Nonlinear determination method and device for two-dimensional joint roughness coefficient based on GBDT

Abstract

The embodiment of the specification provides a two-dimensional joint roughness coefficient nonlinear determination method and device based on GBDT, wherein the method comprises the following steps: obtaining three-dimensional point cloud data of rock joint sample surface topography through scanning; performing simplification processing on the three-dimensional point cloud data and exporting the three-dimensional point cloud data as a CSV file; calculating joint surface topography statistical parameter values and taking the joint surface topography statistical parameter values as machine learning model features; obtaining wall rock strength JCS of the rock joint sample through field experiments, performing direct shear test on the structural surface sample under a certain normal pressure, determining the basic internal friction angle and the peak shear strength of the structural surface sample, and calculating the JRC of the joint surface according to the JRC-JCS shear strength model; establishing a training set and a test set; establishing a GBDT model with a decision tree as a base learner, training and adjusting the model by using the training set, obtaining the trained GBDT model, inputting the test set into the trained GBDT model for testing, and finally taking the average value of the predicted values as the JRC value of the joint surface predicted by the model.

Description

TECHNICAL FIELD

[0001] The present document relates to the technical field of geological engineering, and particularly relates to a two-dimensional joint roughness coefficient nonlinear determination method and device based on GBDT. BACKGROUND

[0002] The structural plane is an important part of the rock mass and plays a major control role in the engineering properties of the rock mass. The roughness of the structural plane of the rock mass has an important influence on the hydraulic and shear strength properties of the unfilled fissured rock mass. The roughness coefficient of the structural plane has a complex nonlinear relationship with its various statistical parameters, and the traditional regression analysis method cannot effectively solve the problem. If a nonlinear prediction model of JRC can be established, the one-sidedness of the description of the structural plane morphology by a single statistical parameter can be overcome to a certain extent, a more comprehensive description of the structural plane morphology can be ensured, the problem that the traditional regression analysis method cannot be effectively applied can be solved, and a new idea is provided for the quantitative determination of JRC.

[0003] In the field of joint roughness determination, common nonlinear methods include grey systems, fuzzy mathematics, artificial neural networks, etc. The most widely used method is artificial neural networks. However, the calculation of artificial neural networks is based on the principle of empirical risk minimization, and over-learning problems are prone to occur for a limited number of samples in real problems. In addition, the solving process is prone to local optimization problems. As an integrated learning method, GBDT has an iterative error correction mechanism that can correct a part of the previous model that has not been captured in each iteration. It can relatively flexibly avoid falling into a global optimal solution, has better stability and generalization ability, and can adapt to the complex structure of data. By combining multiple weak classifiers, the nonlinear relationship in the data can be effectively captured, and small sample, high-dimensional nonlinear problems can be well solved.

[0004] In one technical solution of the prior art, the roughness of the rock mass structural plane is mainly described by geometry and a single statistical parameter, which cannot fully capture the complex characteristics of the rock mass structural plane, and the prediction accuracy is not high. In another technical solution of the prior art, a long profile line of a rock layer is divided into n short profile lines, and the fractal dimension of the n profile lines is obtained by using a fractal mathematical method. The average value of the fractal dimensions is taken as the fractal dimension of the long profile line, and the JRC value of the long profile line is predicted. However, the fractal dimension is sensitive to the scale of the profile line and the calculation is complex, and this method is not conducive to accurate and rapid evaluation of the joint roughness. Compared with the fractal mathematical method, the GBDT model can accurately and rapidly predict the roughness coefficient of the rock mass structure, and has higher processing efficiency and generalization ability. SUMMARY

[0005] The purpose of the present application is to provide a two-dimensional joint roughness coefficient nonlinear determination method and device based on GBDT, which aims to solve the above problems in the prior art.

[0006] The application provides a two-dimensional joint roughness coefficient nonlinear determination method based on GBDT, comprising the following steps:

[0007] Three-dimensional point cloud data of the surface morphology of the rock joint sample is obtained by scanning;

[0008] The three-dimensional point cloud data is simplified, and the profile line point cloud coordinates obtained after the simplification are exported as a CSV file;

[0009] Statistical parameter values of the joint surface morphology are calculated and used as features of a machine learning model;

[0010] The wall rock strength JCS of the rock joint sample is obtained through field experiments, direct shear tests are performed on the structural surface sample under a certain normal pressure, the basic internal friction angle and the peak shear strength are determined, the JRC of the joint surface is calculated according to the JRC-JCS shear strength model, the calculated JRC and the corresponding statistical parameter values are divided into a training set and a test set according to a certain proportion;

[0011] A GBDT model with a decision tree as a base learner is established, the model is trained and parameterized by using the training set, a trained GBDT model is obtained, the test set is input into the trained GBDT model for testing, and finally the average value of the predicted values is the JRC value of the joint surface predicted by the model.

[0012] The application provides a two-dimensional joint roughness coefficient nonlinear determination device based on GBDT, comprising the following steps:

[0013] A scanning module is configured to obtain three-dimensional point cloud data of the surface morphology of the rock joint sample by scanning;

[0014] A processing module is configured to simplify the three-dimensional point cloud data and export profile line point cloud coordinates obtained after the simplification as a CSV file;

[0015] A calculation module is configured to calculate statistical parameter values of the joint surface morphology and use the statistical parameter values as features of a machine learning model;

[0016] A division module is configured to obtain the wall rock strength JCS of the rock joint sample through field experiments, perform direct shear tests on the structural surface sample under a certain normal pressure, determine the basic internal friction angle and the peak shear strength, calculate the JRC of the joint surface according to the JRC-JCS shear strength model, and divide the calculated JRC and the corresponding statistical parameter values into a training set and a test set according to a certain proportion;

[0017] The prediction module is used for establishing a GBDT model of a decision tree-based learner, training and adjusting the model by using a training set, obtaining a trained GBDT model, inputting a test set into the trained GBDT model for testing, and finally obtaining an average value of predicted values as a joint surface JRC value predicted by the model.

[0018] The embodiment of the present application also provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, and the computer program is used for implementing the steps of the GBDT-based two-dimensional joint roughness coefficient nonlinear determination method.

[0019] The embodiment of the present application also provides a computer readable storage medium, and the computer readable storage medium stores an information transmission implementation program, and the program is used for implementing the steps of the GBDT-based two-dimensional joint roughness coefficient nonlinear determination method when executed by a processor.

[0020] The embodiment of the present application can comprehensively reflect joint topographic features, and improve prediction accuracy and generalization ability of a prediction method. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of one or more embodiments of the present application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0022] Figure 1 is a flow chart of the GBDT-based two-dimensional joint roughness coefficient nonlinear determination method of the embodiment of the present application;

[0023] Figure 2 is a flow chart of detailed processing of the GBDT-based two-dimensional joint roughness coefficient nonlinear determination method of the embodiment of the present application;

[0024] Figure 3 is a schematic diagram of 112 profile JRC value data of the embodiment of the present application;

[0025] Figure 4 is a schematic diagram of a GBDT model structure of the embodiment of the present application;

[0026] Figure 5 is a schematic diagram of a comparison between original data and predicted data of the embodiment of the present application;

[0027] Figure 6is a schematic diagram of a GBDT-based two-dimensional joint roughness coefficient nonlinear determination device according to an embodiment of the present application.

[0028] Figure 7 is a schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] In order for those skilled in the art to better understand the technical solutions in the one or more embodiments of the present specification, the technical solutions in the one or more embodiments of the present specification will be described clearly and completely below in conjunction with the drawings in the one or more embodiments of the present specification. Obviously, the described embodiments are only a part of the embodiments of the present specification, rather than all the embodiments. Based on the one or more embodiments of the present specification, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present document.

[0030] Method embodiments

[0031] According to an embodiment of the present application, a GBDT-based two-dimensional joint roughness coefficient nonlinear determination method is provided, Figure 1 is a flowchart of a GBDT-based two-dimensional joint roughness coefficient nonlinear determination method according to an embodiment of the present application, as Figure 1 shown, the GBDT-based two-dimensional joint roughness coefficient nonlinear determination method according to an embodiment of the present application specifically includes:

[0032] Step S101, obtaining three-dimensional point cloud data of the surface topography of the rock joint sample by scanning; specifically including:

[0033] A three-dimensional laser scanner with a scanning interval of 0.1mm is used to scan the rock joint sample to obtain three-dimensional point cloud data of the surface topography of the rock joint sample.

[0034] Step S102, simplifying the three-dimensional point cloud data, and exporting the cross-section line point cloud coordinates obtained after simplification as a CSV file; specifically including:

[0035] A predetermined sampling interval is used to establish a surface morphology three-dimensional model, and the surface morphology three-dimensional model is adjusted to be horizontal to establish an X-Y-Z coordinate system, with the shear direction of the rock joint sample as the X axis, the Y axis perpendicular to the shear direction and the X axis. On this basis, cross-section lines parallel to the shear direction are taken at intervals of 0.4mm, and the cross-section line point cloud coordinates after horizontal calibration are exported as a CSV file.

[0036] Step S103, calculating the joint surface topography statistical parameter value as a feature of a machine learning model; specifically including:

[0037] The five joint surface topography statistical parameter values of root mean square roughness, maximum relative height, height standard deviation, average inclination angle and inclination angle standard deviation are selected, the five joint surface topography statistical parameter values are calculated on the center line of the obtained profile line through corresponding formulas, and the five joint surface topography statistical parameter values are taken as features of the machine learning model. Specifically:

[0038] The least square center line of the profile line is calculated based on the least square method, and a reference straight line parallel to the least square center line is made between the highest point h p and the lowest point h v of the profile line, and the areas A u and A d of the figures enclosed by the profile line and the reference straight line above and below the reference straight line are calculated, respectively.

[0039] The reference straight line is iteratively calculated by being moved upward and parallel to the reference straight line from the lowest point of the profile line until A u and A d are equal, and the reference line at this time is the center line of the profile line.

[0040] The center line of the profile line is taken as the x-axis to establish a reference coordinate system, and the profile line coordinates are obtained, and the statistical parameter values of each profile line are calculated by programming in MATLAB according to the discrete algorithm of each statistical parameter.

[0041] In step S104, the wall rock strength JCS of the rock joint sample is obtained through field experiments, the direct shear test is performed on the structural surface sample under a certain normal pressure, the basic internal friction angle and the peak shear strength are determined, the JRC of the joint surface is calculated according to the JRC-JCS shear strength model; the calculated JRC and the corresponding statistical parameter values are divided into a training set and a test set according to a certain proportion;

[0042] In step S105, a GBDT model is established with a decision tree as a base learner, the model is trained and parameterized with the training set, a trained GBDT model is obtained, the test set is input into the trained GBDT model for testing, and the average value of the final prediction value is the JRC value of the joint surface predicted by the model. The test set is input into the trained GBDT model for testing, which specifically includes:

[0043] The test set is input into the trained GBDT model to obtain a prediction result.

[0044] The prediction value and the true value are evaluated by using a loss function to obtain the accuracy of the GBDT model.

[0045] The method further includes:

[0046] The two-dimensional joint roughness is predicted by the trained GBDT model.

[0047] In summary, the embodiment of the present application can obtain the topographic data of the entire rock mass structure surface at one time through the three-dimensional laser scanner or rock mass structure surface topography instrument, and selects five statistical parameters with high correlation as the description of the joint surface topography, so that the reliability of the model is enhanced, and the prediction accuracy is improved. In addition, GBDT adopts the principle of empirical risk minimization, and iteratively constructs a decision tree to optimize the empirical risk of the training data. Compared with artificial neural networks, GBDT performs more stably on small samples and high-dimensional nonlinear problems, and can effectively adapt to complex data structures.

[0048] The above technical solutions of the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0049] As shown in Figure 2 , the specific processing includes the following:

[0050] S1, scanning the rock joint sample to obtain the surface topography three-dimensional coordinate point cloud data.

[0051] Specifically, a three-dimensional laser scanner with a scanning interval of 0.1mm can be used to scan the rock joint sample.

[0052] S2, as shown in Figure 3 , in order to match the training data with the 112 published profile JRC value data sets, the point cloud data needs to be simplified, a horizontal three-dimensional model is established, and the horizontal calibrated profile point cloud coordinates are exported as a CSV file. Specifically, a surface topography three-dimensional model is established with a predetermined sampling interval, and the surface topography three-dimensional model is adjusted to be horizontal to establish an X-Y-Z coordinate system, with the shear direction of the rock joint sample as the X axis, the Y axis perpendicular to the shear direction and the X axis, and the Z axis perpendicular to both. On this basis, a profile line parallel to the shear direction is taken at an interval of 0.4mm, and the horizontal calibrated profile line point cloud coordinates are exported as a CSV file.

[0053] S3, five statistical parameters are selected as the description of the joint surface topography, the values of the five joint surface topography statistical parameters are calculated on the center line of the obtained profile line through the corresponding formula, and they are used as the features of the machine learning model.

[0054] Specifically, the least squares center line of the profile line can be calculated based on the least squares method, and a reference straight line parallel to the least squares center line is drawn between the highest point h p and the lowest point h v of the profile line, and the areas A u and A d of the figures surrounded by the profile line above and below the reference straight line and the reference straight line are calculated, respectively.

[0055] From the lowest point of the profile line, the reference straight line is moved upward iteratively until A u and A d are equal, the reference line at this time is the center line of the profile line.

[0056] The reference coordinate system is established with the center line of the profile line as the x-axis to obtain the profile line coordinates. The statistical parameter values of each profile line are calculated by programming in MATLAB according to the discrete algorithm of each statistical parameter.

[0057] The selected statistical parameters are the root mean square roughness Rq, the calculation formula of which is:

[0058]

[0059]

[0060] In the formula, y i represents the longitudinal coordinate of the i-th point, N is the number of discrete points of the profile line, y ave The average longitudinal coordinate of the N discrete points is to be solved.

[0061] The maximum relative height h max , the calculation formula of which is:

[0062] h max = h p -h v

[0063] In the formula, h p is the highest point of the profile line, and h v is the lowest point of the profile line.

[0064] The height standard deviation SD h , the calculation formula of which is:

[0065]

[0066] In the formula, represents the characteristic length of the entire profile line, and x i is the horizontal coordinate of the i-th point.

[0067] In the formula, is the average fluctuation of the profile line.

[0068] The average inclination angle i ave , the calculation formula of which is:

[0069]

[0070] The inclination angle standard deviation SD i , the calculation formula of which is:

[0071]

[0072] where N is the number of profile line discrete points; x i is the horizontal coordinate of the i-th point; y i is the vertical coordinate of the i-th point.

[0073] S4, calculate the JRC of the structural surface sample, and divide the calculated JRC and the corresponding statistical parameter value into a training set and a test set according to a ratio of 8:2 or 7:3.

[0074] When calculating the JRC, the wall rock strength JCS is first measured on site using an L-shaped rebound hammer, and a direct shear test is performed on the structural surface sample under a certain normal pressure to determine the basic internal friction angle and the peak shear strength, and then the JRC is calculated according to the following formula:

[0075]

[0076] where: is the basic internal friction angle of the structural surface; τ is the peak shear strength of the structural surface under a normal pressure of σ n .

[0077] S5, establish a GBDT machine learning model and train the model using the training set. When training the model, the weak learner needs to be initialized first, and here the average value of the true JRC value is used. Then the iterative construction of the regression tree is started. The negative gradient of each sample is calculated to fit the residual error between the predicted results of the regression tree and the true JRC value. Then the obtained residual error is taken as the new true value of the sample, and the five statistical parameters of the rock joint topography are taken as the input variables to obtain a new regression tree and update the predicted JRC value of the model. This process is repeated iteratively until the residual error is less than a specified threshold, and then a trained model is obtained.

[0078] Initialize the weak learner, which generally uses the average value of all training sample label values:

[0079]

[0080] where L(y i , c) represents the loss function, c is a constant calculated by the loss function, and argmin c represents searching for c to find the value of c that minimizes the entire loss function, y i is the true value of the i-th sample.

[0081] For the constructed m = 1, 2,..., M, where m is the number of trees, we have:

[0082] For each sample i = 1, 2,..., N, the negative gradient, i.e. the residual error, can be calculated:

[0083]

[0084] where r im is the negative gradient of the ith sample in the mth tree, y i is the true value of the ith sample in the training set, f m-1 (x) is the prediction result of the first m-1 trees, x i represents the feature vector of the sample, i.e. the five statistical parameters describing the rock joint morphology.

[0085] The obtained residual is taken as the new true value of the sample, and the data, i = 1, 2,..., N, is taken as the training data for the next tree, to obtain a new regression tree, whose corresponding leaf node region is, j = 1, 2,..., J. Where J is the number of leaf nodes of the regression tree t.

[0086] For the leaf region j = 1, 2,..., J, the best fitting value is calculated:

[0087]

[0088] In the formula, γ jm represents the best fitting value of leaf node j in the mth tree, γ is the fitting value on the leaf node, R jm represents the sample index set of leaf node j in the mth tree.

[0089] Update the strong learner:

[0090]

[0091] where f m (x) is the model prediction value after the mth iteration, f m-1 (x) is the model prediction value after the first m-1 iterations, γ jm is the best fitting value of leaf node j in the mth tree, I(x∈R jm ) is an indicator function, indicating whether the sample x belongs to the leaf node j of the mth tree. If x is in the leaf node j, the value of the indicator function is 1, otherwise it is 0.

[0092] Iterate this process to obtain the final learner:

[0093]

[0094] Here, f(x) is the final integrated model prediction result, f0(x) is the prediction value of the initial model, which is taken as the average value of the true JRC, M is the number of iterations, which is also the number of trees in the model, and J is the number of leaf nodes of each tree.

[0095] Then, the statistical parameters and the corresponding JRC values are divided into training set and test set in the ratio of 8:2 to train the constructed model, and finally the prediction result is obtained.

[0096] The finally obtained GBDT model structure is shown in Figure 4 .

[0097] S6, inputting the test set into the trained GBDT model for prediction.

[0098] S61, inputting the test set into the GBDT model to obtain the prediction result.

[0099] S62, using a loss function to evaluate the predicted value and the true value to obtain the accuracy of the model, so as to verify the accuracy of the prediction.

[0100] The coefficient of determination R 2 , the root mean square error RMSE and the absolute mean error MAE are used to verify the accuracy of the model, and the formula is:

[0101]

[0102]

[0103]

[0104] wherein, y i is the predicted value of the i point, y is the true value of the i point, i is the i value, and n is the total number of values

[0105] The method of the present application is used to predict the roughness of a certain slope rock mass, and the comparison between the original data and the prediction result is shown in Figure 5 . The coefficient of determination R 2 is 0.9742, the root mean square error RMSE is 0.7234, and the absolute mean error MAE is 0.5675.

[0106] It can be seen that the prediction result is basically consistent with the original data, the root mean square error RMSE and the mean square error MSE are relatively small, and the prediction accuracy is high.

[0107] Device embodiment one

[0108] According to the embodiment of the present application, a two-dimensional joint roughness coefficient nonlinear determination device based on GBDT is provided, Figure 6 is a schematic diagram of the two-dimensional joint roughness coefficient nonlinear determination device based on GBDT of the embodiment of the present application, as shown in Figure 6 , the two-dimensional joint roughness coefficient nonlinear determination device based on GBDT according to the embodiment of the present application specifically comprises:

[0109] a scanning module 60, configured to acquire three-dimensional point cloud data of a surface topography of a rock joint sample through scanning;

[0110] a processing module 62, configured to perform simplification processing on the three-dimensional point cloud data, and export profile line point cloud coordinates obtained after the simplification processing as a CSV file;

[0111] a calculation module 64, configured to calculate a joint surface topography statistical parameter value as a feature of a machine learning model;

[0112] a division module 66, configured to acquire a wall rock strength JCS of the rock joint sample through field experiments, perform a direct shear test on a structure surface sample under a certain normal pressure, determine a basic internal friction angle and a peak shear strength of the structure surface sample, calculate a JRC of the joint surface according to a JRC-JCS shear strength model, and divide the calculated JRC and the corresponding statistical parameter value into a training set and a test set according to a certain proportion;

[0113] a prediction module 68, configured to establish a GBDT model taking a decision tree as a base learner, train and adjust the model by using the training set, obtain a trained GBDT model, input the test set into the trained GBDT model for testing, and finally obtain an average value of predicted values as a JRC value of the joint surface predicted by the model.

[0114] The embodiment of the application is a device embodiment corresponding to the above-mentioned method embodiment, and the specific operations of each module can be understood with reference to the description of the method embodiment, which will not be repeated here.

[0115] Device embodiment two

[0116] The embodiment of the application provides an electronic device, as shown in the figure, comprising a memory 70, a processor 72, and a computer program stored on the memory 70 and executable on the processor 72, wherein the computer program is executed by the processor 72 to implement the steps of the two-dimensional joint roughness coefficient nonlinear determination method based on GBDT as described in the method embodiment. Figure 7 Device embodiment three

[0117] The embodiment of the application provides a computer readable storage medium, wherein the computer readable storage medium stores an implementation program of information transmission, and the program is executed by the processor 72 to implement the steps of the two-dimensional joint roughness coefficient nonlinear determination method based on GBDT as described in the method embodiment.

[0118] The computer readable storage medium described in the embodiment includes but is not limited to ROM, RAM, magnetic disk or optical disk, etc.

[0119]

[0120] ​It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A nonlinear determination method for two-dimensional joint roughness coefficients based on GBDT, characterized in that, include: Three-dimensional point cloud data of the surface morphology of rock jointed samples were obtained by scanning; The three-dimensional point cloud data is simplified, and the point cloud coordinates of the profile line obtained after simplification are exported as a CSV file; Calculate the statistical parameters of the joint surface morphology and use them as features of the machine learning model; The wall strength JCS of rock joint samples was obtained through field experiments. Direct shear tests were conducted on the structural surface samples under a certain normal pressure to determine their basic internal friction angle and peak shear strength. The JRC of the joint surface was calculated according to the JRC-JCS shear strength model. The calculated JRC and the corresponding statistical parameter values ​​were divided into training set and test set according to a certain ratio. A GBDT model with decision tree as the base learner is established, and the model is trained and its parameters are tuned using the training set to obtain a trained GBDT model. The test set is then input into the trained GBDT model for testing, and the average value of the final predicted values ​​is the JRC value of the joint surface predicted by the model.

2. The method according to claim 1, characterized in that, The acquisition of three-dimensional point cloud data of the surface morphology of rock jointed samples by scanning specifically includes: A three-dimensional laser scanner with a scanning interval of 0.1 mm was used to scan the rock joint sample to obtain three-dimensional point cloud data of the surface morphology of the rock joint sample.

3. The method according to claim 1, characterized in that, The process of simplifying the 3D point cloud data and exporting the simplified profile point cloud coordinates to a CSV file specifically includes: A three-dimensional model of the surface morphology is established using a predetermined sampling interval. The three-dimensional model of the surface morphology is then adjusted to establish a horizontal XYZ coordinate system. The shear direction of the rock joint sample is taken as the X-axis, and the Y-axis is perpendicular to the shear direction and the X-axis. Based on this, profile lines parallel to the shear direction are taken at intervals of 0.4 mm. The point cloud coordinates of the horizontally calibrated profile lines are then exported as a CSV file.

4. The method according to claim 1, characterized in that, Calculating the statistical parameters of joint surface morphology and using them as features in a machine learning model specifically includes: Five joint surface morphology statistical parameters were selected: root mean square roughness, maximum relative height, height standard deviation, mean tilt angle, and tilt angle standard deviation. These five joint surface morphology statistical parameters were calculated using the corresponding formulas along the center line of the obtained profile and used as features of the machine learning model.

5. The method according to claim 4, characterized in that, The statistical parameters of the five joint surfaces are calculated using the corresponding formulas along the centerline of the obtained profile. Specifically, these parameters include: The least squares median of the profile line is calculated using the least squares method, at the highest point h of the profile line. p With the lowest point h v Draw any reference line parallel to the least squares median. Calculate the area A of the figure enclosed by the profile lines above and below this reference line and the reference line, respectively. u and A d ; Starting from the lowest point of the self-section line, iterative calculations are performed by continuously moving the reference line upwards in parallel until A is reached. u and A d When they are equal, the reference line is the center line of the section line. A reference coordinate system is established with the centerline of the profile line as the x-axis to obtain the coordinates of the profile line. Based on the discrete algorithm of each statistical parameter, a program is written in MATLAB to calculate the statistical parameter values ​​of each profile line.

6. The method according to claim 1, characterized in that, The method further includes: Two-dimensional joint roughness is predicted using a trained GBDT model.

7. The method according to claim 1, characterized in that, Testing the trained GBDT model by inputting the test set specifically includes: Input the test set into the trained GBDT model to obtain the prediction results; The accuracy of the GBDT model is obtained by evaluating the predicted values ​​and the true values ​​using a loss function.

8. A device for nonlinear determination of two-dimensional joint roughness coefficient based on GBDT, characterized in that, include: The scanning module is used to acquire three-dimensional point cloud data of the surface morphology of rock joint samples through scanning; The processing module is used to simplify the three-dimensional point cloud data and export the simplified profile point cloud coordinates as a CSV file. The calculation module is used to calculate the statistical parameter values ​​of the joint surface morphology and use them as features of the machine learning model; The partitioning module is used to obtain the wall strength JCS of rock joint samples through field experiments. Direct shear tests are conducted on the structural surface samples under a certain normal pressure to determine their basic internal friction angle and peak shear strength. The JRC of the joint surface is calculated according to the JRC-JCS shear strength model. The calculated JRC and the corresponding statistical parameter values ​​are divided into training set and test set according to a certain ratio. The prediction module is used to build a GBDT model with decision tree as the base learner, and to train and tune the model with the training set to obtain a trained GBDT model. The test set is then input into the trained GBDT model for testing. Finally, the average value of the predicted values ​​is the JRC value of the joint surface predicted by the model.

9. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the nonlinear determination method for two-dimensional joint roughness coefficients based on GBDT as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an information transmission implementation program, which, when executed by a processor, implements the steps of the nonlinear determination method for two-dimensional joint roughness coefficients based on GBDT as described in any one of claims 1 to 7.

Images