Fracture network reconstruction method based on linkage analysis of rock mass surface and internal structure

Abstract

The application discloses a crack network reconstruction method based on linkage analysis of rock mass surface and internal structure, which comprises the following steps: obtaining a rock mass surface crack image of a region to be researched, and detecting crack morphology by using an edge detection algorithm; setting a scale when taking a picture, and calculating a crack trace length according to a pixel conversion relationship; calculating a mean value and a standard deviation of a long axis of an elliptical crack corresponding to the trace length; obtaining a rock mass drilling crack image of the region to be researched, and extracting a crack spacing and an occurrence data based on the drilling crack image; calculating Fisher parameters of each group of occurrence data; fitting a spacing probability density function according to the spacing between cracks; calculating a crack line density and a crack volume density according to a mean value of the spacing fitted according to the spacing probability density function; constructing a rock mass three-dimensional elliptical crack network of each group of occurrence by using an RJNS 3D toolbox; superimposing the rock mass three-dimensional elliptical crack networks of all groups of occurrence to obtain a rock mass three-dimensional elliptical crack network.

Description

Technical Field

[0001] This invention relates to the field of geotechnical engineering technology, specifically to a method for reconstructing fracture networks based on the linkage analysis of rock mass surface and internal structure. Background Technology

[0002] The internal structure of a rock mass directly determines its strength characteristics and deformation and failure modes. In rock engineering, in order to ensure the safety of the project, it is often necessary to understand the internal structure of the rock mass in order to take corresponding protective measures or adjust the construction plan to ensure the safe progress of the rock engineering project.

[0003] Because deep rock masses often face challenges such as high degree of fragmentation, strong structural concealment, and high spatiotemporal variability, current technologies mainly determine the internal structure of the rock mass through deterministic and indeterminate structural planes. Due to the extremely concealed nature of fractures in deep rock masses, their exploration is extremely difficult, and conducting investigations one by one is impractical. Therefore, existing technologies have developed methods that construct fracture networks to simulate randomly distributed fractures within the rock mass, thereby describing the structural characteristics and spatial distribution of fractures within the rock mass and assessing the safety of rock engineering projects.

[0004] Currently, most methods for constructing fracture networks involve statistically analyzing fracture data revealed by two-dimensional survey windows or one-dimensional survey lines in the field, using the Monte Carlo method to randomly sample fracture parameters, and establishing a fracture network. The model is mainly based on disk fractures, but most structural surfaces revealed in the geological field are pinch-out, so the disk model cannot accurately reflect the true shape of the fractures.

[0005] In the process of constructing the fracture network, the correction of the fracture model is mainly based on the comparison and correction of two-dimensional window data or one-dimensional line data. Neither of these two methods can comprehensively consider the distribution of fractures on the surface and inside, thus failing to reflect the distribution of deep fractures that are highly concealed and have spatiotemporal variability. Summary of the Invention

[0006] To address the aforementioned shortcomings in existing technologies, the fracture network reconstruction method based on the linkage analysis of rock surface and internal structure provided by this invention solves the problem that the fracture models constructed by existing methods cannot reflect the true shape of rock fractures.

[0007] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0008] A fracture network reconstruction method based on the joint analysis of rock mass surface and internal structure is provided, which includes the following steps:

[0009] S1. Obtain images of rock surface fissures in the area to be studied, and use an edge detection algorithm to detect the fissure morphology; set the scale when taking the picture, and calculate the fissure trace length based on the pixel conversion relationship;

[0010] S2. Draw a frequency distribution histogram of the crack trace length, fit the distribution curve of the trace length, and calculate the mean and standard deviation of the major axis of the elliptical crack corresponding to the trace length based on the distribution pattern of the distribution curve.

[0011] S3. Obtain borehole fracture images of the rock mass in the area to be studied, and extract the spacing and orientation data of the fractures based on the borehole fracture images;

[0012] S4. The CH algorithm is used to determine the optimal number of clusters for fracture attitude. Based on the optimal number of clusters, the FCM clustering algorithm is used to group the fracture attitude. Then, the Fisher parameter of each group of attitude data is calculated.

[0013] S5. Based on the spacing between the fractures, fit the spacing probability density function; calculate the fracture linear density and fracture volume density based on the mean spacing value fitted by the spacing probability density function.

[0014] S6. Based on the mean dip, mean dip angle, mean major axis, standard deviation, Fisher parameter, volume density, and rotation angle corresponding to each group of attitudes, the RJNS was used. 3D The toolbox constructs a three-dimensional elliptical fracture network for each group of rock masses with different orientations;

[0015] S7. Superimpose the three-dimensional elliptical fracture network of the rock mass with all groups of orientations to obtain the three-dimensional elliptical fracture network of the rock mass.

[0016] The beneficial effects of this invention are as follows: This solution considers the distribution characteristics of surface and internal fractures of the rock mass when constructing the fracture network. It establishes an elliptical fracture network by combining the elliptical feature parameters of surface fractures and the distribution parameters of borehole fractures. This solves the linkage problem of reconstructing a three-dimensional fracture network from two-dimensional local measurement windows and one-dimensional limited borehole information. As a result, the constructed model can reflect the true shape of the fractures, thereby ensuring that subsequent rock engineering can make relatively accurate construction plans and improve the safety of rock engineering.

[0017] Furthermore, the formula for calculating the density of the fractured volume is as follows:

[0018]

[0019] Where κ is the ratio of the major axis to the minor axis of the elliptical fracture; C j P is the correction parameter for the density of the j-th group of fractures; 30 P represents the volumetric density of the fracture. 10Let be the one-dimensional linear density of the fracture; |n·j| is the absolute value of the dot product of the average unit normal vector of the j-th fracture group and the unit normal vector of a single fracture surface; u a and σ a These are the mean and variance of the major axis of the elliptical crack, respectively.

[0020] The beneficial effects of the above technical solution are as follows: it takes into account the statistical characteristic value of the major axis of the elliptical fracture derived from the surface fracture trace length data and the influence of the borehole fracture linear density on the volume density of the fracture space. At the same time, the introduction of the correction parameter C can be used to correct and fine-tune the volume density for different fracture groups, thereby ensuring the accuracy required for the project.

[0021] Furthermore, the fracture network reconstruction method based on the joint analysis of rock mass surface and internal structure also includes the correction of the three-dimensional fracture network. The correction methods include:

[0022] A1. Obtain multiple rotation angles, and combine them with the mean dip, mean dip angle, mean major axis, standard deviation, Fisher parameter, and volume density corresponding to each group of attitudes, using RJNS... 3D The toolbox constructs a three-dimensional elliptical fracture network of rock masses with each group of orientations at multiple rotation angles;

[0023] A2. Cut the three-dimensional elliptical fracture network of the rock mass at multiple rotation angles for each group of attitudes, and statistically analyze the corresponding trace lengths through the measurement window to obtain the average trace value of each group of attitudes at multiple rotation angles.

[0024] A3. Compare the mean trace length with the mean true trace length of the rock mass, and select the rotation angle corresponding to the mean trace length with the smallest difference from the mean true trace length as the correction rotation angle for this group of attitudes.

[0025] A4. Construct a three-dimensional elliptical fracture network of the rock mass with each group of attitudes after the rotation angle is corrected, and lay out virtual boreholes on the three-dimensional elliptical fracture network of the rock mass to calculate the number of intersections between boreholes and fractures.

[0026] A5. Calculate the ratio of the number of intersections to the borehole length as the correction linear density;

[0027] A6. Adjust the correction factor and calculate the corrected volume density based on the corrected linear density and the adjusted correction factor;

[0028] A7. Determine whether the difference between the corrected volume density and the actual borehole statistical volume density is less than a preset threshold. If yes, proceed to step A8; otherwise, return to step A6.

[0029] A8. Based on the mean dip, mean dip angle, mean major axis, standard deviation, Fisher parameter, corrected volume density, and corrected rotation angle corresponding to each group of attitudes, the RJNS was used. 3D The toolbox constructs a three-dimensional elliptical fracture network for each group of rock masses after attitude correction;

[0030] A9. Superimpose all groups of attitude-corrected three-dimensional elliptical fracture networks of the rock mass to obtain the corrected three-dimensional elliptical fracture network of the rock mass.

[0031] The beneficial effects of the above technical solution are as follows: Based on the constructed three-dimensional elliptical fracture network of the rock mass, this solution combines the average trace length and the statistical volume density of the actual borehole to correct the rotation angle and correction coefficient C. Then, using the previously obtained parameters and the corrected rotation angle and corrected volume density, the three-dimensional elliptical fracture network of the rock mass is reconstructed. After correcting the two parameters, the statistical characteristic value of the sampled trace length of the model is reduced by 48-59% compared with the actual situation in the field, which greatly improves the accuracy of the fracture network.

[0032] Furthermore, the method for obtaining the Fisher parameter includes:

[0033] Calculate the parameter |R| based on the unit vector of the crack on the crack surface:

[0034]

[0035] l i =sinθ i cosφ i m i =sinθ i sinφ i n i =cosθ i

[0036] Among them, (l i ,m i ,n i ) are the unit normal vectors of the i-th crack on the crack surface in the X, Y, and Z directions, respectively; φ i The tendency of the i-th fracture; θ i Let |R| be the dip angle of the i-th fracture; |R| is the resultant vector of the normal vectors of the N fracture surfaces.

[0037] Calculate the Fisher parameters based on the parameter |R| and the total number of cracks on the crack surface:

[0038]

[0039] Where N is the total number of cracks; K is the Fisher parameter.

[0040] The beneficial effects of the above technical solution are as follows: Fisher distribution is one of the most commonly used theoretical probability distributions for describing the attitude of discontinuities in rock mass. Since this distribution contains only one parameter (Fisher constant) and is an integrable function, it is easy to perform Monte Carlo simulation (MCS) and is widely used. Studies have shown that when N has a large value, the Fisher parameter value can be quickly calculated by the above formula, which is convenient for engineering applications.

[0041] Furthermore, based on the distribution pattern of the distribution curve, the mean and standard deviation of the major axis of the elliptical crack corresponding to the trace length are calculated, including:

[0042] When the distribution curve follows a log-normal distribution, the formulas for calculating the mean and standard deviation of the major axis are as follows:

[0043]

[0044]

[0045] When the distribution curve follows a negative exponential distribution, the formulas for calculating the mean and standard deviation along the major axis are as follows:

[0046]

[0047] When the distribution curve follows a Gamma distribution, the formulas for calculating the mean and standard deviation along the major axis are as follows:

[0048]

[0049]

[0050] Where, μ a and σ a These are the mean and standard deviation of the major axis of the elliptical crack, respectively; μ l and σ l denoted as mean trace length and standard deviation of trace length, respectively, and M is the ratio of the minor axis to the major axis of the elliptical crack.

[0051] The beneficial effects of the above technical solution are: quantitatively determining the statistical characteristic value of the major axis of the elliptical fracture based on different trace length distribution forms, avoiding the derivation of complex and accurate analytical solutions for the length of the elliptical fracture, and improving the reconstruction efficiency and engineering applicability of the three-dimensional fracture network.

[0052] Furthermore, based on the distribution pattern of the distribution curve, the distribution range of the major axis and the ratio of the major and minor axes of the elliptical crack are calculated using the probability density model of the major axis of the elliptical crack and the probability density model of the ratio of the major and minor axes of the elliptical crack in the crack network reconstruction method based on two-dimensional window measurement, so as to obtain the mean and standard deviation of the major axis of the elliptical crack.

[0053] Furthermore, methods for grouping fracture orientation using the FCM clustering algorithm include:

[0054] Initialize the fuzzy weighting exponent m, initialize the number of clusters C as the optimal number of clusters, and randomly initialize the membership matrix U(u i |x j ),i=1.2,…,C; j=1,2,…,n;

[0055] Update the cluster centers for each class:

[0056]

[0057] Where, x j v is the j-th sample point; i and v r μ are the cluster centers of the i-th and r-th clusters, respectively; ij For x j With v i membership degree; d ij (x j ,v i ) is x j With v i The distance;

[0058] Proceed to the next iteration until the cluster center of each cluster no longer changes or the maximum number of iterations is reached, at which point the iteration stops. The label of each sample is the index value of the cluster with the highest membership degree.

[0059] The beneficial effects of the above technical solution are as follows: the algorithm provides more flexible clustering results. It iteratively optimizes the objective function based on the weighted similarity measure of the cluster centers to determine the best cluster, thereby reducing the impact on the clustering advantage value when some fracture attitude data deviate significantly.

[0060] Furthermore, the CH algorithm uses the intra-class deviation matrix to describe the density, and the formula for calculating the inter-class deviation matrix is:

[0061]

[0062] Where C is the number of clusters; k is the current cluster; trB(k) is the trace of the inter-cluster deviation matrix; trW(k) is the trace of the intra-cluster deviation matrix; and CH is the density.

[0063] The beneficial effects of the above technical solution are: the optimal number of fracture clusters can be automatically determined through computer calculation, avoiding the influence of human subjective factors on the number of dominant fracture clusters.

[0064] Furthermore, step S1 further includes:

[0065] S11. Obtain images of surface fractures in the rock mass of the area to be studied;

[0066] S12. In MATLAB software, the Candy algorithm is used to obtain the edge data of the crack image, and the upper and lower thresholds [low, high] are adjusted to obtain the optimal binary image of crack detection.

[0067] The crack image boundary is tracked based on the bwboundaries function, and then the number of pixels at the crack image boundary is extracted based on the bwlabel function and regionprops.

[0068] The scale is set when taking the picture, and the length of the crack trace is calculated according to the pixel conversion parameters. The conversion parameters are the ratio of the actual length of the scale to the pixel length of the scale.

[0069] The beneficial effects of the above technical solution are as follows: Three-dimensional fracture network reconstruction usually requires deriving the fracture probability distribution model and statistical characteristic parameters based on the fracture trace data of the determined sampling section. This solution can acquire and extract surface fracture image data more quickly, process a large amount of fracture trace data in a short time, and indirectly improve the reliability of the fracture network model. Attached Figure Description

[0070] Figure 1 This is a flowchart of a fracture network reconstruction method based on the joint analysis of rock mass surface and internal structure.

[0071] Figure 2 Histograms of trace length statistics and fitted distribution curves of trace length are obtained.

[0072] Figure 3 A schematic diagram of the uncorrected three-dimensional elliptical fracture network of the rock mass obtained from reconstruction.

[0073] Figure 4 A schematic diagram of the corrected three-dimensional elliptical fracture network of the rock mass obtained from reconstruction.

[0074] Figure 5 A comparison diagram of the actual fracture orientation isodense pattern (a) and the simulated fracture orientation isodense pattern (b).

[0075] Figure 6 A comparison of a real fracture orientation rose diagram (a) and a simulated fracture orientation rose diagram (b). Detailed Implementation

[0076] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0077] refer to Figure 1 , Figure 1 This paper demonstrates a fracture network reconstruction method based on the combined analysis of rock mass surface and internal structure; such as Figure 1 As shown, the method includes steps S1 to S7.

[0078] In step S1, an image of the rock surface fissures in the area to be studied is acquired, and the fissure morphology is detected using an edge detection algorithm; the scale is set when taking the picture, and the fissure trace length is calculated based on the pixel conversion relationship.

[0079] In one embodiment of the present invention, step S1 further includes:

[0080] S11. Obtain images of surface fractures in the rock mass of the area to be studied;

[0081] S12. In MATLAB software, the Candy algorithm is used to obtain the edge data of the crack image, and the upper and lower thresholds [low, high] are adjusted to obtain the optimal binary image of crack detection.

[0082] The crack image boundary is tracked based on the bwboundaries function, and then the number of pixels at the crack image boundary is extracted based on the bwlabel function and regionprops.

[0083] The scale is set when taking the picture, and the length of the crack trace is calculated according to the pixel conversion parameters. The conversion parameters are the ratio of the actual length of the scale to the pixel length of the scale.

[0084] In step S2, a frequency distribution histogram of the crack trace length is plotted, and the distribution curve of the trace length is obtained by fitting the plot. Figure 2 The diagram shows a histogram and distribution curve; based on the distribution pattern of the distribution curve, the mean and standard deviation of the major axis of the elliptical crack corresponding to the trace length are calculated;

[0085] In one embodiment of the present invention, calculating the mean and standard deviation of the major axis of the elliptical crack corresponding to the trace length, based on the distribution pattern of the distribution curve, includes:

[0086] When the distribution curve follows a log-normal distribution, the formulas for calculating the mean and standard deviation of the major axis are as follows:

[0087]

[0088]

[0089] When the distribution curve follows a negative exponential distribution, the formulas for calculating the mean and standard deviation along the major axis are as follows:

[0090]

[0091] When the distribution curve follows a Gamma distribution, the formulas for calculating the mean and standard deviation along the major axis are as follows:

[0092]

[0093]

[0094] Where, μ a and σ a These are the mean and standard deviation of the major axis of the elliptical crack, respectively; μ l and σ l denoted as mean trace length and standard deviation of trace length, respectively, and M is the ratio of the minor axis to the major axis of the elliptical crack.

[0095] This scheme can also be based on the distribution pattern of the distribution curve, and use the two-dimensional window-based fracture network reconstruction method (the method disclosed in patent application CN202211274059.0) to select the probability density model of the major axis of the elliptical fracture and the probability density model of the ratio of the major and minor axes of the elliptical fracture to calculate the distribution range of the major axis and the ratio of the major and minor axes of the elliptical fracture, so as to obtain the mean and standard deviation of the major axis of the elliptical fracture.

[0096] In step S3, the JL-IDOI(D) borehole imager is used to acquire borehole fracture images of the rock mass in the area to be studied, and the spacing and orientation data of the fractures in the borehole fracture images are extracted based on the software provided with the JL-IDOI(D) borehole imager.

[0097] In step S4, the CH algorithm is used to determine the optimal number of clusters for fracture orientation. The CH algorithm uses the intra-cluster deviation matrix to describe the density, and the formula for calculating the inter-cluster deviation matrix is ​​as follows:

[0098]

[0099] Where C is the number of clusters; k is the current cluster; trB(k) is the trace of the inter-cluster deviation matrix; trW(k) is the trace of the intra-cluster deviation matrix; and CH is the density.

[0100] The higher the density, the more compact the clusters themselves and the more dispersed the clusters are, which means that the clustering results are better. This scheme can determine the optimal number of groups for the fracture orientation set by trying different numbers of clusters and calculating the CH value.

[0101] Then, based on the optimal number of clusters, the FCM clustering algorithm was used to group the fracture attitude data, and the Fisher parameters of each group of attitude data were calculated:

[0102] Calculate the parameter |R| based on the unit vector of the crack on the crack surface:

[0103]

[0104] l i =sinθ i cosφ i m i =sinθ i sinφ i n i =cosθ i

[0105] Among them, (l i m i n i ) are the unit normal vectors of the i-th crack on the crack surface in the X, Y, and Z directions, respectively; φ i The tendency of the i-th fracture; θ i Let |R| be the dip angle of the i-th fracture; |R| is the resultant vector of the normal vectors of the N fracture surfaces.

[0106] Calculate the Fisher parameters based on the parameter |R| and the total number of cracks on the crack surface:

[0107]

[0108] Where N is the total number of cracks; K is the Fisher parameter.

[0109] In implementation, this scheme preferably uses the FCM clustering algorithm to group fracture orientations, including the following methods:

[0110] Initialize the fuzzy weighting exponent m, initialize the number of clusters C as the optimal number of clusters, and randomly initialize the membership matrix U(u i |x j ),i=1.2,…,C; j=1,2,…,n;

[0111] Update the cluster centers for each class:

[0112]

[0113] Where, x j v is the j-th sample point; i and v r μ are the cluster centers of the i-th and r-th clusters, respectively; ij For x j With v i membership degree; d ij (x j ,v i ) is x j With v i The distance;

[0114] Proceed to the next iteration until the cluster center of each cluster no longer changes or the maximum number of iterations is reached, at which point the iteration stops. The label of each sample is the index value of the cluster with the highest membership degree.

[0115] In step S5, the spacing probability density function is obtained by fitting the spacing between the cracks using MATLAB tools; the crack linear density is calculated based on the mean spacing value fitted by the spacing probability density function: P10 = 1 / E(d), where P 10 Let E(d) be the linear density of the fracture, and E(d) be the mean of the fitted fracture spacing; then the fracture volume density is calculated.

[0116]

[0117] Where κ is the ratio of the major axis to the minor axis of the elliptical fracture; C j P is the correction parameter for the density of the j-th group of fractures; 30 P represents the volumetric density of the fracture. 10 Let be the one-dimensional linear density of the fracture; |n·j| is the absolute value of the dot product of the average unit normal vector of the j-th fracture group and the unit normal vector of a single fracture surface; u a and σ a These are the mean and variance of the major axis of the elliptical crack, respectively.

[0118] In step S6, based on the mean dip, mean dip angle, mean major axis, standard deviation, Fisher parameter, volume density, and rotation angle corresponding to each group of attitudes, the RJNS is used. 3D The toolbox constructs a three-dimensional elliptical fracture network for each group of rock masses; when constructing the fracture network here, the initial correction parameter is 1 and the rotation angle is 0 degrees.

[0119] In step S7, the three-dimensional elliptical fracture networks of the rock mass with all groups of orientations are superimposed to obtain the three-dimensional elliptical fracture network of the rock mass. For details, please refer to [reference needed]. Figure 3 .

[0120] In one embodiment of the present invention, the fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure is characterized by further including the correction of the three-dimensional fracture network, the correction method including:

[0121] A1. Obtain multiple rotation angles (30°, 60°, 90°, 120°, 150°, and 180° in this scheme), and combine them with the mean dip, mean dip angle, mean long axis, standard deviation, Fisher parameter, and volume density corresponding to each group of attitudes, using RJNS... 3D The toolbox constructs a three-dimensional elliptical fracture network of rock masses with each group of orientations at multiple rotation angles;

[0122] A2. Cut the three-dimensional elliptical fracture network of the rock mass at multiple rotation angles for each group of attitudes, and statistically analyze the corresponding trace lengths through the measurement window to obtain the average trace value of each group of attitudes at multiple rotation angles.

[0123] A3. Compare the mean trace length with the mean true trace length of the rock mass, and select the rotation angle corresponding to the mean trace length with the smallest difference from the mean true trace length as the correction rotation angle for this group of attitudes.

[0124] A4. Construct a three-dimensional elliptical fracture network of the rock mass with each group of attitudes after the rotation angle is corrected, and lay out virtual boreholes on the three-dimensional elliptical fracture network of the rock mass to calculate the number of intersections between boreholes and fractures.

[0125] A5. Calculate the ratio of the number of intersections to the borehole length as the correction linear density;

[0126] A6. Adjust the correction factor and calculate the corrected volume density based on the corrected linear density and the adjusted correction factor;

[0127] A7. Determine whether the difference between the corrected volume density and the actual borehole statistical volume density is less than a preset threshold. If yes, proceed to step A8; otherwise, return to step A6.

[0128] A8. Based on the mean dip, mean dip angle, mean major axis, standard deviation, Fisher parameter, corrected volume density, and corrected rotation angle corresponding to each group of attitudes, the RJNS was used. 3D The toolbox constructs a three-dimensional elliptical fracture network for each group of rock masses after attitude correction;

[0129] A9. By superimposing all groups of attitude-corrected three-dimensional elliptical fracture networks of the rock mass, the corrected three-dimensional elliptical fracture network of the rock mass is obtained. For details, please refer to... Figure 4 .

[0130] The following example illustrates the effectiveness of the reconstruction method in this solution:

[0131] Close-range photogrammetry of the tunnel face was continuously carried out in a tunnel under construction in Southwest China. A large amount of surface image crack data was obtained on site, and internal cracks of four boreholes were collected by borehole inspection. A total of 935 crack parameter data were collected as analysis samples. The crack trace parameters were obtained using the method of this scheme, as shown in Table 1.

[0132] Table 1. Statistics of fracture trace parameters

[0133]

[0134] The CH+FCM clustering method used in this scheme was used to quantitatively divide the collected fracture samples into dominant groups. After grouping, two dominant fracture groups (attitude groups) were obtained, as shown in Table 2:

[0135] Table 2 Grouping of Preferred Fractures

[0136]

[0137] Six rotation angles were selected and the correction method of this scheme was applied. The optimal rotation angle was selected as the final corrected rotation angle. The comparison results of the two sets of crack trace lengths after the correction of the rotation angle and the crack trace lengths without the correction of the rotation angle can be referred to Table 3 and Table 4, respectively.

[0138] Table 3, Group 1, Crack Trace Length Correction

[0139]

[0140] Table 4, Group 2, Crack Trace Length Correction

[0141]

[0142] The correction coefficients for the two groups of fractures were adjusted. During the adjustment, nine virtual boreholes were set up on the model. For details of the adjustment, please refer to Tables 5 and 6.

[0143] Table 5. C-adjustment of the first group of dominant fractures.

[0144]

[0145]

[0146] Table 6. C-adjustment of dominant fractures in the second group.

[0147]

[0148]

[0149] Tables 5 and 6 show that as the value of C increases, the fracture density continuously increases, and the number of borehole fracture intersections also shows an increasing trend. Without adjusting the correction parameter C (C=1), the number of intersections between the two sets of fractures and the virtual borehole is generally low, and the calculated fracture linear density is significantly lower than the actual measured situation in the field. As the correction coefficient C increases, the simulated linear density gradually approaches the actual linear density. When the first set of correction coefficients C increases to 2.2, and the second set of correction coefficients C increases to 1.12, the simulated linear density matches the actual linear density optimally.

[0150] After correcting the rotation angle and correction coefficient using a correction method, the final input RJNS is obtained. 3D The modeling parameters for the toolbox are shown in Table 7.

[0151] Table 7. Corrected modeling parameters

[0152]

[0153] The three-dimensional elliptical fracture network of the rock mass constructed using this scheme was used to simulate various parameters within the rock mass, and the simulation results were compared with those obtained from the actual rock mass. The comparison results are shown in Table 8.

[0154] Table 8 Comparison of various indicators of the fracture network model with those of the actual rock mass.

[0155]

[0156] By examining multiple indicators such as model dip, tilt angle, trace length distribution statistical characteristics, and fracture line density, it can be found that the model's indicators match the actual situation well, demonstrating the feasibility and accuracy of the proposed joint reconstruction method. It should be noted that the fracture trace length correction based on rotation angle in this scheme is adjusted in 30° steps. For engineering applications, the correction process is simple and rapid, and the maximum error does not exceed 33%. If further improvement in accuracy is required, a smaller step size for the rotation angle can be adopted.

[0157] The fracture network constructed in this scheme and the attitude data collected from field boreholes were plotted to obtain a fracture attitude isodense map. Figure 5 ) and the path to the rose ( Figure 6 ).

[0158] By comparing the fracture density diagrams, it can be found that the orientation of the two sets of fractures simulated in this scheme matches the actual situation. The fractures tend to be in the NWW and NNE directions, with the fractures tending in the NNE direction being more concentrated and more numerous than the other dominant direction. Comparing the fracture strike rose diagram, the same conclusion can be drawn: the fracture strikes are in the SEE and SSW directions, with the fractures tending in the SEE direction being more concentrated and more numerous than the other dominant direction.

Claims

1. A fracture network reconstruction method based on the joint analysis of rock mass surface and internal structure, characterized in that, Including the following steps: S1. Obtain images of rock surface fissures in the area to be studied, and use an edge detection algorithm to detect the fissure morphology; set the scale when taking the picture, and calculate the fissure trace length based on the pixel conversion relationship; S2. Draw a frequency distribution histogram of the crack trace length, fit the distribution curve of the trace length, and calculate the mean and standard deviation of the major axis of the elliptical crack corresponding to the trace length based on the distribution pattern of the distribution curve. S3. Obtain borehole fracture images of the rock mass in the area to be studied, and extract the spacing and orientation data of the fractures based on the borehole fracture images; S4. The CH algorithm is used to determine the optimal number of clusters for fracture attitude. Based on the optimal number of clusters, the FCM clustering algorithm is used to group the fracture attitude. Then, the Fisher parameter of each group of attitude data is calculated. S5. Based on the spacing between the fractures, fit the spacing probability density function; calculate the fracture linear density and fracture volume density based on the mean spacing value fitted by the spacing probability density function. S6. Based on the mean dip, mean dip angle, mean major axis, standard deviation, Fisher parameter, volume density, and rotation angle corresponding to each group of attitudes, the RJNS was used. 3D The toolbox constructs a three-dimensional elliptical fracture network for each group of rock masses with different orientations; S7. Superimpose the three-dimensional elliptical fracture network of the rock mass with all groups of orientations to obtain the three-dimensional elliptical fracture network of the rock mass.

2. The fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure according to claim 1, characterized in that, The formula for calculating the density of the fractured volume is: in, This is the ratio of the major axis to the minor axis of the elliptical crack; C j For the first j Correction factor for the density of the fractured mass; P 30 The spatial volume density of the fracture; P 10 The one-dimensional linear density of the fracture; For the first j The absolute value of the dot product of the average unit normal vector of the group of fractures and the unit normal vector of a single fracture surface; and These are the mean and standard deviation of the major axis of the elliptical crack, respectively.

3. The fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure according to claim 2, characterized in that, It also includes the modification of the three-dimensional fracture network, and the modification methods include: A1. Obtain multiple rotation angles, and combine them with the mean dip, mean dip angle, mean major axis, standard deviation, Fisher parameter, and volume density corresponding to each group of attitudes, using RJNS... 3D The toolbox constructs a three-dimensional elliptical fracture network of rock masses with each group of orientations at multiple rotation angles; A2. Cut the three-dimensional elliptical fracture network of the rock mass at multiple rotation angles for each group of attitudes, and statistically analyze the corresponding trace lengths through the measurement window to obtain the average trace value of each group of attitudes at multiple rotation angles. A3. Compare the mean trace length with the mean true trace length of the rock mass, and select the rotation angle corresponding to the mean trace length with the smallest difference from the mean true trace length as the correction rotation angle for this group of attitudes. A4. Construct a three-dimensional elliptical fracture network of the rock mass with each group of attitudes after the rotation angle is corrected, and lay out virtual boreholes on the three-dimensional elliptical fracture network of the rock mass to calculate the number of intersections between boreholes and fractures. A5. Calculate the ratio of the number of intersections to the borehole length as the correction linear density; A6. Adjust the correction factor and calculate the corrected volume density based on the corrected linear density and the adjusted correction factor; A7. Determine whether the difference between the corrected volume density and the actual borehole statistical volume density is less than a preset threshold. If yes, proceed to step A8; otherwise, return to step A6. A8. Based on the mean dip, mean dip angle, mean major axis, standard deviation, Fisher parameter, corrected volume density, and corrected rotation angle corresponding to each group of attitudes, the RJNS was used. 3D The toolbox constructs a three-dimensional elliptical fracture network for each group of rock masses after attitude correction; A9. Superimpose all groups of attitude-corrected three-dimensional elliptical fracture networks of the rock mass to obtain the corrected three-dimensional elliptical fracture network of the rock mass.

4. The fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure according to claim 1, characterized in that, The methods for obtaining the Fisher parameter include: Calculate the parameters based on the unit vector of the crack on the crack surface. : ， ， in,( l i , m i , n i ) are respectively the first on the fracture surface i Root fissures X , Y and Z Unit normal vectors in three directions; For the first i The tendency for root fissures; θ i For the first i The inclination angle of the root fracture; for N The resultant vector of the normal vectors of the fracture surfaces; According to parameters Calculate the Fisher parameters based on the total number of cracks on the crack surface: in, N This represents the total number of cracks. K For Fisher parameters.

5. The fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure according to claim 1, characterized in that, Based on the distribution pattern of the distribution curve, the mean and standard deviation of the major axis of the elliptical crack corresponding to the trace length are calculated as follows: When the distribution curve follows a log-normal distribution, the formulas for calculating the mean and standard deviation of the major axis are as follows: When the distribution curve is a negative exponential distribution, the formulas for calculating the mean and standard deviation of the major axis are as follows: ， When the distribution curve follows a Gamma distribution, the formulas for calculating the mean and standard deviation along the major axis are as follows: in, μ a and σ a These are the mean and standard deviation of the major axis of the elliptical crack, respectively; μ l and σ l These are the mean trace length and the standard deviation of trace length, respectively. M It is the ratio of the minor axis to the major axis of the elliptical crack.

6. The fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure according to claim 1, characterized in that, Based on the distribution pattern of the distribution curve, the probability density model of the major axis of the elliptical crack and the probability density model of the ratio of the major and minor axes of the elliptical crack are selected to calculate the distribution range of the major axis and the ratio of the major and minor axes of the elliptical crack, and the mean and standard deviation of the major axis of the elliptical crack are obtained.

7. The fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure according to claim 1, characterized in that, Methods for grouping fracture orientation using the FCM clustering algorithm include: Initialize fuzzy weighted index m Initialize the number of clusters C To determine the optimal number of clusters, randomly initialize the membership matrix. ; Update the cluster centers for each class: ， in, x j For the first j One sample point; v i and v r The first i The cluster and the first r Cluster centers of each cluster; for x j and v i Membership degree; for x j and v i The distance; Proceed to the next iteration until the cluster center of each cluster no longer changes or the maximum number of iterations is reached, at which point the iteration stops. The label of each sample is the index value of the cluster with the highest membership degree.

8. The fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure according to any one of claims 1-7, characterized in that, The CH algorithm uses an intra-class deviation matrix to describe the tightness, and the formula for calculating the tightness is: in, C The number of clusters; k For the current class; The trace of the inter-class deviation matrix; is the trace of the intra-class deviation matrix; CH is the compactness.

9. The fracture network reconstruction method based on the linkage analysis of rock mass surface and internal structure according to any one of claims 1-7, characterized in that, Step S1 further includes: S11. Obtain images of surface fractures in the rock mass of the area to be studied; S12. In MATLAB software, the Canny algorithm is used to obtain the edge data of the crack image, and the upper and lower thresholds [low, high] are adjusted to obtain the optimal binary image of crack detection. The crack image boundary is tracked using the bwboundaries function, and then the number of pixels at the crack image boundary is extracted using the bwlabel and regionprops functions. Based on the scale set during photography, the length of the crack trace is calculated according to the pixel conversion parameter, which is the ratio of the actual length of the scale to the pixel length of the scale.

Images