain the total number of sample points of initial time to the related function maximum absolute value moment, be denoted as N

Then time deviation:

T _offset＝N/f _s (3)

According to the T that previous step calculates _offsetby two signal displacement alignment, so that next step calculates related coefficient;

The correlation matrix of two signals is asked 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 is mathematical expectation.The correlation matrix principal diagonal obtained represents autocorrelation, and it is always 1; Counter-diagonal is the correlativity of two signals, and absolute value is larger, and correlativity is better.Generally two signal correction absolute coefficient are microfacies pass between 0-0.3; For reality closes between 0.3-0.5; It is significant correlation between 0.5-0.8; It is height correlation between 0.8-1.0.By setting cut off value, the method judges whether two signals have high correlation, thus judge that whether it is from same focus, and this cut off value can be changed according to actual needs.

The described mine microquake signal primary wave moment extracting method based on correlation analysis, can be applicable in the seismic source location system of networking distributed mine, accompanying drawing 2 is this system architecture schematic diagram, is described in detail below in conjunction with the embodiment of accompanying drawing 2 to described method.In accompanying drawing 2, round dot is sensor node, comprises microseismic sensors, can monitor mine microquake signal.When vibrations occur focus, the sensor node closed on can collect mine microquake data.The waveform that accompanying drawing 3 is four different shock sensor records receiving mine microquake signal, wherein a, b, c tri-signals come from same focus, and d comes from another focus being different from other three signals.In the present embodiment, determine whether that from the cut off value of same focus be 0.5 according to great amount of samples and correlation experience setting.

First correlation analysis is carried out to a and b signal.Accompanying drawing 4 is a signal and b signal correction function and the alignment effect analogous diagram that is shifted in the inventive method embodiment.Calculate the related function of a and b signal according to formula (1) and map, the related function of trying to achieve specifically distributes and sees accompanying drawing 4.Related function maximal value R is obtained according to formula (2) _max=250.8038.A and b signal two signal time is calculated poor, T according to formula (3) _offset=-0.0393 (s).Two signal correction matrix of coefficients are calculated according to formula (4).In correlation matrix, principal diagonal is always 1, and counter-diagonal i.e. two signal correlations, and the larger then correlativity of absolute value is better.A and the b signal correction matrix of coefficients of trying to achieve is:

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

Taken absolute value by its correlation matrix counter-diagonal, obtain a signal and b signal correlation is 0.5462, be greater than the cut off value 0.4 of setting, then described method judges a signal from b signal from different focus.

Secondly correlation analysis is carried out to a and d signal.Accompanying drawing 5 is a signal and d signal correction function and the alignment effect analogous diagram that is shifted in the inventive method embodiment.Calculate the related function of a and d signal according to formula (1) and map, the related function of trying to achieve specifically distributes and sees accompanying drawing 5.Related function maximal value R is obtained according to formula (2) _max=31.5554.A and b signal two signal time is calculated poor, T according to formula (3) _offset=0.0533 (s).Two signal correction matrix of coefficients are calculated according to formula (4).A and the d signal correction matrix of coefficients of trying to achieve is:

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

Taken absolute value by its correlation matrix counter-diagonal, obtain a signal and d signal correlation is 0.0566, be less than the cut off value 0.4 of setting, then described method judges a signal from d signal from different focus.

Below by reference to the accompanying drawings the specific embodiment of the present invention is described; but these explanations can not be understood to limit scope of the present invention; protection scope of the present invention is limited by the claims of enclosing, and any change on the claims in the present invention basis is all protection scope of the present invention.

Claims (1)

1. based on a mine microquake signal primary wave moment extracting method for correlation analysis, it is characterized in that: mine microquake signal primary wave moment extracting method comprises five steps: one is read mine microquake signal; Two is the related functions solving two mine microquake signals; Three is the related functions according to solving, and obtains mistiming during maximal correlation; Four is be shifted to one of them signal according to the mistiming calculated; Five be to alignment after two mine microquake signals solve its related coefficient, judge its correlativity; Concrete grammar step is as follows:

(1) mine microquake signal is read: two mine microquake signal-obtainings are entered system, definition x _nbe a mine microquake burst, y _nfor another mine microquake burst, require that two sequences must be isometric, setting its sequence length is L, and two signal sampling frequencies are f _s;

(2) solve the related function of two mine microquake signals: solve mine microquake signal correction function according to following formula, when related function obtains maximal value, calculate the mistiming between mine microquake signal according to sampled point and sampling rate; According to the related function formulae discovery related function R of discrete signal sequence _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 related function solved, mistiming during maximal correlation is obtained: according to the mistiming calculated, mine microquake signal is alignd; Mistiming when calculating maximal correlation according to related function, sampled point and sampling rate, obtain related function maximal value R _maxfor:

R _max＝max[|R _xy(m)|]

(4) according to the mistiming calculated, one of them signal is shifted: correlation analysis is carried out to the mine microquake signal after alignment, and sets threshold value to judge whether mine microquake signal comes from same focus; Obtain the total number of sample points of initial time to the related function maximum absolute value moment, be denoted as N, then time deviation T _offset:

T _offset＝N/f _s

(5) its related coefficient is solved to two mine microquake signals after alignment, judges its correlativity:

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 _nsequence overall, Y is y _nsequence overall, E is mathematical expectation; The correlation matrix principal diagonal obtained represents autocorrelation, and it is always 1; Counter-diagonal is the correlativity of two signals, and absolute value is larger, and correlativity is better, and its span is [0,1]; Cut off value can be any value between 0 to 1 according to actual conditions different set; Final determining program is the need of end, if need terminate, ending method, run, return reading data step if need continue, reading new data carries out displacement alignment and correlativity compares.