Coal seam parameter spatial interpolation correction method based on random forest

Abstract

The invention provides a coal seam parameter spatial interpolation correction method based on a random forest, and belongs to the technical field of coal seam parameter spatial interpolation correction, and the method comprises the steps: obtaining a coal seam parameter value of a measurement point in a target coal seam and a corresponding training set residual error based on Kriging interpolation and residual error calculation; carrying out prediction index influence factor calculation; based on a random forest regression model, correcting the coal seam parameter value of the target coal seam according to the index influence factor to obtain a corrected interpolation result; evaluating whether an interpolation result corrected based on the random forest regression model meets a preset requirement or not, if yes, storing the random forest regression model along with the current time, and if not, retraining the random forest regression model; and performing interpolation correction on the to-be-interpolated region. According to the method, the precision and adaptability of coal seam parameter spatial interpolation can be effectively improved, and more reliable data support is provided for coal mine safety production, resource evaluation, gas control and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of coal seam parameter spatial interpolation correction, and in particular to a coal seam parameter spatial interpolation correction method based on random forest. Background Art

[0002] During coal mining, accurate coal seam parameter data is crucial for ensuring safe production, effective resource assessment, and gas control. Traditional coal seam parameter measurement methods, such as borehole measurement and geological exploration, can provide some accurate data. However, due to the complexity of coal seams and the large-scale data requirements, they are usually unable to fully cover the coal seam information of the entire mining area. In addition, the distribution of coal seam parameters has strong spatial variability. Traditional interpolation methods (such as Kriging interpolation and inverse distance weighted interpolation) are easily affected by factors such as noise and data sparsity when processing the spatial distribution of coal seam parameters, resulting in insufficient accuracy and reliability of the interpolation results.

[0003] To overcome these issues, machine learning-based prediction methods have been increasingly applied to coal seam parameter spatial interpolation in recent years. However, most existing technologies rely solely on traditional interpolation methods or single machine learning models, lacking comprehensive optimization and correction methods for coal seam parameter spatial interpolation results. Summary of the Invention

[0004] To address the above issues, the present invention provides a coal seam parameter spatial interpolation correction method based on random forests. This method combines the advantages of traditional interpolation methods and machine learning models, applies machine learning methods to capture the complex nonlinear relationships between index influencing factors, and improves the accuracy of coal seam parameter interpolation. This method is of great significance to safe coal mining and resource assessment, and specifically includes:

[0005] A coal seam parameter spatial interpolation correction method based on random forest includes:

[0006] S1. Based on Kriging interpolation and residual calculation, the coal seam parameter values of the measuring points in the target coal seam and the corresponding training set residuals are obtained;

[0007] S2. Calculate the factors influencing the prediction indicators;

[0008] S3. Based on the random forest regression model, the coal seam parameter values of the target coal seam are corrected according to the index influencing factors to obtain the corrected interpolation results;

[0009] S4, evaluating whether the interpolation result after correction based on the random forest regression model meets the preset requirements. If so, save the random forest regression model and execute S5. If not, retrain the random forest regression model and return to step S3;

[0010] S5. Correct the interpolation in the area to be interpolated.

[0011] Optionally, the coal seam parameter values of the measuring points in the target coal seam and the corresponding training set residuals obtained by Kriging interpolation and residual calculation in the root S1 include:

[0012] S11. Extracting gas pressure, gas content, coal solidity coefficient f, coal gas release initial velocity ΔP, and measuring point locations from the coal seam gas geological map based on Python to form an indicator data set S = (X, Y, Z) corresponding to the measuring point locations and the measured data.

[0013] S12, dividing the indicator data set into interpolation data sets S according to the ratio of 2:2:1 k =(X k ,Y k ,Z k ), training set S t =(X t ,Y t ,Z t ) and the test set S c =(X c ,Y c ,Z c );

[0014] S13, based on the interpolation data set S k =(X k ,Y k ,Z k ), the measurement point position P of the training set t =(X t ,Y t ) and the test point position P of the test set c =(X c ,Y c ), perform Kriging spatial interpolation on the measurement point positions of the training set and the test set to obtain the Kriging interpolation result Z' of the training set t And the Kriging interpolation result Z' of the test set c ;

[0015] S14, according to the measurement point data Z of the training set t And the Kriging interpolation result Z' t , and get the corresponding training set residual r t =Z t -Z' t .

[0016] Optionally, the calculation of the influencing factors of the prediction indicator in S2 includes:

[0017] The fault density is defined as follows:

[0018]

[0019] Among them, E fd is the fault density in the unit area; H is the fault drop, in m; L is the fault strike length, in m; α is the fault dip; β is the angle between the principal stress and the fault strike; S is the unit area, in m 2 ; n is the number of faults in the unit area; the unit area is a square area with a side length of a centered at the measuring point;

[0020] The degree of wrinkle deformation is defined as formula (2):

[0021]

[0022] Among them, F d is the degree of wrinkle deformation in the unit area; h is the elevation distance between adjacent contour lines, in meters; k is the actual length of the contour line with the largest curvature in the unit area, in meters; l is the secant length of the contour line with the largest curvature, in meters; d is the distance between the two contour lines with the largest curvature, in meters;

[0023] The coefficient of variation of coal thickness is defined as formula (3):

[0024]

[0025] Wherein, γ is the coefficient of variation of coal thickness, in %; m is the average coal thickness in the unit area, in m; s is the standard deviation of coal thickness variation;

[0026] Among them, the formula for the standard deviation of coal thickness change is formula (4):

[0027]

[0028] m i is the measured thickness of the coal points, in m; n is the number of coal points.

[0029] Optionally, the random forest regression model in S3 is used to correct the coal seam parameter value of the target coal seam according to the indicator influencing factors, and the corrected interpolation result obtained includes:

[0030] The coal seam depth, fault density, fold deformation degree, coal seam thickness, coal seam inclination, coal thickness variation coefficient, soft layer thickness and spatial position coordinates were input into the random forest regression model, and the training set residual r t As the output of the random forest regression model, the random forest regression model is trained to obtain a trained random forest regression model;

[0031] According to the test set Sc The trained random forest regression model is tested with the data to obtain the corrected residual r c ;

[0032] Correct the Kriging interpolation result to obtain the corrected interpolation result Z p , Z p =Z' c +r c .

[0033] Optionally, the evaluation in S4 is based on whether the interpolation result after the correction of the random forest regression model meets the preset requirements. If so, the random forest regression model is saved and S5 is executed. If not, the random forest regression model is retrained and the process returns to step S3, including:

[0034] Comparison of measurement data Z c And the interpolation result Z p , the evaluation indicators determination coefficient and Pearson correlation coefficient were used to evaluate the combined interpolation results;

[0035] If the coefficient of determination and the Pearson correlation coefficient are both higher than 0.85, the random forest regression model is saved and step S5 is executed;

[0036] Otherwise, retrain the random forest regression model and return to step S3;

[0037] The retraining of the random forest regression model includes: re-adjusting the hyperparameters of the random forest regression model and using random search to find the optimal parameter combination.

[0038] Optionally, the interpolation correction in the to-be-interpolated area in S5 includes:

[0039] S51. Divide the interpolation area into grids with a grid side length of b, and obtain the spatial position P of the interpolation point. N ;

[0040] S52. Calculate the Kriging interpolation result of each interpolation point based on the measured point data S=(X, Y, Z) as Z' N ;

[0041] S53, calculating the influencing factors of the point to be interpolated according to step S3, where the influencing factor calculation unit area is a square area with a side length of a=2b centered at the point to be interpolated;

[0042] S54, the influencing factors and spatial position P of the point to be interpolated N Input the saved random forest regression model and get the corrected residual r N ;

[0043] S55. Apply the residual error correction kriging interpolation result obtained by the random forest regression model to obtain the final interpolation result Z of the interpolation point N =Z' N +r N .

[0044] Compared with the prior art, the above technical solution has at least the following beneficial effects:

[0045] Compared with the traditional Kriging interpolation method, although Kriging interpolation can provide reasonable spatial distribution predictions in most cases, when faced with complex geological conditions and sparse data, the interpolation results may have large errors, especially for the nonlinear changes of parameters and the processing of local abnormal data, which is more difficult, and its accuracy is often not high. In order to overcome this shortcoming, the present invention introduces a random forest regression model to correct the Kriging interpolation results. First, Kriging interpolation provides a preliminary spatial prediction for coal seam parameters, and interpolation is performed using its spatial autocorrelation. Afterwards, by introducing a random forest regression model to capture the nonlinear relationship between coal seam parameters and influencing factors such as coal seam burial depth, geological structure, and soft stratification, adaptive adjustments are made according to different geological environments to correct potential errors, thereby greatly improving the interpolation accuracy. By combining Kriging interpolation and random forest regression models, the present invention not only effectively corrects the shortcomings of Kriging interpolation under complex coal seam conditions, but also improves the accuracy of spatial prediction of coal seam parameters, providing more reliable data support for coal mine safety production, resource assessment, and gas control. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0047] Figure 1 This is the execution process of the coal seam parameter spatial interpolation correction method based on random forest provided by an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of a coal seam parameter spatial interpolation correction method based on random forest provided in an embodiment of the present invention;

[0049] Figure 3 Gas pressure interpolation results using different interpolation methods provided in the embodiments of the present invention;

[0050] Figure 4 Gas content interpolation results using different interpolation methods provided in the embodiments of the present invention;

[0051] Figure 5The interpolation results of the coal solidity coefficient f using different interpolation methods provided in the embodiment of the present invention;

[0052] Figure 6 The interpolation results of the initial gas emission velocity ΔP of coal using different interpolation methods provided in the embodiments of the present invention;

[0053] Figure 7 A gas pressure distribution diagram of a coal seam provided in an embodiment of the present invention;

[0054] Figure 8 A gas content distribution diagram of a coal seam provided in an embodiment of the present invention;

[0055] Figure 9 A distribution diagram of the solidity coefficient f of a coal seam provided in an embodiment of the present invention;

[0056] Figure 10 This is a distribution diagram of the initial velocity ΔP of gas emission in a coal seam provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0057] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] Unless otherwise defined, the technical or scientific terms used in the present invention shall have the usual meaning understood by persons of ordinary skill in the field to which the present invention belongs. The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one", "an" or "the" do not indicate a quantity limitation, but rather indicate the presence of at least one. Words such as "include" or "comprise" mean that the elements or objects preceding the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.

[0059] It should be noted that the terms "up", "down", "left", "right", "front" and "back" used in the present invention are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0060] Compared to traditional kriging interpolation methods, while kriging interpolation can provide reasonable spatial distribution predictions in most cases, it can suffer from significant errors when dealing with complex geological conditions and sparse data. This is particularly problematic for handling nonlinear parameter variations and localized outliers, resulting in low accuracy. To overcome this shortcoming, this paper introduces a random forest regression model to correct the kriging interpolation results.

[0061] Taking an existing coal seam as the engineering background, relevant data is collected and a coal seam parameter spatial interpolation correction method based on random forest is provided. The execution process of this method is as follows Figure 1 The schematic diagram of this method is shown in Figure 2 The specific contents are as follows:

[0062] A coal seam parameter spatial interpolation correction method based on random forest includes:

[0063] S1. Based on Kriging interpolation and residual calculation, the coal seam parameter values of the measuring points in the target coal seam and the corresponding training set residuals are obtained;

[0064] S2. Calculate the factors influencing the prediction indicators;

[0065] S3. Based on a random forest regression model, the coal seam parameter value of the target coal seam is corrected according to the index influencing factors to obtain a corrected interpolation result;

[0066] S4, evaluating whether the interpolation result after correction based on the random forest regression model meets the preset requirements. If so, saving the random forest regression model and executing S5; if not, retraining the random forest regression model and returning to step S3;

[0067] S5. Correct the interpolation in the area to be interpolated.

[0068] A specific implementation method is mainly based on the random forest coal seam parameter spatial interpolation correction method to realize the spatial interpolation of gas pressure, gas content, coal solidity coefficient f value and coal gas release initial velocity ΔP. The specific implementation method is as follows:

[0069] S11. The coal seam parameter values include gas pressure, gas content, coal solidity coefficient f value, and coal gas initial velocity ΔP. The gas pressure, gas content, coal solidity coefficient f value, coal gas initial velocity ΔP, and measuring point locations in the coal seam gas geological map are extracted using Python to form an indicator data set S = (X, Y, Z) corresponding to the measuring point locations and the measured data.

[0070] In this example, Python was used to extract gas pressure, gas content, coal solidity coefficient f, and coal gas initial velocity ΔP from a coal seam gas geological map. The measured data, P = (X, Y), and Z, were used to generate an index dataset S = (X, Y, Z) corresponding to the measured data. 274 sets of gas pressure data, 254 sets of gas content data, and 495 sets of coal solidity coefficient f and coal gas initial velocity ΔP data were extracted.

[0071] S12, dividing the indicator data set into interpolation data sets S according to the ratio of 2:2:1 k =(X k ,Y k ,Z k ), training set S t =(X t ,Y t ,Z t ) and the test set S c =(X c ,Y c ,Z c );

[0072] S13, based on the interpolation data set S k =(X k ,Y k ,Z k ), the measurement point position P of the training set t =(X t ,Y t ) and the test point position P of the test set c =(X c ,Y c ), perform Kriging spatial interpolation on the measurement point positions of the training set and the test set to obtain the Kriging interpolation result Z' of the training set t And the Kriging interpolation result Z' of the test set c ;

[0073] S14, according to the measurement point data Z of the training set t And the Kriging interpolation result Z' t , and get the corresponding training set residual r t =Z t -Z' t .

[0074] S2. Calculate the factors influencing the prediction indicators, including:

[0075] The fault density is defined as follows:

[0076]

[0077] Among them, Efd is the fault density in the unit area; H is the fault drop, in m; L is the fault strike length, in m; α is the fault dip; β is the angle between the principal stress and the fault strike; S is the unit area, in m 2 ; n is the number of faults in the unit area; the unit area is a square area with a side length of a centered at the measuring point;

[0078] The degree of wrinkle deformation is defined as formula (2):

[0079]

[0080] Among them, F d is the degree of wrinkle deformation in the unit area; h is the elevation distance between adjacent contour lines, in meters; k is the actual length of the contour line with the largest curvature in the unit area, in meters; l is the secant length of the contour line with the largest curvature, in meters; d is the distance between the two contour lines with the largest curvature, in meters;

[0081] The coefficient of variation of coal thickness is defined as formula (3):

[0082]

[0083] Wherein, γ is the coefficient of variation of coal thickness, in %; m is the average coal thickness in the unit area, in m; s is the standard deviation of coal thickness variation;

[0084] Among them, the formula for the standard deviation of coal thickness change is formula (4):

[0085]

[0086] m i is the measured thickness of the coal points, in m; n is the number of coal points.

[0087] S3. Based on a random forest regression model, the coal seam parameter value of the target coal seam is corrected according to the index influencing factors to obtain a corrected interpolation result;

[0088] The coal seam depth, fault density, fold deformation degree, coal seam thickness, coal seam inclination, coal thickness variation coefficient, soft layer thickness and spatial position coordinates were input into the random forest regression model, and the training set residual r t As the output of the random forest regression model, the random forest regression model is trained to obtain a trained random forest regression model;

[0089] According to the test set S c The trained random forest regression model is tested with the data to obtain the corrected residual r c ;

[0090] Correct the Kriging interpolation result to obtain the corrected interpolation result Z p , Z p =Z' c +r c .

[0091] In this embodiment, the training set data is used for model training. For gas pressure and gas content, the coal seam depth (A1), fault density (A2), fold deformation degree (A3), coal seam thickness (A4), coal seam inclination (A5), coal thickness variation coefficient (A6), soft layer thickness (A7), and spatial position coordinates are used as model inputs. For the coal solidity coefficient f value and the coal gas release initial velocity ΔP, the fault density (A2), fold deformation degree (A3), soft layer thickness (A7), and spatial position coordinates are used as model inputs. The training set residual r t As the model output, the random forest regression model is trained to obtain the trained random forest regression model. The trained model is tested using the test set data. The influencing factors and spatial positions of gas pressure, gas content, coal solidity coefficient f value and coal gas release initial velocity ΔP are input, and the corrected residual r is output. c , correct the Kriging interpolation result, that is, the final interpolation result is Z p =Z' c +r c .

[0092] S4, evaluating whether the interpolation result after correction based on the random forest regression model meets the preset requirements. If so, saving the random forest regression model and executing S5; if not, retraining the random forest regression model and returning to step S3;

[0093] Comparison of measurement data Z c And the interpolation result Z p , the evaluation indicators determination coefficient and Pearson correlation coefficient were used to evaluate the combined interpolation results;

[0094] If the coefficient of determination and the Pearson correlation coefficient are both higher than 0.85, the random forest regression model is saved and step S5 is executed;

[0095] Otherwise, retrain the random forest regression model and return to step S3. Readjust the hyperparameters of the random forest regression model, such as n_estimators (number of trees), max_depth (maximum depth), min_samples_split (minimum sample split), etc., and use RandomizedSearchCV to find the optimal parameter combination. Finally, retrain the model and evaluate it until the accuracy requirements are met.

[0096] In this embodiment, the comparison measurement data Z c And the interpolation result Z p , the determination coefficient of the application evaluation index (R 2 ) and the Pearson correlation coefficient (R) are used to evaluate the combined interpolation results. If the model evaluation meets the requirements, the random forest regression model is saved for use in correcting the interpolation results. Otherwise, the model is retrained.

[0097] S5. Correct the interpolation in the area to be interpolated.

[0098] The Kriging interpolation values, machine learning interpolation values, and combined interpolation values are compared with the measured values, and the residuals between the interpolation values and the measured values are calculated. Figures 3 to 6 The interpolation results for coal gas pressure, gas content, coal solidity coefficient f, and coal gas initial velocity ΔP using different interpolation methods are shown. These are the interpolation and residual analysis results for these coal seam parameters using three different interpolation methods: Kriging, Random Forest Regression Model (RFR Model), and a combination of Kriging and Random Forest Regression Models (Kriging-RFR Model). This example demonstrates the feasibility and applicability of the proposed random forest-based spatial interpolation correction method for coal seam parameters to the spatial interpolation of prediction indicators.

[0099] Depend on Figures 3 to 6 It can be seen that the Kriging-RFRModel interpolation and the determination coefficient R of gas pressure 2 The correlation coefficient R is 0.89, and the correlation coefficient R is 0.95, which is higher than 0.81 and 0.90 of Kriging and 0.84 and 0.93 of RFR Model. The determination coefficient R of the interpolated gas content of Kriging-RFR Model and the measured gas content is 2 The correlation coefficient R is 0.86, and the correlation coefficient R is 0.93, which is also higher than Kriging's 0.76 and 0.89 and RFRModel's 0.81 and 0.90; the determination coefficient R between the interpolated f-value of Kriging-RFRModel and the measured f-value is 2 The correlation coefficient R is 0.78, and the correlation coefficient R is 0.90, which is also higher than Kriging's 0.19 and 0.45 and RFR Model's 0.76 and 0.88; the determination coefficient R of Kriging-RFRModel interpolated ΔP and measured ΔP is 2 The correlation coefficient R is 0.78, and the correlation coefficient R is 0.90, which is also higher than Kriging's 0.22 and 0.48 and RFR Model's 0.73 and 0.87. Therefore, the proposed coal seam parameter spatial interpolation correction method based on random forest can be used for spatial interpolation of prediction indicators and can improve the spatial interpolation accuracy compared with the traditional Kriging interpolation method.

[0100] The proposed coal seam parameter spatial interpolation correction method based on random forest is applied to obtain the distribution diagram of coal seam gas pressure, gas content, coal solidity coefficient f value and coal gas release initial velocity ΔP, as shown in the following figure: Figures 7 to 10 shown.

[0101] Based on the technical solution provided above, the present invention combines Kriging interpolation and random forest regression models, which not only effectively corrects the shortcomings of Kriging interpolation under complex coal seam conditions, but also improves the accuracy of spatial prediction of coal seam parameters, providing more reliable data support for coal mine safety production, resource assessment and gas control.

[0102] The following points need to be explained:

[0103] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention. Other structures may refer to conventional designs.

[0104] (2) For the sake of clarity, the thickness of layers or regions in the drawings used to describe the embodiments of the present invention are exaggerated or reduced, that is, these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region, or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or intervening elements may be present.

[0105] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to form new embodiments.

[0106] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A coal seam parameter spatial interpolation correction method based on random forest, characterized in that: include: S1. Based on Kriging interpolation and residual calculation, the coal seam parameter values of the measuring points in the target coal seam and the corresponding training set residuals are obtained; S2. Calculate the factors influencing the prediction indicators; S3. Based on the random forest regression model, the coal seam parameter values of the target coal seam are corrected according to the index influencing factors to obtain the corrected interpolation results; S4, evaluating whether the interpolation result after correction based on the random forest regression model meets the preset requirements. If so, save the random forest regression model and execute S5. If not, retrain the random forest regression model and return to step S3; S5. Correct the interpolation in the area to be interpolated.

2. The coal seam parameter spatial interpolation correction method based on random forest according to claim 1 is characterized in that: The coal seam parameter values of the measuring points in the target coal seam and the corresponding training set residuals obtained by Kriging interpolation and residual calculation in the root S1 include: S11. Extracting gas pressure, gas content, coal solidity coefficient f, coal gas release initial velocity ΔP, and measuring point locations from the coal seam gas geological map based on Python to form an indicator data set S = (X, Y, Z) corresponding to the measuring point locations and the measured data. S12, dividing the indicator data set into interpolation data sets S according to the ratio of 2:2:1 k =(X k ,Y k ,Z k ), training set S t =(X t ,Y t ,Z t ) and the test set S c =(X c ,Y c ,Z c ); S13, based on the interpolation data set S k =(X k ,Y k ,Z k ), the measurement point position P of the training set t =(X t ,Y t ) and the test point position P of the test set c =(X c ,Y c ), perform Kriging spatial interpolation on the measurement point positions of the training set and the test set to obtain the Kriging interpolation result Z' of the training set t And the Kriging interpolation result Z' of the test set c ; S14, according to the measurement point data Z of the training set t And the Kriging interpolation result Z' t , and get the corresponding training set residual r t =Z t -Z' t .

3. The coal seam parameter spatial interpolation correction method based on random forest according to claim 2, wherein the step of calculating the influencing factors of the prediction index in S2 comprises: The fault density is defined as follows: Among them, E fd is the fault density in the unit area; H is the fault drop, in m; L is the fault strike length, in m; α is the fault dip; β is the angle between the principal stress and the fault strike; S is the unit area, in m 2 ; n is the number of faults in the unit area; the unit area is a square area with a side length of a centered at the measuring point; The degree of wrinkle deformation is defined as formula (2): Among them, F d is the degree of wrinkle deformation in the unit area; h is the elevation distance between adjacent contour lines, in meters; k is the actual length of the contour line with the largest curvature in the unit area, in meters; l is the secant length of the contour line with the largest curvature, in meters; d is the distance between the two contour lines with the largest curvature, in meters; The coefficient of variation of coal thickness is defined as formula (3): Wherein, γ is the coefficient of variation of coal thickness, in %; is the average coal thickness in the unit area, in meters; s is the standard deviation of coal thickness variation; Among them, the formula for the standard deviation of coal thickness change is formula (4): m i is the measured thickness of the coal points, in m; n is the number of coal points.

4. The coal seam parameter spatial interpolation correction method based on random forest according to claim 3 is characterized in that: In S3, based on the random forest regression model, the coal seam parameter value of the target coal seam is corrected according to the index influencing factors, and the corrected interpolation results obtained include: The coal seam depth, fault density, fold deformation degree, coal seam thickness, coal seam inclination, coal thickness variation coefficient, soft layer thickness and spatial position coordinates were input into the random forest regression model, and the training set residual r t As the output of the random forest regression model, the random forest regression model is trained to obtain a trained random forest regression model; According to the test set S c The trained random forest regression model is tested with the data to obtain the corrected residual r c ; Correct the Kriging interpolation result to obtain the corrected interpolation result Z p , Z p =Z' c +r c .

5. The coal seam parameter spatial interpolation correction method based on random forest according to claim 4 is characterized in that: The evaluation in S4 is based on whether the interpolation result after the correction of the random forest regression model meets the preset requirements. If so, the random forest regression model is saved and S5 is executed. If not, the random forest regression model is retrained and the process returns to step S3, including: Comparison of measurement data Z c And the interpolation result Z p , the evaluation indicators determination coefficient and Pearson correlation coefficient were used to evaluate the combined interpolation results; If the coefficient of determination and the Pearson correlation coefficient are both higher than 0.85, the random forest regression model is saved and step S5 is executed; Otherwise, retrain the random forest regression model and return to step S3; The retraining of the random forest regression model includes: re-adjusting the hyperparameters of the random forest regression model and using random search to find the optimal parameter combination.

6. The coal seam parameter spatial interpolation correction method based on random forest according to claim 5, characterized in that: The correction of interpolation to the area to be interpolated in S5 includes: S51. Divide the interpolation area into grids with a grid side length of b, and obtain the spatial position P of the interpolation point. N ; S52. Calculate the Kriging interpolation result of each interpolation point based on the measured point data S=(X, Y, Z) as Z' N ; S53, calculating the influencing factors of the point to be interpolated according to step S3, where the influencing factor calculation unit area is a square area with a side length of a=2b centered at the point to be interpolated; S54, the influencing factors and spatial position P of the point to be interpolated N Input the saved random forest regression model and get the corrected residual r N ; S55. Apply the residual error correction kriging interpolation result obtained by the random forest regression model to obtain the final interpolation result Z of the interpolation point N =Z' N +r N .

Images