Rock mass structure surface three-dimensional topography generation method and device, storage medium and computer equipment

Abstract

The application relates to the technical field of geotechnical engineering, and particularly discloses a rock mass structure surface three-dimensional morphology generation method and device, a storage medium and computer equipment. The method comprises the following steps: obtaining trace position data corresponding to a rock mass structure surface exposed trace, and constructing an original trace curve based on the trace position data; adopting a spherical basis function to randomly generate a plurality of three-dimensional candidate rough structure surfaces with heterogeneous and anisotropic characteristics; for each three-dimensional candidate rough structure surface, extracting a simulated trace from the three-dimensional candidate rough structure surface according to the coordinate position of the original trace curve; performing weighted combination fitting on the simulated traces of the three-dimensional candidate rough structure surfaces with the original trace curve as a weighted combination target, and determining the weight corresponding to each simulated trace; and weighting and superimposing the three-dimensional candidate rough structure surfaces according to the corresponding weights to generate a rock mass structure surface three-dimensional morphology corresponding to the original trace curve.

Description

Technical Field

[0001] This application relates to the field of geotechnical engineering technology, and in particular to a method and apparatus for generating three-dimensional morphology of rock mass structural surfaces, a storage medium, and a computer device. Background Technology

[0002] In geotechnical and geological engineering, the widely distributed structural surfaces (such as joints, fissures, bedding, and faults) within rock masses are key factors controlling their mechanical and hydraulic behavior. The three-dimensional morphology of these structural surfaces directly determines the rock mass's peak shear strength, dilatational properties, frictional characteristics, and the distribution and permeability of seepage channels. Therefore, accurately obtaining the three-dimensional morphology of rock mass structural surfaces is crucial for tunnel excavation, slope stability analysis, mineral resource assessment, and geological hazard prediction. However, in practical engineering, most structural surfaces are often hidden within the rock mass, and engineers can only observe their outcrops through exposed surfaces such as tunnel walls, outcrops, and core surfaces. How to infer the three-dimensional roughness morphology of structural surfaces at depth based on limited outcrop information has become a long-standing fundamental challenge in rock mass engineering investigation and design.

[0003] To address the aforementioned challenges, existing technologies primarily employ two strategies. One is to stretch the observed exposed traces parallel to their normal direction, directly generating a three-dimensional structural surface with constant thickness or height. While simple to implement, this method ignores the inherent roughness information of the traces themselves, resulting in the complete loss of the original traces' non-uniform distribution and anisotropic details along the stretching direction. The generated structural surface is overly smooth, deviating significantly from the actual roughness. The second strategy is based on statistical methods, simulating the distribution of rock mass structural surfaces through random fields or random fracture networks. While these methods can reproduce the random variations of rock masses, they typically simplify the traces to abstract geometric elements (such as straight line segments or arcs), discarding key morphological features such as twists and turns and unevenness in the actual trace shape. This leads to significant uncontrollable errors between the simulation results and the actual structural surface.

[0004] Despite their significant limitations, the aforementioned existing technologies still possess certain application value in engineering practice. The parallel stretching method boasts extremely high computational efficiency and requires no complex parameters, making it suitable for rapidly and approximate estimation of structural plane attitudes. Statistical methods, on the other hand, can generate multiple feasible random samples of structural planes, providing a data foundation for rock mass reliability analysis and uncertainty assessment. Furthermore, both methods do not rely on complex geological information other than trace lines, facilitating rapid adoption by field technicians. However, in engineering scenarios with extremely stringent requirements for rock mass stability, such as steep slopes and deep-buried tunnels, the shortcomings of existing technologies in terms of accuracy and realism become increasingly apparent. Summary of the Invention

[0005] In view of this, this application provides a method, apparatus, storage medium, and computer equipment for generating three-dimensional morphology of rock mass structural surfaces. It employs spherical basis functions to randomly generate multiple candidate three-dimensional rough structural surfaces with heterogeneous and anisotropic characteristics. The optimal weights are automatically determined through weighted combination fitting with the original trace curves, thus avoiding the drawback of losing roughness information in the normal direction in the parallel stretching method. The resulting superimposed rock mass structural surfaces retain the original non-uniform undulations of the traces while exhibiting true anisotropic roughness characteristics in different directions, greatly improving the fidelity of the reconstructed three-dimensional morphology of the rock mass structural surfaces. Using the measured exposed traces as the direct constraint target, weighted fitting in the least squares sense ensures that the simulated trace superposition results highly match the original trace curves. This overcomes the problem of pure statistical methods simplifying traces to abstract geometric elements and discarding key morphological details, ensuring that the reconstructed results are closely related to real geological information and significantly reducing simulation errors. Furthermore, while retaining the advantages of efficient computation of the parallel stretching method and random sampling of the statistical method, the embodiments of this application realize the intelligent screening and fusion of a large number of three-dimensional candidate rough structural surfaces through an adaptive weight allocation mechanism. It can generate high-precision three-dimensional morphology of rock mass structural surfaces without manual intervention, which is especially suitable for engineering scenarios with stringent requirements for rock mass stability evaluation, such as steep slopes and deep buried tunnels, and provides a reliable foundation for subsequent mechanical analysis and hydraulic coupling calculation.

[0006] According to one aspect of this application, a method for generating the three-dimensional morphology of rock mass structural surfaces is provided, comprising: Obtain the trace location data corresponding to the exposed traces of the rock mass structural surface, and construct the original trace curve based on the trace location data; Using spherical basis functions, multiple three-dimensional candidate rough structure surfaces with heterogeneous and anisotropic characteristics are randomly generated. For each three-dimensional candidate rough structure surface, a simulated trace is extracted from the three-dimensional candidate rough structure surface according to the coordinate position of the original trace curve. Using the original trace curve as the weighted combination target, the simulated traces of each three-dimensional candidate rough structure surface are weighted and combined for fitting to determine the weight corresponding to each simulated trace. Each candidate three-dimensional rough structure surface is weighted and superimposed according to its corresponding weight to generate a three-dimensional morphology of the rock mass structure surface corresponding to the original trace curve.

[0007] According to another aspect of this application, an apparatus for generating three-dimensional morphology of rock mass structural surfaces is provided, comprising: The data acquisition module is used to acquire the trace position data corresponding to the exposed traces of the rock mass structural surface, and to construct the original trace curve based on the trace position data; The simulated trace extraction module is used to randomly generate multiple three-dimensional candidate rough structure surfaces with heterogeneous and anisotropic characteristics using spherical basis functions. For each three-dimensional candidate rough structure surface, a simulated trace is extracted from the three-dimensional candidate rough structure surface according to the coordinate position of the original trace curve. The fitting module is used to perform weighted combination fitting on the simulated traces of each three-dimensional candidate rough structure surface using the original trace curve as the weighted combination target, and to determine the weight corresponding to each simulated trace. The overlay module is used to weight and overlay each three-dimensional candidate rough structure surface according to its corresponding weight to generate a three-dimensional morphology of the rock mass structure surface corresponding to the original trace curve.

[0008] According to another aspect of this application, a storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the above-described method for generating the three-dimensional morphology of rock mass structural surfaces.

[0009] According to another aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the above-described method for generating the three-dimensional morphology of the rock mass structure surface.

[0010] By employing the above technical solution, this application provides a method, apparatus, storage medium, and computer equipment for generating three-dimensional morphology of rock mass structural surfaces. It uses spherical basis functions to randomly generate multiple candidate three-dimensional rough structural surfaces with heterogeneous and anisotropic characteristics. The optimal weights are automatically determined through weighted combination fitting with the original trace curves, thus avoiding the drawback of losing roughness information in the normal direction in the parallel stretching method. The resulting superimposed rock mass structural surfaces retain the original non-uniform undulations of the traces while exhibiting true anisotropic roughness characteristics in different directions, greatly improving the fidelity of the reconstructed three-dimensional morphology of the rock mass structural surfaces. Using the measured exposed traces as the direct constraint target, the least-squares weighted fitting ensures that the simulated trace superposition results highly match the original trace curves. This overcomes the problem of pure statistical methods simplifying traces to abstract geometric elements and discarding key morphological details, ensuring that the reconstructed results are closely related to real geological information and significantly reducing simulation errors. Furthermore, while retaining the advantages of efficient computation of the parallel stretching method and random sampling of the statistical method, the embodiments of this application realize the intelligent screening and fusion of a large number of three-dimensional candidate rough structural surfaces through an adaptive weight allocation mechanism. It can generate high-precision three-dimensional morphology of rock mass structural surfaces without manual intervention, which is especially suitable for engineering scenarios with stringent requirements for rock mass stability evaluation, such as steep slopes and deep buried tunnels, and provides a reliable foundation for subsequent mechanical analysis and hydraulic coupling calculation.

[0011] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0012] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a method for generating a three-dimensional morphology of a rock mass structural surface according to an embodiment of this application is shown. Figure 2 A schematic diagram of an original trace curve provided in an embodiment of this application is shown; Figure 3 A schematic diagram of n three-dimensional candidate rough structure surfaces provided in an embodiment of this application is shown; Figure 4 A schematic diagram of a rock mass structure surface provided in an embodiment of this application is shown; Figure 5 This illustration shows a comparison diagram between an original trace curve and a fitted curve provided in an embodiment of this application; Figure 6 A schematic diagram of the original trace curve of the exposed trace of a certain tunnel structure surface provided in an embodiment of this application is shown; Figure 7 A schematic diagram of the fitting curve of the exposed trace of a certain tunnel structure surface provided in an embodiment of this application is shown; Figure 8 A schematic diagram of a deduced tunnel structure surface provided in an embodiment of this application is shown; Figure 9 A schematic diagram of the original trace curve of a borehole core trace provided in an embodiment of this application is shown; Figure 10 This illustration shows a schematic diagram of a deduced core structure surface from an embodiment of this application; Figure 11 A schematic diagram of a device for generating three-dimensional morphology of rock mass structural surfaces provided in an embodiment of this application is shown. Figure 12 A schematic diagram of the device structure of a computer device provided in an embodiment of this application is shown. Detailed Implementation

[0013] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0014] This embodiment provides a method for generating the three-dimensional morphology of rock mass structural surfaces, such as... Figure 1 As shown, the method includes: Step 101: Obtain the trace location data corresponding to the exposed trace of the rock mass structural surface, and construct the original trace curve based on the trace location data.

[0015] Step 102: Using spherical basis functions, multiple three-dimensional candidate rough structure surfaces with heterogeneous and anisotropic characteristics are randomly generated. For each three-dimensional candidate rough structure surface, a simulated trace is extracted from the three-dimensional candidate rough structure surface according to the coordinate position of the original trace curve.

[0016] Step 103: Using the original trace curve as the weighted combination target, perform weighted combination fitting on the simulated traces of each three-dimensional candidate rough structure surface to determine the weight corresponding to each simulated trace.

[0017] Step 104: The three-dimensional candidate rough structure surfaces are weighted and superimposed according to their corresponding weights to generate the three-dimensional morphology of the rock mass structure surface corresponding to the original trace curve.

[0018] This application provides a method for generating the three-dimensional morphology of rock mass structural surfaces. First, it acquires the location data of the exposed traces of the rock mass structural surfaces and constructs original trace curves based on this data. Here, exposed traces refer to the visible marks formed when the rock mass structural surface intersects with the free surface of the rock mass (such as a tunnel wall, slope surface, or borehole core surface). The spatial location information of these marks can be acquired through three-dimensional laser scanning or photogrammetry, thus obtaining the trace location data. The original trace curves accurately record the geometric shape left by the rock mass structural surface on the rock surface and are the sole source of surface constraints for all subsequent inferences and reconstructions in this application.

[0019] Next, multiple 3D candidate rough surface structures with heterogeneity and anisotropy are randomly generated using spherical basis functions. A spherical basis function is a mathematical method that constructs complex surfaces by randomly generating a large number of spheres in 3D space, each contributing positive or negative to points within a certain radius around it, and then linearly superimposing all contributions. Since the position, radius, and sign of the spheres are randomly set, each generated 3D candidate rough surface exhibits irregular undulations in space (i.e., heterogeneity) and varying degrees of roughness in different directions (i.e., anisotropy). This method allows for the generation of a large number of 3D candidate rough surface structures with diverse shapes at once. For each generated 3D candidate rough surface, attribute values ​​at corresponding positions on the original trace curve are extracted from the surface to form a simulated trace.

[0020] Furthermore, using the original trace curve as the weighted combination objective, a weighted combination fitting is performed on the simulated traces of all three-dimensional candidate roughness surfaces to determine the weight corresponding to each simulated trace. Each simulated trace represents the undulation characteristics of the corresponding three-dimensional candidate roughness surface at the position of the original trace curve, while the actual rock mass structure should be the result of mixing all these three-dimensional candidate roughness surfaces in a certain optimal proportion. A set of non-negative weight values ​​is solved using the least squares method or other optimization algorithms, minimizing the sum of squared residuals between the composite curve obtained by superimposing each simulated trace according to this set of weights and the original trace curve. Thus, the better the simulated trace fits (i.e., the closer its shape is to the original trace curve), the higher the weight can be assigned, while the weight of simulated traces with large differences approaches zero. The entire process automatically completes the selection of the combination that contributes the most to the actual rock mass structure from numerous randomly generated three-dimensional candidate roughness surfaces.

[0021] Finally, the candidate three-dimensional rough structural surfaces are weighted and superimposed according to the corresponding weights determined in the previous step to generate the final three-dimensional morphology of the rock mass structural surface corresponding to the original trace curve. This three-dimensional morphology of the rock mass structural surface not only accurately restores the orientation of the exposed trace on the rock surface, but also reconstructs the rough morphology of the structural surfaces hidden inside the rock mass by leveraging the heterogeneity and anisotropy inherent in the candidate three-dimensional rough structural surfaces generated by the spherical basis function.

[0022] By applying the technical solution of this embodiment, multiple three-dimensional candidate rough structural surfaces with heterogeneous and anisotropic characteristics are randomly generated using spherical basis functions. The optimal weights are automatically determined by weighted combination fitting with the original trace curves, thereby avoiding the drawback of losing roughness information in the normal direction in the parallel stretching method. The resulting superimposed rock mass structural surfaces retain the original non-uniform undulations of the traces and present real anisotropic roughness characteristics in different directions, greatly improving the reconstruction fidelity of the three-dimensional morphology of the rock mass structural surfaces. Using the measured exposed traces as the direct constraint target, the weighted fitting in the least squares sense makes the simulated trace superposition results highly consistent with the original trace curves. This overcomes the problem of pure statistical methods simplifying traces to abstract geometric elements and discarding key morphological details, ensuring that the reconstruction results are closely related to real geological information and significantly reducing simulation errors. Furthermore, while retaining the advantages of efficient computation of the parallel stretching method and random sampling of the statistical method, the embodiments of this application realize the intelligent screening and fusion of a large number of three-dimensional candidate rough structural surfaces through an adaptive weight allocation mechanism. It can generate high-precision three-dimensional morphology of rock mass structural surfaces without manual intervention, which is especially suitable for engineering scenarios with stringent requirements for rock mass stability evaluation, such as steep slopes and deep buried tunnels, and provides a reliable foundation for subsequent mechanical analysis and hydraulic coupling calculation.

[0023] In this embodiment of the application, optionally, each three-dimensional candidate rough structure surface is generated based on the following steps: setting the total number of spheres, the mean and standard deviation of the normal distribution followed by the sphere radii, and the spatial range of the structure surface to be generated; randomly generating the total number of spheres within the spatial range, wherein each sphere has a random spatial position, a random radius, and a random sign, the random radius being obtained by sampling the normal distribution and taking the absolute value, and the random sign being a randomly assigned positive or negative sign; for each spatial point within the spatial range, linearly superimposing the contribution values ​​of all spheres to the spatial point to obtain the attribute value of the spatial point, wherein the contribution value of each sphere is jointly determined by the random radius of the sphere, the distance between the center of the sphere and the spatial point, and the random sign of the sphere; combining the planar coordinates of each spatial point with the corresponding attribute value to form a three-dimensional spatial point, and generating a three-dimensional candidate rough structure surface with non-homogeneous and anisotropic characteristics through all three-dimensional spatial points.

[0024] In this embodiment, three key parameters are first set: the total number of spheres, the mean and standard deviation of the normal distribution followed by the sphere radii, and the spatial extent of the structural surface to be generated. Here, the spherical basis function method is a technique for constructing complex surfaces by randomly stacking a large number of spheres. The total number of spheres determines the refinement of the subsequent surface; the more spheres, the richer the surface details. The mean and standard deviation of the normal distribution control the distribution of sphere radii; for example, the mean determines the average radius, and the standard deviation determines the dispersion of the radius. Since the radius must be positive, this can be ensured by sampling the normal distribution and taking the absolute value during actual generation. The spatial extent defines the region where the structural surface to be generated is located, i.e., the three-dimensional boundary for all subsequent sphere placement and attribute calculations. These three parameters together constitute the initial conditions for the generation process.

[0025] Next, a specified number of spheres are randomly generated within the defined spatial range. Each sphere has three independent random attributes: spatial location, radius, and sign. The spatial location is uniformly and randomly distributed within the spatial range; the radius is obtained by sampling the aforementioned normal distribution and taking the absolute value, thus ensuring that the radius is always non-negative and follows a defined statistical distribution; the sign is randomly assigned as positive or negative, with a positive sign indicating a positive contribution (similar to a bulge) to the surrounding space and a negative sign indicating a negative contribution (similar to a depression). This randomness ensures that each three-dimensional candidate rough structure surface has a unique morphology, providing rich diversity for subsequent screening.

[0026] Then, for each spatial point within the spatial range, the contributions of all spheres to that point can be calculated and linearly superimposed to obtain the attribute value of that point. The contribution value of each sphere is determined by its radius, the distance from its center to the point, and its own sign: the larger the radius, the wider the contribution range; the closer the distance, the stronger the contribution; and the sign determines the positive or negative of the contribution. Generally, the contribution function can be designed as a function that decays with distance (e.g., Gaussian or linear decay). By summing the contributions of all spheres, the final attribute value of the spatial point is obtained, which reflects the relative height or intensity of the point after being affected by the superposition of all random spheres. This process essentially generates a non-uniform scalar field in three-dimensional space.

[0027] Furthermore, the planar coordinates of each spatial point (i.e., the x and y positions on a two-dimensional plane) are combined with the calculated attribute value (i.e., the z value) to form a complete three-dimensional spatial point. All three-dimensional spatial points together constitute a discrete point cloud. Since the position, radius, and sign of the sphere are randomly generated, the attribute values ​​of different spatial points are different, and the rates of change in different directions also differ. Therefore, the three-dimensional candidate rough structure surface composed of these three-dimensional spatial points naturally possesses heterogeneity (i.e., spatial inhomogeneity) and anisotropy (i.e., different degrees of roughness in different directions). Finally, by meshing or interpolating these point clouds, a continuous three-dimensional candidate rough structure surface with realistic roughness characteristics can be generated.

[0028] This application embodiment can efficiently generate an infinite variety of three-dimensional candidate rough structure surfaces by randomly superimposing a large number of spheres with positive and negative signs, without the need for manual design of specific shapes. The sphere radii follow a normal distribution and their positions are completely random, so that the generated three-dimensional candidate rough structure surfaces statistically conform to the randomness characteristics of natural rock mass structure surfaces. Each three-dimensional candidate rough structure surface has its own heterogeneity and anisotropy, providing rich morphological primitives for subsequent approximation of the real trace through weighted fitting. The parameters of the entire generation process are clear (number of spheres, radius distribution, spatial range), which are easy to adjust and control, and can flexibly adapt to rock mass structure surface reconstruction tasks with different scales and roughness requirements.

[0029] Optionally, in this embodiment of the application, step 101 includes: collecting three-dimensional trace position data of the exposed traces of the rock mass structural surface, and converting the three-dimensional trace position data into a two-dimensional coordinate sequence; and generating a continuous curve based on the two-dimensional coordinate sequence as the original trace curve.

[0030] In this embodiment, firstly, three-dimensional trace position data of the exposed traces on the rock mass structural surface are acquired. Specifically, through measurement techniques such as three-dimensional laser scanning or line structured light scanning, the three-dimensional spatial coordinates (i.e., the X, Y, and Z values ​​of each point) of a large number of discrete points on the exposed traces can be obtained. Since the scanning device directly records the spatial position of the object surface, the acquired data is three-dimensional and contains the actual direction, undulation, and depth information of the exposed traces in space.

[0031] Then, the acquired 3D trace position data is converted into a 2D coordinate sequence. In practical engineering, rock mass structural surfaces are often assumed to be analyzable on a certain unfolded or projected plane, such as unfolding the curved surface of a tunnel wall into a plane, or projecting the cylindrical surface of a core into a rectangular surface. The conversion method can be to map the 3D coordinates onto this plane: for tunnel walls, projection along the wall normal; for borehole cores, circumferential angles and axial distances can be converted into planar coordinates. After the conversion, the original 3D points containing depth or curvature information become a 2D coordinate sequence retaining only the planar position (X', Y'). This dimensionality reduction simplifies the problem, allowing the original trace curves to be expressed and compared on a unified 2D plane.

[0032] Finally, based on the aforementioned two-dimensional coordinate sequence, a continuous curve is generated as the original trace curve. Since the collected points are usually discrete, interpolation or fitting algorithms (such as spline interpolation, polynomial fitting, or piecewise linear connection) can be used to connect these points sequentially, forming a smooth and continuous curve passing through all points. This curve faithfully reflects the geometric shape of the exposed trace on the two-dimensional plane, including its curvature, turns, and tortuosity. It will become the target benchmark for extracting the simulated trace and performing weighted fitting in subsequent steps. The superposition results of all three-dimensional candidate rough structure surfaces must be referenced to the shape of this curve.

[0033] This application's embodiments preserve the most authentic spatial information of the original traces by acquiring three-dimensional trace position data, avoiding the loss of depth or curvature details from the outset. Converting the three-dimensional trace position data into a two-dimensional coordinate sequence transforms the trace problem on complex curved surfaces into a planar curve fitting problem, greatly reducing the complexity of subsequent calculations. It also facilitates corresponding sampling with the three-dimensional candidate rough structure surfaces generated by the spherical basis function. The generated original trace curves provide clear and quantifiable targets for subsequent weighted combination fitting, ensuring that the reconstructed rock mass structure surfaces at the surface exposure locations are strictly consistent with the measured results, thereby significantly improving the engineering reliability and interpretability of the method.

[0034] Optionally, in this embodiment of the application, step 102, "for each three-dimensional candidate rough structure surface, extract a simulated trace from the three-dimensional candidate rough structure surface according to the coordinate position of the original trace curve", includes: determining a cutting plane within the spatial range according to the coordinate position of the original trace curve; for each three-dimensional candidate rough structure surface, cutting the three-dimensional candidate rough structure surface through the cutting plane to obtain a curve intersecting the cutting plane, which serves as the simulated trace corresponding to the three-dimensional candidate rough structure surface.

[0035] In this embodiment, firstly, a cutting plane is determined within the spatial range of the three-dimensional candidate rough structure surface based on the coordinate position of the original trace curve. Here, the original trace curve is a continuous curve composed of a two-dimensional coordinate sequence obtained from previous acquisition and conversion; therefore, the coordinates of the original trace curve are a two-dimensional coordinate sequence composed of two-dimensional coordinates (X', Y'). Next, based on the two-dimensional coordinate sequence of the original trace curve, the corresponding X' and Y' axes are determined from the spatial range. Then, according to set rules such as Z'=10, the cutting plane is determined to avoid manual selection of the cutting plane and improve the intelligence of the entire process. The spatial range refers to the preset three-dimensional region when generating all three-dimensional candidate rough structure surfaces. The cutting plane must be located inside or intersect with this region to ensure that the three-dimensional candidate rough structure surface can be effectively cut.

[0036] Next, for each generated 3D candidate rough structure surface, a geometric intersection operation is performed between the 3D candidate rough structure surface and the aforementioned determined intercepting plane to obtain an intersection line. Since the 3D candidate rough structure surface is a continuous surface formed by interpolation of a large number of discrete 3D spatial points, and the intercepting plane is an infinitely extending 2D plane, their intersection will inevitably produce a spatial curve. This curve lies precisely on the intercepting plane; therefore, its geometric shape is entirely determined by the undulations of the 3D candidate rough structure surface at that position on the plane.

[0037] Finally, the intersecting curve is used as the simulated trace corresponding to the 3D candidate rough structure surface. Each simulated trace directly reflects the surface undulation characteristics of the corresponding 3D candidate rough structure surface at the corresponding location. If the 3D candidate rough structure surface has a convexity or depression at that location, the simulated trace will bend or deflect accordingly. Through this interception operation, without complex coordinate mapping or sampling interpolation, a 2D curve that can be directly geometrically compared with the original trace curve can be quickly extracted from each 3D candidate rough structure surface. It should be noted that the same interception plane is used to intercept and obtain the simulated trace for each generated 3D candidate rough structure surface.

[0038] This embodiment of the application utilizes the intersection of the intercept plane directly with the 3D candidate rough structure surface, which has a clear physical meaning and can be efficiently completed through computational geometry algorithms. The extracted simulated trace and the original trace curve can be assumed to be in the same plane, ensuring the consistency of their dimensions and coordinate systems during subsequent weighted combination fitting, and avoiding additional errors introduced by projection or interpolation. For each 3D candidate rough structure surface, the interception operation is independent, which facilitates parallel computation and significantly improves processing efficiency. Since the intercept plane is uniquely determined, no artificial assumptions are required, ensuring the objectivity and repeatability of the simulated trace, and laying a reliable foundation for subsequent accurate weight allocation using the least squares method.

[0039] Optionally, in this embodiment of the application, step 103 includes: taking the original trace curve as the target curve and taking the simulated traces corresponding to each three-dimensional candidate rough structure surface as the curves to be fitted; using the least squares method to solve the target weight set, such that the sum of squared residuals between the fitted curve after all the curves to be fitted are superimposed according to the target weight set and the target curve is less than a preset convergence threshold; and assigning a corresponding weight to each simulated trace according to the target weight set obtained by the solution.

[0040] In this embodiment, firstly, the original trace curve is used as the target curve, and the simulated traces corresponding to each three-dimensional candidate rough structure surface are used as the curves to be fitted. The original trace curve is a two-dimensional curve obtained from actual measurements on the rock mass surface, representing the true geometric shape of the structure surface on the free surface; while each simulated trace is a two-dimensional curve obtained by intercepting the plane corresponding to the three-dimensional candidate rough structure surface, reflecting the undulation characteristics of the three-dimensional candidate rough structure surface at the same plane position. All simulated traces can be compared and combined with the original trace curve as a standard.

[0041] Next, a target weight set is obtained using the least squares method, ensuring that the sum of squared residuals between the fitted curve obtained by superimposing all the curves to be fitted according to this target weight set and the target curve is less than a preset convergence threshold. The least squares method is a classic mathematical optimization technique. Its core is to find an optimal set of non-negative weights that makes the composite curve obtained by weighting and superimposing the simulated traces as close as possible to the target curve. Specifically, for each sampling point, the square of the difference between the height of the composite curve at that point and the height of the original trace curve at that point is calculated. Then, the squared differences of all sampling points are summed to obtain the sum of squared residuals. This sum is continuously reduced by adjusting the weight combination until it falls below a preset convergence threshold (this threshold can be determined according to engineering accuracy requirements, for example, 1% of the total variance of the original trace curve). This process ensures that the superimposed result closely approximates the original trace curve with high accuracy, while avoiding overfitting or numerical instability caused by pursuing an absolute minimum.

[0042] Finally, based on the obtained target weight set, a corresponding weight is assigned to each simulated trace. Each weight value in the target weight set corresponds to a specific simulated trace, reflecting its contribution to approximating the original trace curve: the larger the weight of the simulated trace, the more similar its fluctuation characteristics are to the original trace curve, and therefore it plays a more important role in the final superposition and synthesis; simulated traces with weights close to zero indicate that they deviate significantly from the original trace curve and contribute almost nothing to the final result. In this way, the originally randomly generated 3D candidate rough structure surfaces are automatically filtered and combined through the weights of their simulated traces.

[0043] The embodiments of this application employ the least squares method to solve for the weights, which can guarantee convergence to the globally optimal or near-optimal solution. By setting a preset convergence threshold, a flexible balance can be struck between accuracy and computational efficiency to meet the needs of different engineering scenarios. The weights directly characterize the contribution of each 3D candidate rough structure surface to the real structure surface, facilitating subsequent engineering analysis and interpretation. The entire process is completely data-driven, requiring no human experience intervention, avoiding subjective errors, and significantly improving the objectivity and repeatability of rock mass structure surface reconstruction.

[0044] In another embodiment, after acquiring the three-dimensional trace position data corresponding to the exposed traces of the rock mass structural surface, a three-dimensional original trace curve can be constructed based on this three-dimensional trace position data. Further, based on the aforementioned process, three-dimensional candidate rough structural surfaces with heterogeneous and anisotropic characteristics are generated. According to each three-dimensional coordinate in the original trace curve, the horizontal plane coordinates within those three-dimensional coordinates are determined. For example, if the coordinate axis perpendicular to the ground is the Y-axis, then the horizontal plane coordinates are the coordinates composed of the X-axis and Z-axis. For each three-dimensional candidate rough structural surface, based on the horizontal plane coordinates, the attribute value corresponding to those horizontal plane coordinates is obtained on the current three-dimensional candidate rough structural surface, i.e., the height value Y of the three-dimensional candidate rough structural surface at that location. The horizontal plane coordinates (X, Z) are combined with the extracted height value Y to obtain a new three-dimensional spatial point. Following the point order on the original trace curve, all newly generated three-dimensional spatial points are sequentially connected, and spline interpolation is used for smoothing to form a three-dimensional simulated trace. Then, the three-dimensional simulated traces corresponding to each three-dimensional candidate rough structure surface are used as the curves to be fitted, and the original three-dimensional trace curves are used as the target curves for fitting to obtain the target weight set.

[0045] Optionally, in this embodiment, step 104 includes: determining a spatial region containing all three-dimensional candidate rough structural surfaces, and dividing the horizontal projection of the spatial region into grids to determine the planar coordinates of each grid point; for each three-dimensional candidate rough structural surface, extracting the attribute value of each planar coordinate on the three-dimensional candidate rough structural surface, and multiplying the attribute value by the weight corresponding to the simulated trace on the three-dimensional candidate rough structural surface to obtain the weighted attribute contribution of each planar coordinate; at each planar coordinate, summing the weighted attribute contributions of all three-dimensional candidate rough structural surfaces to obtain the synthetic attribute value at the planar coordinate; combining the planar coordinates of each grid point with the corresponding synthetic attribute value to form a three-dimensional spatial point, constructing a discrete point cloud of the synthetic structural surface from all three-dimensional spatial points, and generating a continuous three-dimensional surface through interpolation fitting as the three-dimensional morphology of the rock mass structural surface corresponding to the original trace curve.

[0046] In this embodiment, firstly, a spatial region containing all three-dimensional candidate roughness surfaces is determined, and the horizontal projection of this spatial region is meshed to determine the planar coordinates of each mesh point. Since each three-dimensional candidate roughness surface is defined within the same three-dimensional spatial region, the projection of this region onto the horizontal plane can be used as a common sampling reference. The projected region is divided into regular meshes (e.g., square meshes) with a certain resolution, and each mesh point corresponds to a planar coordinate. This mesh division determines the fineness of the final synthesized surface: the denser the mesh, the richer the generated three-dimensional topographic details, but the greater the computational cost. All subsequent superposition calculations will be based on these unified mesh points, thereby ensuring the spatial correspondence between different three-dimensional candidate roughness surfaces.

[0047] Next, for each 3D candidate rough surface, the attribute value at each planar coordinate is extracted, and this attribute value is multiplied by the weight corresponding to the simulated trace on the 3D candidate rough surface to obtain the weighted attribute contribution of each planar coordinate. Here, the attribute value refers to the height of the 3D candidate rough surface at that planar coordinate. Since each 3D candidate rough surface is a continuous surface, the corresponding height value can be directly obtained given the planar coordinate. Multiplying the height value by the weight obtained from the previous weighted combination fitting of the 3D candidate rough surface yields the weighted attribute contribution of the 3D candidate rough surface at that grid point. The larger the weight of the 3D candidate rough surface, the greater the proportion of its height value in the final synthesis; conversely, the contribution of the 3D candidate rough surface with a weight close to zero is almost negligible.

[0048] Subsequently, at each planar coordinate, the weighted attribute contributions of all 3D candidate rough structure surfaces are summed to obtain the composite attribute value at that planar coordinate. That is, for any grid point, the weighted height values ​​calculated by all 3D candidate rough structure surfaces at that point are directly added together to obtain the composite height value, or composite attribute value, for that point. Since the weights are non-negative and all 3D candidate rough structure surfaces share the same planar coordinate system, the summation result naturally forms a unified height field. This composite height value preserves the inherent non-homogeneity and anisotropy characteristics of each 3D candidate rough structure surface, and through weight adjustment, causes the overall shape to converge towards the direction constrained by the original trace curve.

[0049] Finally, the planar coordinates of each grid point are combined with the corresponding synthetic attribute values ​​to form a three-dimensional spatial point. A discrete point cloud of the synthetic structural surface is constructed from all these three-dimensional spatial points, and a continuous three-dimensional surface is generated through interpolation fitting, serving as the final three-dimensional morphology. At this point, each grid point has complete three-dimensional coordinates, and all these points constitute a sparse point cloud. To obtain a smooth and continuous structural surface model, interpolation or fitting algorithms (such as bicubic spline interpolation, radial basis function interpolation, or kriging) can be used to reconstruct the surface from the point cloud, filling grid gaps, eliminating possible noise, and ultimately outputting a three-dimensional surface suitable for visualization or subsequent mechanical analysis. At positions corresponding to the original trace curve, the curve cut by this surface closely matches the original trace curve, while in other areas, it reasonably extrapolates the anisotropic morphology of the rock mass structural surface.

[0050] This application's embodiments ensure that all 3D candidate rough structure surfaces are superimposed in the same spatial reference frame through a unified horizontal projection mesh, avoiding coordinate misalignment errors. Weights are directly applied to the attribute values ​​of each 3D candidate rough structure surface, resulting in clear physical meaning and simple calculation. A larger weight indicates a greater impact on the final morphology, consistent with the selection results from the fitting stage. The accumulation process maintains the linear superposition property, ensuring that the final 3D surface is compatible with the characteristics of all 3D candidate rough structure surfaces without nonlinear distortion. By constructing a discrete point cloud and then interpolating and fitting, the final surface is guaranteed to be smooth and continuous, while also allowing flexible control of the output resolution, making it suitable for scenarios with varying accuracy requirements in engineering. This method is fully automated, requiring no manual intervention, significantly improving the efficiency and repeatability of 3D structure surface reconstruction, and providing a reliable basic model for subsequent rock mass stability analysis and seepage simulation.

[0051] Optionally, after step 104, the method further includes: acquiring at least one set of supplementary constraint data, wherein the supplementary constraint data includes at least one of borehole core structural plane orientation, cross-hole seismic wave velocity anomaly distribution, and ground-penetrating radar reflection interface; projecting the generated three-dimensional morphology of the rock mass structural plane onto the acquisition space of the supplementary constraint data, and calculating the residual distribution between the projection result and the supplementary constraint data; constructing a spatial variation penalty function based on the residual distribution, and using the penalty function to correct the weights of each simulated trace; and re-performing weighted superposition based on the corrected weights to generate a three-dimensional morphology of the rock mass structural plane that incorporates the supplementary constraints.

[0052] In this embodiment, firstly, at least one set of supplementary constraint data is acquired. This data may include at least one of the following: borehole core structural plane attitude, cross-hole seismic wave velocity anomaly distribution, and ground-penetrating radar reflection interface. The borehole core structural plane attitude refers to the depth, dip angle, and dip direction of the structural plane recorded during core drilling, providing one-dimensional linear constraints along the borehole trajectory. The cross-hole seismic wave velocity anomaly distribution is obtained by measuring wave velocity between two or more boreholes, reversing the spatial variation of wave velocity within the profile, and indirectly reflecting the wave velocity reduction area caused by the structural plane. The ground-penetrating radar reflection interface uses high-frequency electromagnetic waves to detect underground reflecting layers, obtaining the reflection phase axis on a two-dimensional profile, often corresponding to the boundary of rock strata or structural planes. This supplementary constraint data is independent of surface outcrop traces and can provide spatial location information of structural planes from deep or different physical property perspectives, which is of great value for correcting three-dimensional morphology reconstructed solely from traces.

[0053] Next, the generated 3D morphology of the rock mass structure is projected onto the acquisition space of the supplementary constraint data, and the residual distribution between the projection result and the supplementary constraint data is calculated. Here, projection refers to calculating the theoretically observable response value from the 3D morphology of the rock mass structure based on the observation method of the supplementary data. For example, for borehole core constraints, the depth and attitude of the intersection point of the 3D morphology of the rock mass structure along the borehole trajectory line are extracted to obtain the predicted borehole response; for cross-hole seismic wave velocity, the 3D morphology of the rock mass structure can be converted into a wave velocity model, and then the spatial distribution of wave velocity anomalies can be simulated; for ground-penetrating radar, the reflection interface position of the 3D morphology of the rock mass structure can be intercepted on the profile where the radar line is located. By comparing the projection result with the actual measurement data point by point and calculating the differences at each spatial location (such as depth difference, attitude difference, or wave velocity deviation), the residual distribution is obtained. This residual distribution reveals in which areas and to what extent the current 3D morphology of the rock mass structure deviates from the additional geological constraints.

[0054] Furthermore, based on the residual distribution, a penalty function for spatial variation is constructed, and this penalty function is used to correct the weights of each simulated trajectory. The design principle of the penalty function is: in areas with large residuals, the contribution of the 3D candidate roughness surfaces that cause this deviation in the current 3D morphology of the rock mass structure should be reduced, or the weights of those 3D candidate roughness surfaces that can reduce the residuals should be increased. Since the residuals are spatially variable (e.g., one borehole segment matches well, another poorly), the penalty function should also vary with spatial location. Typically, a weighted sum of squared residuals can be used, assigning different penalty intensities to different regions. This penalty function is then combined with the original weighted combined fitting objective (e.g., adding a penalty term to the original least squares objective) to re-optimize the weights of each simulated trajectory. This ensures that the corrected weights maintain a good fit with the original trajectory curves while better satisfying the supplementary constraint data. This correction process is essentially a multi-objective inversion, fusing multi-source information by adjusting the spatial distribution of weights.

[0055] Finally, the weighted superposition is re-executed based on the corrected weights to generate a 3D morphology of the rock mass structure surface with integrated supplementary constraints. That is, the new weights obtained in the previous step are substituted back into the weighted superposition step, and the synthetic attribute values ​​at each grid point are recalculated to obtain the updated rock mass structure surface. Since the weight correction incorporates information from boreholes, seismic waves, or radar, the newly generated 3D morphology not only matches the original trace curve at the surface outcrop but also closely resembles the measured data at deep borehole locations or geophysical profiles. In a specific embodiment, the above cycle of "projection—residual calculation—weight correction—re-superposition" can be repeated until the residuals of all constraint data are below a preset threshold, ultimately outputting a highly reliable rock mass structure surface with multi-constraint fusion.

[0056] This application's embodiments effectively overcome the multiple solutions and deep uncertainties inherent in retrieving 3D topography solely from surface traces by introducing independent deep data such as borehole data, seismic waves, and ground-penetrating radar data, significantly improving the global reliability of the rock mass structure surface. The spatial variation penalty function can adaptively adjust weights according to the residual magnitude in different regions, making the correction process targeted and avoiding local overfitting that may result from globally uniform correction. This method is entirely based on the aforementioned weighted superposition framework, requiring no change to the core algorithm; it only needs to add a penalty term to the weight solution objective, resulting in low engineering implementation costs. The final generated 3D topography with fused supplementary constraints conforms to both surface observation and deep exploration, providing a more reliable rock mass structure surface for rock mass stability analysis and support design in complex engineering projects such as steep slopes and deep-buried tunnels.

[0057] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, and to fully illustrate the specific implementation process of this embodiment, another method for generating the three-dimensional morphology of rock mass structural surfaces is provided. This method includes: Step 1: Acquisition and Preprocessing of Trace Location Data. Trace lines exposed on rock mass structural surfaces are acquired using 3D laser scanning or line structured light scanning technology to obtain trace location data. This data is then preprocessed to convert the complete trace location data into a two-dimensional coordinate sequence. Based on this two-dimensional coordinate sequence, a continuous curve is generated, such as... Figure 2 This yields the original trace curve.

[0058] Step 2: Generate 3D candidate rough structure surfaces using spherical basis functions. "Spherical series spherical basis functions" is a method for generating spatially non-uniform property fields based on the principle of random superposition. Its principle is to randomly generate... There are spheres, their positions are random, and their radii follow a normal distribution. Take the absolute value The contribution of a sphere to the attribute values ​​of points in its surrounding space is defined using the following function: ; in, The radius of the sphere; For the first The center coordinates of each sphere; A random symbol for a sphere; These are the coordinates of a point in space. For the first The contribution of each sphere at that point in space.

[0059] As spheres are randomly stacked in space, their contribution values ​​accumulate in three-dimensional space, and the final attribute value of a point in space is... The surface is generated by the linear superposition of the contribution values ​​of each sphere. Based on the attribute values ​​of each spatial point, a three-dimensional candidate rough structure surface is generated.

[0060] Step 3: Weighted superposition to generate rock mass structural surfaces. Based on the spherical basis function, a series of three-dimensional candidate rough structural surfaces are generated. A two-dimensional simulated trace is extracted from the corresponding position of each three-dimensional candidate rough structural surface. The least squares method is used to assign different weights to each simulated trace. The simulated traces are superimposed and fitted to generate a curve that highly matches the original trace curve.

[0061] Based on the weight values ​​assigned to each simulated trace during the trace fitting process, the corresponding three-dimensional candidate rough structure surfaces are assigned the same weight, and the rock mass structure surfaces associated with the original trace curves are obtained through weight superposition calculation. For example... Figure 3 As shown, n three-dimensional candidate rough structure surfaces provided in the embodiments of this application are illustrated, and their corresponding weights are as follows: ... . Figure 4 This application illustrates a rock mass structural surface provided in an embodiment, based on... Figure 3 Generate n three-dimensional candidate rough structure surfaces. Figure 5 This illustration shows a comparison between an original trace curve provided in an embodiment of this application and a fitted curve obtained by fitting simulated traces corresponding to the aforementioned n three-dimensional candidate rough structure surfaces.

[0062] In a specific embodiment, the rock mass structural surface is deduced based on the trace location data of the exposed trace of a certain tunnel structural surface: 1. Field Data Preprocessing: The track position data acquired on-site is discretized into the coordinates of 200 evenly distributed points, resulting in the original track curve as shown below. Figure 6 As shown.

[0063] 2. Determine the spherical basis function parameters, generate each 3D candidate rough structure surface, and extract simulated traces from them. The fitting curves generated by each simulated trace are shown below. Figure 7 As shown in Table 1. Specific parameters for the spherical basis functions are listed below.

[0064] Table 1: Parameters of Spherical Basis Functions

[0065] 3. Generation of structural surfaces through fitting: Based on the different weights assigned to the simulated traces of each 3D candidate roughness surface, each 3D candidate roughness surface is assigned a corresponding weight, and a synthetic surface is finally generated through weighted superposition calculation. This surface not only has a high correlation with the original trace curves, but also possesses the anisotropic characteristics unique to real rock mass structural surfaces, such as... Figure 8 As shown, the structural surface of a certain tunnel is derived.

[0066] In another specific embodiment, the rock mass structure surface is deduced based on the core trace of a certain borehole: 1. On-site data preprocessing: Point cloud scanning was performed on the borehole core traces obtained on-site to acquire the trace position data. The resulting original trace curves of the borehole core traces are shown below. Figure 9 As shown.

[0067] 2. Select spherical basis function parameters: Generate multiple 3D candidate rough structure surfaces based on the spherical basis function parameters. See Table 2 for specific spherical basis function parameters.

[0068] Table 2: Spherical Basis Function Parameters

[0069] 3. Generation of structural surfaces through fitting: Based on the least squares method, different weights are assigned to the simulated traces of each three-dimensional candidate rough structure surface. The synthesized contour generated by superposition is matched with the original trace curves of the borehole core traces obtained in the field, thereby providing a basis for the subsequent fitting of the rock mass structure surface.

[0070] Based on the weight values ​​determined in the previous step, each 3D candidate roughness surface is assigned a corresponding weight, and a composite surface is finally generated through weighted superposition calculation. This surface not only has a high correlation with the original borehole core trace, but also possesses the anisotropic characteristics unique to real core structural surfaces, such as... Figure 10 As shown, the deduced core structure of a certain borehole is obtained.

[0071] Furthermore, as Figure 1 In terms of specific implementation, this application provides a device for generating the three-dimensional morphology of rock mass structural surfaces, such as... Figure 11 As shown, the device includes: The data acquisition module is used to acquire the trace position data corresponding to the exposed traces of the rock mass structural surface, and to construct the original trace curve based on the trace position data; The simulated trace extraction module is used to randomly generate multiple three-dimensional candidate rough structure surfaces with heterogeneous and anisotropic characteristics using spherical basis functions. For each three-dimensional candidate rough structure surface, a simulated trace is extracted from the three-dimensional candidate rough structure surface according to the coordinate position of the original trace curve. The fitting module is used to perform weighted combination fitting on the simulated traces of each three-dimensional candidate rough structure surface using the original trace curve as the weighted combination target, and to determine the weight corresponding to each simulated trace. The overlay module is used to weight and overlay each three-dimensional candidate rough structure surface according to its corresponding weight to generate a three-dimensional morphology of the rock mass structure surface corresponding to the original trace curve.

[0072] Optionally, the simulated trace extraction module is used for: Define the total number of spheres, the mean and standard deviation of the normal distribution followed by the sphere radii, and the spatial range of the structural surfaces to be generated; The total number of spheres are randomly generated within the spatial range, wherein each sphere has a random spatial location, a random radius, and a random sign. The random radius is obtained by sampling the normal distribution and taking the absolute value, and the random sign is a randomly assigned positive or negative sign. For each spatial point within the spatial range, the contribution values ​​of all spheres to the spatial point are linearly superimposed to obtain the attribute value of the spatial point. The contribution value of each sphere is determined by the random radius of the sphere, the distance between the center of the sphere and the spatial point, and the random sign of the sphere. The planar coordinates of each spatial point are combined with the corresponding attribute value to form a three-dimensional spatial point. Through all three-dimensional spatial points, a three-dimensional candidate rough structure surface with heterogeneous and anisotropic characteristics is generated.

[0073] Optionally, the data acquisition module is used for: Collect three-dimensional trace position data of the exposed traces of the rock mass structural surface, and convert the three-dimensional trace position data into a two-dimensional coordinate sequence; Based on the two-dimensional coordinate sequence, a continuous curve is generated as the original trace curve.

[0074] Optionally, the simulated trace extraction module is further configured to: Based on the coordinate position of the original trace curve, a cutting plane is determined within the spatial range; For each three-dimensional candidate rough structure surface, the three-dimensional candidate rough structure surface is intercepted by the intercepting plane to obtain the curve intersecting the intercepting plane, which serves as the simulated trace corresponding to the three-dimensional candidate rough structure surface.

[0075] Optionally, the fitting module is used for: The original trace curve is used as the target curve, and the simulated traces corresponding to each three-dimensional candidate rough structure surface are used as the curves to be fitted. The least squares method is used to solve for the target weight set, such that the sum of squared residuals between the fitted curve after all the curves to be fitted are superimposed according to the target weight set and the target curve is less than a preset convergence threshold. Based on the target weight set obtained from the solution, assign corresponding weights to each simulated trace.

[0076] Optionally, the overlay module is used for: Determine the spatial region containing all three-dimensional candidate rough structure surfaces, and mesh the horizontal projection of the spatial region to determine the planar coordinates of each mesh point; For each 3D candidate rough structure surface, the attribute value of each planar coordinate on the 3D candidate rough structure surface is extracted, and the attribute value is multiplied by the weight corresponding to the simulated trace on the 3D candidate rough structure surface to obtain the weighted attribute contribution of each planar coordinate. At each planar coordinate, the weighted attribute contributions of all three-dimensional candidate rough structure surfaces are summed to obtain the composite attribute value at the planar coordinate. The planar coordinates of each grid point are combined with the corresponding synthetic attribute values ​​to form a three-dimensional spatial point. The discrete point cloud of the synthetic structural surface is constructed from all the three-dimensional spatial points, and a continuous three-dimensional surface is generated by interpolation fitting, which serves as the three-dimensional morphology of the rock mass structural surface corresponding to the original trace curve.

[0077] Optionally, the apparatus further includes a weight update module; the weight update module is configured to: After generating the three-dimensional morphology of the rock mass structure surface corresponding to the original trace curve, at least one set of supplementary constraint data is obtained, wherein the supplementary constraint data includes at least one of the following: borehole core structure surface orientation, cross-hole seismic wave velocity anomaly distribution, and ground-penetrating radar reflection interface. The generated three-dimensional topography of the rock mass structure surface is projected onto the acquisition space of the supplementary constraint data, and the residual distribution between the projection result and the supplementary constraint data is calculated. Based on the residual distribution, a penalty function for spatial variation is constructed, and the weights of each simulated trace are corrected using the penalty function. The weighted superposition is re-executed based on the corrected weights to generate the three-dimensional morphology of the rock mass structure surface with integrated supplementary constraints.

[0078] It should be noted that other corresponding descriptions of the functional units involved in the device for generating three-dimensional morphology of rock mass structural surfaces provided in this application embodiment can be found in the following references. Figures 1 to 10 The corresponding descriptions in the method will not be repeated here.

[0079] This application also provides a computer device, which may specifically be a personal computer, a server, a network device, etc. Figure 12 As shown, the computer device includes a bus, a processor, memory, and a communication interface, and may also include an input / output interface and a display device. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores location information. The network interface allows communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps in the various method embodiments.

[0080] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0081] In one embodiment, a computer-readable storage medium is provided, which may be non-volatile or volatile, having stored thereon a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0082] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0083] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0084] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0085] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0086] The embodiments described above are merely examples of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

Claims

1. A method for generating the three-dimensional morphology of rock mass structural surfaces, characterized in that, include: Obtain the trace location data corresponding to the exposed traces of the rock mass structural surface, and construct the original trace curve based on the trace location data; Using spherical basis functions, multiple three-dimensional candidate rough structure surfaces with heterogeneous and anisotropic characteristics are randomly generated. For each three-dimensional candidate rough structure surface, a simulated trace is extracted from the three-dimensional candidate rough structure surface according to the coordinate position of the original trace curve. Using the original trace curve as the weighted combination target, the simulated traces of each three-dimensional candidate rough structure surface are weighted and combined for fitting to determine the weight corresponding to each simulated trace. Each candidate three-dimensional rough structure surface is weighted and superimposed according to its corresponding weight to generate a three-dimensional morphology of the rock mass structure surface corresponding to the original trace curve.

2. The method according to claim 1, characterized in that, Each 3D candidate rough structure surface is generated based on the following steps: Define the total number of spheres, the mean and standard deviation of the normal distribution followed by the sphere radii, and the spatial range of the structural surfaces to be generated; The total number of spheres are randomly generated within the spatial range, wherein each sphere has a random spatial location, a random radius, and a random sign. The random radius is obtained by sampling the normal distribution and taking the absolute value, and the random sign is a randomly assigned positive or negative sign. For each spatial point within the spatial range, the contribution values ​​of all spheres to the spatial point are linearly superimposed to obtain the attribute value of the spatial point. The contribution value of each sphere is determined by the random radius of the sphere, the distance between the center of the sphere and the spatial point, and the random sign of the sphere. The planar coordinates of each spatial point are combined with the corresponding attribute value to form a three-dimensional spatial point. Through all three-dimensional spatial points, a three-dimensional candidate rough structure surface with heterogeneous and anisotropic characteristics is generated.

3. The method according to claim 2, characterized in that, The step of acquiring the trace position data corresponding to the exposed traces of the rock mass structural surface, and constructing the original trace curve based on the trace position data, includes: Collect three-dimensional trace position data of the exposed traces of the rock mass structural surface, and convert the three-dimensional trace position data into a two-dimensional coordinate sequence; Based on the two-dimensional coordinate sequence, a continuous curve is generated as the original trace curve.

4. The method according to claim 3, characterized in that, For each three-dimensional candidate rough structure surface, a simulated trace is extracted from the three-dimensional candidate rough structure surface based on the coordinate position of the original trace curve, including: Based on the coordinate position of the original trace curve, a cutting plane is determined within the spatial range; For each three-dimensional candidate rough structure surface, the three-dimensional candidate rough structure surface is intercepted by the intercepting plane to obtain the curve intersecting the intercepting plane, which serves as the simulated trace corresponding to the three-dimensional candidate rough structure surface.

5. The method according to claim 1, characterized in that, The step of using the original trace curve as the weighted combination target to perform weighted combination fitting on the simulated traces of each three-dimensional candidate rough structure surface, and determining the weight corresponding to each simulated trace, includes: The original trace curve is used as the target curve, and the simulated traces corresponding to each three-dimensional candidate rough structure surface are used as the curves to be fitted. The least squares method is used to solve for the target weight set, such that the sum of squared residuals between the fitted curve after all the curves to be fitted are superimposed according to the target weight set and the target curve is less than a preset convergence threshold. Based on the target weight set obtained from the solution, assign corresponding weights to each simulated trace.

6. The method according to claim 1, characterized in that, The step of weighted superposition of each three-dimensional candidate rough structure surface according to its corresponding weight to generate a three-dimensional morphology of the rock mass structure surface corresponding to the original trace curve includes: Determine the spatial region containing all three-dimensional candidate rough structure surfaces, and mesh the horizontal projection of the spatial region to determine the planar coordinates of each mesh point; For each 3D candidate rough structure surface, the attribute value of each planar coordinate on the 3D candidate rough structure surface is extracted, and the attribute value is multiplied by the weight corresponding to the simulated trace on the 3D candidate rough structure surface to obtain the weighted attribute contribution of each planar coordinate. At each planar coordinate, the weighted attribute contributions of all three-dimensional candidate rough structure surfaces are summed to obtain the composite attribute value at the planar coordinate. The planar coordinates of each grid point are combined with the corresponding synthetic attribute values ​​to form a three-dimensional spatial point. The discrete point cloud of the synthetic structural surface is constructed from all the three-dimensional spatial points, and a continuous three-dimensional surface is generated by interpolation fitting, which serves as the three-dimensional morphology of the rock mass structural surface corresponding to the original trace curve.

7. The method according to claim 1, characterized in that, After generating the three-dimensional morphology of the rock mass structural surface corresponding to the original trace curve, the method further includes: At least one set of supplementary constraint data is obtained, wherein the supplementary constraint data includes at least one of the following: borehole core structural plane orientation, cross-hole seismic wave velocity anomaly distribution, and ground-penetrating radar reflection interface; The generated three-dimensional topography of the rock mass structure surface is projected onto the acquisition space of the supplementary constraint data, and the residual distribution between the projection result and the supplementary constraint data is calculated. Based on the residual distribution, a penalty function for spatial variation is constructed, and the weights of each simulated trace are corrected using the penalty function. The weighted superposition is re-executed based on the corrected weights to generate the three-dimensional morphology of the rock mass structure surface with integrated supplementary constraints.

8. A device for generating three-dimensional morphology of rock mass structural surfaces, characterized in that, include: The data acquisition module is used to acquire the trace position data corresponding to the exposed traces of the rock mass structural surface, and to construct the original trace curve based on the trace position data; The simulated trace extraction module is used to randomly generate multiple three-dimensional candidate rough structure surfaces with heterogeneous and anisotropic characteristics using spherical basis functions. For each three-dimensional candidate rough structure surface, a simulated trace is extracted from the three-dimensional candidate rough structure surface according to the coordinate position of the original trace curve. The fitting module is used to perform weighted combination fitting on the simulated traces of each three-dimensional candidate rough structure surface using the original trace curve as the weighted combination target, and to determine the weight corresponding to each simulated trace. The overlay module is used to weight and overlay each three-dimensional candidate rough structure surface according to its corresponding weight to generate a three-dimensional morphology of the rock mass structure surface corresponding to the original trace curve.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

10. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

Images