Three-dimensional geological modeling method and device for coal mine profile data and storage medium

Abstract

The invention discloses a three-dimensional geologic modeling method and device for coal mine profile data and a storage medium, and the three-dimensional geologic modeling method suitable for a coal mine is found based on related parameters in a Kriging interpolation method by adopting an appropriate method and process. Dividing the spatial data point set into a plurality of verification sets and a plurality of training sets; determining the value range of each parameter of the data point and the calculation step length, calculating each parameter of the training set to obtain a calculation result, and performing verification analysis on the calculation result by using the verification set to obtain an analysis result; performing distribution control on an analysis result, calculating an evaluation value of each parameter under the training set, analyzing each average value under different training sets, and searching a minimum average value; and obtaining the latest evaluation value and each parameter corresponding to the latest evaluation value based on the minimum average value, wherein each parameter is the optimal parameter.

Description

Technical Field

[0001] The invention relates to mine exploration technology, and in particular to a three-dimensional geological modeling method, device and storage medium for coal mine profile data. Background Art

[0002] With the development of 3D modeling technology and mine exploration technology, the parameter selection of interpolation methods in 3D geological modeling in coal mines is particularly important. The same set of data using different parameters will produce completely different results. Kriging interpolation is a widely used 3D geological modeling method. Its core lies in the analysis of variograms. The focus of the range function study is on the three parameters of primary range, secondary range, and nugget value. How to determine these three parameters directly affects the accuracy of 3D geological modeling. At present, the empirical method is mainly used to determine the three parameters. The subjective factors are relatively large, which affects the final scientificity and objectivity of 3D geological modeling.

[0003] Therefore, in the actual work of coal mines, due to the large data limitations, the data near the tunnels are dense and the data within the working surface are sparse, which directly affects the choice during interpolation. Researchers often make choices based on experience during actual use, resulting in a large number of human subjective factors on the basis of known data, which affects the ultimate scientificity and objectivity of three-dimensional geological modeling and ultimately leads to errors.

[0004] In view of this, the present invention is proposed. Summary of the invention

[0005] The main purpose of the present invention is to disclose a three-dimensional geological modeling method, device and storage medium for coal mine profile data, which are used to solve the problem that human subjective factors in the prior art affect the ultimate scientificity and objectivity of three-dimensional geological modeling.

[0006] To achieve the above object, according to one aspect of the present invention, a three-dimensional geological modeling method for coal mine profile data is provided, and the following technical scheme is adopted:

[0007] Determine the spatial data point set, and divide the spatial data point set into several validation sets and several training sets; determine the value range and calculation step of each parameter of the data point, calculate each parameter of the training set, obtain the calculation result, and use the validation set to verify and analyze the calculation result to obtain the analysis result; use distribution control for the analysis result, calculate the size of the evaluation value of each parameter under the training set, analyze the average values ​​under different training sets, and find the minimum average value; obtain the latest evaluation value and the parameters corresponding to the latest evaluation value based on the minimum average value, and the parameters are the optimal parameters; generate contour maps and models at each level based on the optimal parameters, and the calculation formula is as follows:

[0008]

[0009] Among them, Y*(u) is the unknown point information to be obtained, Y(ui) is the known point data, λ i is the weight of the point, and the sum of the weights is 1.

[0010] Furthermore, the determination of the spatial data point set and the division of the spatial data point set into a number of validation sets and a number of training sets include: Step 1: Collect and organize the known information of the coal mine profile, perform statistics in layers, and organize it into a (x, y, z) spatial coordinate format. The shortest straight-line distance of the data distribution in the study area is L min ; The longest straight-line distance of data distribution in the study area is L max ; Step 2: Randomly screen the acquired data and count the spatial coordinate information. The total data point set is N, with a total of n data points; Step 3: Based on the data point set N, the data points are numbered as N1, N2...N n , N1 data is used as validation set V1, and the remaining n-1 data points are used as training set T1. The training set participates in the calculation, and the validation set participates in the validation analysis. Similarly, N2 data is removed as validation set V2, and the remaining n-1 data points are used as training set T2; n groups of validation set V are formed in sequence n , training set T n .

[0011] Furthermore, the parameter value range and calculation step size of the data point are determined, the parameters of the training set are calculated to obtain the calculation results, and the calculation results are verified and analyzed with the verification set to obtain the analysis results, including: Step 4: The mean absolute error MAE1, the root mean square error RMSE1 and the mean square error MSE1 are calculated for the verification set V1 to obtain the mean absolute error verification set UMAE1, the root mean square error verification set URMSE1 and the mean square error verification set UMSE1; Step 5: The mean absolute error verification set UM AE1, root mean square error verification set URMSE1 and mean square error verification set UMSE1 are normalized in sequence, and the three normalized sets are obtained as the mean absolute error verification normalized value NUMAE1, the root mean square error verification normalized value NURMSE1 and the mean square error verification normalized value NUMSE1. The evaluation value P1 is the sum of the three after normalization, P1 = NUMAE1 + NURMSE1 + NUMSE1; Step 6: Execute step 5 in sequence, and n groups of evaluation values ​​P1, P2...P can be calculated in sequence respectively. n , P n =NUMAEn+NURMSEn+NUMSEn.

[0012] Furthermore, the analysis results are distributed controlled, and the evaluation values ​​of the parameters under the training set are calculated, and the average values ​​under different training sets are analyzed to find the minimum average value. Step 7: Calculate the training set T1 using the main variable range MV, and the main variable range MV takes the maximum value L max , the minimum value is L min , the main variable range MV takes an integer and calculates the main variable range step length S MV =1, and the main variable range MV data set MV1 is obtained; Step 8: Based on the main variable range MV data set MV1, the training set T1 is calculated using the secondary variable range SV, and the secondary variable range SV takes the maximum value L max , the minimum value is L min , the secondary variable range SV takes an integer, and the main variable range step length S is calculated SV =1, the secondary range SV is less than the primary range MV, and the secondary range SV data set SV1 is obtained; Step 9: Based on the secondary range data set SV1, the nugget value N is calculated and analyzed. The minimum value of the nugget value N is 0 and the maximum value is 1. The nugget value step length S is calculated. N =0.1, get the nugget value data set N1, compare the nugget value N and the evaluation value P to see if they are monotonically increasing or monotonically decreasing. If they are monotonically increasing or monotonically decreasing, then the minimum value of the lower boundary of the training set T1 is the optimal nugget value parameter N 1bmin , otherwise proceed to step 10; Step 10: Based on the nugget value data set N1 obtained in step 9, find the minimum position of the evaluation value P for the nugget value N, recorded as the initial nugget value minimum value N Cmin , select area [N min -S N , N min +S N ], the calculation step size is selected as S N2 =0.1×S N , calculate the new nugget value data set N 1-1 , analyze data set N 1-1 The gold value when the evaluation value P is the smallest is recorded as the minimum gold value N of the stage Jmin , compare the two gold nugget values ​​N Cmin and N Jmin The evaluation value P, if the minimum value of the nugget is the initial nugget value minimum N Cmin Less than the minimum gold value N of the stage Jmin , then the minimum value of the initial nugget is N Cmin N is the optimal parameter of the nugget value 1min , otherwise proceed to step 11; Step 11: Take the minimum value of the stage nugget value as the new minimum value of the initial nugget value N Cmin2 , select area [N Jmin -S N2 , N Jmin +S N2], the calculation step size is selected as S N 3=0.1×S N2 , calculate the new nugget value data set N 1-2 , analyze data set N 1-2 The gold value when the evaluation value P is the smallest is recorded as the minimum gold value N of the stage Jmin2 , compare the two gold nugget values ​​N Cmin2 and N Jmin2 The evaluation value P, if the minimum value of the nugget is the initial nugget value minimum N Cmin2 Less than the minimum value of the gold nugget N Jmin2 The corresponding evaluation value is the minimum value of the initial nugget value N Cmin2 N is the optimal parameter of the nugget value 1min , otherwise proceed to step 12; Step 12: Set the minimum value of the stage nugget value N Jmin2 As the new minimum value of the initial nugget value N Cmin3 Repeat step 12 until the optimal parameter N for the nugget value under training set T1 is determined. 1min , thus obtaining the evaluation value P under the training set T1 T1 Minimum value P T1min The corresponding three optimal parameter main variable range MV 1min , sub-range SV 1min and the value of the gold nugget N 1min ; Step 13: Repeat steps 7 to 12 to obtain the minimum evaluation value P of each parameter in each training set. T2min , P T3min ...P Tnmin And the corresponding three optimal parameter main variable MV 2min , MV 3min ...MV nmin , sub-range SV 2min SV 3min ...SV nmin and the value of the gold nugget N 2min 、N 3min ...N nmin ; Step 14: Compare the minimum evaluation value P of each parameter under each training set T2min , P T3min ...P Tnmin , determine the minimum evaluation value, which is P min , the corresponding three optimal parameters are the main variable MV min , sub-range SV min and the value of the gold nugget N min .

[0013] Further, the generation of contour maps and models of each layer based on the optimal parameters includes: based on the optimal parameters, the primary variable range MVmin, the secondary variable range SVmin and the nugget value Nmin are calculated to obtain the contour map of the coal seam floor and the structural distribution map of each layer and the three-dimensional geological model, the steps are as follows:

[0014] The covariance and variogram values ​​have the following relationship:

[0015] C(h) = C(0) - γ(h) (2)

[0016] Among them, C(h) is the variogram value, C(0) is the base value, and γ(h) is the covariance value;

[0017] The formula for calculating covariance is:

[0018]

[0019] The relationship between the variogram value and the weight is:

[0020]

[0021] The calculation methods of the evaluation indicators mean absolute error MAE, root mean square error RMSE and mean square error MSE are as follows:

[0022]

[0023] where Y i is the estimated value, Y* is the true value, is the average interpolated elevation of n checkpoints, where n is the number of sample points;

[0024] Let Y*=Y(ui) and obtain the unknown point information. The calculation formula is as follows:

[0025]

[0026] Among them, Y*(u) is the unknown point information to be obtained, Y(ui) is the known point data, λi is the weight of the point, and the sum of the weights is 1.

[0027] According to another aspect of the present invention, a three-dimensional geological modeling device for coal mine profile data is provided, and the following technical solution is adopted:

[0028] The determination module is used to determine the spatial data point set and divide the spatial data point set into several validation sets and several training sets; the first calculation module is used to determine the value range and calculation step of each parameter of the data point, calculate each parameter of the training set, obtain the calculation result, and verify and analyze the calculation result with the validation set to obtain the analysis result; the analysis module is used to use distribution control for the analysis result, calculate the size of the evaluation value of each parameter under the training set, analyze the average values ​​under different training sets, and find the minimum average value; the acquisition module is used to obtain the latest evaluation value and the parameters corresponding to the latest evaluation value based on the minimum average value, and the parameters are the optimal parameters; the second calculation module is used to generate contour maps and models at each level based on the optimal parameters, and the calculation formula is as follows:

[0029]

[0030] Among them, Y*(u) is the unknown point information to be obtained, Y(ui) is the known point data, λ i is the weight of the point, and the sum of the weights is 1.

[0031] Furthermore, the determination module includes: a collection module for collecting and arranging known information of coal mine profiles, performing statistics in layers, and arranging them into (x, y, z) spatial coordinate format. The shortest straight-line distance of data distribution in the study area is L min ; The longest straight-line distance of data distribution in the study area is L max ; The screening module is used to randomly screen the acquired data and count the spatial coordinate information. The total data point set is N, and there are n data points in total; the elimination module is used to number the data points based on the data point set N as N1, N2...N n , N1 data is used as validation set V1, and the remaining n-1 data points are used as training set T1. The training set participates in the calculation, and the validation set participates in the validation analysis. Similarly, N2 data is removed as validation set V2, and the remaining n-1 data points are used as training set T2; n groups of validation set V are formed in sequence n , training set T n .

[0032] According to another aspect of the present invention, a storage medium is provided, and the following technical solution is adopted:

[0033] A storage medium includes the above-mentioned three-dimensional geological modeling method for coal mine profile data.

[0034] The purpose of the present invention is to provide a three-dimensional geological modeling method based on coal mine profile data, based on the relevant parameters in the Kriging interpolation method, using appropriate methods and processes to find a three-dimensional geological modeling method suitable for coal mines. It solves the problem that due to the large data limitations in actual coal mine work, the data near the tunnel is dense and the data in the working surface is sparse, which directly affects the choice during interpolation. In actual use, researchers often make choices based on experience, resulting in a large number of human subjective factors on the basis of known data, which affects the ultimate scientificity and objectivity of three-dimensional geological modeling and ultimately leads to errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0036] Figure 1 A flowchart of a three-dimensional geological modeling method for coal mine profile data according to an embodiment of the present invention;

[0037] Figure 2 A specific flow chart of a three-dimensional geological modeling method for coal mine profile data according to an embodiment of the present invention;

[0038] Figure 3 An algorithm flow chart of a three-dimensional geological modeling method for coal mine profile data according to an embodiment of the present invention;

[0039] Figure 4 This is a structural diagram of a three-dimensional geological modeling device for coal mine profile data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0041] Before explaining the examples, let me explain the abbreviations and key terms in the text:

[0042] The shortest straight-line distance of data distribution in the study area is L min ;

[0043] The longest straight-line distance of the data distribution in the study area is L max ;

[0044] The total set of data points in the study is N;

[0045] Yi is the estimated value, Y* is the true value, is the average interpolated elevation of n checkpoints, where n is the number of sample points;

[0046] Mean absolute error MAE, root mean square error RMSE and mean square error MSE;

[0047] Validation set V1, training set T1;

[0048] Mean absolute error validation set U MAE1 , root mean square error validation set U RMSE1 and mean square error validation set U MSE1 Residual sum of squares verification;

[0049] The mean absolute error validation normalization value NUMAE1, the root mean square error validation normalization value NURMSE1, and the mean square error validation normalization value NUMSE1;

[0050] The evaluation value P is Pn = NU MAEn +NU RMSEn +NU MSEn ;

[0051] The three parameters are the main variable range MV, the secondary variable range SV and the nugget value N;

[0052] The optimal parameter is the main variable MV min , sub-range SV min and the value of the gold nugget N min .

[0053] Figure 1 The present invention is a flowchart of a three-dimensional geological modeling method for coal mine profile data according to an embodiment of the present invention.

[0054] See also Figure 1 As shown, a three-dimensional geological modeling method for coal mine profile data includes:

[0055] S101: Determine a spatial data point set, and divide the spatial data point set into a plurality of validation sets and a plurality of training sets;

[0056] S102: Determine the value range and calculation step of each parameter of the data point, calculate each parameter of the training set, obtain the calculation result, and use the verification set to verify and analyze the calculation result to obtain the analysis result; use distribution control for the analysis result, calculate the size of the evaluation value of each parameter under the training set, analyze the average values ​​under different training sets, and find the minimum average value;

[0057] S103: Obtain the latest evaluation value and parameters corresponding to the latest evaluation value based on the minimum average value, and the parameters are the optimal parameters.

[0058] Specifically, step S101 collects and organizes the known information of the coal mine profile, performs statistics in layers, and organizes all of them into (x, y, z) spatial coordinate format. The shortest straight-line distance of the data distribution in the study area is Lmin; the longest straight-line distance of the data distribution in the study area is Lmax. Then, the organized data is randomly screened, and the spatial coordinate information of the coal seam floor is taken as an example to perform statistics. In step S102, the organized data is randomly screened, and the spatial coordinate information of the coal seam floor is taken as an example to perform statistics on the statistical point set N. The data point set N is divided into a validation set and a training set, and repeated multiple times to obtain multiple validation sets and training sets. The mean absolute error (MAE), root mean square error (RMSE) and mean square error (MSE) are calculated for each validation set, and normalized to obtain an evaluation value P. Repeat the steps until the evaluation values ​​of all validation sets are calculated. In step S105, the main variable range MV and the secondary variable range SV are used for calculation, and optimization is performed on each training set. The optimal parameters (primary range MVmin, secondary range SVmin and nugget value Nmin) are determined by calculating and optimizing the nugget value N. The evaluation values ​​under each training set are compared to determine the minimum evaluation value Pmin and its corresponding three optimal parameters.

[0059] Figure 2 This is a specific flow chart of a three-dimensional geological modeling method for coal mine profile data described in an embodiment of the present invention.

[0060] join Figure 2 As shown in the flowchart of the method for three-dimensional geological modeling of coal mine profile data according to an embodiment of the present invention, the specific process includes the following steps:

[0061] Step 210: determine a set of spatial data points, which are divided into a liver entry verification set and a training set;

[0062] Step 220: Determine the value range and calculation step of each parameter, and perform analysis on each parameter;

[0063] Step 230: using distribution control to calculate the evaluation value of the next parameter in the training set;

[0064] Step 240: Analyze the average values ​​under different training sets and find the minimum average value;

[0065] Step 250: The latest evaluation value corresponding to each parameter is the optimal parameter;

[0066] Step 260: Generate contour maps and models for each layer based on the optimal parameters.

[0067] Specifically, in step 210, the known information of the coal mine profile is collected and sorted, and statistics are performed in layers, and all of them are sorted into (x, y, z) spatial coordinate format, and the shortest straight-line distance of the data distribution in the study area is Lmin; the longest straight-line distance of the data distribution in the study area is Lmax; the sorted data is randomly screened, and the spatial coordinate information of the coal seam bottom is counted by taking the coal seam bottom as an example, and the total data point set is N;

[0068] In step 220, based on the data point set N (a total of n data points), the data points are numbered as N1, N2...Nn, N1 data is used as the validation set V1, and the remaining n-1 data points are used as the training set T1. The training set participates in the calculation, and the validation set participates in the validation analysis.

[0069] In step 230, the method in step 220 is used, and similarly, N2 data is removed as the validation set V2, and the remaining n-1 data points are used as the training set T2; and n groups of validation sets Vn and training sets Tn are obtained; the mean absolute error MAE1, root mean square error RMSE1 and mean square error MSE1 are calculated for the validation set V1, respectively, to obtain the mean absolute error validation set UMAE1, the root mean square error validation set URMSE1 and the mean square error validation set UMSE1; the mean absolute error validation set UMAE1, the root mean square error validation set URMSE1 and the mean square error validation set UMSE1 are normalized in turn, and the result is The normalized set of the three is the normalized value NUMAE1 for mean absolute error verification, the normalized value NURMSE1 for root mean square error verification, and the normalized value NUMSE1 for mean square error verification. The evaluation value P1 is the sum of the three after normalization, P1=NUMAE1+NURMSE1+NUMSE1; repeat the above steps to obtain n groups of evaluation values ​​in turn. Further, the training set T1 is calculated using the main variable range MV. The maximum value of the main variable range MV is Lmax, the minimum value is Lmin, the main variable range MV is an integer, and the main variable range step size SMV=1 is calculated to obtain the main variable range MV data group MV1;

[0070] Furthermore, based on the main range MV data group MV1, the training set T1 is calculated using the secondary range SV. The secondary range SV takes a maximum value of Lmax and a minimum value of Lmin. The secondary range SV takes an integer. The main range step size SSV=1 is calculated. The secondary range SV is smaller than the main range MV, and the secondary range SV data group SV1 is obtained.

[0071] In step 230, further, based on the secondary range data set SV1, the nugget value N is calculated and analyzed, the minimum value of the nugget value N is 0, the maximum value is 1, the nugget value step length SN=0.1 is calculated, and the nugget value data set N1 is obtained. The nugget value N is compared with the evaluation value P to see whether it is monotonically increasing or monotonically decreasing. If it is monotonically increasing or monotonically decreasing, the minimum value of the lower boundary of the training set T1 is the nugget value optimal parameter N1bmin, otherwise, step 10 is performed;

[0072] Furthermore, based on the nugget value data group N1 obtained in step 230, find the minimum position of the evaluation value P for the nugget value N, record it as the initial nugget value minimum NCmin, select the area [Nmin-SN, Nmin+SN], select SN2=0.1×SN as the calculation step, calculate and obtain the new nugget value data group N1-1, analyze the nugget value when the evaluation value P is the smallest in the data group N1-1, record it as the stage nugget value minimum NJmin, compare the evaluation values ​​P of the two nugget values ​​NCmin and NJmin, if the minimum nugget value, the initial nugget value minimum NCmin, is less than the stage nugget value minimum NJmin, then the initial nugget value minimum NCmin is the optimal nugget value parameter N1min.

[0073] If the initial nugget value minimum NCmin is the nugget value and the nugget value is not the optimal parameter N1min, use the stage nugget value minimum as the new initial nugget value minimum NCmin2, select the area [NJmin-SN2, NJmin+SN2], select SN3=0.1×SN2 for the calculation step, calculate the new nugget value data group N1-2, analyze the nugget value when the evaluation value P is the smallest in the data group N1-2, record it as the stage nugget value minimum NJmin2, compare the evaluation values ​​P of the two nugget values ​​NCmin2 and NJmin2, if the initial nugget value minimum NCmin2 is less than the stage nugget value minimum NJ min2 Corresponding evaluation value, the minimum value of the initial nugget value NCmin2 is the optimal parameter N1 of the nugget value min Otherwise, the minimum value of the gold nugget NJmin2 in the current stage is used as the new minimum value of the initial gold nugget NC min3 Repeat step 1 until the optimal parameter N1 of the nugget value under training set T1 is determined. min , thus obtaining the minimum value PT1 in the evaluation value PT1 under the training set T1 min The corresponding three optimal parameter main variables MV1 min , sub-range SV1 min and gold value N1 min .

[0074] In step 250, the above step 240 is repeated to obtain the minimum evaluation value PT2 of each parameter under each training set in turn. min PT3min ...PTn min And the corresponding three optimal parameter main variables MV2 min 、MV3min...MVn min , sub-range SV2 min SV3 min ...SVn min and nugget value N2min, N3 min ...Nn min ; Compare the minimum evaluation value PT2 of each parameter under each training set min PT3 min ...PTn min , determine the minimum evaluation value, which is P min , the corresponding three optimal parameters are the main variable MV min , sub-range SV min and the value of the gold nugget N min .

[0075] In step 260, based on the optimal parameters, the main variable MV min , sub-range SV min and the value of the gold nugget N min The calculation can obtain the coal seam floor contour map, the structural distribution map of each layer and the three-dimensional geological model.

[0076] refer to Figure 3 , Figure 3 This is a logic flow chart of a three-dimensional geological modeling method for coal mine profile data according to an embodiment of the present invention; the specific scheme is as follows:

[0077] 301 Nugget value N, the value is [0,1], the step length is 0.1, and the Nugget value data group is obtained;

[0078] 302 determines whether the nugget value N and the evaluation value P are monotonically increasing or monotonically decreasing;

[0079] 303 The minimum position of the evaluation value P for the nugget value N is the minimum value of the initial nugget value;

[0080] 304 For step length S N Select area [NJ min -S N , N.J. min +S N ], select step size S N2 =0.1×S N ;

[0081] 305 Calculate and obtain a new gold nugget value data set;

[0082] 306 The minimum evaluation value in the analysis data group is the minimum value of the stage nugget value;

[0083] 307 determines whether the maximum value and the minimum value of the gold nugget are equal;

[0084] 308 For step length S N2 ; Select area [NJ min -S N2 , N.J. min +S N2 ], the calculation step size is selected as S N3 =0.1×S N2 ;

[0085] 309 Determine the optimal parameter N1 of the nugget value under training set T1 min .

[0086] Figure 4 This is a structural diagram of a three-dimensional geological modeling device for coal mine profile data according to an embodiment of the present invention.

[0087] A three-dimensional geological modeling device for coal mine profile data includes: 41 a determination module, which is used to determine a spatial data point set and divide the spatial data point set into a number of verification sets and a number of training sets; 42 a first calculation module, which is used to determine the value range and calculation step of each parameter of the data point, calculate each parameter of the training set to obtain a calculation result, and use the verification set to verify and analyze the calculation result to obtain an analysis result; 43 an analysis module, which is used to use distribution control for the analysis result, calculate the size of the evaluation value of each parameter under the training set, analyze each average value under different training sets, and find the minimum average value; 44 an acquisition module, which is used to obtain the latest evaluation value and each parameter corresponding to the latest evaluation value based on the minimum average value, and each parameter is the optimal parameter; wherein the determination module includes: a collection module, which is used to collect and organize known information of the coal mine profile, perform statistics in layers, and organize it into (x, y, z) spatial coordinate format, and the shortest straight-line distance of the data distribution in the study area is L min ; The longest straight-line distance of data distribution in the study area is L max ; The screening module is used to randomly screen the acquired data and count the spatial coordinate information. The total data point set is N, and there are n data points in total; the elimination module is used to number the data points based on the data point set N as N1, N2...N n , N1 data is used as validation set V1, and the remaining n-1 data points are used as training set T1. The training set participates in the calculation, and the validation set participates in the validation analysis. Similarly, N2 data is removed as validation set V2, and the remaining n-1 data points are used as training set T2; n groups of validation set V are formed in sequence n , training set T n .

[0088] The above only describes some exemplary embodiments of the present embodiment by way of illustration. It is undoubted that, for those skilled in the art, the described embodiments can be modified in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A three-dimensional geological modeling method for coal mine profile data, characterized in that: include: Determine a spatial data point set, and divide the spatial data point set into a number of validation sets and a number of training sets; Determine the value range and calculation step of each parameter of the data point, calculate each parameter of the training set, obtain the calculation results, and use the verification set to verify and analyze the calculation results to obtain the analysis results; The analysis results are distributed controlled, and the evaluation values ​​of each parameter under the training set are calculated. The average values ​​under different training sets are analyzed to find the minimum average value. Based on the minimum average value, the latest evaluation value and the parameters corresponding to the latest evaluation value are obtained, and the parameters are the optimal parameters; Based on the optimal parameters, contour maps and models of each layer are generated. The calculation formula is as follows: Among them, Y*(u) is the unknown point information to be obtained, Y(ui) is the known point data, λ i is the weight of the point, and the sum of the weights is 1.

2. The modeling method according to claim 1, characterized in that: The determining of the spatial data point set and dividing the spatial data point set into a plurality of validation sets and a plurality of training sets comprises: Step 1: Collect and organize the known information of the coal mine profile, perform statistics in layers, and organize it into (x, y, z) spatial coordinate format. The shortest straight-line distance of the data distribution in the study area is L min ; The longest straight-line distance of data distribution in the study area is L max ; Step 2: Randomly screen the acquired data and count the spatial coordinate information. The total data point set is N, with a total of n data points; Step 3: Based on the data point set N, number the data points as N1, N2...N n , N1 data is used as validation set V1, and the remaining n-1 data points are used as training set T1. The training set participates in the calculation, and the validation set participates in the validation analysis. Similarly, N2 data is removed as validation set V2, and the remaining n-1 data points are used as training set T2; n groups of validation set V are formed in sequence n , training set T n .

3. The modeling method according to claim 2, characterized in that: The parameter value range and calculation step length of each data point are determined, each parameter of the training set is calculated to obtain a calculation result, and the calculation result is verified and analyzed with a verification set, and the analysis result obtained includes: Step 4: Calculate the mean absolute error MAE1, root mean square error RMSE1 and mean square error MSE1 for the validation set V1, and get the mean absolute error validation set U MAE1 , Root mean square error validation set U RMSE1 and mean square error validation set U MSE1 ; Step 5: Validate the mean absolute error on the set U MAE1 , Root mean square error validation set U RMSE1 and mean square error validation set U MSE1 Normalize them one by one, and get the normalized set of the three as the mean absolute error verification normalized value NU MAE1 , root mean square error verification normalization value NU RMSE1 And the mean square error verification normalization value NU MSE1 , the evaluation value P1 is the sum of the three after normalization, P1 = NU MAE1 +NU RMSE1 +NU MSE1 ; Step 6: Execute step 5 in sequence to calculate n groups of evaluation values ​​P1, P2...P n , P n =NU MAEn +NU RMSEn +NU MSEn .

4. The modeling method according to claim 3, characterized in that: The analysis results are controlled by distribution, and the evaluation values ​​of each parameter under the training set are calculated, and the average values ​​under different training sets are analyzed to find the minimum average value, including: Step 7: Calculate the training set T1 using the main variable MV, and the main variable MV takes the maximum value L max , the minimum value is L min , the main variable range MV takes an integer and calculates the main variable range step length S MV =1, get the main variable range MV data group MV1; Step 8: Based on the main variable range MV data set MV1, the training set T1 is calculated using the secondary variable range SV, and the secondary variable range SV takes the maximum value L max , the minimum value is L min , the secondary variable range SV takes an integer, and the main variable range step length S is calculated SV =1, the secondary range SV is smaller than the primary range MV, and the secondary range SV data set SV1 is obtained; Step 9: Based on the sub-range data set SV1, calculate and analyze the nugget value N. The minimum value of the nugget value N is 0 and the maximum value is 1. Calculate the nugget value step S N =0.1, get the nugget value data set N1, compare the nugget value N and the evaluation value P to see if they are monotonically increasing or monotonically decreasing. If they are monotonically increasing or monotonically decreasing, then the minimum value of the lower boundary of the training set T1 is the optimal nugget value parameter N 1bmin , otherwise proceed to step ten; Step 10: Based on the nugget value data set N1 obtained in step 9, find the minimum position of the evaluation value P for the nugget value N, recorded as the initial nugget value minimum value N Cmin , select area [N min -S N , N min +S N ], the calculation step size is selected as S N2 =0.1×S N , calculate the new nugget value data set N 1-1 , analyze data set N 1-1 The nugget value when the evaluation value P is the smallest is recorded as the minimum nugget value N of the stage Jmin , compare the two gold nugget values ​​N Cmin and N Jmin The evaluation value P, if the minimum value of the nugget is the initial nugget value minimum N Cmin Less than the minimum gold value N of the stage Jmin , then the minimum value of the initial nugget is N Cmin N is the optimal parameter of the nugget value 1min , otherwise proceed to step 11; Step 11: Take the minimum value of the gold nugget as the new minimum value of the initial gold nugget N Cmin2 , select area [N Jmin -S N2 , N Jmin +S N2 ], the calculation step size is selected as S N 3 = 0.1 × S N2 , calculate the new nugget value data set N 1-2 , analyze data set N 1-2 The nugget value when the evaluation value P is the smallest is recorded as the minimum nugget value N of the stage Jmin2 , compare the two gold nugget values ​​N Cmin2 and N Jmin2 The evaluation value P, if the minimum value of the nugget is the initial nugget value minimum N Cmin2 Less than the minimum gold value N of the stage Jmin2 The corresponding evaluation value is the minimum value of the initial nugget value N Cmin2 N is the optimal parameter of the nugget value 1min , otherwise proceed to step 12; Step 12: Set the minimum value of the stage block value N Jmin2 As the new minimum value of the initial nugget value N Cmin3 Repeat step 12 until the optimal parameter N for the nugget value under training set T1 is determined. 1min , thus obtaining the evaluation value P under the training set T1 T1 Minimum value P T1min The corresponding three optimal parameter main variable range MV 1min , sub-range SV 1min and the value of the gold nugget N 1min ; Step 13: Repeat steps 7 to 12 to obtain the minimum evaluation value P of each parameter in each training set. T2min , P T3min ...P Tnmin And the corresponding three optimal parameter main variable MV 2min , MV 3min ...MV nmin , sub-range SV 2min SV 3min ...SV nmin and the value of the gold nugget N 2min 、N 3min ...N nmin ; Step 14: Compare the minimum evaluation value P of each parameter under each training set T2min , P T3min ...P Tnmin , determine the minimum evaluation value, which is P min , the corresponding three optimal parameters are the main variable MV min , sub-range SV min and the value of the gold nugget N min .

5. The modeling method according to claim 4, characterized in that: The generation of contour maps and models of each layer based on the optimal parameters includes: calculating the main variable range MVmin, the secondary variable range SVmin and the nugget value Nmin based on the optimal parameters, so as to obtain the contour map of the coal seam floor and the structural distribution map of each layer and the three-dimensional geological model, the steps are as follows: The covariance and variogram values ​​have the following relationship: C(h) = C(0) - γ(h) (2) Among them, C(h) is the variogram value, C(0) is the base value, and γ(h) is the covariance value; The formula for calculating covariance is: The relationship between the variogram value and the weight is: The calculation methods of the evaluation indicators mean absolute error MAE, root mean square error RMSE and mean square error MSE are as follows: where Y i is the estimated value, Y* is the true value, is the average interpolated elevation of n checkpoints, where n is the number of sample points; Let Y*=Y(ui) and obtain the unknown point information. The calculation formula is as follows: Among them, Y*(u) is the unknown point information to be obtained, Y(ui) is the known point data, λi is the weight of the point, and the sum of the weights is 1.

6. A three-dimensional geological modeling device for coal mine profile data, characterized in that: include: A determination module, used for determining a spatial data point set, and dividing the spatial data point set into a plurality of validation sets and a plurality of training sets; The first calculation module is used to determine the value range and calculation step of each parameter of the data point, calculate each parameter of the training set, obtain the calculation result, and verify and analyze the calculation result with the verification set to obtain the analysis result; The analysis module is used to apply distribution control to the analysis results, calculate the evaluation value of each parameter under the training set, analyze the average values ​​under different training sets, and find the minimum average value; An acquisition module is used to acquire the latest evaluation value and each parameter corresponding to the latest evaluation value based on the minimum average value, and the parameters are the optimal parameters; The second calculation module is used to generate contour maps and models at each level based on the optimal parameters. The calculation formula is as follows: Among them, Y*(u) is the unknown point information to be obtained, Y(ui) is the known point data, λ i is the weight of the point, and the sum of the weights is 1.

7. The modeling device according to claim 6, characterized in that: The determination module comprises: The collection module is used to collect and organize the known information of the coal mine profile, perform statistics in layers, and organize it into (x, y, z) spatial coordinate format. The shortest straight-line distance of the data distribution in the study area is L min ; The longest straight-line distance of data distribution in the study area is L max ; The screening module is used to randomly screen the acquired data and perform statistics on the spatial coordinate information. The total data point set is N, and there are n data points in total; The elimination module is used to number the data points N1, N2...N based on the data point set N. n , N1 data is used as validation set V1, and the remaining n-1 data points are used as training set T1. The training set participates in the calculation, and the validation set participates in the validation analysis. Similarly, N2 data is removed as validation set V2, and the remaining n-1 data points are used as training set T2; n groups of validation set V are formed in sequence n , training set T n .

8. A storage medium, characterized in that: The modeling device comprises the modeling device according to any one of claims 6-7.

Images