A method and apparatus for constructing a structured geological dataset

Abstract

The application provides a structured geological data set construction method and device, and relates to the technical field of geological data processing. The structured geological data set construction method comprises the following steps: based on a preset multi-level grid division rule, performing multi-level grid division and feature extraction on a geological image of a preset region to obtain grid feature data and a grid position corresponding to each grid; and constructing a geological data set of the preset region according to the grid feature data and the grid position corresponding to all the grids. The application can effectively improve the efficiency of geological data set construction.

Description

Technical Field

[0001] This invention relates to the field of geological data processing technology, and more specifically, to a method and apparatus for constructing structured geological datasets. Background Technology

[0002] Geological datasets, as the core foundation for the digital development of the geological field, are widely used in multiple key scenarios such as geological surveys, mineral resource exploration, geological disaster prediction, and geological environmental assessment, providing data support for industry decision-making and scientific research innovation. With the trend of deep integration between geological data processing and the geological field, geological datasets are also a core element for AI model training and optimization.

[0003] However, the construction of existing geological datasets mainly relies on manual work, which is inefficient and struggles to cope with the massive and multi-source nature of geological data, resulting in slow data accumulation and affecting the efficiency of geological dataset construction. Summary of the Invention

[0004] The problem addressed by this invention is how to improve the efficiency of constructing geological datasets.

[0005] To address the above problems, this invention provides a method and apparatus for constructing structured geological datasets.

[0006] In a first aspect, the present invention provides a method for constructing a structured geological dataset, comprising: Based on the preset multi-level grid division rules, the geological image of the preset area is divided into multi-level grids and features are extracted to obtain the grid feature data and grid location corresponding to each grid. The geological dataset of the preset region is constructed based on the grid feature data and grid location corresponding to all the grids.

[0007] Optionally, the step of performing multi-level grid division and feature extraction on the acquired geological image of the preset area to obtain grid feature data and grid location corresponding to each grid includes: Step 1: Perform initial grid division on the geological image according to the multi-level grid division rules to obtain initial grid cells and corresponding grid positions, wherein the multi-level grid division rules include a preset number of rows and columns for the grid division; Step 2: Extract features from the initial grid cells to obtain the corresponding grid feature data; Step 3: Based on the grid feature data and the grid position, obtain the corresponding grid data density through a preset data density relationship; Step 4: When the grid data density is greater than or equal to a preset density threshold, repeat steps 1 to 3 to perform step-by-step grid division on the initial grid unit until the grid data density corresponding to the divided grid is less than the preset density threshold.

[0008] Optionally, the data density relationship satisfies: ; Where a is the grid data density, N is the number of feature information entries in the grid feature data, and Lon max Lon is the maximum longitude at the grid location. min Lat is the minimum longitude in the grid location. max Lat represents the maximum latitude within the grid location. min The minimum longitude at the grid location.

[0009] Optionally, the step of extracting features from the initial grid cells to obtain the corresponding grid feature data includes: Feature representations are extracted based on the initial grid cells, and the feature representations include color features, line features, and symbol features; The feature information corresponding to each feature is determined based on a preset matching rule, wherein the preset matching rule includes a one-to-one correspondence between the color feature, the line feature, the symbol feature and the feature information; The mesh feature data of the initial mesh cell is generated based on the feature information corresponding to all the feature representations.

[0010] Optionally, constructing the geological dataset for the preset region based on the grid feature data and grid locations corresponding to all the grids includes: Construct an initial matrix based on all the grid positions; The geological dataset is generated by mapping the grid feature data corresponding to all grid locations onto the initial matrix.

[0011] Optionally, constructing the initial matrix based on all the grid positions includes: The corresponding grid coordinates of the grid are determined based on the grid position, wherein the grid coordinates include grid row coordinates and grid column coordinates; The initial matrix is ​​constructed based on all the row coordinates and column coordinates.

[0012] Optionally, the step of mapping the feature data corresponding to all the grid locations to the cells of the initial matrix to generate the geological dataset includes: The cell coordinates corresponding to each cell are determined based on the initial matrix; The corresponding cell is determined based on the grid and a preset coordinate matching relationship, wherein the coordinate matching relationship includes a one-to-one correspondence between the grid coordinates of the grid and the cell coordinates of the cell; The geological dataset is generated by mapping the feature data and grid positions corresponding to all the grids to the corresponding cells.

[0013] Optionally, the geological dataset can be stored in any of the following formats: XML, XLS, or XLSX.

[0014] Optionally, the method further includes: The trained AI model is obtained by training the initial AI model to be trained using the geological dataset.

[0015] In a second aspect, the present invention provides a structured geological dataset construction apparatus, comprising: The grid division module is used to perform multi-level grid division and feature extraction on the acquired geological image of a preset area based on preset multi-level grid division rules to obtain the grid feature data and grid location corresponding to each grid. The dataset construction module is used to construct a geological dataset for the preset area based on the grid feature data and grid locations corresponding to all the grids.

[0016] Thirdly, the present invention provides an electronic device, including a memory and a processor; The memory is used to store computer programs; The processor is configured to implement the structured geological dataset construction method as described in the first aspect when executing the computer program.

[0017] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the structured geological dataset construction method as described in the first aspect.

[0018] The beneficial effects of the structured geological dataset construction method and apparatus of the present invention are as follows: By defining standardized requirements such as grid level, size, boundary, and numbering through preset multi-level grid division rules, it replaces the tedious process of traditional manual region division and point-by-point annotation. Automated batch division achieves full coverage processing of geological images within a preset region, quickly completing grid splitting from coarse to fine granular without manual intervention, significantly reducing the time cost of region division and avoiding rework caused by the subjectivity and inconsistency of manual division. The feature extraction mode based on grid units possesses strong batch processing capabilities. Through adapted algorithms, it can simultaneously perform multi-dimensional feature extraction on image sub-regions corresponding to all grids. Compared to the inefficient method of manually identifying and recording features region by region, it significantly improves the feature extraction rate and scalability. Furthermore, it synchronously associates grid location information during extraction, eliminating the need for subsequent manual matching of features and spatial locations. The dataset construction phase uses the unique grid identifier as the core association key, which can directly and automatically integrate the feature data, location information and auxiliary metadata of the entire grid into a standardized geological dataset. This avoids the tedious operation of manually sorting multi-source data and establishing relationships, and reduces the time loss of data entry and verification. At the same time, the standardized grid data structure reduces the cost of data cleaning and format conversion. The overall process achieves efficient connection from image processing to data integration, which greatly shortens the construction cycle of geological datasets. It can quickly respond to the urgent needs of AI models for large-scale training data and effectively solve the core pain points of the traditional manual construction mode being inefficient and unable to cope with the massive amount of geological data. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a method for constructing a structured geological dataset according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the geological image gridding structure according to an embodiment of the present invention; Figure 3 for Figure 2 A schematic diagram of the grid structure corresponding to region A in the middle; Figure 4 This is a schematic diagram of a structured geological dataset construction device according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the accompanying drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0021] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0022] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to"; the term "based on" means "at least partially based on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments"; and the term "optionally" means "optional embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first," "second," etc., mentioned in this invention are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0023] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0024] The names of the messages or information exchanged between the multiple devices in the embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0025] In related technologies, the construction of geological datasets is mainly done manually. However, manual data collection and entry are inefficient and cannot adapt to the massive and multi-source nature of geological data. This results in slow data accumulation and cannot meet the urgent needs of AI models for large-scale, high-quality training data. At the same time, the manual construction model requires a large investment of professional manpower and time, which is not only costly but also makes it difficult to achieve real-time updates and dynamic maintenance of data. These problems reduce the efficiency and quality of geological dataset construction, restrict the deep implementation and large-scale application of AI technology in the geological field, and become a major bottleneck in the industry's digital transformation.

[0026] To address the problems existing in the aforementioned related technologies, embodiments of the present invention provide a method and apparatus for constructing structured geological datasets.

[0027] like Figure 1 As shown in the figure, an embodiment of the present invention provides a method for constructing a structured geological dataset, comprising: S110, based on the preset multi-level grid division rules, performs multi-level grid division and feature extraction on the acquired geological image of the preset area to obtain the grid feature data and grid location corresponding to each grid.

[0028] Specifically, based on predefined multi-level grid division rules (which specify the number of rows and columns in the grid division, as well as the rules for determining whether to perform the next level of grid division), the images of the preset geological area obtained through remote sensing, geological exploration, and other methods are first preprocessed (including image denoising, geometric correction, grayscale normalization, etc., to ensure that the image quality meets the requirements of subsequent processing). Then, multi-level grid division is performed on the preprocessed geological images according to the preset rules. That is, the highest level coarse grid (initial grid unit) is first divided according to the preset number of rows and columns. Then, it is determined whether each coarse grid needs to be further divided. If so, it is divided according to the preset number of rows and columns, and so on, until the divided grid meets the set requirements. By employing extraction algorithms adapted to geological image features (such as feature extraction models based on convolutional neural networks and traditional texture feature extraction algorithms), features are extracted from the image sub-regions corresponding to each grid. Simultaneously, a grid position mapping relationship is established by combining the image coordinate system, and the two-dimensional coordinate information of each grid in the geological image of the preset area (such as the coordinates of the upper left corner vertex and the grid center coordinates) is accurately recorded. Finally, a unique set of grid feature data and accurate grid position information corresponding to each grid are obtained, providing basic data support for subsequent applications such as geological data correlation analysis and anomaly identification.

[0029] S120, construct the geological dataset of the preset area based on the grid feature data and grid location corresponding to all the grids.

[0030] Specifically, after completing the multi-level grid division and full-grid feature extraction of the geological image of the preset area, and the grid location calibration, the construction process of the geological dataset of the preset area is started based on the grid feature data corresponding to all grids (covering multi-dimensional feature parameters such as gray-scale distribution features, texture structure features, geometric morphology features, and mineral composition association features of geological bodies in each grid, and preprocessing operations such as outlier removal and data normalization have been completed) and grid location information (including the two-dimensional coordinate range of the grid in a unified geological coordinate system, hierarchical affiliation, unique grid number, and spatial topological relationship with adjacent grids). First, using the unique grid number as the core association key, a one-to-one mapping relationship is established between grid feature data and grid location information, ensuring that each set of feature data accurately corresponds to its spatial location within a preset area. Then, geological image metadata (such as data acquisition time, detection equipment parameters, image resolution, etc.), grid division rules, feature extraction algorithm parameters, and other auxiliary information are integrated to construct a multi-dimensional data field system. These fields include basic grid information (number, level), spatial location information (coordinate range, topological relationship), core feature data (various geological feature parameters), and auxiliary descriptive information (data source, processing time, etc.). Next, a structured data storage format (such as relational database tables, standardized JSON files, GeoJSON format, etc., adapted to the spatial attributes and feature dimensions of geological data) is adopted to integrate all grid-related data. Data is stored and organized according to unified standards to achieve orderly data management. Finally, through quality control measures such as data integrity verification (checking for issues such as missing grid data or empty fields), consistency verification (verifying the accuracy of the correlation between feature data and location information, and the uniformity of different grid data formats), and validity verification (screening reasonable data that conforms to geological logic and removing abnormal and noisy data), the data is comprehensively verified and optimized. The final result is a geological dataset that covers the entire preset area, has a standardized data structure, accurate spatial location, complete feature information, and reliable quality. This dataset not only retains the fine-grained expression of geological features, but also realizes a strong correlation between feature data and spatial location through grid location, providing high-quality and structured basic data support for subsequent geological engineering applications such as AI model training, geological anomaly identification, and geological body evolution analysis.

[0031] In this embodiment, standardized requirements such as grid level, size, boundary, and numbering are clearly defined through preset multi-level grid division rules, replacing the tedious process of traditional manual region division and point-by-point annotation. Automated batch division achieves full coverage processing of the geological image of the preset region, quickly completing grid splitting from coarse to fine granular without manual intervention, significantly reducing the time cost of region division and avoiding rework caused by the subjectivity and inconsistency of manual division. The feature extraction mode based on grid units has strong batch processing capabilities. Adaptive algorithms can simultaneously perform multi-dimensional feature extraction on all image sub-regions corresponding to all grids. Compared to the inefficient method of manually identifying and recording features region by region, this significantly improves the feature extraction speed and scalability. Furthermore, grid location information is synchronously associated during extraction, eliminating the need for subsequent manual matching of features and spatial locations. The dataset construction phase uses the unique grid identifier as the core association key, which can directly and automatically integrate the feature data, location information and auxiliary metadata of the entire grid into a standardized geological dataset. This avoids the tedious operation of manually sorting multi-source data and establishing relationships, and reduces the time loss of data entry and verification. At the same time, the standardized grid data structure reduces the cost of data cleaning and format conversion. The overall process achieves efficient connection from image processing to data integration, which greatly shortens the construction cycle of geological datasets. It can quickly respond to the urgent needs of AI models for large-scale training data and effectively solve the core pain points of the traditional manual construction mode being inefficient and unable to cope with the massive amount of geological data.

[0032] Optionally, the step of performing multi-level grid division and feature extraction on the acquired geological image of the preset area to obtain grid feature data and grid location corresponding to each grid includes: Step 1: Perform initial grid division on the geological image according to the multi-level grid division rules to obtain initial grid cells and corresponding grid positions, wherein the multi-level grid division rules include a preset number of rows and columns for the grid division.

[0033] Specifically, based on preset multi-level grid division rules (the core of which includes the preset number of rows and columns for the initial grid division, and clarifies the grid boundary alignment, numbering rules, and coordinate system reference to ensure the standardization and uniformity of the division), the preprocessed geological image is first subjected to initial grid division. The entire geological image is uniformly divided according to the preset number of rows and columns, resulting in several non-overlapping initial grid units that completely cover the geological image, such as... Figure 2 As shown, if the preset number of rows is 11 and the number of columns is 15, then the image is initially divided into a grid. Figure 2The geological map is divided into 11 × 15 = 165 initial grid units. After the division, based on the unified and accurate calibration of the geological image, the grid position information corresponding to each initial grid unit is determined (including the relative positions between grids and the latitude and longitude of the grid location). For example, after the initial grid division, the position information of the first initial grid unit in the upper left corner can include: L1 - row 1 - column 1, corresponding to the longitude range of 114.929°E to 114.944°E and the latitude range of 35.040°N to 35.050°N. Thus, the cell position information of the displacement corresponding to each initial grid is obtained, realizing a one-to-one correspondence between the initial grid unit and the grid position.

[0034] Step 2: Extract features from the initial grid cells to obtain the corresponding grid feature data.

[0035] Specifically, for each initial grid cell, an extraction algorithm adapted to geological features (such as texture feature extraction algorithm, morphological feature analysis model, etc.) is used to extract multi-dimensional features from the corresponding image sub-region, and obtain grid feature data corresponding to each initial grid cell. This data covers key information such as the color distribution parameters, line structure features, and symbol features of the geological body. In addition, preprocessing such as data noise reduction and preliminary screening of outliers are completed simultaneously during the extraction process.

[0036] Step 3: Based on the grid feature data and the grid position, obtain the corresponding grid data density through a preset data density relationship.

[0037] Specifically, based on the acquired grid feature data and grid location information, the grid data density of each initial grid cell is quantitatively calculated using a preset data density relationship (which predefines the calculation method of data density, such as using the number of effective feature parameters, feature information entropy value, or effective feature dimension ratio within a unit grid as the core calculation index), thereby obtaining the grid data density corresponding to each initial grid cell.

[0038] Step 4: When the grid data density is greater than or equal to a preset density threshold, repeat steps 1 to 3 to perform step-by-step grid division on the initial grid unit until the grid data density corresponding to the divided grid is less than the preset density threshold.

[0039] Specifically, the calculated grid data density is compared with a preset density threshold. When the grid data density of an initial grid unit is greater than or equal to the preset density threshold, it indicates that the geological features within that grid unit are relatively dense or complex, requiring further refinement. In this case, the initial grid unit is used as the new processing object, and steps 1 to 3 are repeated. That is, the grid unit is re-divided according to the multi-level grid division rules (using the preset row and column rules or the next level of subdivision rules), resulting in new secondary grid units and their grid positions are marked. Grid feature data is extracted from the secondary grid units, and the corresponding grid data density is calculated. This process is repeated, subdividing and verifying the grid units level by level, until the grid data density of all grid units obtained after division is less than the preset density threshold. The grid division operation is then stopped, ultimately forming a multi-scale, adaptive grid division result, where each grid unit corresponds to unique grid feature data and grid position information. Figure 3 As shown, Figure 2 For the region corresponding to grid A, when it is determined that grid A needs to be divided into grids at different levels, the number of rows and columns of the corresponding grid division is first generated according to the multi-level grid division rules. There are 6 rows and 6 columns. Then, the feature representation of the newly divided secondary grid is extracted to obtain the grid feature data corresponding to each secondary grid.

[0040] In this optional embodiment, an initial grid is first created according to a preset number of rows and columns. Then, the grid data density is calculated based on grid feature data and location. The dense data areas are then progressively subdivided according to a preset density threshold (until the grid data density falls below the threshold). This achieves "on-demand adaptation" of the grid division, ensuring that key areas with rich geological features and high data density receive finer-grained grid subdivisions to accurately capture the detailed features of these areas. This avoids the loss of key information caused by uniform division and avoids over-subdivision of sparse data areas, reducing meaningless feature extraction and data storage redundancy, and significantly lowering computational resource consumption and data processing costs. Simultaneously, the entire process is centered on data density, requiring no manual intervention in subdivision levels and ranges. This achieves automated dynamic adjustment of grid division and feature extraction, improving processing efficiency while ensuring the relevance and effectiveness of the grid data. The resulting grid data possesses both fine-grained feature representation of key areas and overall data compactness and rationality, providing high-quality and efficient basic data support for the subsequent construction of geological datasets. This effectively meets the dual requirements of AI model training for data accuracy and scalability.

[0041] Optionally, the data density relationship satisfies: ; Where a is the grid data density, N is the number of feature information entries in the grid feature data, and Lon max Lon is the maximum longitude at the grid location. min Lat is the minimum longitude in the grid location. max Lat represents the maximum latitude within the grid location. min The minimum longitude at the grid location.

[0042] Optionally, the step of extracting features from the initial grid cells to obtain the corresponding grid feature data includes: Feature representations are extracted based on the initial grid cells, and the feature representations include color features, line features, and symbol features; The feature information corresponding to each feature is determined based on a preset matching rule, wherein the preset matching rule includes a one-to-one correspondence between the color feature, the line feature, the symbol feature and the feature information; The mesh feature data of the initial mesh cell is generated based on the feature information corresponding to all the feature representations.

[0043] In this optional embodiment, color features are achieved by extracting the RGB color space distribution, grayscale histogram statistical features, and hue-saturation-brightness (HSV) space parameters of pixels within the grid, accurately capturing the color differences of geological bodies due to differences in mineral composition and water content; line features are achieved by using edge detection algorithms (such as Canny operator, Sobel operator) and Hough transform techniques to identify and extract parameters such as the direction, length, curvature, density, and intersection of geological structural lines (such as fault lines and stratigraphic boundaries) within the grid, reflecting the spatial structural features of geological bodies; symbol features are achieved by using template matching, morphological recognition, and other methods to locate and extract pre-defined standardized geological symbols (such as stratigraphic codes, mineral markers, borehole location markers, etc.) in the geological image, clarifying the geological semantic information contained within the grid. Subsequently, based on pre-defined standardized matching rules (which have been calibrated according to geological industry standards and have established a unique mapping relationship between color features, line features, symbol features, and specific geological feature information, such as specific RGB value ranges corresponding to sandstone strata, lines with specific orientations and lengths corresponding to thrust faults, and specific morphological symbols corresponding to lead-zinc mineralization points), the three types of feature representations extracted from each initial grid cell are matched one by one to determine the specific geological meaning and quantitative parameters corresponding to each type of feature representation (such as color features corresponding to stratigraphic lithology types, line features corresponding to tectonic activity intensity, and symbol features corresponding to mineral resource markers). Finally, the feature information corresponding to all feature representations is structurally integrated and transformed into quantitative data or classification labels in a unified format, forming exclusive grid feature data for the initial grid cell covering multi-dimensional information such as lithology, structure, and mineral markers of geological bodies. This provides accurate and standardized core feature support for subsequent grid data density calculation, adaptive subdivision, and geological dataset construction.

[0044] For example, such as Figure 2As shown in the figure, in grid A, the color features of the grid cells can be extracted to obtain the corresponding coarse-grained granite and fine-grained granite. Then, by using preset matching rules, the corresponding feature information of the coarse-grained and fine-grained granite is matched. Specifically, the coarse-grained granite belongs to acidic intrusive rocks, with core characteristics including a grain size greater than 5 mm, mostly exhibiting an equigranular structure, some with a porphyritic structure, and complete, clear, and easily identifiable mineral crystals. Its main mineral components are quartz, potassium feldspar, and acidic plagioclase; minor minerals include biotite and amphibole; and accessory minerals are mostly zircon and apatite. This rock formed deep underground in a slowly cooling environment. The low cooling rate allowed the minerals ample time to crystallize and grow, resulting in a uniform color, mostly grayish-white or light pinkish-red. It is hard and dense, with high compressive strength and is not easily weathered. It is commonly used in building decorative panels, municipal curbs, and load-bearing stone for large-scale projects. In geological exploration, it is also a typical indicator for identifying deep-seated intrusive bodies. Fine-grained granite is a fine-grained variety of acidic intrusive rocks. Its key characteristic is that the grain size is less than 2mm, with an overall fine-grained equigranular structure. The grains are fine and uniform, making it difficult to distinguish individual mineral particles with the naked eye. The mineral composition is consistent with coarse-grained granite, mainly composed of quartz, potassium feldspar, and acidic plagioclase, with less biotite and other minor minerals, which have even finer grains. It forms in shallow underground parts, cools rapidly, and has limited time for mineral crystallization, making it difficult to grow. It has a dense and delicate appearance, and its color is mostly light gray or pink. It is hard and has low water absorption, strong wear and corrosion resistance, and a high surface finish after processing. It is suitable for precision building decoration, carving materials and instrument and equipment bases. It is also a representative rock type of shallow acidic intrusive rocks. Similarly, the red line features were extracted as cassiterite-quartz-chlorite veins. Through preset matching rules, the corresponding feature information of cassiterite-quartz-chlorite veins was obtained. Cassiterite-quartz-chlorite veins are a medium-low temperature hydrothermal vein-like mineral aggregate, mainly composed of cassiterite (containing tin ore minerals), quartz (vein framework), and chlorite (alteration minerals). They are often found in the fissures of the surrounding rocks of granite bodies. The veins are mostly irregularly banded or reticulated, and their color ranges from dark green to grayish green depending on the chlorite content. They are dense in texture, with cassiterite mostly embedded in the quartz matrix in fine grains, which is an important indicator of tin mineralization. They are often associated with sulfides such as pyrite. The dark blue line features were extracted as normal faults. Through preset matching rules, the feature information of the faults was obtained, namely, the hanging wall of the fault is more eroded than the footwall, and there are older strata exposed. The information of patchy dotted patterns was extracted, and through preset matching rules, it was found that there is an ancient weathering crust present. Thus, the above-mentioned characteristic information constitutes the grid characteristic data of grid unit A, that is, there are five characteristic information in grid unit A, namely, geological characteristic information such as coarse-grained granite, fine-grained granite, cassiterite-quartz-chlorite vein, normal fault and paleoweathering crust.

[0045] Optionally, constructing the geological dataset for the preset region based on the grid feature data and grid locations corresponding to all the grids includes: Construct an initial matrix based on all the grid positions; The geological dataset is generated by mapping the grid feature data corresponding to all grid locations onto the initial matrix.

[0046] In this optional embodiment, based on the grid location information corresponding to all grids (including the spatial coordinates, hierarchical affiliation, and topological relationships of the grids within the preset area), an initial matrix is ​​constructed that is perfectly adapted to the spatial distribution of the grids. The structure of the initial matrix thus accurately reproduces the spatial layout of the grids, reserving only the storage dimension for feature data. Subsequently, the grid feature data (including quantitative information or classification labels covering multiple dimensions such as color, lines, and symbols) corresponding to each grid location is mapped and filled into the corresponding elements of the initial matrix one by one according to the index correspondence between grid location and matrix elements. This transforms each element of the matrix from an empty placeholder into a structured data unit containing complete feature information of the grid at that location (for example, an element stores the lithology type, structural line parameters, mineral markers, and other feature sets of the corresponding grid). The resulting matrix-form dataset is the geological dataset of the preset area. It retains the spatial location association of the grids through matrix indexing and tightly binds the feature data with the spatial location, forming a structured data carrier that combines spatial topological relationships and multi-dimensional feature information, facilitating data retrieval and processing in subsequent applications such as geological feature analysis and AI model training.

[0047] Optionally, constructing the initial matrix based on all the grid positions includes: The corresponding grid coordinates of the grid are determined based on the grid position, wherein the grid coordinates include grid row coordinates and grid column coordinates; The initial matrix is ​​constructed based on all the row coordinates and column coordinates.

[0048] In this optional embodiment, for each geological grid obtained after multi-level division, according to the preset multi-level grid division rules (including the starting reference of grid rows and columns, and the coordinate encoding logic of different levels of grids), starting from the preset reference point of the geological image (usually the upper left corner of the image), a unique grid coordinate is assigned to each grid according to the spatial arrangement order of the grids in the region. The grid row coordinate corresponds to the grid's arrangement number in the vertical direction, and the grid column coordinate corresponds to the grid's arrangement number in the horizontal direction, ensuring that grids in different spatial locations have non-repeating combinations of row and column coordinates, and realizing a precise one-to-one binding between grid position and grid coordinate. It should be noted that to distinguish the coordinates of the upper and lower level grids, hierarchical identifiers and hierarchical relationship information can be incorporated into the coordinate system. First, a unique hierarchical number is assigned to each level of the multi-level grid (e.g., L1 represents the first level coarse grid, L2 represents the second level fine grid, and the smaller the hierarchical number, the higher level grid it corresponds to). Then, the coordinates of the upper level grid are used as the "prefix identifier" for the coordinates of the lower level grid. Based on the row / column coordinates of the upper level grid, the row / column number of the current level grid within the upper level grid is added to the coordinates of the lower level grid. For example, the coordinates of the upper level grid (L1 level) can be represented as "L1-row x-column y" (representing the first level, row x, column y coarse grid), while the coordinates of the lower level grid (L2 level) within the upper level grid are recorded as "L2-row x-column y-current level row m-current level column n" (where "row x-column y" is the coordinate of the upper level grid to which it belongs, and "current level row m-current level column n" is the arrangement sequence number of the lower level grid within the corresponding upper level grid). This coding method of "hierarchical number + previous level coordinate prefix + current level row and column" can clearly distinguish between different levels of grids and directly locate the previous level grid corresponding to a certain lower level grid through the coordinate prefix, thus avoiding confusion between the coordinates of grids of different levels.

[0049] Furthermore, the maximum value of the row coordinates corresponding to all grid cells is counted (i.e., the highest-level index of the row coordinates in all grids). This determines the coordinates of the cells in the initial matrix that match it. That is, the coordinates of each grid cell correspond one-to-one with the coordinates of the cells in the constructed matrix. Thus, by constructing an empty two-dimensional initial matrix, the multi-level grid structure is reproduced. Each row index of the initial matrix matches the grid row coordinates, and each column index also matches the grid column coordinates. This makes the row and column structure of the initial matrix completely adapt to the spatial distribution pattern of all grids, thus building a spatially corresponding structural carrier for the subsequent mapping and filling of grid feature data.

[0050] Optionally, the step of mapping the feature data corresponding to all the grid locations to the cells of the initial matrix to generate the geological dataset includes: The cell coordinates corresponding to each cell are determined based on the initial matrix; The corresponding cell is determined based on the grid and a preset coordinate matching relationship, wherein the coordinate matching relationship includes a one-to-one correspondence between the grid coordinates of the grid and the cell coordinates of the cell; The geological dataset is generated by mapping the feature data and grid positions corresponding to all the grids to the corresponding cells.

[0051] In this optional embodiment, firstly, using a preset reference point of the initial matrix (usually the top left corner of the matrix) as the origin of the coordinate system, each row index of the initial matrix is ​​defined as the cell row coordinate, and each column index is defined as the cell column coordinate. This assigns a unique cell coordinate (i.e., a combination of row and column coordinates; a hierarchical coordinate division method can also be used to determine cell coordinates at different levels), ensuring that the coordinates of different cells are not repeated, thus achieving precise binding between the cell's position in the matrix and its cell coordinates. Subsequently, based on a preset coordinate matching relationship (this relationship predetermines a one-to-one correspondence between the grid coordinates and the cell coordinates, for example, the grid's row coordinates are equal to the cell's row coordinates, the grid's column coordinates are equal to the cell's column coordinates, or a unique association is established through a preset mapping rule), the grid coordinates of each grid are matched and queried to locate the corresponding target cell in the initial matrix. Finally, the complete feature data (covering multi-dimensional quantitative feature information or classification labels such as color, lines, and symbols) and precise grid location information (including the spatial coordinates and hierarchical affiliation of the grid in the geological image) corresponding to each grid are mapped and filled into the corresponding cells one by one according to the coordinate matching results. This makes each cell of the initial matrix a structured data unit containing the spatial location and multi-dimensional feature information of the corresponding grid. In the end, a geological dataset with both spatial correlation and feature completeness is generated, providing structured and easily accessible data support for subsequent applications such as geological data analysis and AI model training.

[0052] Optionally, the geological dataset can be stored in any of the following formats: XML, XLS, or XLSX.

[0053] In this optional embodiment, the geological dataset can be flexibly stored in any of the XML, XLS, or XLSX formats, fully adapting to the core requirements of multi-source heterogeneity, multi-dimensional extension, and cross-level correlation of geological data. Among them, the XLS / XLSX format, relying on the row and column advantages of general structured tables, can naturally achieve the ordered and standardized storage of massive geological data. Through cell format standardization (such as data type definition, background color annotation, and annotation supplementation), automatic data identification and semantic annotation can be completed. With the help of the multi-sheet association and referencing mechanism, it can also efficiently achieve gridded mapping of geological data at different levels and dimensions, natively supporting three-dimensional spatial data storage and flexible access to time-dimensional extensions, perfectly matching the "multi-level gridded iteration" data preparation and management model; the XML format, with its extensible markup language, ... The three formats can accurately describe the semantic attributes, hierarchical relationships, and spatial association rules of geological data through custom tags, adapting to the structured representation needs of complex and heterogeneous geological data, and are especially suitable for scenarios that require finely defined data association logic. All three formats are compatible with the core requirements of "precise data-space mapping" and "cross-scale data association" in the patented technology, supporting seamless integration of gridded data with geospatial coordinate systems. Furthermore, they can be optimized through data structure to adapt to the input specifications of mainstream AI platforms, providing standardized data support for subsequent hierarchical driving tests of AI models. At the same time, the universality of the three formats can reduce the technical threshold for use by small and medium-sized geological research units.

[0054] Optionally, the method further includes: The trained AI model is obtained by training the initial AI model to be trained using the geological dataset.

[0055] In this optional embodiment, firstly, relying on the constructed multi-level gridded geological dataset (which is generated through an iterative process from coarse to fine, covering spatial coordinates of grids from level one to N, multi-source heterogeneous geological feature data, standardized annotation information, and cross-scale correlation data, and based on a general structured table to achieve ordered storage of spatiotemporal multidimensional data, and has completed data validity verification and normalization processing), an AI model suitable for geological professional tasks (such as mainstream platforms like DeepSeek, iFlytek Xinghuo, and Tongyi Qianwen, which support structured data parsing and hierarchical reasoning) is selected as the initial AI model, and the training is first performed according to the grid level correlation and... The data distribution pattern is used to divide the dataset into training, validation, and test sets. Data augmentation (such as grid feature combination expansion and noise simulation) is used to optimize the data generalization ability. The preprocessed dataset is then input into the model, and hyperparameters (learning rate, number of iterations, etc.) and targeted loss functions are set to adapt to geological feature recognition. The training process of "hierarchical and step-by-step verification" is initiated. The model first learns the mapping relationship of macroscopic geological features based on L1 level coarse grid data. After verification, iteratively learns the microscopic features of L2 level, L3 level, and other fine grids. The model inference results are compared and verified with standardized prior knowledge in the structured knowledge base at each level of training to ensure professional accuracy. After the model completes full-level training and passes the accuracy test of feature recognition at each level, its generalization ability and stability are verified through a test set. Finally, a trained AI model is obtained, thus overcoming the core bottleneck of "high professional threshold and poor specialization effect" of general-purpose large language models in the geological field. It enables efficient parsing of geological data, cross-scale correlation analysis, and accurate identification of professional features, supporting the intelligent implementation of geological tasks such as fault structure identification, rock and mineral classification, mineral exploration pattern analysis, and borehole design assistance. At the same time, through standardized processes and low-cost training modes, it provides professional-grade AI capabilities for small and medium-sized geological research units that lack top-notch AI R&D resources, promoting the deep implementation and large-scale application of AI technology in the geosciences.

[0056] like Figure 4 As shown, an embodiment of the present invention provides a structured geological dataset construction device 300, comprising: The grid division module 410 is used to perform multi-level grid division and feature extraction on the acquired geological image of the preset area based on the preset multi-level grid division rules to obtain the grid feature data and grid position corresponding to each grid. The dataset construction module 420 is used to construct a geological dataset of the preset area based on the grid feature data and grid location corresponding to all the grids.

[0057] The structured geological dataset construction device of this embodiment is used to implement the structured geological dataset construction method described above. Its advantages over the prior art are the same as the advantages of the structured geological dataset construction method compared with the prior art, and will not be repeated here.

[0058] like Figure 5 As shown, an electronic device 500 provided in this embodiment of the invention includes a memory 510 and a processor 520; the memory 510 is used to store a computer program; the processor 520 is used to implement the structured geological dataset construction method as described above when the computer program is executed.

[0059] Alternatively, an electronic device 500 includes a memory 510 and a processor 520 coupled to the memory 510; the memory 510 is configured to store a computer program; and the processor 520 is configured to perform the following operations when the computer program is executed: Based on the preset multi-level grid division rules, the geological image of the preset area is divided into multi-level grids and features are extracted to obtain the grid feature data and grid location corresponding to each grid. The geological dataset of the preset region is constructed based on the grid feature data and grid location corresponding to all the grids.

[0060] This invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the structured geological dataset construction method described above.

[0061] Alternatively, a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the following operations: Based on the preset multi-level grid division rules, the geological image of the preset area is divided into multi-level grids and features are extracted to obtain the grid feature data and grid location corresponding to each grid. The geological dataset of the preset region is constructed based on the grid feature data and grid location corresponding to all the grids.

[0062] The present invention will now be described an electronic device 500 that can serve as a server or client of the present invention, which is an example of a hardware device that can be applied to various aspects of the present invention. Electronic device 500 is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 500 can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0063] Electronic device 500 includes a computing unit that can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) or a computer program loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The computing unit, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0064] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc. In this application, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention according to actual needs. Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units can be implemented in hardware or as software functional units.

[0065] While the present invention has been disclosed above, its scope of protection is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and all such changes and modifications will fall within the scope of protection of the present invention.

Claims

1. A method for constructing a structured geological dataset, characterized in that, include: Based on the preset multi-level grid division rules, the geological image of the preset area is divided into multi-level grids and features are extracted to obtain the grid feature data and grid location corresponding to each grid. The geological dataset of the preset region is constructed based on the grid feature data and grid location corresponding to all the grids.

2. The method for constructing a structured geological dataset according to claim 1, characterized in that, The process of performing multi-level grid division and feature extraction on the acquired geological image of the preset area to obtain grid feature data and grid location corresponding to each grid includes: Step 1: Perform initial grid division on the geological image according to the multi-level grid division rules to obtain initial grid cells and corresponding grid positions, wherein the multi-level grid division rules include a preset number of rows and columns for the grid division; Step 2: Extract features from the initial grid cells to obtain the corresponding grid feature data; Step 3: Based on the grid feature data and the grid position, obtain the corresponding grid data density through a preset data density relationship; Step 4: When the grid data density is greater than or equal to a preset density threshold, repeat steps 1 to 3 to perform step-by-step grid division on the initial grid unit until the grid data density corresponding to the divided grid is less than the preset density threshold.

3. The method for constructing a structured geological dataset according to claim 2, characterized in that, The data density relationship satisfies: ； Where a is the grid data density, N is the number of feature information entries in the grid feature data, and Lon max Lon is the maximum longitude at the grid location. min Lat is the minimum longitude in the grid location. max Lat represents the maximum latitude within the grid location. min The minimum longitude at the grid location.

4. The method for constructing a structured geological dataset according to claim 2, characterized in that, The step of extracting features from the initial grid cells to obtain the corresponding grid feature data includes: Feature representations are extracted based on the initial grid cells, and the feature representations include color features, line features, and symbol features; The feature information corresponding to each feature is determined based on a preset matching rule, wherein the preset matching rule includes a one-to-one correspondence between the color feature, the line feature, the symbol feature and the feature information; The mesh feature data of the initial mesh cell is generated based on the feature information corresponding to all the feature representations.

5. The method for constructing a structured geological dataset according to claim 1, characterized in that, The step of constructing the geological dataset for the preset region based on the grid feature data and grid locations corresponding to all the grids includes: Construct an initial matrix based on all the grid positions; The geological dataset is generated by mapping the grid feature data corresponding to all grid locations onto the initial matrix.

6. The method for constructing a structured geological dataset according to claim 5, characterized in that, The step of constructing the initial matrix based on all the grid positions includes: The corresponding grid coordinates of the grid are determined based on the grid position, wherein the grid coordinates include grid row coordinates and grid column coordinates; The initial matrix is ​​constructed based on all the row coordinates and column coordinates.

7. The method for constructing a structured geological dataset according to claim 6, characterized in that, The step of mapping the feature data corresponding to all the grid positions to the cells of the initial matrix to generate the geological dataset includes: The cell coordinates corresponding to each cell are determined based on the initial matrix; The corresponding cell is determined based on the grid and a preset coordinate matching relationship, wherein the coordinate matching relationship includes a one-to-one correspondence between the grid coordinates of the grid and the cell coordinates of the cell; The geological dataset is generated by mapping the feature data and grid positions corresponding to all the grids to the corresponding cells.

8. The method for constructing a structured geological dataset according to any one of claims 1 to 7, characterized in that, The geological dataset can be stored in any of the following formats: XML, XLS, or XLSX.

9. The method for constructing a structured geological dataset according to any one of claims 1 to 7, characterized in that, Also includes: The initial AI model to be trained is trained using the geological dataset to obtain a trained AI model.

10. A structured geological dataset construction device, characterized in that, include: The grid division module is used to perform multi-level grid division and feature extraction on the acquired geological image of a preset area based on preset multi-level grid division rules, so as to obtain the grid feature data and grid location corresponding to each grid. The dataset construction module is used to construct a geological dataset for the preset area based on the grid feature data and grid locations corresponding to all the grids.

Images