Construction method of overground and underground integrated geological dynamic grid structure

Abstract

The invention discloses a method for constructing an overground and underground integrated geological dynamic grid structure, and belongs to the field of geological data grid structure construction.The method comprises the steps that geological space data are collected; presetting a grid size, and dividing the geological space data based on the grid size to obtain an initial grid model; clustering the geological spatial data by using a DBSCAN clustering algorithm, and removing noise data points to obtain a plurality of clustering regions; dynamically adjusting the grid size in each clustering region in the initial grid model based on the data point density of each clustering region to obtain a primary refined grid model; on the basis of gradient refinement of a clustering region junction in the primary refinement grid model, a secondary refinement grid model is obtained; and correcting data values of data points in a ground and underground transition zone in the secondary refined grid model by using convolution operation to obtain a final ground and underground integrated grid model. The problem that a traditional grid structure cannot adapt to spatial data distribution of different areas is solved.

Description

Technical Field

[0001] The present invention belongs to the field of geological data grid structure construction, and in particular relates to a method for constructing an above-ground and underground integrated geological dynamic grid structure. Background Art

[0002] Existing 3D geological modeling technology has certain shortcomings when dealing with the integration of ground and underground geological information. Traditional 3D modeling methods often model ground and underground geological information separately, resulting in certain differences in the spatial position and accuracy of the model. This separate modeling method not only reduces the overall consistency and reliability of the model, but also limits the effectiveness of the model in practical applications, especially when it comes to the interaction and relationship between ground and underground facilities, it is difficult to provide a comprehensive geological view and decision support.

[0003] An existing technical solution is the 3D geological model visualization technology based on drilling data: based on geological drilling data and geophysical exploration methods, a 3D visualization model of underground strata is constructed through data fusion and interpolation modeling. The HT framework of Tupu software is used for Web-based 3D visualization. Disadvantages: large amount of generated data, drastic changes in the bottom layer, and large errors in regional models and cutting surfaces.

[0004] An existing technical solution is the theoretical framework and construction method of DEM value-added for the origin of geomorphology: introducing value-added DEM to achieve dynamic association between spatial dimension and time dimension, breaking through the limitation of traditional DEM that only reflects surface elevation. Disadvantages: high requirements of spatial data structure in data storage efficiency, spatial analysis complexity and visualization, and the diversity and complexity of the demand for digital landform modeling and analysis in the field of geomorphology. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for constructing an integrated above-ground and underground geological dynamic grid structure, which solves the problem that the traditional grid structure cannot adapt to the spatial data distribution of different regions.

[0006] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a method for constructing an above-ground and underground integrated geological dynamic grid structure, comprising: Collect geospatial data; Preset the grid size, and divide the geological space data based on the grid size to obtain an initial grid model; The DBSCAN clustering algorithm is used to cluster the geological spatial data and remove noise data points to obtain several clustering areas; Dynamically adjust the grid size in each clustering area in the initial grid model based on the data point density of each clustering area to obtain the initial refined grid model; Refine the junction of clustering regions in the initial refined grid model based on gradients to obtain a secondary refined grid model; Use convolution operations to correct the data values of data points in the transition zone between the ground and underground in the secondary refined grid model to obtain a final integrated above-ground and underground grid model.

[0007] The beneficial effects of the present invention are as follows: The grid is adaptively divided according to the density distribution of spatial data, treating dense and sparse regions differently, avoiding waste of computing resources or insufficient accuracy caused by a unified grid granularity. Dynamically improve the spatial resolution, realize the adaptive grid construction with high resolution in dense regions and low resolution in sparse regions, improve the modeling efficiency and spatial expression ability, and improve the accuracy and rationality of spatial data modeling. By gradient-refining the clustering boundary region, the mutation problem at the boundary of different clustering regions is avoided, making the model have better continuity and reality in the regional transition part, and enhancing the spatial consistency of the model. The calculation content includes the mixed input and processing of point cloud, grid, and vector data, which is easy to integrate with various data types and suitable for various scenarios such as smart cities, underground pipe networks, and mine modeling.

[0008] Furthermore, the expression for the data point density of each clustering region is:

[0009] where is the data point density of the th clustering region; is the number of data points in the th clustering region; is the volume of the convex hull of the th clustering region.

[0010] The beneficial effect of the above further solution is that by calculating the ratio of the number of data points to the area of the region, a comparable spatial density index is obtained, avoiding errors caused by evaluating only the number of points or area alone, and enabling the quantitative measurement of the density within the region.

[0011] Furthermore, the expression for dynamically adjusting the grid size in each clustering region in the initial grid model based on the data point density of each clustering region is:

[0012] where is the adjusted grid size of the th clustering region; is the preset grid size; is the data point density of the th clustering region; is the weighting coefficient given based on geographical features; is the correlation coefficient for controlling the influence degree of density on the grid size; is the control parameter of the logarithmic mapping for balancing the smoothness of grid adjustment.

[0013] The beneficial effects of the above further solution are as follows: By dynamically calculating the grid size of each region, the grid granularity can be flexibly adjusted according to the data point density in different regions, realizing adaptive grid division based on data distribution. Considering various factors such as geography, function, and usage frequency in different regions, the recognition ability of the model for spatial heterogeneity and regional complexity is improved, and the adaptability of the model to spatial heterogeneity is enhanced.

[0014] Further, the secondary refined grid model is obtained by refining the junction of the clustering regions in the initially refined grid model based on gradients, specifically as follows: A1. Set the neighborhood radius, determine the boundary points of the current clustering region and the boundary points of the current adjacent clustering region, and determine the neighborhood of the boundary points of the current clustering region and the neighborhood of the boundary points of the current adjacent clustering region according to the neighborhood radius; A2. Integrate the neighborhood of the boundary points of the current clustering region and the neighborhood of the boundary points of the current adjacent clustering region to obtain the junction region; A3. Calculate the gradient values of each data point in the junction region; A4. Extract the data points with gradient values greater than the gradient threshold as the data points to be refined; A5. For the data points to be refined, refine the grid size of the data points based on gradients; A6. Return to step A3 until the gradient values of all data points are less than the gradient threshold, and enter step A7; A7. Return to A1 to refine the junction points between the current clustering region and the next adjacent clustering region until the junction points between the current clustering region and all adjacent clustering regions are refined, and enter step A8; A8. Return to A1 to refine the junction points between the next clustering region and the adjacent clustering regions until the junction points between all clustering regions and all adjacent clustering regions are refined, obtaining the secondary refined grid model.

[0015] The beneficial effects of the above further scheme are: by calculating the gradient value of the data point to dynamically identify the mutation points in the junction area, and then performing local grid refinement processing, it can effectively alleviate the grid jump phenomenon at the cluster boundary and enhance the expression continuity and accuracy at the junction of the cluster area. The introduction of gradient thresholds and multi-round iteration mechanisms can capture and optimize areas with drastic changes in spatial attributes in a more fine-grained manner, so that the model has a higher response sensitivity and adaptability to heterogeneous changes such as geology and landforms. The boundary area after gradient refinement has higher consistency and smoothness, providing more reasonable input for subsequent operations such as convolution correction and spatial interpolation, and improving the overall reality expression effect and post-processing stability.

[0016] Furthermore, the expression of the gradient value of the data point in A3 is:

[0017] in, For data points The gradient value of is the symbol of partial derivative; is the attribute value of the data point; is the horizontal coordinate of the data point; is the ordinate of the data point; is the vertical coordinate of the data point.

[0018] The beneficial effects of the above further scheme are: quantitatively reflecting the degree of change of data points in different directions, accurately measuring the trend of spatial attribute changes, and providing a scientific basis for boundary refinement. In the case of uneven data density or the presence of outliers, the stability and accuracy of the gradient value are effectively improved, providing basic indicators for multi-scale refinement.

[0019] Furthermore, the expression of the grid size based on the gradient-refined data points is:

[0020]

[0021] in, For the The grid size after the data points to be refined are based on the gradient refinement; For the The grid size of the data points to be refined; For the Data points to be refined The gradient value at ; For the The horizontal coordinate of the data point to be refined; For the The vertical coordinate of the data point to be refined; For the The vertical coordinate of a data point to be refined Is a positive number Is an adjustment factor for controlling the intensity of the influence of the gradient on the grid density Is a harmonic function that fuses the local actual value and the regional gradient change trend Is a data point After fusing the local actual value and the regional gradient change trend, the updated attribute value Is a weight coefficient for controlling the proportion of the original value and the gradient trend Is an attribute value extraction function for data points Is a data point The attribute value of Is the average gradient of the junction area

[0022] The beneficial effects of the above further solution are as follows: By introducing a harmonic function to fuse the original spatial attribute value and the regional gradient change, dynamic perception and response to regions with different spatial change rates are achieved, thereby reasonably adjusting the grid granularity and improving the adaptability of the model to geological structure differences. The setting of parameters such as the weight coefficient and the adjustment factor makes the method have good adjustability. It enhances the spatial analysis ability of the junction area and optimizes the spatial continuity expression of the above-ground and underground transition areas

[0023] Furthermore, the method of using convolution operation to correct the data values of data points in the transition zone between the ground and the underground in the secondary refined grid model to obtain the final integrated above-ground and underground grid model is specifically as follows Identify the transition zone between the ground and the underground in the secondary refined grid model; the transition zone is the overlapping part of the underground clustering area and the ground clustering area in the secondary refined grid model Calculate the local density of each data point in the transition zone Based on the local density of each data point in the transition zone, respectively determine the convolution kernel size of the convolution operation for each data point in the transition zone For each data point in the transition zone, perform convolution operation iteration respectively until the maximum iteration number is reached or the difference between the results of two adjacent iterations is less than the preset threshold, and take the convolution operation iteration output result of each data point in the transition zone as the modified data value of each data point in the transition zone to obtain the final integrated above-ground and underground grid model

[0024] The beneficial effects of the above further solution are as follows: It can effectively eliminate data mutation, jump, and fault phenomena, achieve natural connection and smooth transition of the model in the spatial transition region, and improve the data continuity and smoothness of the ground and underground transition zones. Dynamically adjusting the convolution kernel size based on local density enables more detailed local correction in data-dense regions, while avoiding overfitting in sparse regions, thus more accurately retaining local geological features and structural contours. Preventing over-iteration and information loss, and enhancing numerical stability.

[0025] Further, the expression for the local density of each data point in the transition zone is:

[0026] Where, is the local density of the th data point in the transition zone; is the number of data points within the neighborhood of the th data point in the transition zone; is the neighborhood volume of the th data point in the transition zone; is the neighborhood radius of the data points in the transition zone.

[0027] The beneficial effects of the above further solution are as follows: The calculation of local density realizes the perception of the local spatial distribution characteristics of each data point, enhancing the spatial response ability of the model.

[0028] Further, the expression for the convolution kernel size of the convolution operation for each data point in the transition zone is:

[0029] Where, is the convolution kernel size of the convolution operation for the th data point in the transition zone; is a constant used to control the convolution kernel scale; is the local density of the th data point in the transition zone; is a parameter used to control the influence degree of density on the convolution kernel size.

[0030] The beneficial effects of the above further solution are as follows: Adjusting the convolution kernel size of each data point based on local density makes the convolution kernel smaller in high-density regions to retain details and larger in low-density regions to enhance smoothness. By parameter-regulating the influence degree of density on the convolution kernel, the continuity and stability control of the convolution scale are achieved, thus improving the flexibility and adaptability of the entire transition zone modeling.

[0031] Further, the expression for the convolution operation is:

[0032] Among them, is the output result of the -th iterative convolution operation, indicating that the value at the data point is modified to ; is the convolution kernel function for weighting neighborhood points according to distance; is the output result of the -th iterative convolution operation; is the number of data points within the neighborhood of the -th data point in the transition band; is the Euclidean distance between the -th data point and the -th data point in the transition band; is the parameter for controlling weight decay; is the abscissa of the -th data point; is the ordinate of the -th data point; is the vertical coordinate of the -th data point; is the abscissa of the -th data point; is the ordinate of the -th data point; is the vertical coordinate of the -th data point; is the convolution kernel size of the convolution operation for the -th data point in the transition band.

[0033] The beneficial effect of the above further solution is that the convolution operation based on distance weighting ensures smooth data transition, effectively realizes the spatial weight decay mechanism between data points, and improves the expression continuity of the data in the transition band. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is the flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0035] The following describes the specific embodiments of the present invention to facilitate those skilled in the art to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions created using the concept of the present invention are within the scope of protection.

[0036] As Figure 1As shown in the figure, in one embodiment of the present invention, a method for constructing an integrated on - ground and underground geological dynamic grid structure includes: Collect geological spatial data; Preset the grid size, and divide the geological spatial data based on the grid size to obtain an initial grid model; Use the DBSCAN clustering algorithm to cluster the geological spatial data, and remove the noise data points to obtain several clustering regions; Dynamically adjust the grid size in each clustering region in the initial grid model based on the data point density of each clustering region to obtain a first refined grid model; Refine the junction of the clustering regions in the first refined grid model based on the gradient to obtain a second refined grid model; Use the convolution operation to correct the data values of the data points in the transition zone between the ground and the underground in the second refined grid model to obtain the final integrated on - ground and underground grid model.

[0037] In this embodiment, the geological spatial data includes: ① Three - dimensional spatial coordinate data: The spatial position coordinates (X, Y, Z) of each geological point, which are used to express the spatial position of the geological body on the surface or underground.

[0038] ② Geological attribute data: The attribute values corresponding to each point or cell, such as physical parameters like lithology type (such as sandstone, shale, etc.), density, resistivity, porosity, etc.

[0039] ③ Stratigraphic number or stratification identifier: Whether it belongs to the on - ground / underground identifier.

[0040] ④ Terrain and landform data: Information such as digital elevation model (DEM), contour lines, surface slope, etc.

[0041] ⑤ Underground structure data: Data describing the underground rock formation structure, such as borehole records, seismic exploration results, geological profiles, etc.

[0042] ⑥ Data source information: Data acquisition methods (such as lidar, remote sensing, drilling, geological survey), acquisition accuracy, timestamp, coordinate system information (such as WGS84, CGCS2000, etc.).

[0043] When constructing the grid structure with the above - mentioned data, the three - dimensional spatial coordinate data is used to obtain the coordinates of the data points, and the stratigraphic number or stratification identifier is used to identify whether the data point belongs to the on - ground or underground; while in different application scenarios, the corresponding data is selected as the attribute value of the data point from the geological attribute data, terrain and landform data, and underground structure data, and the data source information is used to identify the relevant information of the attribute value.

[0044] For example, when constructing a density grid model, the attribute value of each data point is density; when constructing a resistivity grid model, the attribute value of each data point is resistivity; when constructing a surface slope grid model, the attribute value of each data point is surface slope, and so on; At the same time, terrain and geomorphic data are also hidden constraints for constructing the grid model, which are used to limit the positions of data points.

[0045] The finally generated grid model has application value in the following aspects: In 3D geological modeling, by integrating surface and underground space information, the finally generated integrated aboveground and underground grid model can be used to construct a more refined 3D geological structure model, providing a data basis for geological surveys, mineral resource assessments, etc.

[0046] In remote sensing and spatial analysis applications, during the processing of remote sensing images and lidar data, the finally generated integrated aboveground and underground grid model can be used as a spatial reference framework for data fusion and feature extraction, improving the efficiency and accuracy of subsequent analysis.

[0047] In this embodiment, in the integrated aboveground and underground spatial data, the data density varies greatly in different regions. Using the traditional unified grid accuracy often cannot effectively reflect the spatial characteristics of the data.

[0048] Using the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, the grid division is adaptively adjusted according to the density distribution of the data. By clustering based on the density distribution of data points, dense regions and sparse regions can be identified, and noise points can be removed, which can handle irregular cluster shapes in spatial data. It can automatically adapt to the data characteristics of different regions, avoiding the limitations of preset grid sizes in traditional methods.

[0049] The expression for the data point density of each clustering region is:

[0050] Where is the data point density of the th clustering region; is the number of data points in the th clustering region; is the convex hull volume of the th clustering region.

[0051] The expression for dynamically adjusting the grid size in each clustering region of the initial grid model based on the data point density of each clustering region is:

[0052] Among them, is the adjusted grid size of the th clustering region; is the preset grid size; is the data point density of the th clustering region; is the weighting coefficient given based on geographical features; is the correlation coefficient that controls the influence degree of density on the grid size; is the control parameter of the logarithmic mapping for balancing the smoothness of grid adjustment.

[0053] In this embodiment, during the process of adaptive grid accuracy, a mathematical relationship is established between the density information in the clustering result and the adjustment of the grid size. The clustering result (i.e., density information) is mapped onto the grid accuracy.

[0054] In this embodiment, the higher the density, it indicates that the points in this area are more concentrated, representing that the data in this area is more dense, and the grid accuracy should be higher. In order to better express the integrated spatial data above and below the ground, an adaptive grid model is designed according to the characteristics of the clustering result combined with geographical information. Guided by geographical information, the rationality and accuracy of grid division are effectively improved. It can dynamically adapt to the complexity of different geographical environments and provide more accurate data analysis results.

[0055] Different geological clusters can be identified according to the clustering result. Semantic annotation is performed according to the spatial distribution characteristics of the clusters, and additional geographical features are introduced into the formula. Using geographical information (such as terrain undulation, building distribution, underground pipeline location, etc.), different regions can have different density mapping rules, which can affect the grid size.

[0056] This method for adjusting the adaptive grid accuracy based on clustering density under the guidance of various geographical information can improve the accuracy and efficiency of spatial data processing. A finer grid is used for complex areas with a larger data density, while a larger grid is used for open areas or relatively uniform underground areas.

[0057] At the junction of the clustering regions in the above-mentioned initially refined grid model based on gradient refinement, a secondary refined grid model is obtained, specifically as follows: A1. Set the neighborhood radius, determine the boundary points of the current clustering region and the boundary points of the current adjacent clustering region, and determine the neighborhood of the boundary points of the current clustering region and the neighborhood of the boundary points of the current adjacent clustering region according to the neighborhood radius; A2. Integrate the neighborhood of the boundary points of the current clustering region and the neighborhood of the boundary points of the current adjacent clustering region to obtain the junction region; A3. Calculate the gradient values of each data point in the junction region; A4. Extract data points with gradient values greater than the gradient threshold as the data points to be refined; A5. For the data points to be refined, based on the gradient, refine the grid size of the data points; A6. Return to step A3 until the gradient values of all data points are less than the gradient threshold, and then proceed to step A7; A7. Return to A1 to refine the intersection points between the current clustering region and the next adjacent clustering region until the refinement of the intersection points between the current clustering region and all adjacent clustering regions is completed, and then proceed to step A8; A8. Return to A1 to refine the intersection points between the next clustering region and its adjacent clustering regions until the refinement of the intersection points between all clustering regions and all their adjacent clustering regions is completed, obtaining a secondary refined grid model.

[0058] In this embodiment, the intersection region is usually a place where data changes significantly, and fine gradient calculation and refinement processing are required to improve the accuracy. The intersection region refers to the boundary region between two clustering regions, which usually has large data changes. The steps to identify the intersection region are as follows: Definition of the clustering region boundary: For every two clustering regions, take the boundary points in the clustering result, and by calculating the neighborhood information of the boundary points of each clustering region, determine which points are on the boundary of the clustering region. Next, use these boundary points for further refinement calculation.

[0059] Gradient change detection: In the intersection region, by calculating the gradient value of each point on the boundary , determine which regions change significantly. The gradient change at the intersection is relatively drastic, which is usually the place that needs to be refined.

[0060] In the intersection region, refinement is implemented according to the gradient information to ensure that the intersection region has sufficient accuracy to accurately depict the transition between the two clustering regions.

[0061] Gradient refinement: When the gradient is greater than the threshold, refine the grid accuracy in the intersection region. The refinement process can increase the grid points through interpolation methods, such as cubic spline interpolation or bilinear interpolation, to make the data smoother and improve the resolution.

[0062] Since the intersection region between clustering regions usually has a complex spatial structure, the refinement process can be carried out iteratively to further improve the accuracy and eliminate unnecessary computational burdens.

[0063] First iteration: For the intersection region between each pair of clustering regions, initially refine according to the gradient value, increase the grid density, and process the regions with large changes.

[0064] Second iteration: After the first iteration, recalculate the gradient of the junction region and further optimize the refinement region based on the new gradient value. If necessary, increase or decrease the precision. For regions with small gradient changes, the precision can be reduced to save computational effort.

[0065] Multiple rounds of iteration: After each round of iteration refinement, check the precision change of the refinement region. If the refinement effect meets the expectation, the iteration can be stopped; otherwise, the iteration can continue until the precision requirement is met.

[0066] The expression for the gradient value of the data point in A3 is:

[0067] Where, is the gradient value of the data point ; is the partial derivative symbol; is the attribute value of the data point; is the abscissa of the data point; is the ordinate of the data point; is the vertical coordinate of the data point.

[0068] The expression for the grid size of the data point refined based on the gradient is:

[0069]

[0070] Where, is the grid size of the th data point to be refined after gradient refinement; is the grid size of the th data point to be refined; is the gradient value of the th data point to be refined ; is the abscissa of the th data point to be refined; is the ordinate of the th data point to be refined; is the vertical coordinate of the th data point to be refined; is a positive number; is an adjustment factor for controlling the intensity of the influence of the gradient on the grid density; is a harmonic function that fuses the local actual value and the regional gradient change trend; is the updated attribute value of the data point after fusing the local actual value and the regional gradient change trend; is the weight coefficient for controlling the proportion of the original value and the gradient trend; is the attribute value extraction function of the data point; is the data point 's attribute value; is the average gradient of the boundary region.

[0071] In this embodiment, is a very small positive number used to avoid division by zero.

[0072] The method of using convolution operation to correct the data values of data points in the transition zone between the ground and the underground in the secondary refined grid model to obtain the final integrated above-ground and underground grid model is as follows: Identify the transition zone between the ground and the underground in the secondary refined grid model; the transition zone is the overlapping part of the clustering region underground and the clustering region on the ground in the secondary refined grid model; Calculate the local density of each data point in the transition zone; Based on the local density of each data point in the transition zone, determine the convolution kernel size of the convolution operation for each data point in the transition zone respectively; For each data point in the transition zone, perform convolution operation iteration respectively until the maximum iteration number is reached or the difference between the results of two adjacent iterations is less than the preset threshold, and use the convolution operation iteration output result of each data point in the transition zone as the modified data value of each data point in the transition zone to obtain the final integrated above-ground and underground grid model.

[0073] In this embodiment, the integrated above-ground and underground spatial data usually includes the overlapping part between the ground and the underground, and the data has high complexity. To adapt to the updated grid size and refine the data accuracy, the data can be locally modified through convolution operation. It can flexibly adapt to the data characteristics of different regions and avoid excessive noise interference. It can improve the spatial resolution and accuracy of the data, especially in complex regions such as the overlapping region of the underground structure and the ground. By analyzing the spatial position and height information of different clusters in the DBSCAN results, the overlapping region between the ground and the underground is identified and the region that needs to be modified by convolution is marked. This region is usually the transition zone between two clusters after DBSCAN clustering. When processing the overlapping region between the ground and the underground, the size of the convolution kernel is dynamically adjusted according to the local density of the point cloud data, and noise is removed and the data is refined through convolution operation. For the overlapping part, the size of the convolution kernel and the step size should be dynamically optimized according to different ground and underground structures. Through convolution operation, noise points are removed, valid data is retained, and the data accuracy is improved.

[0074] It is necessary to calculate the local density of each point in the transition zone after clustering in order to adjust the size of the convolution kernel. For areas with higher density (such as areas where there are many overlapping points on the ground and underground), the size of the convolution kernel should be smaller. For areas with lower density (such as the boundary area between the ground and underground), the size of the convolution kernel should be larger.

[0075] Let the size of the convolution kernel be inversely proportional to the local density The purpose of the convolution operation is to smooth the data through local weighted averaging and remove noise. After the size of the convolution kernel is determined, the convolution operation can be used for local smoothing.

[0076] The expression for the local density of each data point in the transition zone is:

[0077] where is the local density of the th data point in the transition zone; is the number of data points in the neighborhood of the th data point in the transition zone; is the neighborhood volume of the th data point in the transition zone; is the neighborhood radius of the data points in the transition zone.

[0078] The expression for the size of the convolution kernel of the convolution operation for each data point in the transition zone is:

[0079] where is the size of the convolution kernel of the convolution operation for the th data point in the transition zone; is a constant used to control the scale of the convolution kernel; is the local density of the th data point in the transition zone; is a parameter used to control the degree of influence of density on the size of the convolution kernel.

[0080] The expression for the convolution operation is:

[0081] where is the output result of the th iteration of the convolution operation, indicating that the value at the data point is modified to ; is the convolution kernel function used to weight the neighborhood points according to the distance; is the output result of the th iteration of the convolution operation; is the The number of data points within the neighborhood of a data point; For the th data point in the transition band and the th data point in the transition band, the Euclidean distance; A parameter for controlling weight decay; For the th data point, the abscissa; For the th data point, the ordinate; For the th data point, the vertical coordinate; For the th data point, the abscissa; For the th data point, the ordinate; For the th data point, the vertical coordinate; For the th data point in the transition band, the convolution kernel size of the convolution operation.

Claims

1. A method for constructing an integrated on - ground and underground geological dynamic grid structure, characterized in that, Including: Collecting geological spatial data; Presetting a grid size and dividing the geological spatial data based on the grid size to obtain an initial grid model; Using the DBSCAN clustering algorithm to cluster the geological spatial data and removing noise data points to obtain several clustering regions; Dynamically adjusting the grid size in each clustering region in the initial grid model based on the data point density of each clustering region to obtain a first refined grid model; Refining the junction of clustering regions in the first refined grid model based on gradients to obtain a second refined grid model; Using a convolution operation to correct the data values of data points in the transition zone between the ground and the underground in the second refined grid model to obtain a final integrated above-ground and underground grid model.

2. The method for constructing the integrated above-ground and underground geological dynamic grid structure according to claim 1, characterized in that, The expression for the data point density of each said clustering region is: Among them, is the data point density of the th clustering region; is the number of data points in the th clustering region; is the convex hull volume of the th clustering region.

3. The method for constructing the integrated above-ground and underground geological dynamic grid structure according to claim 1, characterized in that, The expression for dynamically adjusting the grid size in each clustering region in the initial grid model based on the data point density of each clustering region is: Among them, is the adjusted grid size of the th clustering region; is the preset grid size; is the data point density of the th clustering region; is the weighting coefficient given based on geographical features; is the correlation coefficient that controls the influence degree of density on the grid size; is the control parameter of the logarithmic mapping for balancing the smoothness of grid adjustment.

4. The method for constructing an integrated above-ground and underground geological dynamic grid structure according to claim 1, characterized in that The method of refining the junction of clustering regions in the first refined grid model based on gradients to obtain a second refined grid model is specifically: A1. Set a neighborhood radius, determine the boundary points of the current clustering region and the boundary points of the current adjacent clustering region, and determine the neighborhood of the boundary points of the current clustering region and the neighborhood of the boundary points of the current adjacent clustering region according to the neighborhood radius; A2. Integrate the neighborhood of the boundary points of the current clustering region and the neighborhood of the boundary points of the current adjacent clustering region to obtain a junction region; A3. Calculate the gradient values of each data point in the junction region; A4. Extract the data points with gradient values greater than the gradient threshold as the data points to be refined; A5. For the data points to be refined, refine the grid size of the data points based on the gradient; A6. Return to step A3 until the gradient values of all data points are less than the gradient threshold, and enter step A7; A7. Return to A1 to refine the junction points between the current clustering region and the next adjacent clustering region until the junction points between the current clustering region and all adjacent clustering regions are refined, and enter step A8; A8. Return to A1 to refine the junction points between the next clustering region and the adjacent clustering region until the junction points between all clustering regions and all adjacent clustering regions are refined to obtain a second refined grid model.

5. The method for constructing an integrated on - ground and underground geological dynamic grid structure according to claim 4, characterized in that, The expression for the gradient value of the data point in A3 is: wherein, is the gradient value of the data point ; is the partial derivative symbol; is the attribute value of the data point; is the abscissa of the data point; is the ordinate of the data point; is the vertical coordinate of the data point.

6. The method for constructing an integrated above-ground and underground geological dynamic grid structure according to claim 4, wherein The expression for refining the grid size of the data point based on the gradient is: Among them, is the grid size after gradient refinement for the th data point to be refined; is the grid size of the th data point to be refined; is the gradient value at the th data point to be refined ; is the abscissa of the th data point to be refined; is the ordinate of the th data point to be refined; is the vertical coordinate of the th data point to be refined; is a positive number; is an adjustment factor for controlling the intensity of the influence of the gradient on the grid density; is a harmonic function for fusing the local actual value and the regional gradient change trend; is the updated attribute value after the data point fuses the local actual value and the regional gradient change trend; is a weight coefficient for controlling the proportion of the original value and the gradient trend; is an attribute value extraction function for the data point; is the attribute value of the data point ; is the average gradient of the junction region.

7. The method for constructing an integrated on - ground and underground geological dynamic grid structure according to claim 1, characterized in that, The method of using a convolution operation to correct the data values of data points in the transition zone between the ground and the underground in the second refined grid model to obtain a final integrated above-ground and underground grid model is specifically: Identifying the transition zone between the ground and the underground in the second refined grid model; the transition zone is the overlapping part of the underground clustering region and the ground clustering region in the second refined grid model; Calculating the local density of each data point in the transition zone; Based on the local density of each data point in the transition zone, respectively determine the convolution kernel size of the convolution operation for each data point in the transition zone; For each data point in the transition zone, perform convolution operation iteration respectively until the maximum iteration number is reached or the difference between the results of two adjacent iterations is less than the preset threshold, and use the output result of the convolution operation iteration of each data point in the transition zone as the modified data value of each data point in the transition zone to obtain a final integrated above-ground and underground grid model.

8. The method for constructing an integrated above-ground and underground geological dynamic grid structure according to claim 7, wherein The expression for the local density of each data point in the transition zone is as follows: Among them, is the local density of the -th data point in the transition zone; is the number of data points within the neighborhood of the -th data point in the transition zone; is the neighborhood volume of the -th data point in the transition zone; is the neighborhood radius of the data points in the transition zone.

9. The method for constructing the integrated on - ground and underground geological dynamic grid structure according to claim 7, characterized in that, The expression for the convolution kernel size of the convolution operation for each data point in the transition zone is as follows: Among them, is the convolution kernel size of the convolution operation for the -th data point in the transition band; is a constant used to control the scale of the convolution kernel; is the local density of the -th data point in the transition band; is a parameter used to control the degree of influence of density on the convolution kernel size.

10. The method for constructing an integrated above-ground and underground geological dynamic grid structure according to claim 7, characterized in that, The expression for the convolution operation is as follows: Among them, is the output result of the -th iterative convolution operation, indicating that the value at the data point is modified to ; is the convolution kernel function for weighting neighborhood points according to distance; is the output result of the -th iterative convolution operation; is the number of data points within the neighborhood of the -th data point in the transition band; is the Euclidean distance between the -th data point and the -th data point in the transition band; is the parameter for controlling weight decay; is the abscissa of the -th data point; is the ordinate of the -th data point; is the vertical coordinate of the -th data point; is the abscissa of the -th data point; is the ordinate of the -th data point; is the vertical coordinate of the -th data point; is the convolution kernel size of the convolution operation for the -th data point in the transition band.

Images