CN111126439A - Method for generating coal mine underground two-dimensional mine earthquake imaging training data set - Google Patents

Method for generating coal mine underground two-dimensional mine earthquake imaging training data set Download PDF

Info

Publication number
CN111126439A
CN111126439A CN201911164767.7A CN201911164767A CN111126439A CN 111126439 A CN111126439 A CN 111126439A CN 201911164767 A CN201911164767 A CN 201911164767A CN 111126439 A CN111126439 A CN 111126439A
Authority
CN
China
Prior art keywords
wcmp
wacofmtr
wacofmtrpaclus
wavelet coefficient
wacofmtrpa
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.)
Withdrawn
Application number
CN201911164767.7A
Other languages
Chinese (zh)
Inventor
陈卿
刘冬冬
李克荣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xuzhou University of Technology
Original Assignee
Xuzhou University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xuzhou University of Technology filed Critical Xuzhou University of Technology
Priority to CN201911164767.7A priority Critical patent/CN111126439A/en
Publication of CN111126439A publication Critical patent/CN111126439A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computing Systems (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • Image Processing (AREA)

Abstract

The invention provides a coal mine underground two-dimensional mine earthquake imaging training data set generation method, which comprises the following steps: determining grid subdivision parameters of a target area and parameters of a seismic source and a detector; determining a wavelet decomposition scale; determining a target area wavelet coefficient vector format; determining the upper limit and the lower limit of the wavelet coefficient of the target area and the sampling interval; determining the splitting multiple of the matrix WaCofMtrPa; decomposing the WaCofMtrPa according to the splitting multiple NWCMP; generating all possible WaCofMtr according to the WaCofMtrPaClus, and obtaining seismic records through forward modeling; the method can provide a method of selecting a limited number of models in a known model space S that can represent the characteristics of the model space.

Description

Method for generating coal mine underground two-dimensional mine earthquake imaging training data set
Technical Field
The invention relates to a coal mine underground two-dimensional mineral earthquake imaging training data set generation method.
Background
The mine earthquake imaging can be used for obtaining information such as geological structure, stress distribution and the like of a target area under a coal mine, and is a basic technology for forecasting and forecasting underground disasters. The current mine earthquake imaging thinking is to equate the imaging problem to a multi-objective optimization problem, namely, an optimal target area model representation vector s is found so that a target function | d is achievedTrue-dTheo||2Minimum where dTrueFor measured seismic records, dTheoFor theoretical seismic records, s and dTheoThe relationship between is dTheoF(s) is a forward means such as a constant density acoustic wave equation and a variable density acoustic wave equation. Since the dimension of S is usually high, the variation range of the possible values of each dimension is also large, so the whole model space S (i.e. the space formed by all possible model representations) will be very large, which means that even if local optimization methods are used to find the best in the model space, the computation time is considerable. For example, the target region is subdivided into 1000 × 1000 grids, each grid having a range of wave velocities of only v1And v 22 possibilities, then the model space will be 21000000One possible value constitutes.
Different from the current mine earthquake imaging thought, the mine earthquake imaging based on the neural network uses a large amount of known dTrueS pairs of networks are trained to establish dTrueAnd s, inputting a new actual measurement seismic record after training is finished so as to obtain an imaging result. Compared with the traditional method, the neural network-based mineral earthquake imaging does not need forward calculation and iterative optimization, so that the inversion speed is extremely high, and the method is a promising inversion means. However, whether the neural network-based mineral earthquake imaging can obtain correct results depends on the quality of the training data set, namely a large number of correct d which can represent the spatial features of the modelTrueAnd s pairs, in actual conditions, the geological structure and stress distribution condition of a target area are difficult to obtain directly, so a large number of high-quality training data sets are not easy to obtainTheo as a real seismic record dTrueTo train the neural network. Obviously, the number of models selected from S should be as small as possible, and at the same time, should represent the features of S as much as possible, and the models selected directly from S cannot accommodate the contradiction between number and quality (for example, in the above example, even if only 2 possible values are set per grid wave velocity, the number of selected models will reach 21000000One).
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a method for generating a coal mine underground two-dimensional mine earthquake imaging training data set, which can provide a method for selecting a limited number of models which can represent model space characteristics in a known model space S.
In order to achieve the purpose, the invention provides a coal mine underground two-dimensional mine earthquake imaging training data set generation method, which comprises the following steps:
step 1: determining grid subdivision parameters of a target area and parameters of a seismic source and a detector;
selecting a square target area, and dividing the target area into m multiplied by m grids, wherein m is 2nN is an integer greater than 0; recording a wave velocity distribution matrix after subdivision is VS, wherein each element in VS represents the wave velocity in a grid of a target area, the wave velocity values in the same grid are the same, and recording parameters such as a seismic source position, a wavelet parameter and a detector position as SSP;
step 2: determining a wavelet decomposition scale;
recording the wavelet decomposition scale as WaDeSca, wherein the WaDeSca satisfies the condition that WaDeSca is more than or equal to 1 and less than or equal to n and is an integer, and selecting a Haar wavelet equivalent to a Daubechies wavelet with the vanishing moment of 1 as a wavelet decomposition base;
and step 3: determining a target area wavelet coefficient vector format;
recording wavelet coefficient vectors of the target region as WaCofMtr, wherein VS corresponds to WaCofMtr one by one due to the fact that the same object is only represented in different domains, namely one wave velocity distribution vector corresponds to one wavelet coefficient vector;
the WaCofMtr process obtained from VS is recorded as WaCofMtr ═ S _ TO _ W (VS), the reverse process is recorded as VS ═ W _ TO _ S: (WaCofMtr), the length of the vector WaCofMtr is consistent with the number of elements in VS, and both are 22n＝m2Except for the maximum scale WaDeSca, the wavelet coefficient of each scale consists of a horizontal wavelet coefficient vector, a vertical wavelet coefficient vector and an oblique wavelet coefficient vector, the lengths of the three vectors are consistent and are related to the current decomposition scale, and the length formula is as follows: 22(n-CWS)Wherein CWS is the current scale; maximum scale WaDesca contains a scale coefficient vector AWaDeScaAnd corresponding three wavelet coefficient vectors HWaDeSca、VWaDeScaAnd DWaDeScaThe lengths of the four vectors also satisfy equation 22(n-CWS)(ii) a Giving a coordinate to each element in WaCofMtr for identifying each element, from left to right, the coordinate vector corresponding to each element in WaCofMtr is WaCofMtrCod ═ 1,2,3, L, m2]；
And 4, step 4: determining the upper limit and the lower limit of the wavelet coefficient of the target area and the sampling interval;
recording the wavelet coefficient parameter matrix of the target area as WaCofMtrPa, and recording the 1 st row 1,2,3, L, n of the WaCofMtrPa matrixWCMP,L,m2Wavelet coefficient coordinates, for identifying which column in the WaCofMtr the current column parameters apply to, line 2 ul1,L,ulWCMP,L,
Figure BDA0002287134850000031
Upper limit for wavelet coefficient values, line 3 ll1,L,llWCMP,L,
Figure BDA0002287134850000032
For wavelet coefficient value lower bound, line 4 it1,L,itWCMP,L,
Figure BDA0002287134850000033
Line 5 nu for wavelet coefficient sampling interval1,L,nuWCMP,L,
Figure BDA0002287134850000034
The subscript WCMP satisfies the condition that WCMP is more than or equal to 1 and less than or equal to m2And is an integer;
and 5: determining the splitting multiple of the matrix WaCofMtrPa;
let WaCofMtrPa be NWCMP, which is the division multiple of WaCofMtrPa
Figure BDA0002287134850000035
Wherein n isNIs an integer greater than 0, and NWCMP is required to satisfy NWCMP < mnw;
step 6: decomposing the WaCofMtrPa according to the splitting multiple NWCMP;
the set of matrixes after WaCofMtrPa decomposition is WaCofMtrPaClus, and the number of matrixes in the WaCofMtrPaClus is nWCMPCInitializing WaCofMtrPaClus, storing the matrix WaCofMtrPa into a matrix set WaCofMtrPaClus, wherein only 1 matrix exists in the matrix set, and judging whether i is smaller than n or not when an initialization count value i is 0NIf it is less than nNExecuting i to i +1, increasing the number of matrixes in the WaCofMtrPaClus to 2 times, and judging whether i is smaller than n againNIf it is less than nNAnd executing i to i +1 again, increasing the number of matrixes in the WaCofMtrPaClus to 2 times, and sequentially circulating until i is larger than or equal to nNCompleting the decomposition process and returning to WaCofMtrPaClus;
and 7: generating all possible WaCofMtr according to the WaCofMtrPaClus, and obtaining seismic records through forward modeling;
the procedure of WaCofMtr obtained from WaCofMtrPaClus is denoted as WCMP _ TO _ WCM, and then:
{WaCofMtr1,WaCofMtr2,…,WaCofMtrmnw}＝WCMP_TO_WCM(WaCofMtrPaClus)
the set of models to be selected from the model space S is then:
{VS1,VS2,…,VSmnw}＝W_TO_S({WaCofMtr1,WaCofMtr2,…,WaCofMtrmnw})；
recording that the process of obtaining the seismic records through normal evolution of the constant density acoustic wave equation is FW, the theoretical seismic record set corresponding to each model is as follows:
{TheSei1,TheSei2,…,TheSeimnw}＝FW({VS1,VS2,…,VSmnw})；
since WaCofMtrPaClus is a collection of matrices, the above process of generating forward seismic records may be performed in parallel on a multi-core computer or cluster of computers to increase computational speed.
In step 4, the WCMP column in WaCofMtrPa represents the sampling range of the WCMP column element in WaCofMtr, and the rule is as follows:
if ulWCMP＝llWCMPThen nu WCMP1, indicates that there is only one possible value of the WCMP column element in WaCofMtr, and its value is equal to ulWCMP；
If ulWCMP＞llWCMPThen, then
Figure BDA0002287134850000041
Denotes the WCMP column element in WaCofMtr by itWCMPFor intervals from an upper limit ulWCMPAt the beginning, at the upper limit ulWCMPTo the lower limit llWCMPValue within the range, sampling interval itWCMPSatisfy itWCMPGreater than 0, and an upper limit ulWCMPNot less than the lower limit llWCMP。
The invention provides a method for selecting a limited number of models capable of representing model space characteristics from a known model space S, which can conveniently solve the problem of master model forward modeling by utilizing the parallel capability of a multi-core computer or a computer cluster.
Drawings
FIG. 1 is a schematic representation of the WaCofMtr format of the present invention;
FIG. 2 is a diagram of a wavelet parameter matrix format in the present invention;
FIG. 3 is a schematic view of the WaCofMtrPa decomposition process in the present invention;
FIG. 4 is a schematic diagram of the splitting of a wavelet coefficient parameter matrix in the present invention.
Detailed Description
The present invention is further described below.
A coal mine underground two-dimensional mine earthquake imaging training data set generation method comprises the following steps:
step 1: determining grid subdivision parameters of a target area and parameters of a seismic source and a detector;
selecting a square target area for subsequent decomposition, and dividing the target area into m × m grids, wherein m is 2nN is an integer greater than 0; recording the wave velocity after subdivisionThe distribution matrix is VS, each element in VS represents the wave velocity in the grid of the target area, the wave velocity values in the same grid are the same, and the parameters such as the seismic source position, the wavelet parameters, the detector position and the like are recorded as SSP;
step 2: determining a wavelet decomposition scale;
recording the wavelet decomposition scale as WaDeSca, wherein the WaDeSca satisfies the condition that WaDeSca is more than or equal to 1 and less than or equal to n and is an integer, and selecting a Haar wavelet equivalent to a Daubechies wavelet with the vanishing moment of 1 as a wavelet decomposition base;
and step 3: determining a target area wavelet coefficient vector format;
recording wavelet coefficient vectors of the target region as WaCofMtr, wherein VS corresponds to WaCofMtr one by one due to the fact that the same object is only represented in different domains, namely one wave velocity distribution vector corresponds to one wavelet coefficient vector;
the WaCofMtr process from VS is recorded as WaCofMtr ═ S _ TO _ W (VS), the reverse process is recorded as VS ═ W _ TO _ S (WaCofMtr), the WaCofMtr format is specified as shown in FIG. 1, the length of the vector WaCofMtr is consistent with the number of elements in VS, and the lengths are all 22n＝m2In addition to the maximum dimension WaDeSca, the wavelet coefficients of each dimension are composed of a horizontal wavelet coefficient vector, a vertical wavelet coefficient vector, and an oblique wavelet coefficient vector (e.g., H at dimension 1 as shown in FIG. 1)1、V1And D1) The lengths of the three vectors are consistent and related to the current decomposition scale, and the length formula is as follows: 22(n-CWS)Wherein CWS is the current scale; the maximum scale WaDeSca contains a scale coefficient vector AWaDeScaAnd corresponding three wavelet coefficient vectors HWaDeSca、VWaDeScaAnd DWaDeScaThe lengths of the four vectors also satisfy equation 22(n-CWS)(ii) a Giving a coordinate to each element in WaCofMtr for identifying each element, from left to right, the coordinate vector corresponding to each element in WaCofMtr is WaCofMtrCod ═ 1,2,3, L, m2]；
And 4, step 4: determining the upper limit and the lower limit of the wavelet coefficient of the target area and the sampling interval;
the wavelet coefficient parameter matrix of the target region is WaCofMtrPa, the format of which is shown in FIG. 2, and the 1 st row 1,2,3, L, n of the WaCofMtrPa matrixWCMP,L,m2Wavelet coefficient coordinates, for identifying which column in the WaCofMtr the current column parameters apply to, line 2 ul1,L,ulWCMP,L,
Figure BDA0002287134850000051
Upper limit for wavelet coefficient values, line 3 ll1,L,llWCMP,L,
Figure BDA0002287134850000052
For wavelet coefficient value lower bound, line 4 it1,L,itWCMP,L,
Figure BDA0002287134850000053
Line 5 nu for wavelet coefficient sampling interval1,L,nuWCMP,L,
Figure BDA0002287134850000054
The subscript WCMP satisfies the condition that WCMP is more than or equal to 1 and less than or equal to m2And is an integer;
and 5: determining the splitting multiple of the matrix WaCofMtrPa;
from WaCofMtrPa, if the models are generated in the manner indicated by WaCofMtrPa, the number of models is
Figure BDA0002287134850000061
Mnw will be a large value even if the lower scale coefficients are set to 0 using the model's wavelet domain sparsity, so it is necessary to consider how to solve the problem of large number of forward models using the parallel capability of the computer's multi-core or computer clusters.
Let WaCofMtrPa be NWCMP, which is the division multiple of WaCofMtrPa
Figure BDA0002287134850000062
Wherein n isNIs an integer greater than 0, and NWCMP is required to satisfy NWCMP < mnw;
step 6: decomposing the WaCofMtrPa according to the splitting multiple NWCMP;
let WaCofMtrPa decomposed matrix set be WaCofMtrPaClus,the number of matrixes in the WaCofMtrPaClus is nWCMPCThe decomposition flow is as shown in fig. 3, WaCofMtrPaClus is initialized, the initialization count value i is 0, and whether i is smaller than n is judgedNIf it is less than nNExecuting i to i +1, increasing the number of matrixes in the WaCofMtrPaClus to 2 times, and judging whether i is smaller than n againNIf it is less than nNAnd executing i to i +1 again, increasing the number of matrixes in the WaCofMtrPaClus to 2 times, and sequentially circulating until i is larger than or equal to nNCompleting the decomposition process and returning to WaCofMtrPaClus;
in fig. 3, initializing WaCofMtrPaClus means storing the matrix WaCofMtrPa into a matrix set WaCofMtrPaClus, where there are only 1 matrix in the matrix set.
In fig. 3, the specific operation of increasing the number of matrices in WaCofMtrPaClus by 2 times is as follows:
selecting the mnw largest matrix in WaCofMtrPaClus, recording the matrix as WaCofMtrPa', and selecting nu in WaCofMtrPaWCMPThe largest column, assuming this column is the nth columnWCMPThe WaCofMtrPa 'can be split into 2 matrixes by taking the middle point of the upper limit and the lower limit of the WaCofMtrPa' as a boundary, and a wavelet coefficient parameter matrix splitting schematic diagram is shown in FIG. 4;
thus, the number of matrices in WaCofMtrPaClus will be nWCMPCIs changed into nWCMPC+1, repeat the above steps until nWCMPCTo 2nWCMPC。
And 7: generating all possible WaCofMtr according to the WaCofMtrPaClus, and obtaining seismic records through forward modeling;
the procedure of WaCofMtr obtained from WaCofMtrPaClus is denoted as WCMP _ TO _ WCM, and then:
{WaCofMtr1,WaCofMtr2,…,WaCofMtrmnw}＝WCMP_TO_WCM(WaCofMtrPaClus)
the set of models to be selected from the model space S is then:
{VS1,VS2,…,VSmnw}＝W_TO_S({WaCofMtr1,WaCofMtr2,…,WaCofMtrmnw})；
recording that the process of obtaining the seismic records through normal evolution of the constant density acoustic wave equation is FW, the theoretical seismic record set corresponding to each model is as follows:
{TheSei1,TheSei2,…,TheSeimnw}＝FW({VS1,VS2,…,VSmnw})；
since WaCofMtrPaClus is a collection of matrices, the above process of generating forward seismic records may be performed in parallel on a multi-core computer or cluster of computers to increase computational speed.
In step 4, the WCMP column in WaCofMtrPa represents the sampling range of the WCMP column element in WaCofMtr, and the rule is as follows:
if ulWCMP＝llWCMPThen nu WCMP1, indicates that there is only one possible value of the WCMP column element in WaCofMtr, and its value is equal to ulWCMP。
If ulWCMP＞llWCMPThen, then
Figure BDA0002287134850000071
Denotes the WCMP column element in WaCofMtr by itWCMPFor intervals from an upper limit ulWCMPAt the beginning, at the upper limit ulWCMPTo the lower limit llWCMPValue within the range, sampling interval itWCMPSatisfy itWCMPGreater than 0, and an upper limit ulWCMPNot less than the lower limit llWCMP。
The invention provides a method for selecting a limited number of models capable of representing model space characteristics from a known model space S, which can conveniently solve the problem of master model forward modeling by utilizing the parallel capability of a multi-core computer or a computer cluster.

Claims (2)

1. A coal mine underground two-dimensional mine earthquake imaging training data set generation method is characterized by comprising the following steps:
step 1: determining grid subdivision parameters of a target area and parameters of a seismic source and a detector;
is selected byA square target area, subdivided into a grid of m × m, where m is 2nN is an integer greater than 0; recording a wave velocity distribution matrix after subdivision is VS, wherein each element in VS represents the wave velocity in a grid of a target area, the wave velocity values in the same grid are the same, and recording parameters such as a seismic source position, a wavelet parameter and a detector position as SSP;
step 2: determining a wavelet decomposition scale;
recording the wavelet decomposition scale as WaDeSca, wherein the WaDeSca satisfies the condition that WaDeSca is more than or equal to 1 and less than or equal to n and is an integer, and selecting a Haar wavelet equivalent to a Daubechies wavelet with the vanishing moment of 1 as a wavelet decomposition base;
and step 3: determining a target area wavelet coefficient vector format;
recording wavelet coefficient vectors of the target region as WaCofMtr, wherein VS corresponds to WaCofMtr one by one due to the fact that the same object is only represented in different domains, namely one wave velocity distribution vector corresponds to one wavelet coefficient vector;
the WaCofMtr process obtained from VS is recorded as WaCofMtr ═ S _ TO _ W (VS), the reverse process is recorded as VS ═ W _ TO _ S: (WaCofMtr), the length of the vector WaCofMtr is consistent with the number of elements in VS, and both are 22n＝m2Except for the maximum scale WaDeSca, the wavelet coefficient of each scale consists of a horizontal wavelet coefficient vector, a vertical wavelet coefficient vector and an oblique wavelet coefficient vector, the lengths of the three vectors are consistent and are related to the current decomposition scale, and the length formula is as follows: 22(n-CWS)Wherein CWS is the current scale; the maximum scale WaDeSca contains a scale coefficient vector AWaDeScaAnd corresponding three wavelet coefficient vectors HWaDeSca、VWaDeScaAnd DWaDeScaThe lengths of the four vectors also satisfy equation 22(n-CWS)(ii) a Giving a coordinate to each element in WaCofMtr for identifying each element, from left to right, the coordinate vector corresponding to each element in WaCofMtr is WaCofMtrCod ═ 1,2,3, L, m2]；
And 4, step 4: determining the upper limit and the lower limit of the wavelet coefficient of the target area and the sampling interval;
recording the wavelet coefficient parameter matrix of the target area as WaCofMtrPa, and recording the 1 st row 1,2,3, L, n of the WaCofMtrPa matrixWCMP,L,m2Is a waveletCoefficient coordinates to identify which column, line 2 ul, in the WaCofMtr the current column parameters apply to1,L,ulWCMP,L,
Figure FDA0002287134840000011
Upper limit for wavelet coefficient values, line 3 ll1,L,llWCMP,L,
Figure FDA0002287134840000012
For wavelet coefficient value lower bound, line 4 it1,L,itWCMP,L,
Figure FDA0002287134840000013
Line 5 nu for wavelet coefficient sampling interval1,L,nuWCMP,L,
Figure FDA0002287134840000014
The subscript WCMP satisfies the condition that WCMP is more than or equal to 1 and less than or equal to m2And is an integer;
and 5: determining the splitting multiple of the matrix WaCofMtrPa;
let WaCofMtrPa be NWCMP, which is the division multiple of WaCofMtrPa
Figure FDA0002287134840000021
Wherein n isNIs an integer greater than 0, and NWCMP is required to satisfy NWCMP < mnw;
step 6: decomposing the WaCofMtrPa according to the splitting multiple NWCMP;
the set of matrixes after WaCofMtrPa decomposition is WaCofMtrPaClus, and the number of matrixes in the WaCofMtrPaClus is nWCMPCInitializing WaCofMtrPaClus, storing the matrix WaCofMtrPa into a matrix set WaCofMtrPaClus, wherein only 1 matrix exists in the matrix set, and judging whether i is smaller than n or not when an initialization count value i is 0NIf it is less than nNExecuting i to i +1, increasing the number of matrixes in the WaCofMtrPaClus to 2 times, and judging whether i is smaller than n againNIf it is less than nNAnd executing i +1 again, and increasing the number of matrixes in the WaCofMtrPaClus to 2 timesSequentially circulating until i is more than or equal to nNCompleting the decomposition process and returning to WaCofMtrPaClus;
and 7: generating all possible WaCofMtr according to the WaCofMtrPaClus, and obtaining seismic records through forward modeling;
the procedure of WaCofMtr obtained from WaCofMtrPaClus is denoted as WCMP _ TO _ WCM, and then:
{WaCofMtr1,WaCofMtr2,…,WaCofMtrmnw}＝WCMP_TO_WCM(WaCofMtrPaClus)
the set of models to be selected from the model space S is then:
{VS1,VS2,…,VSmnw}＝W_TO_S({WaCofMtr1,WaCofMtr2,…,WaCofMtrmnw})；
recording that the process of obtaining the seismic records through normal evolution of the constant density acoustic wave equation is FW, the theoretical seismic record set corresponding to each model is as follows:
{TheSei1,TheSei2,…,TheSeimnw}＝FW({VS1,VS2,…,VSmnw})；
since WaCofMtrPaClus is a collection of matrices, the above process of generating forward seismic records may be performed in parallel on a multi-core computer or cluster of computers to increase computational speed.
2. The method for generating the two-dimensional mine seismic imaging training data set under the coal mine tunnel according to claim 1, wherein in step 4, the WCMP column in WaCofMtrPa represents the sampling range of the WCMP column element in WaCofMtr, and the rule is as follows:
if ulWCMP＝llWCMPThen nuWCMP1, indicates that there is only one possible value of the WCMP column element in WaCofMtr, and its value is equal to ulWCMP；
If ulWCMP＞llWCMPThen, then
Figure FDA0002287134840000031
Denotes the WCMP column element in WaCofMtr by itWCMPIs a spaceFrom the upper limit ulWCMPAt the beginning, at the upper limit ulWCMPTo the lower limit llWCMPValue within the range, sampling interval itWCMPSatisfy itWCMPGreater than 0, and an upper limit ulWCMPNot less than the lower limit llWCMP。
CN201911164767.7A 2019-11-25 2019-11-25 Method for generating coal mine underground two-dimensional mine earthquake imaging training data set Withdrawn CN111126439A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911164767.7A CN111126439A (en) 2019-11-25 2019-11-25 Method for generating coal mine underground two-dimensional mine earthquake imaging training data set

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911164767.7A CN111126439A (en) 2019-11-25 2019-11-25 Method for generating coal mine underground two-dimensional mine earthquake imaging training data set

Publications (1)

Publication Number Publication Date
CN111126439A true CN111126439A (en) 2020-05-08

Family

ID=70496585

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911164767.7A Withdrawn CN111126439A (en) 2019-11-25 2019-11-25 Method for generating coal mine underground two-dimensional mine earthquake imaging training data set

Country Status (1)

Country Link
CN (1) CN111126439A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113944511A (en) * 2021-09-28 2022-01-18 武汉理工大学 Multi-stope environment generation method for underground mine

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113944511A (en) * 2021-09-28 2022-01-18 武汉理工大学 Multi-stope environment generation method for underground mine

Similar Documents

Publication Publication Date Title
Wang et al. IK-SVD: dictionary learning for spatial big data via incremental atom update
Tao et al. Bayesian tensor approach for 3-D face modeling
CN104122585B (en) Seismic forward simulation method based on elastic wave field resolution of vectors and low-rank decomposition
Jouvet et al. Deep learning speeds up ice flow modelling by several orders of magnitude
EA004551B1 (en) Method for seismic facies interpretation using textural analysis and neutral networks
CN110031895B (en) Multipoint geostatistical stochastic inversion method and device based on image stitching
CN114724012B (en) Tropical unstable wave early warning method and device based on space-time cross-scale attention fusion
CN114117906A (en) Multi-scale unsupervised seismic wave velocity inversion method based on observation data self-encoding
CN109272029B (en) Well control sparse representation large-scale spectral clustering seismic facies partitioning method
Nama et al. Parameters optimization of geotechnical problem using different optimization algorithm
US20110282636A1 (en) Variable Resolution Uncertainty Expert System for Digital Bathymetry Database
Babovic et al. Error correction of a predictive ocean wave model using local model approximation
Homma et al. A physics-based Monte Carlo earthquake disaster simulation accounting for uncertainty in building structure parameters
CN111025385B (en) Seismic data reconstruction method based on low rank and sparse constraint
CN114970290B (en) Ground transient electromagnetic method parallel inversion method and system
CN115238565A (en) Resistivity model reconstruction network training method, electromagnetic inversion method and device
Dietrich et al. A fast and exact method for multidimensional Gaussian stochastic simulations: Extension to realizations conditioned on direct and indirect measurements
CN113780346B (en) Priori constraint classifier adjustment method, system and readable storage medium
CN111126439A (en) Method for generating coal mine underground two-dimensional mine earthquake imaging training data set
CN112100906B (en) Data-driven large-scale density modeling method, computing device and storage medium
CN112285776A (en) Seismic velocity automatic picking method based on deep learning
CN112444850A (en) Seismic data velocity modeling method, storage medium, and computing device
CN113238284B (en) Gravity and magnetic fast forward modeling method
Adam et al. Adaptive Haar wavelets for the angular discretisation of spectral wave models
Lehmann et al. Seismic hazard analysis with a Factorized Fourier Neural Operator (F-FNO) surrogate model enhanced by transfer learning

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
WW01 Invention patent application withdrawn after publication

Application publication date: 20200508

WW01 Invention patent application withdrawn after publication