Three-dimensional geological modeling parameter optimization method and system based on deep learning

Abstract

The invention provides a three-dimensional geological modeling parameter optimization method and system based on deep learning, and relates to the field of geological prospecting and modeling.The method comprises the steps that drilling data and remote sensing data of a geological prospecting area are obtained, and multi-source geological data of three-dimensional geological modeling are formed; generating geological profile map data of the geological features according to the drilling data, generating landform profile map data corresponding to the topographic features according to the remote sensing data, and extracting boundary information of the geological exploration area; carrying out gridding processing on a three-dimensional space, dividing a geologic body of the geological exploration area into a plurality of small grid units, and defining geological modeling parameters; the preliminary three-dimensional geological grid model is constructed according to the geological profile map data, the landform contour map data and the boundary information, the size and the resolution of the grid are dynamically adjusted according to the geological change area, the three-dimensional geological model is obtained through optimization and adjustment, the modeling precision is improved, manual intervention is reduced, and computing resources are saved.

Description

Technical Field

[0001] The present invention relates to the field of geological exploration and modeling, and in particular to a three-dimensional geological modeling parameter optimization method and system based on deep learning. Background Art

[0002] 3D geological modeling technology is widely used in mineral exploration, oil and gas exploration, groundwater research, engineering geology and other fields. Its core goal is to provide accurate descriptions of important information such as geological structure, lithology distribution, reservoir characteristics, etc. through modeling and analysis of underground geological information. Traditional 3D geological modeling methods rely on exploration data (such as drilling data, seismic reflection data, geophysical data, etc.) and the experience of geological experts. The geological model is constructed by manually selecting modeling parameters. The selection and optimization of modeling parameters are highly subjective and easily affected by data quality, computational complexity and uncertainty, resulting in low model accuracy, lengthy optimization process, and even difficulty in handling large-scale data sets.

[0003] In recent years, the application of artificial intelligence technologies such as deep learning and reinforcement learning in various industries has made significant progress, but their application in the field of geological modeling is still in the exploratory stage. How to automatically optimize modeling parameters through advanced computing methods is a difficult problem in current geological modeling. Summary of the invention

[0004] In view of this, the purpose of the present invention is to propose a three-dimensional geological modeling parameter optimization method and system based on deep learning. By constructing a deep learning model and combining the characteristics of geological data, the various parameters involved in the modeling process are optimized, the modeling accuracy is improved, manual intervention is reduced, and computing resources are saved.

[0005] To achieve the above object, the present invention provides the following technical solutions: Based on the above objectives, in a first aspect, the present invention provides a method for optimizing three-dimensional geological modeling parameters based on deep learning, comprising the following steps: Acquire drilling data and remote sensing data in geological exploration areas to form multi-source geological data for 3D geological modeling; Preprocess the acquired multi-source geological data, and use spatial location coding to calibrate and match the drilling data and remote sensing data; Generate geological profile data of geological features based on drilling data, generate geomorphic contour map data corresponding to terrain features based on remote sensing data, and extract boundary information of geological exploration areas; Based on the spatial position coding, the generated geological profile data and geomorphic contour map data are gridded in three-dimensional space, and the geological body in the geological exploration area is divided into multiple small grid units by discretization, and the geological modeling parameters including lithology, porosity, mineral composition and fault position are defined in each grid unit; Construct a preliminary 3D geological grid model based on geological profile data, geomorphic contour data, and boundary information, and dynamically adjust the size and resolution of the grid according to the geological change area; Generative adversarial networks are used to optimize geological modeling parameters in the three-dimensional geological grid model, train generators and discriminators, and perform local and global optimization based on the optimized geological modeling parameters to adjust the three-dimensional geological model. A real-time feedback mechanism is set up to compare and verify the output results of the 3D geological model with the actual exploration data, and to correct errors in the model in a timely manner.

[0006] As a further solution of the present invention, the drilling data is geological sample data obtained by drilling during geological exploration, including depth, lithology, porosity, mineral composition, stratigraphic boundaries and fault locations, wherein the drilling data also records the drilling depth and drilling location through GPS positioning, and each set of drilling data is stored in a tabular form.

[0007] As a further solution of the present invention, the remote sensing data is remote sensing data obtained by image preprocessing of remote sensing images containing surface spectral information, temperature, terrain changes, and vegetation coverage data collected by satellite remote sensing, aerial remote sensing and / or unmanned aerial vehicle equipment through sensors, wherein the remote sensing data is stored in the form of raster data and contains terrain features and surface features.

[0008] As a further solution of the present invention, the borehole data and remote sensing data are calibrated and matched using spatial position coding, including the following steps: De-noising and removing redundant data from acquired remote sensing data and drilling data; The acquired borehole data and remote sensing data are converted to the same coordinate system and spatial resolution, and the borehole data and remote sensing data are calibrated and matched using spatial position coding for remote sensing. The data are cropped and geometrically corrected so that the spatial extent of the remote sensing data matches that of the borehole data; For each drilling point in the drilling data , the drilling point position is converted into a feature vector through spatial position encoding For raster data in remote sensing data, each pixel corresponds to a location , generating a spatial code for each pixel ; Among them, the drilling data position coding formula is:

[0009] In the formula, is the geographic coordinate of the drilling data, is the longitude of the drilling point, is the latitude of the drilling point; and The range of longitude and latitude, which is the maximum longitude in the drilling data and maximum latitude The corresponding minimum longitude and minimum latitude difference; The remote sensing data position encoding formula is:

[0010] In the formula, is the spatial coordinate of the remote sensing data, The remote sensing image The longitude of the pixel, For remote sensing images The latitude of pixels; and Same as the borehole data location code, indicating the range of longitude and latitude in geographic coordinates, which is the maximum longitude in remote sensing data and maximum latitude The corresponding minimum longitude and minimum latitude difference; The location codes of the drilling data and remote sensing data are compared and matched, the spatial information is converted into feature vectors, and the K nearest neighbor algorithm is used to match them according to the distance of the location code:

[0011] In the formula, Indicates drilling data and remote sensing data The spatial distance between Feature vector encoding the drilling data location, indicating the drilling point spatial location; The feature vector of the remote sensing data position encoding represents the pixel point in the remote sensing image

[0012] The borehole data are mapped onto the spatial grid of remote sensing data, and the borehole data and remote sensing data are interpolated using matching spatial position codes.

[0013] As a further solution of the present invention, generating geological profile data of geological features according to the drilling data comprises the following steps: Get the drilling position in the drilling data and corresponding geological characteristics information , and generate a position code for each drilling data point ; According to drilling depth Extract geological features at different depths for each borehole and generate data on the relationship between depth and geological features; define a weight function for each borehole data point based on the spatial distance between the borehole data point and the target point ;in:

[0014] In the formula, The drilling data point and the target point The spatial distance is the weight attenuation factor, with a value of 2, indicating the inverse distance square weight; Use weighted interpolation formula to calculate the geological characteristic value at each location in the profile , interpolated by weighted average, according to the drilling depth Generate complete geological profile data; The interpolation results are converted into three-dimensional graphics, according to the geological characteristic values and drilling depth Generate a geological profile, and further generate a surface profile, a lithology distribution map, and a depth analysis map based on the geological profile.

[0015] As a further solution of the present invention, the geomorphic contour map data corresponding to the terrain features is generated according to the remote sensing data, and the boundary depth information of the geological exploration area is extracted, including the following steps: Generate elevation data raster map based on remote sensing data, extract contour lines of different heights based on gradient contour lines, and form a landform contour map; Calculate the slope and aspect based on remote sensing data to determine the potential boundaries of the geological exploration area, and extract the boundaries of the geological exploration area that meet the exploration conditions based on the set slope threshold and elevation threshold; The spatial boundaries of the borehole data are merged with the boundaries of the geological exploration area extracted from remote sensing data using spatial overlap to generate a final geological exploration area boundary map, marking potential exploration areas.

[0016] As a further solution of the present invention, the generated geological profile data and landform contour map data are subjected to three-dimensional grid processing based on spatial position coding, including the following steps: Obtaining the spatial coordinates of the three-dimensional geological profile and the geomorphic contour map, converting the three-dimensional geological profile and the geomorphic contour map into three-dimensional spatial coordinates, and defining a spatial position code for each grid cell; Define the size and range of the grid cell, discretize the geological area into a three-dimensional grid structure according to the spatial coordinate range, each grid cell corresponds to a spatial coordinate point, and generate a grid cell coordinate matrix to assign a spatial code to each grid cell.

[0017] As a further solution of the present invention, the three-dimensional geological profile and the topographic contour map are converted into three-dimensional spatial coordinates, wherein the profile data is ,in, is the profile depth of the drilling point; the contour data is ,in, For remote sensing images When defining the size and range of the grid cells, if the range of the geological area is , when divided into multiple small grid units, the size of the grid unit is set to , then the number of grid cells is: .

[0018] As a further solution of the present invention, each grid unit corresponds to a spatial coordinate point , for each grid cell Allocate space encoding:

[0019] In the formula, is the corresponding grid coordinate value.

[0020] As a further solution of the present invention, constructing a preliminary three-dimensional geological grid model includes the following steps: Convert the data of geological profiles and geomorphic contours into a standard three-dimensional coordinate format, parse the boundary information, and represent each boundary in a three-dimensional coordinate system; Determine the spatial scope of the geological area and preliminarily divide the geological area into uniform grid units along the coordinate axis direction of the three-dimensional coordinate system; According to the preliminary divided grid, the three-dimensional coordinates corresponding to each grid unit and the geological profile data are preliminarily connected so that each grid unit corresponds to a geological point; The geological change rate within each grid unit is calculated and compared with the set change rate threshold. If the geological change rate exceeds the change rate threshold, the grid is subdivided, and local fine grid areas are automatically generated according to the geological change area to construct a preliminary three-dimensional geological grid model.

[0021] As a further solution of the present invention, when training the generator and the discriminator, an adversarial loss function is used to train the generator and the discriminator, the input of the generator is random noise and initial geological parameters, and the output is the optimized geological modeling parameters; the input of the discriminator is the generated geological parameters and the real geological parameters, and the output is a true or false judgment.

[0022] As a further solution of the present invention, the error in the model is corrected, including: calculating the error between the output value of the three-dimensional model and the actual exploration data, when the error is greater than the feedback threshold, starting the real-time feedback mechanism, and completing the error correction through least squares incremental optimization or adaptive grid adjustment.

[0023] In a second aspect, the present invention provides a three-dimensional geological modeling parameter optimization system based on deep learning, comprising: Data acquisition module, used to obtain drilling data and remote sensing data in geological exploration areas, forming multi-source geological data for 3D geological modeling; The data preprocessing module is used to preprocess the acquired multi-source geological data and calibrate and match the drilling data and remote sensing data using spatial position coding; A geological feature extraction module is used to generate geological profile data of geological features based on drilling data, generate geomorphic contour map data corresponding to terrain features based on remote sensing data, and extract boundary information of geological exploration areas; A three-dimensional grid processing module is used to perform three-dimensional grid processing on the generated geological profile data and geomorphic contour map data based on spatial position coding, and to divide the geological body in the geological exploration area into multiple small grid units by discretization, and to define geological modeling parameters including lithology, porosity, mineral composition and fault position in each grid unit; A grid model building module is used to build a preliminary three-dimensional geological grid model based on geological profile data, geomorphic contour map data and boundary information, and dynamically adjust the size and resolution of the grid according to the geological change area; The modeling parameter optimization module is used to optimize the geological modeling parameters in the three-dimensional geological grid model by using the generative adversarial network, train the generator and the discriminator, and perform local and global optimization based on the optimized geological modeling parameters to adjust and obtain the three-dimensional geological model; The feedback adjustment module is used to set up a real-time feedback mechanism, which compares and verifies the output results of the three-dimensional geological model with the actual exploration data and corrects the errors in the model in a timely manner.

[0024] Compared with the prior art, the method and system for optimizing three-dimensional geological modeling parameters based on deep learning proposed in the present invention have the following beneficial effects: 1. Improved geological modeling accuracy and reliability. The present invention can accurately optimize the modeling parameters in the three-dimensional geological grid model through an optimization module based on a generative adversarial network. The iterative training of the generator and the discriminator ensures the high accuracy of the geological modeling parameters, thereby improving the reliability of the model. The real-time feedback mechanism enables the system to automatically correct the errors in the model based on the comparison between the actual exploration data and the model output. The model will be continuously adjusted according to the feedback to ensure the consistency of the geological modeling accuracy with the field data and avoid errors caused by human intervention.

[0025] 2. Reduced manual intervention and improved work efficiency. The method of the present invention reduces manual intervention and subjective judgment in the traditional geological modeling process by introducing deep learning and automated algorithms. The system can automatically process data, optimize models and adjust grids, significantly improving modeling efficiency. Through automated data preprocessing, feature extraction and optimization of generative adversarial networks, high-precision three-dimensional geological modeling can be completed in a short time. Compared with traditional methods, a large amount of manual labor and data processing time are reduced.

[0026] 3. Effectively integrate different types of data and realize the synergy of multi-source data fusion and deep learning. The method of the present invention can effectively integrate multiple data such as drilling data, remote sensing data, geological profiles and geomorphic contours. Through spatial position coding technology, data from different sources can be accurately connected to ensure the consistency and interoperability of various types of data. By optimizing geological modeling parameters through the generative adversarial network of deep learning, it can automatically learn and adapt to the complex laws in geological data, providing a more accurate and flexible modeling solution than traditional methods.

[0027] 4. Dynamic grid division and parameter adjustment are achieved through adaptive grid processing. The system of the present invention can dynamically adjust the size and resolution of the grid according to the changes in geological characteristics. In areas with large or complex geological changes, the grid can be refined to increase the resolution; in areas with small geological changes, the system will automatically reduce the resolution of the grid to avoid unnecessary waste of calculations. Multiple geological modeling parameters including lithology, porosity, mineral composition and fault location are defined in each grid unit to ensure the accuracy and details of the geological model.

[0028] 5. Visualization and interactivity are enhanced through intuitive 3D visualization and interactive operation. The system of the present invention provides an intuitive 3D geological model display to help users clearly understand the geological structure and ore body distribution. Users can view geological features at different levels in a 3D view, which improves the ease of use and comprehensibility of the model. The system of the present invention supports user interaction with the 3D geological model. Users can adjust the viewing angle, zoom, and switch between different geological levels in real time, which facilitates in-depth analysis and decision-making.

[0029] 6. It can respond quickly to changes in actual exploration data, and achieve real-time feedback and rapid optimization through local and global optimization in parallel. The real-time feedback mechanism of the present invention can monitor the output results of the three-dimensional geological model in real time, and compare it with the actual exploration data, so as to timely discover and correct errors in the model. This real-time feedback loop greatly improves the adaptability and accuracy of the model. Through the training of the generative adversarial network, the system can optimize both globally and locally, making the adjustment of the entire geological model more comprehensive and refined, and the optimization process more efficient.

[0030] In summary, the three-dimensional geological modeling parameter optimization method and system based on deep learning of the present invention not only shows significant advantages in modeling accuracy, work efficiency and automation, but also solves many problems in traditional geological modeling methods through deep learning optimization, real-time feedback and multi-source data fusion. It provides a more intelligent, efficient and accurate solution for geological exploration, mineral resource development, environmental monitoring and other fields, and has broad application prospects and huge market potential.

[0031] These and other aspects of the present application will be more concise and understandable in the following description of the embodiments. It should be understood that the above general description and the following detailed description are only exemplary and explanatory and cannot limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the following briefly introduces the drawings required for use in the exemplary embodiments or related technical descriptions. The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the drawings: Figure 1 The present invention is a flowchart of a method for optimizing three-dimensional geological modeling parameters based on deep learning according to an embodiment of the present invention.

[0033] Figure 2 The present invention is a flowchart of calibrating and matching drilling data and remote sensing data in a three-dimensional geological modeling parameter optimization method based on deep learning in an embodiment of the present invention.

[0034] Figure 3 The present invention is a flowchart of generating geological profile data of geological features based on drilling data in a three-dimensional geological modeling parameter optimization method based on deep learning in an embodiment of the present invention.

[0035] Figure 4 The present invention is a flowchart of generating landform contour map data corresponding to terrain features according to remote sensing data in a three-dimensional geological modeling parameter optimization method based on deep learning in an embodiment of the present invention.

[0036] Figure 5 The present invention is a flowchart of a three-dimensional space gridding process in a three-dimensional geological modeling parameter optimization method based on deep learning according to an embodiment of the present invention.

[0037] Figure 6 The present invention provides a flowchart for constructing a preliminary three-dimensional geological grid model in a three-dimensional geological modeling parameter optimization method based on deep learning in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0039] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0040] The traditional 3D geological modeling method relies on exploration data (such as drilling data, seismic reflection data, geophysical data, etc.) and the experience of geological experts. The geological model is constructed by manually selecting modeling parameters. The selection and optimization of modeling parameters are highly subjective and easily affected by data quality, computational complexity and uncertainty, resulting in low model accuracy, lengthy optimization process, and even difficulty in handling large-scale data sets. The present invention proposes a 3D geological modeling parameter optimization method and system based on deep learning. By constructing a deep learning model and combining the characteristics of geological data, the various parameters involved in the modeling process are optimized, the modeling accuracy is improved, manual intervention is reduced, and computing resources are saved.

[0041] See also Figure 1 As shown, an embodiment of the present invention provides a three-dimensional geological modeling parameter optimization method based on deep learning, and the method includes the following steps: Step S10: Acquire drilling data and remote sensing data in the geological exploration area to form multi-source geological data for three-dimensional geological modeling.

[0042] In this step, the borehole data is the geological sample data obtained by drilling during geological exploration, including depth, lithology, porosity, mineral composition, stratigraphic boundaries and fault locations, wherein the borehole data also records the borehole depth and borehole location through GPS positioning, and each set of borehole data is stored in a table. Remote sensing data is remote sensing images collected by satellite remote sensing, aerial remote sensing and / or unmanned aerial vehicle equipment through sensors, including surface spectral information, temperature, terrain changes, and vegetation coverage data, and remote sensing data obtained by image preprocessing, wherein the remote sensing data is stored as raster data, including terrain features and surface features.

[0043] Step S20: pre-process the acquired multi-source geological data, and use spatial position coding to calibrate and match the drilling data and remote sensing data.

[0044] In this step, see Figure 2 As shown, the borehole data and remote sensing data are calibrated and matched using spatial position coding, including the following steps: Step S201, denoising and removing redundant data from the acquired remote sensing data and drilling data; Step S202, converting the acquired drilling data and remote sensing data to the same coordinate system and spatial resolution, using spatial position coding for remote sensing to calibrate the drilling data and remote sensing data, and performing matching data clipping and geometric correction to make the spatial range of the remote sensing data match the drilling data; Step S203: for each drilling point in the drilling data, convert the drilling point position into a feature vector through spatial position coding. For the raster data in the remote sensing data, each pixel point corresponds to a position, and a spatial code is generated for each pixel point. Step S204: compare and match the location codes of the drilling data and the remote sensing data, convert the spatial information into a feature vector, and use the K nearest neighbor algorithm to match according to the distance of the location code: Step S205: Map the drilling data onto the spatial grid of the remote sensing data, and interpolate the drilling data and the remote sensing data using matching spatial position codes.

[0045] In this embodiment, for each drilling point in the drilling data , the drilling point position is converted into a feature vector through spatial position encoding For raster data in remote sensing data, each pixel corresponds to a location , generating a spatial code for each pixel ; Among them, the drilling data position coding formula is:

[0046] In the formula, is the geographic coordinate of the drilling data, is the longitude of the drilling point, is the latitude of the drilling point; and The range of longitude and latitude, which is the maximum longitude in the drilling data and maximum latitude The corresponding minimum longitude and minimum latitude difference; The remote sensing data position encoding formula is:

[0047] In the formula, is the spatial coordinate of the remote sensing data, For remote sensing images The longitude of the pixel, For remote sensing images The latitude of pixels; and Same as the borehole data location code, indicating the range of longitude and latitude in geographic coordinates, which is the maximum longitude in remote sensing data and maximum latitude The corresponding minimum longitude and minimum latitude difference; The location codes of the drilling data and remote sensing data are compared and matched, the spatial information is converted into feature vectors, and the K nearest neighbor algorithm is used to match them according to the distance of the location code:

[0048] In the formula, Indicates drilling data and remote sensing data The spatial distance between Feature vector encoding the drilling data location, indicating the drilling point spatial location; The feature vector of the remote sensing data position encoding represents the pixel point in the remote sensing image .

[0049] For example, when obtaining the drilling data and remote sensing data of the geological exploration area, it is assumed that the exploration area contains 3 drilling data and surface remote sensing image data, as follows: Drilling data: including 3 drilling locations and depth data, the drilling depth is 500 meters, the location is =(100, 200); =(150, 250); = (200, 300); that is: Borehole 1: depth 500 meters, location (longitude 100, latitude 200); Borehole 2: depth 500 meters, location (longitude 150, latitude 250); Borehole 3: Depth 500 m, location (longitude 200, latitude 300).

[0050] Then, the drillhole data includes the following attributes: • Depth: 500 m (maximum depth of each borehole); • Lithology: Sandstone; • Porosity: 0.3; •Mineral composition: contains 10% quartz; • Stratigraphic boundaries: include sedimentary and metamorphic layers; • Fault location: The fault was found at a depth of 200 meters; These drilling data are stored in a table format, with each row representing the location, depth, and lithology information of a drilling point.

[0051] Remote sensing data: Contains surface terrain data with a resolution of 10 meters. The image covers regional terrain data, including surface terrain features such as elevation changes, vegetation cover, temperature and surface spectral information. The data of remote sensing images is raster data, and each pixel represents a geographical location of the remote sensing data.

[0052] Then, in the structure of remote sensing data, the data of each pixel includes: • Latitude and longitude coordinates; • Spectral information (e.g., reflectance in infrared, visible, and near-infrared bands); • Terrain height data (in meters); • Vegetation cover (e.g. degree of green plant cover).

[0053] When preprocessing multi-source geological data and performing calibration and matching, the borehole data and remote sensing data are aligned to the same coordinate system and matched through spatial position encoding. The specific processing process is as follows: First, remove invalid data from the borehole data, such as repeated borehole records or invalid measurement data, remove noise from the remote sensing data, and smooth the image through filtering and denoising algorithms. Then, convert the coordinate system and spatial resolution, and unify the longitude and latitude coordinates of the borehole data (such as 100, 200) with the coordinate system of the remote sensing data. Assume that remote sensing data and borehole data have different spatial resolutions. Remote sensing data has a resolution of 10 meters, while borehole data is a discrete point, so they need to be mapped to the same spatial resolution; then perform spatial position encoding. When encoding the position of the borehole data, generate a spatial position encoding based on the longitude and latitude of the borehole. For example, for the borehole 1 position (100, 200), assuming that the maximum longitude is 200, the minimum longitude is 100, the maximum latitude is 300, and the minimum latitude is 200, then the spatial position encoding of borehole 1 is:

[0054] For drill 2 position (150, 250):

[0055] For drilling hole 3 position (200, 300):

[0056] Then the position encoding of remote sensing data is performed. For example, a pixel point (longitude 105, latitude 205) in a remote sensing image will be encoded as:

[0057] Then, the borehole data and remote sensing data are matched, and the spatial codes of the borehole data and remote sensing data are matched by the K-nearest neighbor algorithm (KNN). For each borehole data point, the distance between it and each pixel of the remote sensing data is calculated to find the closest remote sensing data point and match it. Assuming that the code of borehole 1 is (0, 0), we calculate its distance (Euclidean distance) to the code of a pixel point in the remote sensing image (such as (0.05, 0.05)):

[0058] In this way, the borehole data points are paired with the pixels in the remote sensing data. The matched borehole data are mapped to the spatial grid of the remote sensing data, and the data are interpolated according to the spatial position code to fill the blank areas in each grid that are not directly covered by the borehole data. For example, the data at the borehole 1 location (such as lithology, porosity, etc.) can be mapped to the relevant grid of the remote sensing image through spatial interpolation methods (such as bilinear interpolation or Kriging interpolation), thereby forming a complete three-dimensional geological model.

[0059] Through the above steps, combined with deep learning algorithms and multi-source geological data processing methods, three-dimensional geological models can be efficiently and accurately constructed and optimized. The system docks and interpolates borehole data and remote sensing data through spatial position coding, which can not only solve the multi-source data fusion problem in traditional geological modeling, but also improve the automation and accuracy of the model, thereby supporting more complex geological exploration tasks.

[0060] Step S30: Generate geological profile data of geological features according to the drilling data, generate landform contour map data corresponding to the terrain features according to the remote sensing data, and extract boundary information of the geological exploration area.

[0061] In this step, see Figure 3 As shown, generating geological profile data of geological features based on drilling data includes the following steps: Step S301: Obtain the drilling position in the drilling data and corresponding geological feature information, and generate a location code for each drilling data point; Step S302, extracting geological features for different depth positions of each borehole according to the borehole depth, generating relationship data between the depth and the geological features; defining a weight function for each borehole data point according to the spatial distance between the borehole data point and the target point; Step S303, using a weighted interpolation formula to calculate the geological characteristic value of each position in the profile, interpolating by weighted average value, and generating complete geological profile data according to the drilling depth; Step S304: convert the interpolation result into a three-dimensional graph, generate a geological profile according to the geological characteristic values ​​and the drilling depth, and further generate a surface profile, a lithology distribution map and a depth analysis map according to the geological profile.

[0062] In this embodiment, the drilling position in the drilling data is obtained. and corresponding geological characteristics information , and generate a position code for each drilling data point ; According to the drilling depth Extract geological features at different depths for each borehole and generate data on the relationship between depth and geological features; define a weight function for each borehole data point based on the spatial distance between the borehole data point and the target point .

[0063] in:

[0064] In the formula, The drilling data point and the target point The spatial distance is the weight decay factor, and its value is 2, which means the inverse distance square weight.

[0065] Use weighted interpolation formula to calculate the geological characteristic value at each location in the profile , interpolated by weighted average, according to the drilling depth Generate complete geological profile data; convert interpolation results into three-dimensional graphics, according to geological characteristic values and drilling depth Generate geological cross sections.

[0066] See also Figure 4 As shown, the geomorphic contour map data corresponding to the terrain features is generated according to the remote sensing data, and the boundary depth information of the geological exploration area is extracted, including the following steps: Step S311, generating an elevation data grid map according to remote sensing data, extracting contour lines of different heights based on gradient contour lines, and forming a landform contour map; Step S312: Calculate the slope and slope aspect according to the remote sensing data, determine the potential boundary of the geological exploration area, and extract the boundary of the geological exploration area that meets the exploration conditions according to the set slope threshold and elevation threshold; Step S313: Use spatial overlap to merge the spatial boundaries of the borehole data with the boundaries of the geological exploration area extracted from the remote sensing data to generate a final geological exploration area boundary map, marking the potential exploration area.

[0067] Exemplarily, when generating geological profile data and geomorphic contour data, the borehole location and geological feature information are obtained as follows: • Borehole 1: Location (100, 200), geological characteristics: sandstone, porosity 0.3, mineral content 10% quartz; • Borehole 2: Location (150, 250), geological characteristics: sandstone, porosity 0.32, mineral content 12% quartz; •Drill hole 3: Location (200, 300), geological characteristics: sandstone, porosity 0.28, mineral content 9% quartz.

[0068] Generate spatial position code for each drill hole: • Borehole 1 position code: (0, 0); • Borehole 2 position code: (0.5, 0.5); •Drill hole 3 position code: (1, 1).

[0069] Then, when extracting geological features according to the borehole depth and defining the weight function, the depth information of boreholes 1, 2, and 3 is 500 meters, and the geological feature data of each borehole depth is extracted: such as lithology, porosity, and mineral composition at different depths. Calculate the spatial distance between the borehole data point and the target point. Assume that the target point is any position in the geological profile (such as (125,225)). Calculate the spatial distance between the borehole point and the target point. Assume that the distance is D1, D2, and D3. Assume that the distance between the target point and borehole 1 is 10 meters, the distance between the target point and borehole 2 is 5 meters, and the distance between the target point and borehole 3 is 15 meters, then: • Drill 1 Weight: ; • Drill 2 Weight: ; • Drilling 3 weights: .

[0070] When calculating the profile according to weighted interpolation, it is assumed that Z1, Z2, and Z3 are the geological characteristic values ​​(such as porosity, mineral content, etc.) of borehole 1, borehole 2, and borehole 3 at the target depth respectively. The geological characteristic values ​​of the target point (125, 225) are calculated according to the weights to generate complete geological profile data, showing the relationship between depth and geological characteristics. The profile will show the lithology distribution and porosity changes at different depths. Generate three-dimensional geological profile data based on the interpolation results, and construct a three-dimensional profile through depth and geological characteristic values. Generate the following images at the same time: Surface contour map: shows the surface topographic relief of the exploration area.

[0071] Lithology distribution map: shows the changes in lithology in different areas, such as sandstone, clay, etc.

[0072] Depth analysis map: Analyze the distribution of geological features at different depth levels.

[0073] Then, contour lines of different heights are extracted using contour lines to form a geomorphic outline map, showing the topographic features of the exploration area. Areas with slopes greater than 10° have higher exploration value. Using spatial overlapping technology, the spatial boundaries of the drilling data are merged with the boundaries of the geological exploration area extracted from remote sensing data to generate the final geological exploration area boundary map, marking the potential exploration areas.

[0074] Through the above steps, multi-source data (drilling data and remote sensing data) can be integrated into a detailed three-dimensional geological model, helping explorers to more accurately evaluate the geological characteristics and resource potential of the exploration area.

[0075] Step S40, based on the spatial position coding, the generated geological profile data and landform contour map data are gridded in three-dimensional space, the geological body in the geological exploration area is divided into multiple small grid units by discretization, and geological modeling parameters including lithology, porosity, mineral composition and fault position are defined in each grid unit.

[0076] In this step, see Figure 5 As shown, the three-dimensional gridding process is performed on the generated geological profile data and landform contour map data based on the spatial position coding, including the following steps: Step S401, obtaining the spatial coordinates of the three-dimensional geological profile map and the geomorphic contour map, converting the three-dimensional geological profile map and the geomorphic contour map into three-dimensional spatial coordinates, and defining a spatial position code for each grid unit; Step S402: define the size and range of the grid unit, discretize the geological area into a three-dimensional grid structure according to the spatial coordinate range, each grid unit corresponds to a spatial coordinate point, and generate a grid unit coordinate matrix, and assign a spatial code to each grid unit.

[0077] Among them, the three-dimensional geological profile and landform contour map are converted into three-dimensional spatial coordinates, among which the profile data is ,in, is the profile depth of the drilling point; the contour data is ,in, For remote sensing images When defining the size and range of the grid cells, if the range of the geological area is , when divided into multiple small grid units, the size of the grid unit is set to , then the number of grid cells is: .

[0078] Each grid cell corresponds to a spatial coordinate point , for each grid cell Allocate space encoding:

[0079] In the formula, is the corresponding grid coordinate value.

[0080] For example, when obtaining spatial coordinates and converting them into three-dimensional coordinates, the positions in the borehole data can be converted into a three-dimensional coordinate system (X, Y, Z), where: X, Y are longitude and latitude positions; Z is the depth of the borehole. For borehole 1 (position 100, 200, depth 500 meters), its three-dimensional coordinates are (100, 200, 500), and the elevation information in the remote sensing data is converted into Z coordinates, and each pixel position (such as longitude and latitude position) is paired with the corresponding elevation value. Assume that the range of the geological exploration area is (50,150) to (250,350); divide the area into smaller grid cells. For example, the size of the grid unit is set to 10 meters x 10 meters, and the number of grids in the geological area is: the number of grids on the X axis is 20, the number of grids on the Y axis is 20, and the total number of grids is: 20 × 20 = 400 small grid units; each grid unit has a spatial code, such as: the position code of grid (1, 1) is (1, 1), the position code of grid (1, 2) is (1, 2), and so on.

[0081] When determining the spatial extent of a geological region, the geological region is divided into multiple uniform grids according to the range of the three-dimensional coordinate system. For example, if the boundaries of the region are 50-250 meters (X axis) and 150-350 meters (Y axis), it is divided into 20 × 20 grid cells. Each grid cell is preliminarily connected to the data in the geological profile, and the grid (1, 1) receives the data of borehole 1 and the elevation data in the remote sensing image.

[0082] Step S50: construct a preliminary three-dimensional geological grid model based on the geological profile data, the geomorphic contour map data and the boundary information, and dynamically adjust the size and resolution of the grid according to the geological change area.

[0083] In this step, see Figure 6 As shown in the figure, constructing a preliminary three-dimensional geological grid model includes the following steps: Step S501, converting the data of the geological profile and the landform contour map into a standard three-dimensional coordinate format, parsing the boundary information, and representing each boundary in a three-dimensional coordinate system; Step S502, determining the spatial range of the geological region, and preliminarily dividing the geological region into uniform grid units along the coordinate axis direction of the three-dimensional coordinate system; Step S503: preliminarily connect the three-dimensional coordinates corresponding to each grid unit with the geological profile data according to the preliminarily divided grids, so that each grid unit corresponds to a geological point; Step S504, calculate the geological change rate in each grid unit, and compare it with the set change rate threshold. If the geological change rate exceeds the change rate threshold, increase the grid subdivision, automatically generate local fine grid areas according to the geological change area, and construct a preliminary three-dimensional geological grid model.

[0084] The adversarial loss function is used to train the generator and discriminator to optimize the geological modeling parameters. The generator generates geological data that is as realistic as possible through learning, and the discriminator recognizes the authenticity of the generated data through learning. Ultimately, through adversarial training, the generator can output more realistic and accurate geological modeling parameters.

[0085] Step S60: Optimize the geological modeling parameters in the three-dimensional geological grid model using a generative adversarial network, train the generator and the discriminator, perform local and global optimization based on the optimized geological modeling parameters, and adjust to obtain a three-dimensional geological model.

[0086] In this step, when training the generator and the discriminator, the adversarial loss function is used to train the generator and the discriminator. The input of the generator is random noise and initial geological parameters, and the output is the optimized geological modeling parameters; the input of the discriminator is the generated geological parameters and the real geological parameters, and the output is a true or false judgment.

[0087] The system of the present invention provides an intuitive three-dimensional geological model display, helping users to clearly understand the geological structure and ore body distribution. Users can view geological features at different levels in a three-dimensional view, which improves the ease of use and comprehensibility of the model. The system of the present invention supports the interaction between users and three-dimensional geological models. Users can adjust the viewing angle, zoom, and switch different geological levels in real time, which is convenient for in-depth analysis and decision-making.

[0088] Step S70: Setting a real-time feedback mechanism to compare and verify the output results of the three-dimensional geological model with the actual exploration data, and to correct the errors in the model in a timely manner.

[0089] In this step, the error in the model is corrected, including: calculating the error between the output value of the 3D model and the actual exploration data. When the error is greater than the feedback threshold, the real-time feedback mechanism is activated to complete the error correction through least squares incremental optimization or adaptive grid adjustment.

[0090] The real-time feedback mechanism of the present invention can monitor the output results of the three-dimensional geological model in real time, and compare it with the actual exploration data, so as to timely discover and correct the errors in the model. This real-time feedback loop greatly improves the adaptability and accuracy of the model. Through the training of the generative adversarial network, the system can optimize both globally and locally, making the adjustment of the entire geological model more comprehensive and refined, and the optimization process more efficient.

[0091] The three-dimensional geological modeling parameter optimization method based on deep learning of the present invention can accurately optimize the modeling parameters in the three-dimensional geological grid model through an optimization module based on a generative adversarial network. The iterative training of the generator and the discriminator ensures the high precision of the geological modeling parameters, thereby improving the reliability of the model. The real-time feedback mechanism enables the system to automatically correct the errors in the model according to the comparison between the actual exploration data and the model output. The model will be continuously adjusted according to the feedback to ensure the consistency of the geological modeling accuracy with the field data and avoid errors caused by manual intervention. Through the introduction of deep learning and automated algorithms, manual intervention and subjective judgment in the traditional geological modeling process are reduced. The system can automatically process data, optimize models and adjust grids, significantly improving modeling efficiency. Through automated preprocessing of data, feature extraction and optimization of generative adversarial networks, high-precision three-dimensional geological modeling can be completed in a short time. Compared with traditional methods, a large amount of manual labor and data processing time are reduced.

[0092] The method of the present invention can effectively integrate multiple data such as drilling data, remote sensing data, geological profiles and geomorphic contours. Through spatial position coding technology, data from different sources can be accurately connected to ensure the consistency and interoperability of various types of data. The geological modeling parameters are optimized through the generative adversarial network of deep learning, which can automatically learn and adapt to the complex laws in geological data, and provide a more accurate and flexible modeling solution than traditional methods. Through the changes in geological characteristics, the system can dynamically adjust the size and resolution of the grid. In areas with large or complex geological changes, the grid can be refined to increase the resolution; in areas with small geological changes, the system will automatically reduce the resolution of the grid to avoid unnecessary waste of calculations. Multiple geological modeling parameters including lithology, porosity, mineral composition and fault location are defined in each grid unit to ensure the accuracy and details of the geological model.

[0093] It should be noted that the above figures are only schematic illustrations of the processes included in the method according to an exemplary embodiment of the present invention, and are not intended to be limiting. It is easy to understand that the processes shown in the above figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0094] It should be understood that, although described in a certain order, these steps are not necessarily performed in sequence in the above order. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, a part of the steps of the present embodiment may include a plurality of steps or a plurality of stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a part of the steps or stages in other steps or other steps.

[0095] In a second aspect of the embodiments of the present invention, the present invention further provides a three-dimensional geological modeling parameter optimization system based on deep learning, comprising: Data acquisition module, used to obtain drilling data and remote sensing data in geological exploration areas, forming multi-source geological data for 3D geological modeling; The data preprocessing module is used to preprocess the acquired multi-source geological data and calibrate and match the drilling data and remote sensing data using spatial position coding; A geological feature extraction module is used to generate geological profile data of geological features based on drilling data, generate geomorphic contour map data corresponding to terrain features based on remote sensing data, and extract boundary information of geological exploration areas; A three-dimensional grid processing module is used to perform three-dimensional grid processing on the generated geological profile data and geomorphic contour map data based on spatial position coding, and to divide the geological body in the geological exploration area into multiple small grid units by discretization, and to define geological modeling parameters including lithology, porosity, mineral composition and fault position in each grid unit; A grid model building module is used to build a preliminary three-dimensional geological grid model based on geological profile data, geomorphic contour map data and boundary information, and dynamically adjust the size and resolution of the grid according to the geological change area; The modeling parameter optimization module is used to optimize the geological modeling parameters in the three-dimensional geological grid model by using the generative adversarial network, train the generator and the discriminator, and perform local and global optimization based on the optimized geological modeling parameters to adjust and obtain the three-dimensional geological model; The feedback adjustment module is used to set up a real-time feedback mechanism, which compares and verifies the output results of the three-dimensional geological model with the actual exploration data and corrects the errors in the model in a timely manner.

[0096] Through the above detailed steps, the three-dimensional geological modeling parameter optimization system based on deep learning of the present invention is used to execute the steps of the three-dimensional geological modeling parameter optimization method based on deep learning in the above embodiment, which will not be repeated here. The three-dimensional geological modeling parameter optimization system based on deep learning of the present invention not only shows significant advantages in modeling accuracy, work efficiency and degree of automation, but also solves many problems in traditional geological modeling methods through technologies such as deep learning optimization, real-time feedback and multi-source data fusion. It provides more intelligent, efficient and accurate solutions for the fields of geological exploration, mineral resource development, environmental monitoring, etc., and has broad application prospects and huge market potential.

[0097] The above are exemplary embodiments disclosed in the present invention, but it should be noted that various changes and modifications may be made without departing from the scope disclosed in the embodiments of the present invention as defined in the claims. The functions, steps and / or actions of the method claims according to the disclosed embodiments described herein do not need to be performed in any particular order. In addition, although the elements disclosed in the embodiments of the present invention may be described or required in individual form, they may also be understood as multiple unless explicitly limited to the singular.

[0098] It should be understood that, as used herein, the singular form "a" or "an" is intended to include the plural form as well, unless the context clearly supports an exception. It should also be understood that, as used herein, "and / or" refers to any and all possible combinations of one or more of the items listed in association. The serial numbers of the embodiments disclosed in the above embodiments of the present invention are for description only and do not represent the advantages and disadvantages of the embodiments.

[0099] A person skilled in the art should understand that the discussion of any of the above embodiments is only exemplary and is not intended to imply that the scope of the disclosure of the embodiments of the present invention (including the claims) is limited to these examples; under the concept of the embodiments of the present invention, the technical features in the above embodiments or different embodiments can also be combined, and there are many other changes in different aspects of the above embodiments of the present invention, which are not provided in detail for the sake of simplicity. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention should be included in the protection scope of the embodiments of the present invention.

Claims

1. A three-dimensional geological modeling parameter optimization method based on deep learning, characterized in that: The method comprises the following steps: Acquire drilling data and remote sensing data in geological exploration areas to form multi-source geological data for 3D geological modeling; Preprocess the acquired multi-source geological data, and use spatial location coding to calibrate and match the drilling data and remote sensing data; Generate geological profile data of geological features based on drilling data, generate geomorphic contour map data corresponding to terrain features based on remote sensing data, and extract boundary information of geological exploration areas; Based on the spatial position coding, the generated geological profile data and geomorphic contour map data are gridded in three-dimensional space, and the geological body in the geological exploration area is divided into multiple small grid units by discretization, and the geological modeling parameters including lithology, porosity, mineral composition and fault position are defined in each grid unit; Construct a preliminary 3D geological grid model based on geological profile data, geomorphic contour data, and boundary information, and dynamically adjust the size and resolution of the grid according to the geological change area; Generative adversarial networks are used to optimize geological modeling parameters in the three-dimensional geological grid model, train generators and discriminators, and perform local and global optimization based on the optimized geological modeling parameters to adjust the three-dimensional geological model. A real-time feedback mechanism is set up to compare and verify the output results of the 3D geological model with the actual exploration data, and to correct errors in the model in a timely manner.

2. The three-dimensional geological modeling parameter optimization method based on deep learning according to claim 1, characterized in that: The drilling data is geological sample data obtained by drilling during geological exploration, including depth, lithology, porosity, mineral composition, stratigraphic boundaries and fault locations. The drilling data also records the drilling depth and drilling location through GPS positioning, and each set of drilling data is stored in a table form.

3. The three-dimensional geological modeling parameter optimization method based on deep learning according to claim 2, characterized in that: The remote sensing data is remote sensing images collected by satellite remote sensing, aerial remote sensing and / or unmanned aerial vehicle equipment through sensors, which contain surface spectral information, temperature, terrain changes, and vegetation coverage data, and are obtained by image preprocessing. The remote sensing data is stored in the form of raster data and contains terrain features and surface features.

4. The three-dimensional geological modeling parameter optimization method based on deep learning according to claim 3, characterized in that: The borehole data and remote sensing data are calibrated and matched using spatial location coding, including the following steps: De-noising and removing redundant data from acquired remote sensing data and drilling data; The acquired borehole data and remote sensing data are converted to the same coordinate system and spatial resolution, and the borehole data and remote sensing data are calibrated and matched using spatial position coding for remote sensing. The data are cropped and geometrically corrected so that the spatial extent of the remote sensing data matches that of the borehole data; For each drilling point in the drilling data, the drilling point position is converted into a feature vector through spatial position coding. For the raster data in the remote sensing data, each pixel point corresponds to a position, and a spatial code is generated for each pixel point; The location codes of the drilling data and remote sensing data are compared and matched, the spatial information is converted into feature vectors, and the K nearest neighbor algorithm is used to match them according to the distance of the location code: The borehole data are mapped onto the spatial grid of remote sensing data, and the borehole data and remote sensing data are interpolated using matching spatial position codes.

5. The three-dimensional geological modeling parameter optimization method based on deep learning according to claim 4, characterized in that: Generating geological profile data of geological features based on drilling data includes the following steps: Obtain the borehole location and corresponding geological feature information in the borehole data, and generate a location code for each borehole data point; Extract geological features at different depths of each borehole according to the drilling depth, and generate relationship data between depth and geological features; define a weight function for each borehole data point according to the spatial distance between the borehole data point and the target point; The weighted interpolation formula is used to calculate the geological characteristic value of each position in the profile, and the weighted average value is interpolated to generate complete geological profile data according to the drilling depth; The interpolation results are converted into three-dimensional graphics, and a geological profile is generated according to the geological characteristic values ​​and the drilling depth. The surface contour, lithology distribution map and depth analysis map are further generated according to the geological profile.

6. The method for optimizing three-dimensional geological modeling parameters based on deep learning according to claim 5, characterized in that: Generate geomorphic contour map data corresponding to terrain features according to remote sensing data, and extract boundary depth information of geological exploration areas, including the following steps: Generate elevation data raster map based on remote sensing data, extract contour lines of different heights based on gradient contour lines, and form a landform contour map; Calculate the slope and aspect based on remote sensing data to determine the potential boundaries of the geological exploration area, and extract the boundaries of the geological exploration area that meet the exploration conditions based on the set slope threshold and elevation threshold; The spatial boundaries of the borehole data are merged with the boundaries of the geological exploration area extracted from remote sensing data using spatial overlap to generate a final geological exploration area boundary map, marking potential exploration areas.

7. The three-dimensional geological modeling parameter optimization method based on deep learning according to claim 6, characterized in that: The generated geological profile data and landform contour map data are subjected to three-dimensional grid processing based on spatial position coding, including the following steps: Obtaining the spatial coordinates of the three-dimensional geological profile and the geomorphic contour map, converting the three-dimensional geological profile and the geomorphic contour map into three-dimensional spatial coordinates, and defining a spatial position code for each grid cell; Define the size and range of the grid cell, discretize the geological area into a three-dimensional grid structure according to the spatial coordinate range, each grid cell corresponds to a spatial coordinate point, and generate a grid cell coordinate matrix to assign a spatial code to each grid cell.

8. The method for optimizing three-dimensional geological modeling parameters based on deep learning according to claim 1, characterized in that: The three-dimensional geological profile and landform contour map are converted into three-dimensional spatial coordinates, where the profile data is ,in, is the profile depth of the drilling point; the contour data is ,in, The remote sensing image When defining the size and range of the grid cells, if the range of the geological area is , when divided into multiple small grid units, the size of the grid unit is set to , then the number of grid cells is: .

9. The method for optimizing three-dimensional geological modeling parameters based on deep learning according to claim 8, characterized in that: Constructing a preliminary 3D geological grid model includes the following steps: Convert the data of geological profiles and geomorphic contours into a standard three-dimensional coordinate format, parse the boundary information, and represent each boundary in a three-dimensional coordinate system; Determine the spatial scope of the geological area and preliminarily divide the geological area into uniform grid units along the coordinate axis direction of the three-dimensional coordinate system; According to the preliminary divided grid, the three-dimensional coordinates corresponding to each grid unit and the geological profile data are preliminarily connected so that each grid unit corresponds to a geological point; The geological change rate within each grid unit is calculated and compared with the set change rate threshold. If the geological change rate exceeds the change rate threshold, the grid is subdivided, and local fine grid areas are automatically generated according to the geological change area to construct a preliminary three-dimensional geological grid model.

10. A three-dimensional geological modeling parameter optimization system based on deep learning, characterized in that: The system is used to execute the three-dimensional geological modeling parameter optimization method based on deep learning as described in any one of claims 1 to 9, comprising: Data acquisition module, used to obtain drilling data and remote sensing data in geological exploration areas, forming multi-source geological data for 3D geological modeling; The data preprocessing module is used to preprocess the acquired multi-source geological data and calibrate and match the drilling data and remote sensing data using spatial position coding; A geological feature extraction module is used to generate geological profile data of geological features based on drilling data, generate geomorphic contour map data corresponding to terrain features based on remote sensing data, and extract boundary information of geological exploration areas; A three-dimensional grid processing module is used to perform three-dimensional grid processing on the generated geological profile data and geomorphic contour map data based on spatial position coding, and to divide the geological body in the geological exploration area into multiple small grid units by discretization, and to define geological modeling parameters including lithology, porosity, mineral composition and fault position in each grid unit; A grid model building module is used to build a preliminary three-dimensional geological grid model based on geological profile data, geomorphic contour map data and boundary information, and dynamically adjust the size and resolution of the grid according to the geological change area; The modeling parameter optimization module is used to optimize the geological modeling parameters in the three-dimensional geological grid model by using the generative adversarial network, train the generator and the discriminator, and perform local and global optimization based on the optimized geological modeling parameters to adjust and obtain the three-dimensional geological model; The feedback adjustment module is used to set up a real-time feedback mechanism, which compares and verifies the output results of the three-dimensional geological model with the actual exploration data and corrects the errors in the model in a timely manner.

Images