A dynamic modeling method of a geological structure three-dimensional model

Abstract

The present application relates to the technical field of three-dimensional modeling, and particularly relates to a dynamic modeling method of a geological structure three-dimensional model. The content comprises: obtaining multi-source data and preprocessing, introducing a voxel semantic fusion algorithm based on variational optimization, mapping the preprocessed multi-source data to a target three-dimensional space grid after extracting semantic information, and obtaining an optimal semantic fusion vector; based on the optimal semantic fusion vector, generating a standard voxel data pool and constructing a geological structure three-dimensional model; listening to data changes and calculating the position of new data points, combining the standard voxel data pool to obtain a spatial update area, and modeling the spatial update area to realize model updating; after completing the modeling of the spatial update area, the boundary continuity is optimized. The problems of multi-source geological data cannot be directly used for the structure construction and semantic fusion of the three-dimensional model, lack of dynamic response to new data, and existence of geometric discontinuity and structural logic discontinuity at the boundary are solved.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional modeling technology, and in particular to a dynamic modeling method for three-dimensional geological structures. Background Technology

[0002] Currently, 3D geological modeling serves as a key supporting technology for multiple application fields such as geological surveys, resource exploration, underground engineering planning, and risk assessment. Its core lies in constructing a spatial model with realistic geological structural characteristics by integrating various data sources, including borehole, seismic, remote sensing, and geological logging data. However, existing modeling techniques largely rely on manual interpretation and static modeling processes, making it difficult to meet the demands of scenarios involving continuously updated geological data and complex, frequently changing geological structures. Firstly, geological data itself is highly heterogeneous; different data sources differ in spatial coordinate systems, data accuracy, and semantic definitions. Traditional modeling platforms typically merge data through manual conversion and limited rules, resulting in low data fusion efficiency and poor semantic consistency, thus affecting model quality. Secondly, existing modeling methods generally employ a global reconstruction approach, recalculating the entire model whenever data changes. This not only consumes significant computational resources but is also unsuitable for rapid decision-making in practical engineering. The need for real-time modeling, especially during drilling operations where new data is frequently added and the model requires rapid local responses, cannot be met by traditional methods in terms of timeliness and flexibility. Third, once the model is built, adding new data or making local adjustments later often results in geometric abrupt changes, discontinuous boundaries, or lithological abrupt change zones at the junctions of old and new models, severely impacting the model's reliability in engineering applications. Finally, existing methods lack support for modeling mechanisms of geological process evolution and tectonic continuity, failing to simulate the evolutionary patterns of natural processes such as sequence deposition and tectonic deformation from a geological perspective, resulting in models with high geometric accuracy but insufficient geological interpretation capabilities.

[0003] In summary, traditional geological structural model construction methods still have technical problems such as the inability to directly use multi-source geological data for the structural construction and semantic fusion of 3D models, the lack of dynamic response to new data, and the existence of geometric discontinuities and incoherent structural logic at the boundaries. Summary of the Invention

[0004] This invention provides a dynamic modeling method for three-dimensional geological structures to solve the technical problems that multi-source geological data cannot be directly used for the structural construction and semantic fusion of three-dimensional models, lack of dynamic response to new data, and geometric discontinuity and incoherent construction logic at the boundaries.

[0005] The present invention provides a dynamic modeling method for a three-dimensional geological structure model, specifically comprising the following technical solutions:

[0006] A dynamic modeling method for a three-dimensional geological structure includes the following steps:

[0007] S1. Acquire multi-source data and preprocess it to obtain preprocessed multi-source data; introduce a voxel semantic fusion algorithm based on variational optimization, extract semantic information from the preprocessed multi-source data, map it to the target three-dimensional space grid, and obtain the optimal semantic fusion vector; based on the optimal semantic fusion vector, generate a standard voxel data pool and construct a three-dimensional model of geological structure;

[0008] S2. Monitor data changes and process newly added data points using the aforementioned variational optimization-based voxel semantic fusion algorithm to obtain the location of the new data points. Combine this with the standard voxel data pool to obtain the spatial update region, and model the spatial update region to achieve model update.

[0009] S3. After completing the modeling of the spatial update region, a tensor field is constructed in the boundary transition region, and the weight factors of the boundary points are obtained. Based on the weight factors of the boundary points, a Bezier interpolation surface is constructed to optimize the boundary continuity and realize the dynamic modeling of the three-dimensional geological structure.

[0010] Preferably, S1 specifically includes:

[0011] In the implementation of the voxel semantic fusion algorithm based on variational optimization, semantic vectors are constructed based on preprocessed multi-source data, and scaling and rotation matrices are introduced to map the semantic vectors to the voxel elements of the target 3D spatial grid, thereby obtaining the coordinates of the data points in the standard spatial coordinate system; the target 3D spatial grid is a 3D mesh semantic space generated by using voxel elements as the basic modeling unit.

[0012] Preferably, S1 specifically includes:

[0013] In the implementation of the voxel semantic fusion algorithm based on variational optimization, semantic consistency terms and spatial continuity regularization terms are generated based on the coordinates of data points in the standard spatial coordinate system, and a consistency energy function of preprocessed multi-source data in the semantic dimension is constructed.

[0014] Preferably, S1 specifically includes:

[0015] In the implementation of the voxel semantic fusion algorithm based on variational optimization, the consistency energy function is optimized to obtain the optimal semantic fusion vector. The optimal semantic fusion vector is then bound to the corresponding spatial coordinates and uniformly stored as a three-dimensional voxel unit with semantic labels, thus constructing a standard voxel data pool.

[0016] Preferably, S2 specifically includes:

[0017] A semantic confidence factor for voxels is introduced, and the spatial influence radius of the new data point is calculated based on the spatial relationship between the location of the new data point and the existing point set in the standard voxel data pool.

[0018] Preferably, S2 specifically includes:

[0019] All voxel points whose distance to the new data point is less than the spatial influence radius are grouped into a set and used as the spatial update region for which local reconstruction needs to be performed. Control points are selected based on the spatial update region, and a three-dimensional response function is constructed with the control points as the kernel function center. The three-dimensional response function is evaluated over the entire spatial update region, and the structural boundary is determined to complete the modeling of the spatial update region.

[0020] Preferably, S3 specifically includes:

[0021] By introducing a joint vector of geological attributes at boundary points, a tensor field is constructed, and an evolution tensor is defined.

[0022] Preferably, S3 specifically includes:

[0023] Based on the evolution tensor, a time-sensitive factor is introduced, and combined with the geological structure dip angle, the weight factor of the boundary points is calculated.

[0024] Preferably, S3 specifically includes:

[0025] Based on the weighting factors of the boundary points, the control weights of the Bézier control points are calculated, and the control weights of the Bézier control points are used as input to construct a Bézier interpolation surface on the boundary surface, thereby dynamically updating the three-dimensional geological structure model.

[0026] The beneficial effects of the technical solution of the present invention are:

[0027] 1. By designing a voxel semantic fusion algorithm based on variational optimization, this invention introduces for the first time a unified method for multi-source data structures centered on "semantic vectors" in the preprocessing stage of 3D modeling. Traditional 3D geological modeling often relies on manual format conversion and spatial registration of data such as borehole data, remote sensing images, and 2D seismic profiles, resulting in problems such as semantic inconsistency and scale mismatch. The semantic vector defined in this invention encapsulates lithology, time, confidence level, and spatial variation characteristics into a unified multi-dimensional representation, and, combined with scaling and rotation matrices, achieves automatic mapping of data from different sources to a unified 3D voxel mesh.

[0028] 2. Existing 3D modeling methods are mostly based on static scenes. Once the model is built, it is difficult to flexibly respond to the addition of new data, especially in geological engineering scenarios where new drilling data is frequently added, resulting in low efficiency. In this invention, by calculating the spatial influence radius of new data points and combining it with semantic confidence factors, a spatial update region that can be locally updated as data changes is constructed. Within the spatial update region, a 3D response function is used to simulate structural undulations with a periodic modulation term (cosine wave), and a semantic difference attenuation function is used to control the response range, realizing probabilistic modeling of the geological structures surrounding the new data. It integrates three-dimensional factors of spatial frequency, semantic difference, and spatial distance, possessing high expressive power and accurately capturing complex structural features such as stratigraphic undulations and tectonic folds, avoiding the waste of resources in full model reconstruction, while improving the real-time performance and regional focus of the 3D geological structure model update.

[0029] 3. A tensor field is proposed to be constructed in the boundary transition region and the evolution trend of the boundary points is characterized, so as to accurately capture the changes in spatial attribute gradient. Based on the tensor square trace intensity, geological structure dip angle and time decay, the control weight of the Bezier control points is determined, and a Bezier interpolation surface is constructed on the boundary points. This solves the most difficult problems in 3D modeling of geological structures, such as "unsmooth splicing of old and new models and interruption of structural logic". Attached Figure Description

[0030] Figure 1 This is a flowchart of a dynamic modeling method for a three-dimensional geological structure model according to the present invention. Detailed Implementation

[0031] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0033] The following description, in conjunction with the accompanying drawings, details the specific scheme of the dynamic modeling method for a three-dimensional geological structure provided by this invention.

[0034] See attached document Figure 1 The diagram illustrates a flowchart of a dynamic modeling method for a three-dimensional geological structure model according to an embodiment of the present invention. The method includes the following steps:

[0035] S1. Acquire multi-source data and preprocess it to obtain preprocessed multi-source data; introduce a voxel semantic fusion algorithm based on variational optimization, extract semantic information from the preprocessed multi-source data, map it to the target three-dimensional space grid, and obtain the optimal semantic fusion vector; based on the optimal semantic fusion vector, generate a standard voxel data pool and construct a three-dimensional model of geological structure;

[0036] Multi-source data was acquired from borehole records, remote sensing images, seismic profiles, and geological event logs, including borehole data, remote sensing images, 2D seismic profiles, and historical geological data. Borehole data included lithological sections, density, and porosity listed by depth; remote sensing images reflected surface albedo and vegetation distribution; 2D seismic profiles recorded reflective interface intensity; and historical geological data included fault activity time, slip rate, and deposition rate. The multi-source data underwent preprocessing, such as outlier handling, standardization, and normalization, to obtain preprocessed multi-source data. All preprocessing steps were performed... The techniques used are well-known to those skilled in the art and will not be elaborated upon here. To ensure that the subsequent 3D model is built on a unified data space, a 3D mesh semantic space, namely a 3D voxel mesh, is formed using voxels as the basic modeling unit, and serves as the target 3D space mesh. A voxel semantic fusion algorithm based on variational optimization is introduced to automatically identify the semantic types of the preprocessed multi-source data and fuse their structural distribution in the target 3D space mesh. This maps the preprocessed multi-source data to the target 3D space mesh, forming a voxel mesh structure compatible with geological evolution logic. The specific implementation process is as follows:

[0037] First, for the preprocessed multi-source data, a semantic vector is constructed for each data point in each data source. Its representation is as follows:

[0038] ,

[0039] in, Indicates the first The first data source The semantic vector of each data point; Indicates the first The first data source The lithology code for each data point includes the lithology category to which the data point belongs (e.g., shale, sandstone, etc.). The code is a set of discrete values ​​obtained through the geological exploration drilling process. Indicates the first The first data source The time labels of each data point (such as the time of geological event occurrence) are obtained based on preprocessed multi-source data; Indicates the first The first data source The confidence factor for each data point is determined using statistical methods such as least squares regression and maximum likelihood estimation (MLE), and its value range is [value range missing]. ; Indicates the first The first data source Density gradient estimation related to spatial continuity for each data point can reflect the attribute variation trend between the current point and its spatial neighborhood, and is obtained through density gradient estimation methods such as least squares fitting. Furthermore, the aforementioned semantic vectors are mapped to voxel elements of the target 3D spatial mesh, completing the projection from the original data coordinate system to the target voxel coordinate system, specifically as follows:

[0040] ,

[0041] in, It is the first The first data source The coordinates of the nth data point in the standard spatial coordinate system, which serves as a unified spatial reference for all data fusion. This means that through data transformation, the original data points are mapped from different source coordinate systems to a common three-dimensional coordinate system (i.e., the target three-dimensional spatial grid). Here, the nth data point from all data sources... All data points are mapped to the first data point. On individual factors; These are the coordinates of a point in the original data, i.e., the first... The first data source The coordinates of each data point are three-dimensional spatial coordinates defined in the coordinate system of the original data. Each data point represents geological data from different sources, such as borehole data, seismic profiles, and remote sensing images. It is the first The scaling matrix of the first data source is used to adjust the scaling of the second data source. The scale of each data source is set to be consistent with the scale in the standard spatial coordinate system. This is to ensure that the spatial dimensions of different data sources are consistent when merging. The scale is calculated based on the actual scale differences of the data sources. For example, the resolution of remote sensing images may be different from the spatial accuracy of borehole data, and scaling operations must be used to match them. It is a rotation matrix, representing the rotation of the first... The direction in which the coordinates of a data source are rotated to the standard spatial coordinate system describes the relative angle between the spatial orientation of the data source and the standard spatial coordinate system. This is calculated using known spatial calibration points or known geographic orientation information. It is the first The translation vector of the nth data source is used to transform the nth data source. The coordinate systems of each data source are translated to the standard spatial coordinate system, which is calculated through spatial calibration points or known positional differences, to ensure that the geographical location of the preprocessed multi-source data is consistent with its location in the standard spatial coordinate system.

[0042] After the transformation is complete, construct the semantic consistency energy function of the preprocessed multi-source data. The consistency energy function is used to align and merge multiple data sources in the semantic space; the consistency energy function is defined as follows:

[0043] ,

[0044] in, This refers to the number of data sources; It is the number of voxels; It is the first The first data source The original semantic vector of the nth data point, here the nth The data point and the first Individual element correspondence; It is the first The fusion semantic vector of individual elements, with initial values ​​of all data sources for the first element. The average semantic vector of individual elements; It is the first The first data source The coordinates of each data point in the standard spatial coordinate system; It is the first The first data source The coordinates of each data point in the standard spatial coordinate system; For the first The central reference point of each data source is calculated using the following formula: ; Represents semantic fusion residuals; The square of the semantic fusion residual is used to measure the degree of consistency between the original semantic vector and the fused semantic vector. It is the first The data point and the first The squared Euclidean distance between the central reference points of each data source is used as a distance penalty weighting factor to adjust the influence of the data points. These are regularization term weights, used to balance the constraints of space compactness and semantic consistency. They are determined through cross-validation, and the reference value range is [value range missing]. ; This is a distance decay rate control factor used to control the impact of distance between neighboring points on continuity. It is obtained based on the local reconstruction error minimization method using Gaussian kernel regression, and the reference value range is [value missing]. ; It is a semantic consistency term, representing the first... The data source in the first Semantic information provided on individual prime points and fused semantic vectors Consistency is used to reflect spatial credibility; It is a spatial continuity regularization term that can constrain spatial changes.

[0045] Furthermore, the uniformity energy function By optimizing using the existing stochastic gradient descent method, the optimal representation of each voxel in the semantic dimension, i.e., the optimal semantic fusion vector, is obtained. By binding the optimal semantic fusion vector with its corresponding spatial coordinates and storing them uniformly as three-dimensional voxel units with semantic labels, a standard voxel data pool is constructed. This achieves voxelization and serves as the foundational spatial semantic structure for subsequent modeling and computation. It is the first The coordinates of individual voxels; based on a standard voxel data pool, a three-dimensional model of the geological structure is constructed using existing computer graphics techniques.

[0046] S2. Monitor data changes and process newly added data points using the aforementioned variational optimization-based voxel semantic fusion algorithm to obtain the location of the new data points. Combine this with the standard voxel data pool to obtain the spatial update region, and model the spatial update region to achieve model update.

[0047] After voxelization is complete, the dynamic model reconstruction phase begins. The goal of this phase is to automatically trigger local updates to the 3D geological structure model by monitoring data changes, rather than reconstructing the entire model, thus conserving computational resources. When a new data point is detected, its coordinates are first obtained using a voxel semantic fusion algorithm based on variational optimization. Then, analyze the spatial relationship between the location of the new data points and the existing point sets in the standard voxel data pool to determine the spatial influence radius of the new data points. The specific calculation formula is as follows:

[0048] ,

[0049] in, It is the spatial influence radius of the newly added data point, which represents the maximum spatial distance at which the newly added data point affects the model update in the three-dimensional voxel field, i.e., the local boundary that needs to trigger model reconstruction; This is the influence weighting coefficient, used to control the overall amplification or compression of the influence range of new data points on the existing model. It is a manually set influence coefficient, and the reference value range is [insert range here]. ; It is the first The coordinates of a voxel are spatial points extracted from the standard voxel data pool; This represents the spatial straight-line distance between newly added data points and existing model points, used to measure their proximity; It is the first The semantic confidence factor of individual elements is calculated based on the confidence of semantic fusion residuals, with a reference value range of [value missing]. ;

[0050] Furthermore, the distance to the newly added data points will be less than the spatial influence radius. All voxel points constitute a set and serve as the spatial update region for which local reconstruction needs to be performed. Based on the spatial update region, key points are selected as control points using the existing semantic feature-based K-means method. Then, using the control points as the kernel function centers, a three-dimensional response function is constructed. The three-dimensional response function is updated throughout the entire spatial region. The value is evaluated and compared with the geological validity probability threshold set based on prior geological knowledge to determine the structural boundary, complete the modeling of the spatial update region, and achieve the initial update of the three-dimensional geological structure model; the specific formula of the three-dimensional response function is as follows:

[0051] ,

[0052] in, It is a structure function, i.e., a three-dimensional response function, representing a three-dimensional point. The probability of the existence of the construct or the intensity of the response; It is the number of control points; Indicates the first The lithological dominant weights for each control point are determined based on prior geological knowledge and an expert system, with a reference range of [value range missing]. ; It is the first The coordinates of the control points; These are spatial frequency coefficients, representing the spatial frequency coefficients of the structure along... The directional tendency, used to modulate wave patterns, is automatically obtained using existing principal component analysis (PCA) methods, and its value range is [missing information]. ; Let be the semantic difference degree, representing the th Each control point and the current point The intensity of the difference in the geological attribute space (such as lithology, age, etc.), that is, the L2 norm of the corresponding optimal semantic fusion vector difference; It is a periodic modulation term used to reflect the spatial periodic changes inside the structure and to simulate natural phenomena such as strata undulation and folding. It is the squared Euclidean distance, used to describe the distance between the current point and the first point. Spatial distance between control points; It is a comprehensive weight decay factor, used to characterize the first... Each control point and the current point The semantic-spatial joint influence distance between them; It is a decay term used to prevent the three-dimensional response function from being over-amplified when it is far from the control point, so as to control the influence range of the kernel function and ensure smooth transition and rapid decay;

[0053] Three-dimensional response function The value is defined as the probability of the existence of the structure, and is expressed as isosurfaces in the three-dimensional model of the geological structure. As the structural boundary of the construct, it enables model updates; where... It is a threshold for the probability of geological validity set based on prior geological knowledge, such as This indicates that there is a 50% probability that the structure exists.

[0054] S3. After completing the modeling of the spatial update region, a tensor field is constructed in the boundary transition region, and the weight factors of the boundary points are obtained. Based on the weight factors of the boundary points, a Bezier interpolation surface is constructed to optimize the boundary continuity and realize the dynamic modeling of the three-dimensional geological structure.

[0055] After modeling the spatial update region, boundary continuity optimization is performed to address the spatial discontinuity at the boundary, aiming to eliminate geometric abrupt changes at the boundary connections of the updated model. First, the boundary transition region is obtained using the existing normal vector difference determination method, and a tensor field is constructed within this transition region. To describe the spatial variation trend, the evolution tensor of each point is defined as:

[0056] ,

[0057] in, It is the first The symmetric positive definite tensor constructed at each boundary point, i.e. the evolution tensor; It is the first Joint vector of geological attributes at each boundary point Spatial gradient matrix, joint vector of geological properties , It is the first Lithological classification at each boundary point It is the first The geological formation time of each boundary point , All data were taken from preprocessed multi-source data; Indicates transpose; It is a tensor regularization term used to prevent the gradient field from being zero at certain points, which would cause the tensor to degenerate into a rank-deficient matrix. It is a unit array. It is a perturbation factor that adjusts the tensor rank, and can take values ​​of... It is used to prevent matrix singularities. The evolution tensor describes the direction and magnitude of changes in the local structure and is used to guide the construction of boundary surfaces.

[0058] Based on this, the weight factor for each boundary point is calculated. The weighting factor will be used to control the fitting curvature in Bézier surface interpolation, and its expression is as follows:

[0059] ,

[0060] in, It is the first The control strength of each boundary point in the Bézier surface fitting process, i.e., the weighting factor; It is the square of the evolution tensor norm, representing the squared trace of the evolution tensor, used to reflect the intensity of changes in local properties; It is a time-sensitive factor used to control the decay effect of time on the weighting factor. The larger the time-sensitive factor, the more it emphasizes the suppression effect of time on the weighting factor of the boundary point. It can be determined by existing deposition rate normalization methods. No. The geological dip angle of each boundary point is used to reflect the tendency of tectonic deformation and is obtained by measuring an angle measuring instrument; This is a structural morphology adjustment coefficient used to control the degree to which geological orientation affects smooth fitting. Its value is determined based on the specific application scenario, with a reference range of values. ; The weighting terms representing tensor strength and time decay reflect that points with drastic attribute changes and short time intervals should have higher fitting priority. It is a geometric enhancement term.

[0061] Finally, based on the weighting factor of each boundary point The existing Gaussian kernel function weighted average method is used to determine the control weights of the Bézier control points. The existing Bézier surface technology is used to construct a Bézier interpolation surface on the boundary surface with the control weights of the Bézier control points as input. The Bézier difference surface is added as a new component and dynamically updated to the existing three-dimensional model to realize the dynamic modeling of the three-dimensional geological structure.

[0062] In summary, a dynamic modeling method for three-dimensional geological structures has been developed.

[0063] The order of the embodiments is for illustrative purposes only and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0064] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0065] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A dynamic modeling method for a three-dimensional geological structure model, characterized in that, Includes the following steps: S1. Acquire multi-source data and preprocess it to obtain preprocessed multi-source data; introduce a voxel semantic fusion algorithm based on variational optimization, extract semantic information from the preprocessed multi-source data, map it to the target three-dimensional space grid, and obtain the optimal semantic fusion vector; based on the optimal semantic fusion vector, generate a standard voxel data pool and construct a three-dimensional model of geological structure; S2. Monitor data changes and process newly added data points using the aforementioned variational optimization-based voxel semantic fusion algorithm to obtain the location of the new data points. Combine this with the standard voxel data pool to obtain the spatial update region, and model the spatial update region to achieve model update. S3. After completing the modeling of the spatial update region, a tensor field is constructed in the boundary transition region, and the weight factors of the boundary points are obtained. Based on the weight factors of the boundary points, a Bezier interpolation surface is constructed to optimize the boundary continuity and realize the dynamic modeling of the three-dimensional geological structure.

2. The dynamic modeling method for a three-dimensional geological structure model according to claim 1, characterized in that, S1 specifically includes: In the implementation of the voxel semantic fusion algorithm based on variational optimization, semantic vectors are constructed based on preprocessed multi-source data, and scaling and rotation matrices are introduced to map the semantic vectors to the voxel elements of the target 3D spatial grid, thereby obtaining the coordinates of the data points in the standard spatial coordinate system; the target 3D spatial grid is a 3D mesh semantic space generated by using voxel elements as the basic modeling unit.

3. The dynamic modeling method for a three-dimensional geological structure model according to claim 2, characterized in that, S1 specifically includes: In the implementation of the voxel semantic fusion algorithm based on variational optimization, semantic consistency terms and spatial continuity regularization terms are generated based on the coordinates of data points in the standard spatial coordinate system, and a consistency energy function of preprocessed multi-source data in the semantic dimension is constructed.

4. The dynamic modeling method for a three-dimensional geological structure model according to claim 3, characterized in that, S1 specifically includes: In the implementation of the voxel semantic fusion algorithm based on variational optimization, the consistency energy function is optimized to obtain the optimal semantic fusion vector. The optimal semantic fusion vector is then bound to the corresponding spatial coordinates and uniformly stored as a three-dimensional voxel unit with semantic labels, thus constructing a standard voxel data pool.

5. The dynamic modeling method for a three-dimensional geological structure model according to claim 1, characterized in that, S2 specifically includes: A semantic confidence factor for voxels is introduced, and the spatial influence radius of the new data point is calculated based on the spatial relationship between the location of the new data point and the existing point set in the standard voxel data pool.

6. The dynamic modeling method for a three-dimensional geological structure model according to claim 5, characterized in that, S2 specifically includes: All voxel points whose distance to the new data point is less than the spatial influence radius are grouped into a set and used as the spatial update region for which local reconstruction needs to be performed. Control points are selected based on the spatial update region, and a three-dimensional response function is constructed with the control points as the kernel function center. The three-dimensional response function is evaluated over the entire spatial update region, and the structural boundary is determined to complete the modeling of the spatial update region.

7. The dynamic modeling method for a three-dimensional geological structure model according to claim 1, characterized in that, S3 specifically includes: By introducing a joint vector of geological attributes at boundary points, a tensor field is constructed, and an evolution tensor is defined.

8. The dynamic modeling method for a three-dimensional geological structure model according to claim 7, characterized in that, S3 specifically includes: Based on the evolution tensor, a time-sensitive factor is introduced, and combined with the geological structure dip angle, the weight factor of the boundary points is calculated.

9. The dynamic modeling method for a three-dimensional geological structure model according to claim 8, characterized in that, S3 specifically includes: Based on the weighting factors of the boundary points, the control weights of the Bézier control points are calculated, and the control weights of the Bézier control points are used as input to construct a Bézier interpolation surface on the boundary surface, thereby dynamically updating the three-dimensional geological structure model.

Images