2.5 D statistical optimal distributed modeling method for geologic structure of thin coal seam group

Abstract

The invention discloses a 2.5 D statistical optimal distributed modeling method for a geologic structure of a thin coal seam group, and belongs to the technical field of geologic modeling and mineral resource development. The method comprises the steps that drilling space information and sequence constraints of all stratums are obtained based on drilling data, elevation estimation values and corresponding estimation value variances of all points in estimation value grids of all the stratums are obtained through independent Kriging interpolation operation of all the stratums, and constraint conditions are constructed according to the sequence of all the stratums; and constructing a target function of'minimizing the sum of variance with the original value 'by utilizing the elevation estimation value and the corresponding variance on each stratum estimation value grid point, and solving the elevation value of each stratum on the grid point in the estimation value range. The method fuses adjacent horizon information and spatial statistical constraints, can more accurately describe and express complex three-dimensional spatial relationships such as stratum pinching, crossing and merging in the thin coal seam group, remarkably improves the spatial fidelity of the model, realizes automatic and quantitative correction of the stratum sequence in strip mine full-stratum modeling, eliminates manual intervention errors, and improves the modeling efficiency. And the modeling efficiency and the geometric accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of geological modeling and mineral resource development technology, and in particular to a 2.5D statistical optimal distributed modeling method for the geological structure of thin coal seam groups. Background Technology

[0002] Thin coal seam groups refer to coal-bearing stratigraphic units in geological profiles composed of multiple thin coal seams with small interlayer spacing and dense vertical stacking. Their core characteristics lie in the large number of coal seams, their thinness, and the significant and frequent alternation of thickness in the non-coal rock layers (intercalation, roof and floor), resulting in a highly complex and anisotropic stratigraphic structure. The structural complexity of thin coal seam groups presents a severe challenge to high-precision 3D geological modeling, with the core difficulties concentrated in:

[0003] (1) The stratigraphic relationships are extremely complex: Due to the large number of thin coal seams and their spatial undulations, coupled with the frequent pinch-out and lenticular formation of thin non-coal strata (especially weak strata such as mudstone and siltstone), a large number of intersections, cuts, and mergers occur between stratigraphic interfaces. This complex spatial topological relationship is very likely to produce topological errors such as self-intersection and overlap during the modeling process.

[0004] (2) Strong spatial variability: The thickness, lithology and occurrence of thin coal seams and interbedded rock layers may change significantly over short distances, requiring the model to have extremely high local adaptability.

[0005] 2. Mainstream Modeling Techniques and Their Limitations: For modeling thin coal seam groups, current mainstream techniques all face significant bottlenecks:

[0006] (1) Explicit modeling.

[0007] Precision control is difficult: relying on manual interpretation and connection of stratigraphic interfaces (such as faults and pinch-out lines), it is difficult to accurately control the geometry and contact relationship at the intersections of dense and complex stratigraphic intersections in thin coal seam groups, often requiring a lot of empirical processing.

[0008] High costs of manual intervention: When dealing with complex geological structures (e.g., more than 10 layers), geological engineers need to perform tedious manual checks and edits to ensure the model's topological rationality and basic accuracy. This process is extremely time-consuming, usually taking several days or even weeks, becoming a major bottleneck in the modeling workflow, and the results are greatly affected by subjective factors.

[0009] (2) Implicit modeling

[0010] Theoretical accuracy advantage: By implicitly defining all stratigraphic interfaces through spatial interpolation functions (such as radial basis functions RBF), it can theoretically handle complex stratigraphic intersections more naturally, and has high accuracy potential.

[0011] Algorithm complexity constraint: Its global optimization characteristics result in extremely high space complexity (typically O(n^2)). 3 (where n is the number of control points). In the scenario of modeling thin coal seam groups with all geological features (all coal seams + all non-coal interlayers + faults), the number of control points n is huge, and the consumption of computational resources (memory, time) grows exponentially, making it difficult to apply to the full stratigraphic modeling of large-scale thin coal seam groups in practice. Summary of the Invention

[0012] To address the above shortcomings, the purpose of this invention is to provide a 2.5D statistical optimal distributed modeling method for the geological structure of thin coal seam groups. The aim is to perform layered distributed modeling according to the stratigraphic sequence while maintaining key three-dimensional spatial relationships (such as stratigraphic undulations and contact relationships), thereby significantly reducing the dimensionality processed simultaneously.

[0013] The technical solution adopted in this invention is: a 2.5D statistical optimal distributed modeling method for thin coal seam geological structures, the key technical points of which include the following steps:

[0014] Step 1: Data preparation, obtain borehole data of the open-pit mine area, calculate the estimated range and grid division, and determine the N stratigraphic layers to be modeled;

[0015] Step 2: Kriging spatial interpolation. Kriging interpolation is performed independently for each stratum to generate the elevation estimate and interpolation variance of each grid point in the grid covering the estimation range for each stratum.

[0016] Step 3: Construct constraints using the stratigraphic order from Step 2, and construct the objective function using the elevation estimates and corresponding variances of the estimated grid points of each stratigraphic unit calculated in Step 2 as the processing data and weight terms. Use the sequential quadratic programming gradient optimization algorithm to solve for the elevation values ​​of each stratigraphic unit at the grid points within the estimation range.

[0017] Step 4: Generate the top and bottom triangular mesh surfaces of the strata from the optimized elevation values ​​of each stratum grid point, and perform Boolean operations to cut out the 3D stratum model.

[0018] In the above scheme, the construction constraint condition for the construction sequence of each stratum in step 3 is: avoid cross-penetration constraint: ensure that the bottom plate of the stratum is always above the top plate of the next stratum.

[0019] In the above scheme, the formula corresponding to the constraint condition is:

[0020]

[0021] In the formula, Let Z{i} be the optimal elevation value. Let Z{i-1} be the optimal elevation value. This is the tolerance for the elevation difference between point Z{i} and the upper layer (theoretically, the value is 0, meaning the current layer's bottom slab and the lower layer's top slab are the same). The optimized output is a grid point (x... j y j The optimal elevation values ​​of all strata i on the surface.

[0022] In the above scheme, the objective function for constructing the optimization in step 3 should satisfy the following constraints: the current top plate elevation is the same as and greater than the bottom plate elevation of the previous stratum, and the current bottom plate elevation is equal to the top plate elevation of the next stratum.

[0023] In the above scheme, the objective function formula in step 3 is:

[0024]

[0025] In the formula, Let Z{i} be the optimal elevation value. Let Z{i} be the Kriging elevation value. Let Z{i} be the square of the Kriging estimate variance:

[0026] In the above scheme, the three-dimensional stratum model generation method in step 5 is to construct a triangular mesh of the top and bottom plates using optimized stratum elevation data, and then perform Boolean difference set operation based on the vertical stretching of the top and bottom plates into a three-dimensional model to obtain the three-dimensional model of the corresponding stratum.

[0027] The beneficial effects of this invention are as follows: This 2.5D statistically optimal distributed modeling method for the geological structure of thin coal seam groups, while maintaining key three-dimensional spatial relationships (such as stratigraphic undulations and contact relationships), performs layered distributed modeling according to the stratigraphic sequence, significantly reducing the dimensionality processed simultaneously. Compared to 3D implicit modeling, distributed computing effectively decomposes and significantly reduces the overall spatial complexity of the algorithm, solving the computational feasibility problem of full stratigraphic modeling of large-scale thin coal seam groups; compared to traditional 2D profile modeling, it overcomes the defect of traditional 2D methods that ignore vertical stratigraphic correlations. In the layered modeling process, this method integrates information from adjacent stratigraphic levels and spatial statistical constraints, enabling more accurate characterization and expression of complex three-dimensional spatial relationships such as stratigraphic pinch-outs, intersections, and mergers within thin coal seam groups, significantly improving the spatial fidelity of the model, realizing automated and quantitative correction of stratigraphic sequence in open-pit mine full stratigraphic modeling, eliminating human intervention errors, and improving modeling efficiency and geometric accuracy. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a schematic diagram of the stratigraphic sequence structure for open-pit mine geological modeling in an embodiment of the present invention.

[0030] Figure 2 This is a flowchart of the open-pit mine geological modeling and stratigraphic modeling method in an embodiment of the present invention;

[0031] Figure 3 This is a schematic diagram illustrating the trend of spatial complexity as a function of grid size in an embodiment of the present invention;

[0032] Figure 4 This is a schematic diagram of an unoptimized coal seam in an embodiment of the present invention;

[0033] Figure 5 This is a schematic diagram of the optimized coal seam in an embodiment of the present invention;

[0034] Figure 6 This is a schematic diagram comparing real and virtual drilling in an embodiment of the present invention;

[0035] Figure 7 This is a schematic diagram comparing the depth of each layer of the top and bottom plates in a real drilling and virtual drilling embodiment of the present invention. Detailed Implementation

[0036] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the following description is provided in conjunction with the accompanying drawings. Figures 1-7 The present invention will be further described in detail below with reference to specific embodiments.

[0037] The 2.5D statistical optimal distributed modeling method for thin coal seam geological structures used in this embodiment includes the following steps:

[0038] Step 1: Based on the borehole data from the mining area, including spatial coordinates and lithological information identifying stratigraphic attributes, determine the stratigraphic sequence to be modeled. Specifically: determine the stratigraphic sequence based on the lithological information in the borehole data. For borehole data with clearly defined stratigraphic attributes and numbers, directly confirm the stratigraphic sequence according to the numbers. For borehole data without specific stratigraphic numbers, perform custom partitioning and define stratigraphic numbers.

[0039] Step 2: Perform ordinary kriging spatial interpolation or universal kriging (if a trend term exists) independently for each stratum. Select either a spherical or linear variogram model. Fit the optimal parameters based on the actual data to obtain the interpolation results and interpolation variance at the grid points of the corresponding estimation range for each stratum.

[0040] The borehole data from the mining area is analyzed to construct a borehole dataset including borehole number (project number), borehole XYZ coordinates, lithological information, etc. Based on the spatial coordinate data of the mining area, the boundary of the modeling area is determined, generating a regular planar grid, and the grid is divided. In this embodiment, the estimated grid is divided in the XY direction at a specified step size (e.g., 50 meters). After determining the stratigraphic sequence based on the lithological information in the borehole data, virtual points are added to the strata not reached in each borehole. The rule for adding virtual points is to take the bottom plate of the previous stratum or the top plate of the next stratum in the sequence.

[0041] Based on the parsed borehole dataset, the data is grouped by borehole number and lithology information. The grouped data is then sorted in ascending order by borehole entry elevation, and the first data point is taken to obtain the corresponding stratum borehole top plate dataset. The grouped data is then sorted in descending order by borehole exit elevation, and the first data point is taken to obtain the corresponding stratum borehole bottom plate dataset. Kriging space interpolation is then performed on each stratum separately based on the top plate dataset and bottom plate dataset to obtain the elevation interpolation and interpolation variance of each coordinate point within the estimated range grid.

[0042] Step 3: Construct constraints using the stratigraphic order from Step 2. Use the elevation estimates and corresponding variances of the estimated grid points of each stratigraphic unit calculated in Step 2 as processing data and weight terms to construct the objective function of "minimizing the sum of variances with the original values". Use the Sequential Quadratic Programming (SLSQP) gradient optimization algorithm to solve for the elevation values ​​of each stratigraphic unit at the grid points within the estimation range.

[0043] Step 3.1: Construct constraints and objective function.

[0044] The constraints in this embodiment are as follows: ensure that the top plate of formation i is always below the bottom plate of formation i-1, avoiding cross-penetration. The formula is:

[0045]

[0046] In the formula, the optimized output is the grid point (x j y j The optimal elevation values ​​of all strata i on the surface.

[0047] An objective function is constructed by uniformly sampling grid points. This objective function uses the elevation values ​​of each layer at each grid point as variables and minimizes the sum of the variances between the optimized values ​​and the original values ​​of all grid points at each layer.

[0048] The objective function is to minimize the sum of the variances between the optimized value and the original value, that is, the current top elevation of the formation is the same as and greater than the bottom elevation of the previous formation, and the current bottom elevation is equal to the top elevation of the next formation.

[0049] In this embodiment, based on all strata i = 1, 2, ..., N at the same coordinate point within the estimation range grid, at grid point (x j y j Kriging interpolation results and interpolation variance σ 2 i (x j y j The objective function that minimizes the variance of the original values ​​is constructed as follows:

[0050]

[0051] In the formula, the decision variable is the optimized grid point elevation value. The first term: a weighted least squares term, which optimizes the elevation. Approaching Kriging's valuation The weights are the inverse of the interpolation variance (higher variance points allow for greater deviation).

[0052] Step 3.2: Use a gradient-type mathematical optimization algorithm to solve for the optimal elevation values ​​of each stratum at all grid points within the estimation range.

[0053] A gradient-based mathematical optimization algorithm is used to solve for the elevation values ​​of each stratum at the grid points within the estimation range. The algorithm can be either Sequential Quadratic Programming (SLSQP) or the Interior-Point method, supporting large-scale constrained optimization. The objective function gradient is... right The partial derivatives are:

[0054] The estimated range is divided into grid blocks and distributed to computing nodes. The M estimated grids are further divided into K×K sub-regions, and each node independently optimizes the elevation of all strata within its sub-region. Distributed multi-process parallel processing is used to improve computational speed.

[0055] The optimized model can provide a reliable basis for open-pit mine resource assessment and mining design, and reduce manual processing costs.

[0056] In this embodiment, after parsing the borehole data, the Kriging estimation ensures that the estimation in the XY plane is optimal (2D), but the Kriging estimation result is 3D (elevation value in the Z direction). Then, an optimization adjustment is made in the Z direction (to ensure the formation order). That is, the Kriging estimation is first used to ensure that the 2D of the XY plane is optimal, and then optimization is made based on the Z direction. It is between the 2D and 3D direct estimation and is defined as 2.5D.

[0057] Step 4: Generate a three-dimensional geological structure model.

[0058] Optimize the grid point elevations using step 3 Constrained Delaunay triangulation (CDT) is applied layer by layer to construct the triangular mesh of the top and bottom plates. Based on the triangular mesh data of the top and bottom plates, the generalized triangular prism (GTP) volume elements are used to stretch vertically to form the columnar model of the top and bottom plates. The Boolean difference result of the top and bottom plate columnar model is the three-dimensional model of stratum i.

[0059] Depend on Figure 3 As can be seen, in the process of geological modeling of thin coal seams, the method in this embodiment can significantly reduce the spatial complexity of computation, reduce memory overhead, and improve computational efficiency. When the estimated grid size decreases, the number of grid points requiring estimation increases, leading to a larger computational load. Figure 3 As can be seen, the computational space complexity increases quadratically with the change in grid size, which is advantageous compared to the exponential growth of traditional methods.

[0060] from Figure 4 It can be seen that there are many obvious intersections in the unoptimized strata, such as: "Upper 6 Coal" and "6-1 Coal" intersect at point ①, "Lower 6 Coal" and "7 Coal" intersect at points ② and ③, and "7 Coal" and "8 Coal" intersect at point ④. From Figure 5 It can be seen that the optimized strata no longer intersect at the previously intersecting points ①, ②, ③, and ④, and there are no intersecting phenomena in other parts.

[0061] This embodiment evaluates the quality of a three-dimensional geological model based on borehole data by reserving a portion of real boreholes during modeling, and then extracting virtual boreholes from the same locations on the three-dimensional geological model. The degree of agreement between the virtual boreholes and the real boreholes is used as the reliability of the corresponding points in the model.

[0062] Calculation of the degree of fit: Calculate the overlapping and non-overlapping portions of the real borehole and the model borehole. Divide the thickness of the overlapping portion by the actual depth of the real borehole to obtain the degree of fit between the virtual and real boreholes. For example... Figure 6As shown, the actual borehole depth is L, and the overlap is a+b+c+d+e. Then, the confidence level of the model at this borehole location is (a+b+c+d+e) / L, where a is the overlap thickness between the actual and virtual boreholes in the first stratum from top to bottom, b is the overlap thickness between the actual and virtual boreholes in the second stratum from top to bottom, c is the overlap thickness between the actual and virtual boreholes in the third stratum from top to bottom, d is the overlap thickness between the actual and virtual boreholes in the fourth stratum from top to bottom, and e is the overlap thickness between the actual and virtual boreholes in the fifth stratum from top to bottom.

[0063] Referring to the above method, the modeling error is quantitatively described as shown in Table 1:

[0064] Table 1 is a diagram showing the reliability comparison between the traditional method and the method in this embodiment.

[0065] Bore ID Conventional method (unadjusted) reliability This method reliability 0-1 72％ 89.9％ 10-1 86％ 93％ 10-5 92％ 97％ 1-1 81％ 91％ 1-3 85％ 90％

[0066] Five holes were selected to calculate the model's reliability. The calculations showed that the average reliability reached over 89%.

[0067] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A 2.5D statistical optimal distributed modeling method for thin coal seam geological structures, characterized in that, Includes the following steps: Step 1: Data preparation, obtain borehole data of the open-pit mine area, calculate the estimated range and grid division, and determine the N stratigraphic layers to be modeled; Step 2: Kriging spatial interpolation. Kriging interpolation is performed independently for each stratum to generate the elevation estimate and interpolation variance of each grid point in the grid covering the estimation range for each stratum. Step 3: Construct constraints using the stratigraphic order from Step 1, and construct the objective function using the elevation estimates and corresponding variances of the estimated grid points of each stratigraphic unit calculated in Step 2 as the processing data and weight terms. Use the sequential quadratic programming gradient optimization algorithm to solve for the elevation values ​​of each stratigraphic unit at the grid points within the estimation range. Step 4: Generate the top and bottom triangular mesh surfaces of the strata from the optimized elevation values ​​of each stratum grid point, and perform Boolean operations to cut out the 3D stratum model.

2. The 2.5D statistical optimal distributed modeling method for thin coal seam geological structures according to claim 1, characterized in that, The construction constraints for the strata sequence described in step 3 are as follows: Avoid cross-penetration constraint: ensure that the bottom plate of the stratum is always above the top plate of the next stratum.

3. The 2.5D statistical optimal distributed modeling method for thin coal seam geological structures according to claim 2, characterized in that, The formula corresponding to the aforementioned constraint is: In the formula, Let Z{i} be the optimal elevation value. Let Z{i-1} be the optimal elevation value. The tolerance for the elevation difference between point Z{i} and the upper layer is theoretically 0, meaning that when the bottom slab of the current layer is the same as the top slab of the lower layer, the optimized output is a grid point (x). j y j The optimal elevation values ​​of all strata i on the surface.

4. The 2.5D statistical optimal distributed modeling method for thin coal seam geological structures according to claim 1, characterized in that, The objective function for constructing the optimization in step 3 should satisfy the following constraints: the current top elevation of the formation is the same as and greater than the bottom elevation of the previous formation, and the current bottom elevation of the formation is equal to the top elevation of the next formation.

5. The 2.5D statistical optimal distributed modeling method for thin coal seam geological structures according to claim 4, characterized in that, The objective function formula in step 3 is: In the formula, Let Z{i} be the optimal elevation value. Let Z{i} be the Kriging elevation value. Let Z{i} be the square of the variance of the Kriging estimate at point Z{i}.

6. The 2.5D statistical optimal distributed modeling method for thin coal seam geological structures according to claim 1, characterized in that, The method for generating the three-dimensional stratum model in step 5 is as follows: construct the triangular mesh of the top and bottom plates using the optimized stratum elevation data, stretch the top and bottom plates vertically into a three-dimensional model, and then perform Boolean difference set operation to obtain the three-dimensional model of the corresponding stratum.

Images