Geological exploration achievement intelligent analysis and visual display system

Abstract

The application belongs to the technical field of geological data visualization, and specifically discloses an intelligent analysis and visualization display system for geological exploration results, which comprises the following steps: extracting a report text to generate a trace reference and a bottom layer byte offset through a knowledge graph construction module; processing a two-dimensional profile drawing by a topological analysis module, and injecting a topological placeholder node at a stratum missing position; matching a text semantic with a spatial adjacency graph by an isomorphic matching module and writing the matching result into an attribute; generating a three-dimensional voxel set by an interpolation module, calculating a spatial interpolation weight to construct a multi-source trace index array; and driving a weighted display and interactive pickup of the text and a three-dimensional view based on the array and the weight by a visualization trace module. The application overcomes the data isolation defect of geological text and a three-dimensional model, establishes a reliable addressing path between a bottom layer stream file and a three-dimensional space, realizes bidirectional collaborative positioning and tracing, and accurately restores a complex stratum topological structure.

Description

Technical Field

[0001] This application relates to the field of geological data visualization technology, specifically to an intelligent analysis and visualization system for geological exploration results. Background Technology

[0002] Geological exploration results are typically stored and delivered in various heterogeneous data formats, including text reports, two-dimensional geological profiles, and three-dimensional geological models. Existing geological data visualization technologies primarily focus on reconstructing the geometry of underground space based on two-dimensional vector drawings. However, in practical applications, data isolation is prevalent between exploration report text and three-dimensional spatial models. The three-dimensional model is usually presented as an independent visual result, lacking a physical addressing relationship between its internal primitives and the underlying original text data stream. When technicians examine a specific geological body in the three-dimensional view, the system struggles to automatically locate and extract the original report text paragraphs supporting the attributes of that three-dimensional entity; similarly, when reviewing professional text reports, it cannot dynamically drive the three-dimensional view to highlight the corresponding spatial geometric location. This static display mode, lacking an underlying mapping relationship, hinders bidirectional collaborative tracing between text records and three-dimensional spatial locations.

[0003] Furthermore, due to the lack of a joint analysis mechanism for heterogeneous data, existing spatial reconstruction methods have limitations in handling complex geological topological relationships. Particularly when encountering stratigraphic gaps caused by geological evolution, conventional 2D drawings often fail to explicitly represent the missing strata graphically. Because no topological compensation mechanism is established for stratigraphic gaps, the system typically can only perform direct surface-to-surface connections based on existing geometric boundaries when performing spatial interpolation on adjacent 2D profiles. This approach ignores the continuity of geological sequences, easily leading to unnatural geometric fractures or topological sequence disorder during 3D modeling. It fails to accurately reconstruct transitional morphologies such as stratigraphic pinch-outs that conform to actual physical sedimentary patterns, thus reducing the rigor and usability of digitally displaying complex geological structures. Summary of the Invention

[0004] To address this, this application provides an intelligent analysis and visualization system for geological exploration results, which solves the problems of existing technologies where the separation of geological report text and three-dimensional spatial model data makes it difficult to achieve bidirectional collaborative traceability, and the inaccurate reconstruction of topological structure in areas with missing strata.

[0005] To achieve the above objectives, this application provides the following technical solution: A smart analysis and visualization system for geological exploration results includes: The knowledge graph construction module is used to construct a text semantic directed acyclic graph and a knowledge graph of geological entity relationships based on geological exploration report text. It extracts the document identifier and starting byte offset of each node in the corresponding standardized text stream file, generates text source references and encapsulates them as node attributes into the node attribute structure of the corresponding node. The topology parsing module is used to generate an initial spatial adjacency graph based on the input two-dimensional vector geological profile map, perform a topological comparison between the initial spatial adjacency graph and the geological entity relationship knowledge graph, and inject topological placeholder nodes at the missing strata locations to generate a final spatial geometric adjacency graph. The isomorphic matching module is used to calculate the subgraph isomorphic node mapping relationship between the text semantic directed acyclic graph and the final spatial geometric adjacency graph, and write the geological entity attributes and the text source reference into the attribute structure of the corresponding node according to the mapping relationship; The spatial interpolation module is used to perform interpolation operations on adjacent two-dimensional profiles that have completed attribute writing to generate a three-dimensional voxel set, calculate the spatial interpolation weight of the interpolated voxels relative to the adjacent two-dimensional profiles, read the text source references in the corresponding nodes of the adjacent two-dimensional profiles, and jointly encapsulate each text source reference with the corresponding spatial interpolation weight to generate a multi-source source index array. The visualization and tracing module is used to render the three-dimensional voxel set to generate a three-dimensional view. When receiving an interactive instruction for the interpolated voxel, it reads the multi-source tracing index array, locates and displays the corresponding text paragraph according to the document identifier and the starting byte offset, and displays the text paragraph with weights based on the spatial interpolation weight.

[0006] By adopting the above technical solution, a reliable addressing association can be established between the underlying text stream file and the three-dimensional visual space due to the joint mechanism of multi-source tracing index array and topology placeholder nodes. In the specific processing, the system obtains the starting byte offset by calculating the physical byte length occupied by the character, thereby obtaining physical address data independent of the front-end typesetting format.

[0007] Based on this, the spatial geometric relationships of the two-dimensional profiles are extracted and topologically compared with a knowledge graph formed from expert experience extracted from the text. When encountering stratigraphic gaps caused by unconformities, topological placeholder nodes with zero area are injected at the corresponding spatial locations. These placeholder nodes maintain the continuity of the geological sequence, giving the discrete two-dimensional geometric slices global spatial semantics.

[0008] As heterogeneous data processing advances, the system performs subgraph isomorphic matching between the directed acyclic graph of the text and the geometric adjacency graph of the cross-section, attaching text source references to spatial nodes to complete attribute association. In the subsequent generation of 3D voxels, the system assigns weights based on the Euclidean spatial distance between the interpolated voxels and adjacent 2D cross-sections, and uses a multi-source source index array to record the degree of influence of each data source cross-section on the voxels. This helps improve the data discontinuity situation where 3D interpolated voxels lack direct text correspondences.

[0009] Finally, during interactive picking, the system adjusts the opacity of the text display background by extracting interpolation weights, allowing users to intuitively identify the weight distribution ratio of the current 3D location by multiple sides of the exploration text. Therefore, it achieves bidirectional weighted collaborative localization between geological exploration text data and the 3D geological model, as well as a reasonable reconstruction of complex stratigraphic structures.

[0010] Preferably, the knowledge graph construction module divides the relationship edges corresponding to stratigraphic superposition relationships into a stratigraphic superposition relationship edge set, divides the relationship edges corresponding to magmatic intrusion relationships, fault cutting relationships, or other relationships that cause existing geological bodies to intersect or cut into an intersecting and cutting relationship edge set, sets an intersecting and cutting relationship flag with a first value for the relationship edges in the intersecting and cutting relationship edge set, and sets an intersecting and cutting relationship flag with a second value different from the first value for the relationship edges in the stratigraphic superposition relationship edge set; The knowledge graph construction module is also used to determine the starting character position number of the text mention corresponding to the target geological entity in the standardized text stream file, and to calculate the cumulative value of the byte length occupied by each character before the starting character position number according to the preset character encoding format, and to determine the obtained cumulative value as the starting byte offset.

[0011] By adopting the above technical solution, since the physical bytes occupied by a single character differ in different character encoding formats, the system uses the cumulative length of statistical bytes as the underlying physical offset, reducing the alignment error generated by cross-platform text readers when locating the underlying file stream. By setting a binary relationship flag to distinguish between interlacing cuts and stratigraphic overlays, the state traversal logic during graph querying can be simplified to some extent, thereby helping to improve the efficiency of determining node topological relationships.

[0012] Preferably, when generating the initial spatial adjacency graph, the topology parsing module extracts the endpoint coordinates of the linear primitives in the two-dimensional vector geological profile, performs node capture and merging operations on the endpoints of line segments whose distance is less than the set spatial tolerance based on the set spatial tolerance, performs boundary closure repair and extracts closed planar geological primitives according to the connection relationship of the merged line segments, establishes each closed planar geological primitive as a planar primitive node, and establishes spatial adjacency edges according to the shared boundary between planar primitives or the contact relationship formed within the spatial tolerance range, thereby generating the initial spatial adjacency graph.

[0013] By adopting the above technical solution, for geometric defects such as incomplete intersection or breakage of lines in actual geological drawings, endpoint capture and coordinate merging are performed using preset spatial tolerance, which can filter out some geometric drawing errors, assist in completing polygon topology repair and closed surface extraction, thereby improving the fault tolerance of the spatial adjacency graph construction process.

[0014] Preferably, when performing the topology comparison, the topology parsing module traverses the adjacent planar primitive node pairs in the initial spatial adjacency graph, and determines whether the direct spatial contact between the adjacent planar primitive node pairs can be explained by the interpenetration and cutting relationship corresponding to the interpenetration and cutting relationship marker, based on the relationship record of the geological entities corresponding to the adjacent planar primitive nodes in the geological entity relationship knowledge graph. When the direct spatial contact cannot be explained by the interpenetration and cutting relationship, the stratigraphic superposition relationship path between the geological entities corresponding to the adjacent planar primitive nodes is retrieved from the stratigraphic superposition relationship edge set of the geological entity relationship knowledge graph; When the stratigraphic overlay path contains at least one intermediate geological entity other than the geological entity corresponding to the adjacent planar primitive node, the at least one intermediate geological entity is identified as an unexpressed intermediate stratigraphic entity that needs topological compensation at the current direct spatial contact location.

[0015] By employing the above technical solution, a computational approach is provided for identifying geological unconformities through map path traversal retrieval. The system extracts missing intermediate stratigraphic entities by comparing the differences between standard stratigraphic stacking paths and actual two-dimensional spatial contact paths. This approach transforms the traditional qualitative assessment of missing stratigraphic elements into a more concrete map path search and matching process.

[0016] Preferably, the topology parsing module establishes corresponding topology placeholder nodes; The topological placeholder node is a virtual topological node that does not have actual planar geological elements. Its node attribute structure includes at least the geological entity identifier corresponding to the unexpressed intermediate stratum entity, the virtual node marker, the geometric area attribute with a value of zero, and the geometric anchoring information. The geometric anchoring information is determined based on the contact boundary between the corresponding adjacent planar primitive node pairs before the topological placeholder node is injected. After establishing the topological placeholder nodes, the topology parsing module deletes the original direct spatial adjacency edges between the adjacent planar primitive node pairs, and connects each topological placeholder node in series with the adjacent planar primitive node pairs according to the geological entity arrangement order in the stratigraphic superposition path, so as to generate the final spatial geometric adjacency graph.

[0017] By adopting the above technical solution, a virtual node data structure containing area attributes with a value of zero and geometric anchoring information is defined, which enables the repair of geological topological sequence maps with faults without introducing additional two-dimensional geometric area errors. This is beneficial to maintaining the integrity of the corresponding number of nodes during subsequent isomorphic matching calculations.

[0018] Preferably, the isomorphic matching module extracts the color channel values ​​of the primitive nodes in the two-dimensional vector geological profile as primitive fill color parameters, and calculates the square root of the sum of the squares of the differences between the primitive fill color parameters and the preset standard color values ​​in the color space as the color distance; When the color distance is less than or equal to a preset threshold, the standard geological entity identifier corresponding to the standard color value is added to the candidate geological entity set. The candidate geological entity set is used as a node mapping constraint to search for the largest common subgraph mapping that satisfies the edge relationship consistency constraint and completes the attribute writing.

[0019] By employing the above technical solution, and calculating the root of the square of the three color channels in Euclidean space, the error of primitive color deviation from the standard color chart can be quantitatively evaluated. Limiting the color distance to a threshold for extracting candidate feature sets reduces the exhaustive state space of the subgraph isomorphism search algorithm, thereby improving the execution efficiency of node attribute writing and matching.

[0020] Preferably, the spatial interpolation module determines the correspondence between the outlines of similar geological entities in adjacent two-dimensional geological profiles based on the geological entity identifiers carried by the spatial geometric nodes in the adjacent two-dimensional geological profiles that have completed attribute writing, extracts the source outline and target outline that establish the correspondence, performs outline lofting operation on the source outline and the target outline to establish a three-dimensional continuous boundary representation model, and performs three-dimensional spatial voxelization operation on the three-dimensional continuous boundary representation model to generate the three-dimensional voxel set; When the spatial geometric node corresponding to the source contour or target contour is the topological occupant node, the spatial interpolation module reads the geometric anchoring information of the topological occupant node and constructs a degraded source contour or degraded target contour with zero area based on the geometric anchoring information. This causes the geological entity contour on the other side, which is the same as the geological entity corresponding to the topological occupant node and has an actual area, to continuously shrink towards the degraded source contour or degraded target contour, so as to generate a pinch-out transition three-dimensional geometry in which the stratum thickness gradually decreases towards the position defined by the geometric anchoring information.

[0021] By employing the aforementioned technical solution, the boundary convergence deformation of the 3D surface loft is controlled by extracting the degraded contour, driving the physical volume of the ground surface element to undergo volume reduction calculations towards the zero-area geometric anchor position. This computational mechanism simulates the pinch-out morphology in geological structures, making the generated 3D discrete voxels more closely resemble the actual physical sedimentary deformation trends.

[0022] Preferably, the spatial interpolation module extracts the center point coordinates of the interpolation voxel and calculates the first Euclidean distance from the center point to the first adjacent two-dimensional geological profile projection reference plane and the second Euclidean distance to the second adjacent two-dimensional geological profile projection reference plane, respectively. When the first Euclidean distance or the second Euclidean distance is less than the preset minimum distance tolerance constant, the spatial interpolation weight corresponding to the side with a distance less than the minimum distance tolerance constant is determined to be 1, and the spatial interpolation weight corresponding to the other side is determined to be 0. When both the first Euclidean distance and the second Euclidean distance are greater than or equal to the minimum distance tolerance constant, the negative power exponent values ​​of the first Euclidean distance and the second Euclidean distance are calculated respectively, and the ratio of the negative power exponent value corresponding to any side to the sum of the negative power exponent values ​​of both sides is determined as the spatial interpolation weight corresponding to that side.

[0023] By adopting the above technical solution, the allocation rate of voxels affected by the spatial influence of the interface is calculated based on inverse distance weighted logic. A minimum distance tolerance constant is set to intervene in the calculation, effectively avoiding division-by-zero overflow anomalies in the underlying floating-point operation process. The proportion of negative exponent values ​​in the calculation allows the generated spatial interpolation weights to reflect the relevance of text references inversely, providing a basis for the numerical expression of source citations.

[0024] Preferably, the interaction command is a picking interaction command for the interpolation voxel. When the visualization tracing module receives the picking interaction command, it back-projects the click position in the screen coordinate system to the three-dimensional world coordinate system to generate a three-dimensional virtual picking ray, calculates the intersection state between the three-dimensional virtual picking ray and the bounding boxes of each voxel in the three-dimensional view, and determines the voxel at the intersection point closest to the virtual camera as the target interpolation voxel when there are intersecting voxels. The visualization tracing module reads the multi-source tracing index array of the target interpolation voxel and sends the document identifier, starting byte offset and text fragment length contained in the valid text tracing reference to the built-in document reader. The built-in document reader determines the corresponding standardized text stream file based on the document identifier, locates the starting position of the target text based on the starting byte offset, reads the target text segment based on the text segment length, performs forward and backward delimiter scanning with the target text segment as the center, and determines and displays the geological exploration report text paragraph corresponding to the target text segment.

[0025] By employing the above technical solution and applying raycasting and bounding box collision detection logic, screen interaction commands are converted into voxel retrieval commands in a three-dimensional virtual space. The application of physical offset combined with forward and backward delimiter character scanning technology reduces the interference of redundant spaces and line breaks within the document on the reading program, thereby enabling targeted locking of long text target paragraphs.

[0026] Preferably, the visualization tracing module, for any valid text tracing reference in the multi-source tracing index array, obtains the spatial interpolation weight corresponding to the text tracing reference, calculates the difference between a preset upper opacity threshold and a preset lower opacity threshold, adds the product of the difference and the spatial interpolation weight to the lower opacity threshold to obtain the target highlight background opacity coefficient corresponding to the text tracing reference, and provides the target highlight background opacity coefficient as the transparency channel value to the graphics renderer of the built-in document reader to perform highlight background rendering on the corresponding geological exploration report text paragraph; The visualization tracing module is also used to listen for text selection or text positioning interaction events in the built-in document reader. After obtaining the document identifier and byte position range corresponding to the text paragraph selected by the user, it queries the three-dimensional interpolation voxels containing the text tracing references corresponding to the document identifier and byte position range according to the pre-established text tracing reference reverse index, and reads the spatial interpolation weights corresponding to the selected text paragraphs in the multi-source tracing index array of each queried three-dimensional interpolation voxel. The spatial interpolation weights are used to control the highlighting degree of the corresponding three-dimensional interpolation voxels in the three-dimensional view.

[0027] By employing the aforementioned technical solution, a linear interpolation model with extrema is calculated, and the spatial interpolation weights are mapped to the executable transparency channel coefficients of the graphics rendering engine. This allows the shading depth of the front-end reading interface to reflect changes in spatial distance. Through text event listening combined with a reverse indexing mechanism, reverse data flow transmission driving the rendering of a 3D model matrix using 2D text input is achieved, constructing a relatively complete bidirectional collaborative visualization interaction process.

[0028] Compared with the prior art, this application has at least the following beneficial effects: 1. Based on further analysis and research into the problems of existing technologies, this application recognizes that data isolation exists between two-dimensional text and three-dimensional spatial models in the current display of geological exploration results, making accurate bidirectional tracing and collaborative positioning difficult. This application introduces a joint mechanism of multi-source tracing index arrays and topological placeholder nodes to calculate the physical byte offset of the underlying text and perform isomorphic matching with the spatial nodes of the two-dimensional profile. Simultaneously, spatial interpolation weights are calculated and recorded during the three-dimensional voxel generation stage, achieving the effect of establishing a reliable addressing path between the underlying text stream file and the three-dimensional visual space. This mechanism overcomes the correspondence fault problem between heterogeneous data and realizes bidirectional collaborative tracing between geological exploration report text and three-dimensional geological models.

[0029] 2. This application addresses the topological representation of missing stratigraphic regions by comparing and retrieving stratigraphic overlay paths under the constraints of a knowledge graph. Missing intermediate geological entities are identified, and zero-area topological placeholder nodes are injected at the corresponding locations. In subsequent 3D spatial interpolation calculations, the system reads the geometric anchoring information of these placeholder nodes as degenerate contours, constraining the continuous spatial contraction of adjacent geological entity contours. This processing method can repair geological topological sequences without increasing geometric errors in 2D drawings, and can generate stratigraphic pinch-out morphologies that conform to physical sedimentary laws during 3D modeling, improving the accuracy of reconstructing complex stratigraphic structures.

[0030] 3. In the collaborative processing of interpolation weights and visualization rendering, this application obtains spatial interpolation weights by calculating the proportion of negative exponents of the distances from interpolated voxels to adjacent 2D profiles, and uses a minimal distance tolerance constant to avoid division-to-zero anomalies at the underlying level. The system further maps this weight to an alpha channel coefficient, and combines text interaction events and a reverse lookup mechanism to perform bidirectional weighted highlight rendering on text paragraphs in the document reader and interpolated voxels in the 3D view. This scheme transforms abstract spatial distance relationships into intuitive visual shading depth, allowing users to judge the current 3D position or the proportion of influence from multiple data sources on the text paragraph based on color depth when operating the interface. Attached Figure Description

[0031] To more intuitively illustrate the prior art and this application, exemplary drawings are provided below. It should be understood that the specific shapes and structures shown in the drawings should not generally be regarded as limiting conditions for implementing this application; for example, based on the technical concept disclosed in this application and the exemplary drawings, those skilled in the art are able to easily make conventional adjustments or further optimizations to the addition / reduction / classification, specific shapes, positional relationships, connection methods, size ratios, etc. of certain units (components).

[0032] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram of the knowledge graph construction process of the present invention; Figure 4 This is a schematic diagram of the topology parsing process of the present invention; Figure 5 This is a schematic diagram of the isomorphic matching process of the present invention; Figure 6 This is a schematic diagram of the spatial interpolation and multi-source tracing index array generation process of the present invention; Figure 7 This is a schematic diagram of the material mapping and multi-terminal bidirectional traceability process of the present invention.

[0033] Among them, 100 is the knowledge graph construction module; 200 is the topology parsing module; 300 isomorphic matching module; 400 is the spatial interpolation module; and 500 is the visualization tracing module. Detailed Implementation

[0034] The present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] In the description of this application: unless otherwise stated, "a plurality of" means two or more. The terms "first," "second," "third," etc., in this application are intended to distinguish the objects referred to and do not have any special meaning in terms of technical connotation (e.g., they should not be construed as an emphasis on importance or order). Expressions such as "including," "comprising," and "having" also mean "not limited to" (certain units, components, materials, steps, etc.).

[0036] The terms used in this application, such as "upper," "lower," "left," "right," and "middle," are generally used to indicate the general relative positional relationship for the purpose of intuitive understanding by referring to the accompanying drawings, and are not absolute limitations on the positional relationship in the actual product.

[0037] See attached document Figure 1 This invention provides an intelligent analysis and visualization system for geological exploration results, comprising: The knowledge graph construction module 100 is used to extract stratigraphic and structural entities from geological exploration report text to establish a geological entity relationship knowledge graph. According to geological relationship rules, it assigns corresponding interpenetration and cutting relationship markers to geological contact relationships. Based on the geological exploration report text, it constructs a text semantic directed acyclic graph and extracts the document identifier and starting byte offset of each node in the text semantic directed acyclic graph in the corresponding standardized text stream file. It generates text source references and encapsulates the text source references as node attributes into the node attribute structure of the corresponding node. The topology parsing module 200 is used to perform node capture and merging operations on the input two-dimensional vector geological profile map based on the set spatial tolerance. It generates an initial spatial adjacency map by performing boundary closure repair, performs joint topological comparison between the initial spatial adjacency map and the geological entity relationship knowledge graph, injects missing strata topological placeholder nodes with zero geometric area at the strata missing positions caused by non-interpenetrating cutting, and generates a final spatial geometric adjacency map that meets the preset isomorphic prior conditions. The isomorphic matching module 300 is used to extract the primitive fill color parameters of the two-dimensional vector geological profile as candidate node mapping constraints, calculate the subgraph isomorphic node mapping relationship between the text semantic directed acyclic graph and the final spatial geometric adjacency graph, and write the geological entity attributes and text source references in the text semantic directed acyclic graph into the geometric primitive node attribute structure of the final spatial geometric adjacency graph according to the subgraph isomorphic node mapping relationship. Among them, for nodes with actual planar geometry, the extended attribute field of the corresponding geometric polygon is written, and for nodes with missing strata topology occupants, the corresponding virtual node attribute structure is written, thus completing the attribute attachment. The spatial interpolation module 400 is used to perform contour interpolation operations on adjacent two-dimensional profiles that have completed attribute attachment to generate a three-dimensional voxel set. It uses the inverse distance weighted interpolation calculation method to solve the spatial interpolation weight of the interpolated voxels in the three-dimensional voxel set relative to the adjacent two-dimensional profiles, and generates a multi-source traceability index array that integrates spatial interpolation weights within the interpolated voxel attributes. The visualization and tracing module 500 is used to parse the geometric coordinates and lithological material properties of a three-dimensional voxel set to perform data-driven material mapping rendering to generate a three-dimensional view. When the front-end interface detects a pick-up interaction command for the interpolated voxel, it reads the multi-source tracing index array corresponding to the interpolated voxel. Based on the document identifier and starting byte offset in the array, it uses the built-in document reader to locate and display the corresponding geological exploration report text paragraph. It also calculates the opacity of the text highlight background color based on the spatial interpolation weight in the multi-source tracing index array to complete the text background color rendering on the interface.

[0038] See attached document Figure 2 This invention provides a method for intelligent analysis and visualization of geological exploration results, executed by the aforementioned intelligent analysis and visualization system for geological exploration results, comprising the following steps: S100, the knowledge graph construction module 100 extracts the stratigraphic and structural entities in the geological exploration report text to establish a geological entity relationship knowledge graph. According to the geological relationship rules, it assigns corresponding interpenetration and cutting relationship markers to the geological contact relationships. Based on the geological exploration report text, it constructs a text semantic directed acyclic graph. It extracts the document identifier and starting byte offset of each node in the text semantic directed acyclic graph in the standardized text stream file, generates text source references, and encapsulates the text source references as node attributes into the node attribute structure of the corresponding node. S200, the topology parsing module 200 performs node capture and merging operations on the input two-dimensional vector geological profile map based on the set spatial tolerance, performs boundary closure repair to generate an initial spatial adjacency map, performs joint topological comparison between the initial spatial adjacency map and the geological entity relationship knowledge graph, injects missing strata topological placeholder nodes with zero geometric area at the strata missing positions caused by non-interpenetrating cutting, and generates a final spatial geometric adjacency map that meets the preset isomorphic prior conditions; S300, the isomorphic matching module 300 extracts the primitive fill color parameters of the two-dimensional vector geological profile as candidate node mapping constraints, calculates the subgraph isomorphic node mapping relationship between the text semantic directed acyclic graph and the final spatial geometric adjacency graph, and writes the geological entity attributes and text source references in the text semantic directed acyclic graph into the geometric primitive node attribute structure of the final spatial geometric adjacency graph according to the subgraph isomorphic node mapping relationship. Among them, for nodes with actual planar geometry, the extended attribute field of the corresponding geometric polygon is written, and for nodes with missing strata topology occupants, the corresponding virtual node attribute structure is written, thus completing the attribute attachment. S400, the spatial interpolation module 400 performs contour interpolation operation on adjacent two-dimensional profiles that have completed attribute attachment to generate a three-dimensional voxel set. For the interpolated voxels in the three-dimensional voxel set, the inverse distance weighted interpolation calculation method is used to solve the spatial interpolation weight of the interpolated voxels relative to the adjacent two-dimensional profiles. The starting byte offset on the adjacent two-dimensional profile nodes is read, and a multi-source traceability index array that integrates spatial interpolation weights is generated inside the interpolated voxel attributes. The S500 visualization traceability module 500 parses the geometric coordinates and lithological material properties of the three-dimensional voxel set, executes data-driven material mapping rendering to generate a three-dimensional view, and when it detects a pick interaction command for the interpolated voxel, it reads the multi-source traceability index array corresponding to the interpolated voxel, locates and displays the corresponding geological exploration report text paragraph, calculates the opacity of the text highlight background color based on the spatial interpolation weight in the multi-source traceability index array, and completes the text background color rendering on the interface.

[0039] To further clarify the implementation of each technical aspect of the present invention, the following will provide a detailed description of the implementation of each functional module involved above and its internal processing flow.

[0040] See attached document Figure 3 In step S100, the knowledge graph construction module 100 performs geological entity identification, entity standardization, and entity relationship extraction on the geological exploration report text, forming a geological entity relationship knowledge graph and a directed acyclic graph of text semantics for subsequent spatial matching, and establishing text source tracing references between text semantic nodes and the original report text locations. Specifically, this includes the following sub-steps: S110, the knowledge graph construction module 100 reads the geological exploration report text set within the entire work area, performs text parsing on the geological exploration reports of each region, and uses a named entity recognition model to identify the stratigraphic entities, igneous rock entities, and structural entities within them.

[0041] In this embodiment, the named entity recognition model includes a feature extraction layer and a sequence labeling layer. The geological text character sequence obtained through text preprocessing is input into the named entity recognition model, and each input character is converted into a fixed-dimensional vector representation. The feature extraction layer uses a bidirectional long short-term memory network to encode the forward and backward contextual features of the character sequence. The sequence labeling layer uses a conditional random field to jointly decode the output of the feature extraction layer based on the transition constraints between adjacent labels, obtaining the entity label corresponding to each character.

[0042] The named entity recognition model uses BIO sequence labeling to mark entity boundaries and entity categories. Its output includes entity text fragments identified from geological exploration report text, entity boundary locations, and corresponding geological entity categories. During model training, manually annotated historical geological exploration report corpora are used as training samples, and the negative log-likelihood function is used as the loss function. The model parameters are updated through backpropagation until the preset training termination condition is met.

[0043] The knowledge graph construction module 100 standardizes the entity text fragments obtained from named entity recognition based on the standard stratigraphic vocabulary, structural vocabulary, and entity synonym mapping table. It maps entity mentions with different names but pointing to the same geological object to the same standard geological entity and assigns a unique geological entity identifier to each standard geological entity. .

[0044] When the same standard geological entity appears repeatedly in the same geological exploration report or different geological exploration reports, standard geological entity nodes are not created repeatedly. Instead, the textual mention information corresponding to each occurrence is recorded separately, so that the same... The corresponding standard geological entity node can be associated with one or more text locations.

[0045] Based on standardized geological entities, the knowledge graph construction module 100 establishes a geological entity relationship knowledge graph according to the geological age relationships, overlying relationships, contact relationships, intrusive relationships, and tectonic cutting relationships recorded in the geological exploration report: ; In the formula, This represents the globally generated geological entity relationship knowledge graph. The set of nodes representing the standard geological entity categories within the work area. This represents the set of connecting edges between entity nodes.

[0046] The knowledge graph construction module 100 further identifies stratigraphic relation words, contact relation words, and relation words representing intrusion or tectonic cutting relationships from the geological exploration report text, determining the geological entities and their semantic relationships involved in subsequent spatial matching. The geological entities involved in spatial matching are used as a set of text semantic nodes. The stratigraphic superposition and interpenetration relationships between geological entities are used as the semantic edge set. Construct a directed acyclic graph of textual semantics: ; For stratigraphic superposition relationships, directed edges are set according to a pre-defined unified stratigraphic age direction; for magmatic intrusion relationships or tectonic cutting relationships, directed edges are set according to the direction from the intruded or cut geological entity to the geological entity that carried out the intrusion or cutting, and the relationship type identifier is written in the corresponding semantic edge. This gives the edge direction and edge type in the text semantic directed acyclic graph a unified expression rule, which is then used for attribute graph matching with the spatial geometric adjacency graph.

[0047] S120, the knowledge graph construction module 100, based on the geological relationship extraction results and preset geological relationship rules, constructs a geological entity relationship knowledge graph. The relationships in the data are classified, and interleaving and cutting relationship markers are set for the corresponding relationships.

[0048] The knowledge graph construction module 100 is configured with a geological attribute rule base, which includes at least a standard stratigraphic terminology, a structural terminology, and relational rules used to characterize stratigraphic superposition, magmatic intrusion, and tectonic cutting relationships. The knowledge graph construction module 100 uses standardized... The results of extracting geological entity categories and relationships between entities are matched with the geological attribute rule base to determine the types of relationships between entities.

[0049] When the relationship between two geological entities is identified as magmatic intrusion, fault cutting, or other relationships that cause existing geological bodies to intersect or cut, the corresponding relationship edges are assigned to the set of intersecting and cutting relationship edges. .

[0050] Specifically, for igneous rock entities with planar spatial representations, these entities are retained as directed acyclic graphs for textual semantics. The semantic nodes in the graph enable them to participate in subsequent node matching between the text semantic graph and the spatial geometry graph.

[0051] For fault structures represented by line primitives in a two-dimensional vector geological profile, in a geological entity relationship knowledge graph The corresponding fault structure entities and their relationship records are retained in the text semantic directed acyclic graph used for subsequent isomorphic matching. In this paper, the cutting relationship between fault structures and related planar geological entities is represented as the interlacing cutting type edge between corresponding geological entity nodes, without requiring a one-to-one node mapping between linear fault structure entities and planar geometric nodes.

[0052] When the relationship between entities is determined to be a normal stratigraphic superposition relationship based on stratigraphic relationship, geological age relationship, or overlying relationship, and no corresponding interpenetrating or cutting relationship is identified, the relationship edge is assigned to the stratigraphic superposition relationship edge set. .

[0053] The knowledge graph construction module 100 sets Boolean-type interleaving and cutting relationship flags for the above-mentioned relationship edges. Among them, for the set of interlacing and cutting relationship edges. Set the relationship edge in the middle. For stratigraphic superposition boundary sets Set the relationship edge in the middle. Accordingly: ; In the formula, This represents the total set of connection edges in the knowledge graph. This represents the set of extracted stratigraphic overlap edges. This represents the set of extracted interlacing and cutting edges. This represents the set union operator.

[0054] Interlacing and cutting relationship markers Stored as a relation edge attribute in the geological entity relation knowledge graph. In this context, it is used to distinguish between contact relationships formed by normal stratigraphic stacking and those formed by magma intrusion, fault cutting, or other interpenetrating cutting processes during subsequent topological comparison.

[0055] Knowledge graph construction module 100 is a directed acyclic graph of text semantics. When writing semantic edges, directed cycle detection is performed on the semantic edges to be written.

[0056] If we add the semantic edge to be written, the text semantic directed acyclic graph If a directed path originating from any node and returning to that node along a directed edge is formed, then the semantic edge to be written is considered to have created a directed cycle. The corresponding relationship is recorded as a text relationship conflict record, and the edge is not added to the set of valid semantic edges participating in subsequent graph matching. If no directed cycle is formed, then write the semantic edge into... Through the above processing, it can be used for subsequent isomorphic matching. Preserve the structure of the directed acyclic graph.

[0057] The knowledge graph construction module 100 also targets directed acyclic graphs for text semantics. Geological entity nodes in the text generate text tracing references.

[0058] Specifically, geological exploration reports in PDF, DOC, DOCX, or other formats are parsed into standardized text stream files encoded in UTF-8, and each standardized text stream file is assigned a document identifier associated with the corresponding geological exploration report. Text semantic analysis, entity location recording, and subsequent byte offset calculation are all based on the same version of the standardized text stream file.

[0059] For text semantic directed acyclic graph For any geological entity node in the document, the knowledge graph construction module 100 determines the starting character position of the entity mention corresponding to that node in the standardized text stream file. The total number of bytes occupied by all characters preceding the starting character is counted according to the UTF-8 encoding rules to obtain the starting byte offset of the entity reference in the standardized text stream file. .

[0060] Start byte offset Calculate according to the following formula: ; in, Indicates the starting byte offset. This indicates the character position number of the target entity in the parsed text stream. Indicates the first in the text stream One character, Indicates the first The physical length of bytes occupied by each character under the preset character encoding format.

[0061] because The addressing object is a standardized text stream file, therefore This does not represent the physical byte location within the original PDF, DOC, or DOCX container file. Subsequent text tracing will be performed based on... Determine the appropriate standardized text stream file, and based on... Locate the starting position of the corresponding text to distinguish the different occurrences of the same geological entity name in the same or different reports.

[0062] The knowledge graph construction module 100 further obtains the length of the text fragments corresponding to the target entity mentions in the standardized text stream file. and will , and Encapsulated as a text source citation : ; In the formula, Indicates from The indicated position begins with the length of bytes used to determine the range of the corresponding target text segment to be read. Used to determine the standardized text stream file in which the text resides. Used to determine the starting position of the target text segment. Used to determine the reading range of the target text segment.

[0063] The knowledge graph construction module 100 will generate Write text semantics to a directed acyclic graph The node attribute structure corresponding to the geological entity node is used to realize the association between the geological entity node and the specific text location in the standardized text stream file.

[0064] When the same standard geological entity has multiple valid textual references in one or more geological exploration reports, corresponding references are generated for each textual reference. and multiple The source citations are stored in the node attribute structure of the corresponding geological entity node in the form of a text source citation list. Therefore, in subsequent two-dimensional spatial attribute attachment and three-dimensional voxel source tracing processes, the specific... Locate the corresponding geological exploration report and the location of the target text within it.

[0065] See attached document Figure 4 In step S200, the topology parsing module 200 performs geometric topology repair on the two-dimensional vector geological profile and combines it with the geological entity relationship knowledge graph obtained in step S100. The consistency of the repaired spatial adjacency relationships is checked; for spatial direct contact positions that cannot be explained by interpenetration and cutting relationships, but should have intermediate stratigraphic entities according to the stratigraphic stacking relationship, missing stratigraphic topological placeholder nodes with zero geometric area are injected to form the final spatial geometric adjacency graph for subsequent attribute graph matching. Specifically, it includes the following sub-steps: S210, the topology parsing module 200 reads the two-dimensional vector geological profile file and parses the linear primitives, area primitives, layer identifiers, area primitive fill color parameters, and text annotations associated with or adjacent to the area primitives.

[0066] For linear primitives in a 2D vector geological profile, the endpoint coordinates of each line segment are extracted, and the extracted endpoints form an initial set of geometric nodes. Layer identifiers, text annotations, and area primitive fill color parameters, which can be used to identify geological entity categories, are matched with a standard stratigraphic thesaurus, a structural thesaurus, and a color mapping dictionary used in geological maps, respectively, to establish a candidate set of geological entities for the subsequently formed area primitive nodes. .

[0067] When a layer identifier or text annotation can uniquely identify the corresponding standard geological entity, the corresponding geological entity identifier will be... Write When existing attributes can only identify multiple possible geological entities, in Multiple corresponding entries are retained in the middle. Furthermore, spatial adjacency relationships are used to further narrow down the candidate range in subsequent joint topology comparison and isomorphic matching processes.

[0068] For gaps at the endpoints of line segments in two-dimensional vector geological profiles caused by digitization or mapping errors, the topology parsing module 200 determines the gaps according to the spatial tolerance threshold. Perform geometry node snapping and merging.

[0069] Let the endpoints of the two line segments in the initial set of geometric nodes be respectively and The Euclidean distance between the two in the two-dimensional profile coordinate system Calculate according to the following formula: ; In the formula, This represents the calculated two-dimensional Euclidean distance between two nodes. This represents the first node to be calculated in the set of line segment endpoints. This represents the second node to be calculated in the set of line segment endpoints. Represents a node The x-coordinate in a two-dimensional projected coordinate system Represents a node The ordinate in a two-dimensional projected coordinate system Represents a node The x-coordinate in a two-dimensional projected coordinate system Represents a node The ordinate in a two-dimensional projected coordinate system.

[0070] when At that time, the endpoint and Endpoints are identified as those within the same spatial tolerance range and are captured by the same representative node. For cases with only two endpoints to be merged, the arithmetic mean of the coordinates of the two endpoints can be used as the coordinates of the merged representative node. For cases with multiple interconnected endpoints within the same tolerance range, the endpoints can be grouped into the same node group to be merged, and the average of the coordinates of each endpoint within that node group can be used to determine the coordinates of the merged representative node.

[0071] Spatial tolerance threshold The spatial tolerance threshold is determined based on the scale of the two-dimensional vector geological profile and the allowable digitization error. In one embodiment, the spatial tolerance threshold is... Calculate according to the following formula: ; In the formula, The denominator is the scale of the drawing. This is a constant for the error limit of digital drafting, typically ranging from 0.1 to 0.3 mm. Those skilled in the art can make specific adjustments based on the accuracy requirements of the drawings.

[0072] When the coordinate units in a two-dimensional vector geological profile are... When the calculated actual length units are inconsistent, The length unit used after conversion to a two-dimensional profile coordinate system is used as the spatial tolerance threshold. .

[0073] After completing the geometric node capture and merging, the topology parsing module 200 performs boundary closure repair based on the merged line segment connection relationships and extracts planar geological primitives from the closed boundaries. Each closed planar geological primitive is established as a planar primitive node.

[0074] For any two planar primitives, when they share a common boundary, or when their boundaries are within a spatial tolerance threshold... When a proximity relationship is formed within the area that can be identified as the same geological contact edge after closure and repair, a spatial adjacent edge is established between the corresponding planar primitive nodes, and the corresponding contact boundary information is saved.

[0075] For fault lines in a two-dimensional vector geological profile, if the fault line intersects with the contact edge between two adjacent planar elements, or if the distance between the fault line and the contact edge is within the spatial tolerance threshold... If a spatially adjacent edge forms a continuous proximity relationship within the specified range, it is pre-marked as a structural cutting candidate edge. This structural cutting candidate edge marking is only used to indicate that the spatial contact may be affected by fault structures; the actual relationship type is determined in step S220 by combining the geological entity relationship knowledge graph. Further confirmation.

[0076] The topology parsing module 200 establishes an initial spatial adjacency graph based on the extracted planar primitive nodes and spatial adjacency edges: ; in, This represents the set of planar primitive nodes obtained by boundary closure repair in a two-dimensional vector geological profile. This represents the set of spatially adjacent edges between polygon primitive nodes. Each polygon primitive node is associated with at least one corresponding geometric polygon. In addition to primitive fill color parameters; each spatial adjacent edge records at least the nodes it connects to and the corresponding contact boundary information, and may include construction cut candidate edge markers.

[0077] S220, the topology resolution module 200 generates the initial spatial adjacency graph in step S210. Knowledge graph relating to geological entities generated in step S100 Joint topological comparison is performed to determine whether the direct spatial contact in the initial spatial adjacency map can be explained by the stratigraphic superposition or interpenetration relationship between the corresponding geological entities.

[0078] The topology parsing module traverses the initial spatial adjacency graph 200 times. Adjacent face primitive node pairs Read them separately and And based on the candidate set Knowledge Graph of Geological Entity Relationships The corresponding standard geological entity nodes and relationship records can be queried in the database.

[0079] when and Each contains only one At that time, the local topology check of the current node pair is performed using the corresponding combination of geological entities.

[0080] when or Contains multiple At that time, candidate geological entity combinations are formed respectively, and it is tested whether each candidate combination can be formed. The current spatial adjacency relationship is interpreted based on the interpenetration or stratigraphic stacking relationship. Candidate combinations without any valid relationship paths are eliminated; candidate combinations that can form a valid topological interpretation are retained and updated accordingly. and .

[0081] If the geological entity combination corresponding to the current node can be uniquely determined after the above verification, the relationship type determination will continue to be performed according to the combination; if there are still multiple candidate combinations with different relationship interpretations, the current spatial adjacent edge will be retained and temporarily recorded as an edge with unknown relationship, and the injection of missing stratum topology placeholder nodes will not be triggered in the current local comparison.

[0082] If node or node No valid ,or and None of the candidate combinations formed can be in If a valid relation interpretation is obtained, the node is retained. With nodes The existing spatial adjacency relationships between them are determined, and the corresponding spatial adjacency edges are recorded as edges with unknown relationships.

[0083] For adjacent node pairs that can determine the corresponding geological entity combination The topology analysis module 200 first determines whether the direct spatial contact between the two can be explained by the interpenetration and cutting relationship.

[0084] Specifically, in the knowledge graph of geological entity relationships interlacing and cutting relationship edge set Query node and nodes Record the relationships between corresponding geological entities and read whether the current spatial adjacent edges have structural cutting candidate edge markers.

[0085] when When there is an interpenetration and cutting relationship between corresponding geological entities, the current direct spatial contact is determined to be interpretable by the interpenetration and cutting relationship, the spatial adjacency relationship between node u and node v is preserved, and the corresponding spatial adjacency edge is written into the interpenetration and cutting type identifier.

[0086] when If there is a fault cutting relationship record corresponding to the current node, and the current spatial adjacent edge also has a structural cutting candidate edge mark generated by the fault line in the two-dimensional vector geological profile, the corresponding spatial adjacent edge will also be determined as an interlacing cutting type edge.

[0087] Only construction cut candidate edge markers exist, but cannot be obtained from... If the corresponding construction relationship record is obtained, the edge is not directly identified as an interlacing cutting type edge, but its candidate state is retained or recorded as an edge with unknown relationship.

[0088] When node With nodes When the direct spatial contact between them cannot be explained by the interpenetration and cutting relationship, the topology analysis module 200 uses the stratigraphic stacking edge set. The stratigraphic superposition relationship path between the corresponding geological entities is retrieved.

[0089] Set nodes and nodes The currently identified corresponding standard geological entities are as follows: and Following the unified stratigraphic age direction specified in step S100, in Search and A directed connected path between them. If and They are connected only by a single edge representing a direct stacking relationship, meaning the path does not contain any edges other than those representing stacking relationships. and For any intermediate geological entities other than those mentioned above, it is determined that the two are directly adjacent strata in the standard stratigraphic sequence, and the nodes are preserved. With nodes The spatial adjacency relationship between them is determined, and the corresponding spatial adjacency edges are written into the stratigraphic stacking type identifier.

[0090] If the retrieved stratigraphic overlay relationship path is: ; In the formula, Represents a node The standard geological entity mapped Represents a node The standard geological entity mapped Indicates that it is located in the knowledge graph and One or more intermediate standard geological entities between them This represents the number of intermediate geological entities and is a positive integer. If the above path exists, then the initial spatial adjacency graph is determined. Nodes in With nodes There are direct spatial contacts that are inconsistent with the standard stratigraphic sequence, and intermediate stratigraphic entities are assembled. This is identified as the set of intermediate strata entities not represented at this contact location in the current two-dimensional vector geological profile.

[0091] The unrepresented state of intermediate stratigraphic entities in a 2D profile can correspond to stratigraphic pinch-outs, sedimentary hiatuses, erosion defects, or omissions in 2D mapping. This step determines the intermediate stratigraphic entities requiring topological compensation based on the stratigraphic stacking relationship path, without determining their specific geological origin solely based on these topological differences.

[0092] S230, for the set of intermediate strata entities determined in step S220 The topology resolution module 200 initializes the spatial adjacency graph. In order to compensate for the differences in node and edge structure between the text semantic relationship and the two-dimensional spatial adjacency relationship, missing stratum topology placeholder nodes are injected at the corresponding spatial direct contact positions.

[0093] Corresponding to each intermediate stratigraphic entity Create a placeholder node for the missing stratum topology in the graph data structure. Missing stratigraphic topological placeholder nodes are virtual topological nodes that do not possess actual areal geological features. Their node attributes must include at least the following: Corresponding geological entity identifiers Virtual node marker Geometric area attributes and geometric anchoring information .

[0094] For missing stratigraphic topology placeholder nodes ,set up: ; ; in, This represents the geometric area corresponding to the missing topological occupant node. For ordinary planar primitive nodes formed by actual closed polygons... Set to 0.

[0095] Geometric anchoring information Based on the node before injection With nodes The contact boundary between them is determined. In one embodiment, the geometric midpoint of the contact boundary is used as a point-like geometric anchor point; in another embodiment, the contact boundary itself is preserved as a degenerate linear profile. Used to determine the termination position of the contour with zero area in the subsequent three-dimensional contour interpolation process, so that the corresponding stratigraphic contour with actual area can shrink to the position of the missing stratigraphic topological placeholder node.

[0096] After establishing the missing stratum topology placeholder nodes, the topology parsing module 200 deletes the nodes. With nodes The existing direct spatial adjacency edges between them, and according to the geological entity relationship knowledge graph. The corresponding stratigraphic sequence will be related to The corresponding missing stratum topology placeholder node is connected in series with the original node. With nodes between.

[0097] When the standard stratigraphic superposition relationship is: ; Then the original spatial adjacency relationship will be: Updated to: ; in, Represents polygon primitive nodes The corresponding first-stratum entity node in the knowledge graph. Represents polygon primitive nodes The corresponding second-stratum entity node in the knowledge graph. This indicates that in the spatial topology, it corresponds to The first instantiated placeholder node for missing stratum topology. This indicates that in the spatial topology, it corresponds to The second missing stratum topology placeholder node instantiated.

[0098] The injection operation only alters the node and adjacent edge structure used for topological representation, without generating planar geometry with non-zero area for missing stratigraphic topological occupant nodes. Thus, while preserving the actual planar geometry of the original two-dimensional vector geological profile from artificial expansion, it retains the intermediate node structure required by the standard stratigraphic sequence in the spatial adjacency map.

[0099] After completing the node injection and spatial adjacency reconstruction for all locations to be compensated, the topology resolution module 200 generates the final spatial geometric adjacency graph: ; in, Including the original spatial adjacency graph The planar primitive nodes and newly injected missing strata topological placeholder nodes in the graph; This includes spatial adjacency edges that are retained or re-established after topology comparison and node injection.

[0100] For those who can base The relation records in the graph determine the spatial adjacency edges of semantic relations, according to the directed acyclic graph of the text semantics. The consistent direction rule sets the edge direction, and writes the stratigraphic stacking type identifier or the interlacing and cutting type identifier according to the relationship type.

[0101] For spatial adjacency edges whose semantic relationship or direction cannot be determined after step S220, the unknown relationship identifier is retained, and the stratigraphic stacking type or interpenetration and cutting type is not forcibly assigned; the unknown relationship edges are only used to constrain the spatial adjacency relationship between the corresponding nodes in the graph matching process of the subsequent step S300, and are not used as a mandatory judgment condition for edge type consistency and edge direction consistency.

[0102] Thus, the final spatial geometric adjacency graph Directed acyclic graph with text semantics The stratigraphic superposition and interpenetration relationships involved in subsequent matching are expressed using consistent edge direction and edge type, and the intermediate node structure that can be determined by the standard stratigraphic sequence is compensated by missing stratigraphic topological occupant nodes.

[0103] See attached document Figure 5 In step S300, the isomorphic matching module 300 further limits the candidate geological entity range of the planar primitive nodes based on the primitive fill color parameters in the two-dimensional vector geological profile map, and uses a text semantic directed acyclic graph. As a schematic diagram, it represents the final spatial geometric adjacency graph. As the target graph, node mapping relationships are calculated under node attribute constraints and edge relationship constraints. Based on the obtained node mapping relationships, the geological entity attributes and text source references carried by the text semantic nodes are attached to the corresponding actual areal primitive nodes or missing stratigraphic topological placeholder nodes. Specifically, this includes the following sub-steps: S310, the isomorphic matching module 300 reads the final spatial geometric adjacency graph: ; And traverse the node set This is to establish node attribute constraints for graph matching.

[0104] For virtual node flags ordinary planar primitive nodes The isomorphic matching module 300 reads the fill color of the geometric polygon associated with the node and obtains the RGB three-channel values. , and And represent it as the primitive fill color parameter: ; for For missing strata topology placeholder nodes, since these nodes do not correspond to areal primitives with actual areas, fill color extraction is not performed. Instead, the fill color is used as written in step S230 when creating the missing strata topology placeholder node. As a deterministic node attribute constraint.

[0105] The isomorphic matching module 300 is equipped with a color mapping dictionary for geological maps, which records standard color values ​​and one or more standard geological entity identifiers. The correspondence between them. For ordinary planar primitive nodes. Fill its primitives with color parameters Compare with the standard color values ​​in the color mapping dictionary for geological maps. Compare them.

[0106] Set standard color value Then the node primitive fill color parameters Compared with standard color values RGB color distance between Calculate according to the following formula: ; In the formula, This represents the Euclidean color distance between the calculated node-extracted color and the dictionary standard color. This represents the ordinary polygon primitive node that is currently to be compared in color. Represents a node Fill color parameters of primitives extracted from 2D vector graphics paper. The color mapping dictionary used to represent geological maps contains known standard color values. Represents a node The primitive fill color parameter is the red channel component value in the RGB color space. Represents a node The green channel component value of the primitive fill color parameter in the RGB color space. Represents a node The primitive fill color parameter is the blue channel component value in the RGB color space. This represents the red channel component value of the standard color values ​​in the color mapping dictionary used for geological maps, expressed in the RGB color space. This represents the green channel component value of the standard color values ​​in the color mapping dictionary used for geological maps, expressed in the RGB color space. This represents the blue channel component value of the standard color value in the RGB color space from the color mapping dictionary used for geological maps.

[0107] Let the RGB color distance threshold be When the calculated color distance satisfies At that time, the isomorphic matching module 300 will use the standard color value Corresponding standard geological entity identifier Add node Color candidate geological entity set middle.

[0108] in, This represents the maximum allowed RGB color distance threshold for determining a successful color match. This refers to a standard geological entity identifier used in a knowledge graph to uniquely represent attributes of a specific geological category. This indicates that the node is assigned after color difference tolerance filtering. A set of labels containing candidate geological attributes. RGB color distance threshold. The color fidelity of the original two-dimensional vector geological profile drawing or the allowable color difference range caused during the digital format conversion process is preset according to the original drawing.

[0109] For step S210, which has already been established Ordinary planar primitive nodes, isomorphic matching module 300 according to and The value of is used to update the set of candidate geological entities for graph matching.

[0110] when and When neither is empty, and their intersection is not empty, then: ; As an updated set of candidate geological entities.

[0111] when empty, and When not empty, As .

[0112] when Not empty When empty, retain the original As This prevents color matching failures from directly excluding candidate geological entities determined by layer identifiers, text annotations, or other attributes.

[0113] when and If neither node is empty but their intersection is empty, record that the node has a color attribute conflict, and retain the value generated in step S210. As The node disambiguation will continue based on subsequent topological constraints.

[0114] when and When all elements are empty, the text semantic directed acyclic graph is transformed. In the current two-dimensional profile matching The set serves as a provisional candidate range for the node, allowing the node to be constrained primarily by topological relationships in subsequent graph matching, rather than being pre-excluded due to a lack of color, layer, or text annotation information.

[0115] for Missing stratigraphic topology placeholder nodes ,set up: ; The isomorphic matching module 300 will As a node Node mapping constraints in step S320.

[0116] S320, the isomorphic matching module 300 converts the text semantics into a directed acyclic graph: As a schematic diagram, the final spatial geometric adjacency graph will be: As the target graph, a subgraph isomorphic search is performed under constraints on node attributes, node adjacency relationships, edge directions, and edge types.

[0117] Among them, the text semantic directed acyclic graph Nodes in At least carry the corresponding And geological entity categories; final spatial geometric adjacency graph Ordinary planar primitive nodes in Carrying the determination in step S310 The missing stratum topology placeholder node carries a definite .

[0118] During the graph matching state expansion process, when attempting to establish text semantic nodes With spatial geometric nodes When mapping between nodes, a node feasibility check is performed first.

[0119] For ordinary planar primitive nodes When text semantic nodes of belong When, the node attribute constraints are satisfied; when Not belonging to At that time, the judgment and If the node attribute constraints are not met, no mapping will be established for the current node.

[0120] For missing stratigraphic topology placeholder nodes ,when If the node attribute constraints are satisfied, then no node is established; otherwise, no node is established. and Node mapping between them.

[0121] If the node attribute constraints are satisfied, a feasibility check of the edge relationship is further performed. This is done for any text semantic node with an established mapping relationship. and its corresponding spatial geometric nodes If the text semantics are a directed acyclic graph There are connections in With the current node The semantic edges, then the final spatial geometric adjacency graph The corresponding spatial geometric nodes and There should be spatially adjacent edges between them.

[0122] when and When spatially adjacent edges have a defined relationship type and direction, the relationship type and direction of the spatially adjacent edges should be consistent with... and The semantic edges between them have the same relationship type and direction.

[0123] The relationship types include at least stratigraphic superposition type and interpenetration and cutting type.

[0124] when and When the spatial adjacent edges between them are unknown relation edges determined in step S220, only the following checks are performed. and Whether there is an adjacency relationship between them is not taken as a mandatory condition for the current mapping to be valid, and the type and direction of the unknown edge are not used as mandatory conditions for the current mapping to be valid.

[0125] When both the node attribute constraints and the corresponding edge relationship constraints are satisfied, and Add to the current candidate mapping state; if any mandatory constraint is not satisfied, terminate the current candidate pairing and continue to examine other node combinations.

[0126] Therefore, in the graph matching process, with The candidate range of nodes is restricted by spatial adjacency relationships, and the consistency of relationships is restricted by the determined stratigraphic stacking type, interpenetration and cutting type, and corresponding edge direction.

[0127] In one embodiment, the isomorphic matching module 300 performs the above-mentioned subgraph isomorphic search using the VF2 state space search method. During the state space expansion process, candidate node pairs are screened according to the aforementioned node feasibility conditions and edge relationship feasibility conditions; for candidate node pairs that do not meet the conditions, the corresponding search branch is not expanded further.

[0128] When there exists a set of node mappings that can cover the text semantics of a directed acyclic graph If all text semantic nodes in the graph are selected, and the node feasibility condition and edge relationship feasibility condition mentioned above are satisfied, then the set of mappings is determined as a complete subgraph isomorphic mapping.

[0129] For text semantic directed acyclic graph Nodes in and the final spatial geometric adjacency graph Nodes in Using a binary mapping matrix Record the mapping relationship of the selected nodes, where: ; Representing text semantic nodes With spatial geometric nodes Establish mapping relationships; ; Representing text semantic nodes With spatial geometric nodes No mapping relationship has been established.

[0130] For a complete subgraph isomorphism mapping, the pattern graph Each semantic node in the text that participates in the matching corresponds to A geometric node in the primary mapping space. Target graph. Other spatial geometric nodes that do not have corresponding text semantic nodes do not need to be covered by the complete subgraph isomorphism mapping.

[0131] When the same standard geological entity corresponds to multiple separate planar geometric instances in the same two-dimensional vector geological profile due to folding, faulting, or other geological structural effects, it is not required that a text semantic node be in... Simultaneously establish a master mapping with multiple spatial geometric nodes.

[0132] In text semantic nodes With the geometric nodes of the main mapping space After establishing a valid mapping, the isomorphic matching module 300 performs... Search for other common planar primitive nodes in the middle .like belong ,and The stratigraphic relationship or interpenetration relationship relative to its adjacent nodes and If the corresponding local semantic relations are compatible, then... Record to text semantic node The corresponding set of equivalent geometric instances.

[0133] The equivalent geometric instance set is used for attribute synchronization attachment in step S330, without changing... The term is used to represent the node correspondence in the subgraph isomorphic master mapping.

[0134] If no coverage is obtained after completing the search according to the above constraints If all text semantic nodes are isomorphically mapped to a complete subgraph, then the isomorphic matching module 300 obtains the text semantic directed acyclic graph under the conditions of feasibility of identical nodes and edge relationships. Adjacency graph of final spatial geometry The maximum common subgraph mapping between them is used to obtain the largest set of nodes that can satisfy the constraints.

[0135] When there is only one candidate result of the maximum common subgraph with the maximum number of mapped nodes, that candidate result is determined as the final node mapping relationship.

[0136] When multiple candidate results for the maximum common subgraph with the same number of mapping nodes exist, the candidate results are compared in the following order: the candidate result with more matching edges of the determined relation type is selected first; if the aforementioned numbers are the same, the candidate result with a higher degree of consistency of node candidate constraints is selected; if they still cannot be distinguished, the candidate result with a smaller cumulative RGB color distance of the corresponding ordinary planar primitives is selected; if they are still the same, the candidate result with fewer local stratigraphic sequence conflicts is selected.

[0137] The mapping of the largest common subgraph node selected according to the above rules is also recorded as follows: For those not included Text semantic nodes or spatial geometric nodes in the text are not required to establish a corresponding relationship in the current step.

[0138] S330, the isomorphic matching module 300 determines the node mapping relationship based on step S320. Directed acyclic graph of text semantics The geological entity attributes and text source references of the corresponding text semantic nodes are written into the final spatial geometric adjacency graph. The corresponding spatial geometric nodes in the diagram.

[0139] for any matrix element ,when At that time, read the text semantic nodes. The node attributes read must include at least the geological entity identifier. Geological entity categories and one or more textual source citations .

[0140] When text semantic nodes When lithological attributes are also recorded, the corresponding lithological attributes are read as well. The lithological attributes are extracted by the knowledge graph construction module 100 from the text description corresponding to the geological entity; in one embodiment, the corresponding lithological attribute can be determined based on the lithological descriptive words in the sentence where the entity is mentioned or within a preset context. For geological entities not recorded in the original geological exploration report or for which clear lithological attributes could not be extracted, the lithological attribute field is saved as a null value, and no specific lithology is artificially generated based on this.

[0141] isomorphic matching module 300 according to Determine the corresponding spatial geometric nodes and according to The node type performs attribute writing.

[0142] for Ordinary polygon primitive nodes, text semantic nodes of Geological entity categories, lithological properties, and textual source citations are written into the text. Extended attribute fields of the associated geometric polygon.

[0143] When text semantic nodes When multiple text references are involved, the multiple references generated in step S100 will be... Instead of merging multiple original text locations of the same geological entity into a single text location, the extended attribute field is written as a list of text source references.

[0144] for The missing strata topology placeholder nodes will be used to define the text semantic nodes. of The geological entity category, lithological properties, and textual source references are written into the node attribute structure of the corresponding virtual node. Since the missing stratigraphic topology placeholder node does not correspond to a geometric polygon with actual area, it is not required to create a planar geometric extension attribute field for it.

[0145] For text semantic nodes The corresponding set of equivalent geometric instances The isomorphic matching module 300 will determine the already determined Geological entity categories, lithological properties, and textual source citations are written synchronously. The nodes of ordinary planar primitives in the graph allow multiple independent planar geometric instances that belong to the same standard geological entity but are formed due to spatial segmentation to share the same geological semantic source.

[0146] When step S320 uses the maximum common subgraph mapping as... At that time, only for For spatial geometric nodes with established node mapping relationships and their equivalent geometric instances confirmed by local relation constraints, the above attribute attachment is performed; for spatial geometric nodes without established valid mappings, their original geometric attributes are retained. Do not force the writing of unconfirmed text semantic attributes.

[0147] After completing the attribute attachment, the final spatial geometric adjacency graph is obtained. Both matched ordinary planar primitive nodes and missing strata topological placeholder nodes can be identified through their node attributes. Geological entity categories and textual source citations; among them, ordinary planar primitive nodes retain their actual two-dimensional geometric data, missing stratigraphic topological occupant nodes are retained, along with zero-area attributes and geometric anchoring information. This is so that subsequent steps S400 can perform contour correspondence and spatial interpolation between adjacent two-dimensional geological profiles.

[0148] See attached document Figure 6 In step S400, the spatial interpolation module 400 performs contour lofting and voxelization operations on adjacent two-dimensional geological profiles with completed attribute attachments to generate a three-dimensional voxel set. It calculates the spatial interpolation weight of each voxel relative to the two side profiles and encapsulates the weights with the starting byte offset inherited from the two side profiles to generate a multi-source tracing index array embedded within each voxel. The specific processing flow of the spatial interpolation module 400 includes the following sub-steps: S410, the spatial interpolation module 400 reads two adjacent two-dimensional geological profile maps with spatial positioning parameters. Each two-dimensional geological profile map is associated with at least the coordinates of the profile start point, the coordinates of the profile end point, and the elevation datum information. The spatial interpolation module 400 establishes a coordinate transformation relationship from the local coordinate system of the two-dimensional profile to a unified three-dimensional geographic coordinate system based on the spatial positioning parameters. It transforms the vertices of the primitives within the two-dimensional profile to the unified three-dimensional geographic coordinate system, obtaining adjacent two-dimensional geological profiles with definite spatial locations. It extracts the geometric polygons in the drawings that have completed attribute attachment as the source contour and the target contour.

[0149] The spatial interpolation module 400 first uses the geological entity identifier attached in step S330. Determine the set of outlines of similar geological entities in the source and target profiles; when the same When multiple geometric profiles correspond to any given profile, the spatial distance between the geometric centroids of the profiles, their arrangement along the profile direction, and the adjacent strata are considered. The consistency of the sequence establishes a contour correspondence. Among the candidate contours that meet the sequence consistency condition, the contour with the smaller geometric centroid spatial distance is selected as the corresponding contour. Subsequently, the spatial interpolation module 400 performs a geometric contour lofting operation on the source contour and the target contour that have established a correspondence, and constructs a three-dimensional continuous boundary representation model connecting the two side profiles.

[0150] In actual geological modeling, similar geological entities on two adjacent profiles may exhibit topological heterogeneity, such as branching or merging (e.g., one side is a single polygon, while the other side branches into multiple polygons). When the number of source contours and target contours is inconsistent, the spatial interpolation module 400 calculates the geometric centroid based on the area weighting of primitives, uses the area weighting centroid of the corresponding contour set as an auxiliary skeleton node, and establishes skeleton connection relationships between the auxiliary skeleton nodes and each branch contour. Subsequently, the skeleton connection relationship is used as a constraint for contour vertex pairing and triangulation connection, decomposing the one-to-many or many-to-one contour correspondence into multiple local lofting units from the real contours to the auxiliary skeleton nodes, thereby constructing a continuous transition surface.

[0151] When a node on one side is a missing stratum topology occupant node generated in step S230, the spatial interpolation module 400 reads its geometric anchoring information. and will As a degraded target profile or degraded source profile with zero area, the geological profile with actual area on the other side gradually shrinks towards the degraded profile, thus forming a three-dimensional transition geometry in which the thickness of the strata continuously decreases towards the pinch-out position.

[0152] To accommodate subsequent discretization rendering and physical attribute querying requirements, the spatial interpolation module 400 performs 3D spatial voxelization operations on the 3D continuous boundary representation model, discretizing it into a regularly arranged set of tiny cubic meshes to generate a continuous set of 3D geological entity voxels. Compared to traditional continuous surface meshes, the discrete voxel structure can provide independent voxel attribute storage units for each tiny region within 3D space. These attribute storage units can be set in system memory, video memory buffers, 3D textures, or other data structures that can establish an index correspondence with voxels, thus providing the underlying data carrier for subsequent multi-attribute fusion refined to local spatial locations.

[0153] For the specific implementation of 3D geometric contour lofting and voxelization spatial operations, those skilled in the art can call the Poisson surface reconstruction and spatial mesh generation interface in the standard 3D graphics computing engine. This is a well-known technology in the field and will not be elaborated here.

[0154] S420, for each independent interpolation voxel in the three-dimensional geological entity voxel set, the spatial interpolation module 400 calculates its spatial interpolation weight. The spatial interpolation module 400 extracts the coordinates of the geometric center point of the current interpolation voxel and calculates the spatial Euclidean distance from this center point to the two adjacent two-dimensional profile projection reference planes. The current interpolation voxel is set as... The Euclidean distance from its center point to the first two-dimensional profile plane is... The Euclidean distance to the second two-dimensional cross-sectional plane is The spatial interpolation module 400 uses the inverse distance weighted interpolation method to calculate the spatial interpolation weight value of the voxel affected by the known profile data on both sides.

[0155] In actual calculations, if a voxel happens to lie on a certain cross-sectional plane, the distance parameter will approach zero, causing the denominator in the inverse distance weighting formula to become zero, thus triggering algorithm crashes. To avoid this logical dead zone, the spatial interpolation module 400 presets a very small distance tolerance constant. Specifically, this minimum distance tolerance constant The value of is usually determined by the mesh resolution during voxelization.

[0156] In an optional embodiment, The value can be taken as 0.1% to 1% of the side length of a single cube mesh. Those skilled in the art can set it conventionally based on the spatial span and computational accuracy requirements of the actual 3D model. At that time, the spatial interpolation module 400 directly determines the spatial interpolation weights on the first side. Second-side spatial interpolation weights ;when At that time, the judgment , In the normal non-adhesive state, i.e. and At that time, the spatial interpolation module 400 calculates the spatial interpolation weights according to the following formula: ; ; In the formula, Indicates the current interpolation voxel The spatial interpolation weights corresponding to the first two-dimensional profile are Indicates the current interpolation voxel The spatial interpolation weights corresponding to the second-side two-dimensional profile, This represents the spatial Euclidean distance from the center point of the voxel to the first side profile. This represents the spatial Euclidean distance from the center point of the voxel to the second side profile. This represents the power exponent parameter of the inverse distance weight. The value of is usually set to 2 to reflect the physical characteristic that geological properties decrease with the square of spatial distance.

[0157] S430, the spatial interpolation module 400 performs joint encapsulation of heterogeneous data based on the calculated weight parameters, generating a multi-source tracing index array. The three-dimensional geological body located between two profiles is generated by joint interpolation and deduction of the data from both profiles; its physical properties do not originate from a single, definitive text source. The spatial interpolation module 400 reads the first starting byte offset stored in the underlying attribute structure of the corresponding node of the first two-dimensional profile. And the second starting byte offset stored in the underlying attribute structure of the corresponding node on the second two-dimensional profile. .

[0158] The spatial interpolation module 400 reads the first text source reference stored in the attribute structure of the corresponding node of the first two-dimensional profile. And the second text source reference stored in the node attribute structure corresponding to the second side two-dimensional profile. Each text source citation includes at least the following: , and This is used to determine the standardized text stream file containing the target text and its specific text location.

[0159] If, during the reading process, a section node is found to lack a valid text source, including cases where the corresponding missing stratigraphic topology placeholder node is not associated with report text or where text attachment is incomplete due to missing input files, the spatial interpolation module 400 will reference the text source of the corresponding side. The value is assigned to null. At this time, the joint encapsulation operation only retains the valid side index identifier with the entity source, but its spatial interpolation weight is still retained according to the result calculated by the original formula, so as to maintain the overall normalization balance of the three-dimensional spatial interpolation weight system.

[0160] The spatial interpolation module 400 encapsulates the read multi-source text start byte offsets with the corresponding spatial interpolation weights solved in step S420 into a key-value pair data structure. The spatial interpolation module 400 generates a multi-source tracing index array. Its data structure format is defined as follows: ; In the formula, This represents a multi-source traceability index array. This indicates the text source reference corresponding to the first two-dimensional profile. This indicates the text source reference corresponding to the second two-dimensional profile. and These represent the spatial interpolation weights of the two side profiles for the current interpolation voxel.

[0161] The spatial interpolation module 400 will generate a multi-source tracing index array Write the corresponding independent interpolation voxel In the associated voxel attribute storage structure; in implementations that use GPUs to perform 3D rendering or computation, the voxel attribute storage structure can be further mapped to the video memory buffer.

[0162] By embedding a multi-source traceability index array, the three-dimensional voxels generated by interpolation not only have the ability to trace their original source, but also quantitatively record the spatial interpolation weights of different source text data on the current voxel's spatial position.

[0163] See attached document Figure 7 In step S500, the visualization tracing module 500 reads the lithological parameters of the front-end 3D voxel mesh and calls the physical rendering material library to complete the 3D scene rendering. It listens for ray intersection picking events within the 3D viewport to read the multi-source tracing index array and locate the corresponding text. Simultaneously, it listens for text selection events in the document reader to reverse-query the associated 3D voxels based on the text tracing references. Based on the read spatial interpolation weights, it dynamically calculates the corresponding highlighting effects for the text reader and the 3D voxels, achieving weighted bidirectional collaborative localization between the text and the 3D model. The specific processing flow of the visualization tracing module 500 includes the following sub-steps: S510, the visualization and tracing module 500 reads the set of voxels of a 3D geological entity and extracts the inheritance or interpolation relationships of each voxel. The system uses geological entity categories and lithological properties to perform data-driven material mapping. The visualization and tracing module 500 has a pre-built geological material library based on physically based rendering at its underlying level.

[0164] For the geological entity category identified by the current voxel, the visualization tracing module 500 retrieves matching rendering parameters such as albedo maps, roughness maps, and normal maps from the material library. If no matching item is found for the geological entity category identified by the current voxel in the material library, the visualization tracing module 500 uses the system's default solid color material for fallback rendering to prevent abnormal interruptions in the rendering pipeline due to missing material parameters. The visualization tracing module 500 submits the integrated voxel mesh data and material parameters to the underlying graphics application interface to complete the global visualization graphics rendering of the 3D geological scene. The execution flow of the 3D scene rasterization and material shading pipeline based on the underlying graphics interface can be implemented using conventional 3D graphics engines by those skilled in the art; this is well-known technology in the field and will not be elaborated upon here.

[0165] S520, the visualization tracing module 500 listens for user interaction click events within the 3D visualization viewport to trigger multi-source text collaborative localization logic. When a click input in the screen coordinate system is received, the visualization tracing module 500 inversely projects the 2D screen coordinates to the 3D world coordinate system, generating a virtual picking ray. The visualization tracing module 500 calculates the intersection state of this ray with the bounding boxes of each voxel in the 3D geological scene. If the calculation result shows that the virtual picking ray does not intersect with any voxel bounding box, the visualization tracing module 500 determines that the picking action has not hit the target and directly terminates the current collaborative localization process; if an intersection state exists, the visualization tracing module 500 determines the target interpolation voxel where the nearest intersection point to the virtual camera is located.

[0166] The visualization tracing module 500 reads the multi-source tracing index array stored in the voxel attribute storage structure associated with the target interpolated voxel. The visualization and tracing module contains 500 analytical arrays. and and further read each In , and .

[0167] For the parsed valid text source citations, the visual source tracing module 500 sends a message containing [text missing] to the built-in multi-window document reader via the system inter-process communication protocol. , and The document reader first uses text positioning instructions. Identify the corresponding standardized text stream file and its associated original geological exploration report, then move the read pointer of the standardized text stream file to... The indicated byte position, and according to Read the corresponding text segment.

[0168] To accurately locate complete sentences or paragraphs related to the description of geological entities, the document reader, based on... and After obtaining the target text fragment, forward and backward character scanning can be performed with the text fragment as the center. The scanning stops when a carriage return, line feed, or special delimiter is detected, thereby determining the specific rendering boundary of the target text, parsing and locking the corresponding original geological exploration report text paragraph.

[0169] If the starting byte offset read from one side is null, it indicates that the physical properties of the voxel are affected by local stratigraphic pinch-out or missing nodes in the drawing. The visualization tracing module 500 will then intercept the addressing command sent to the document reader on that side to avoid file pointer out-of-bounds errors and system crashes.

[0170] To achieve a complete closed loop, the visual tracing module 500 simultaneously listens for text selection or text positioning interaction events in the built-in document reader. When the user selects a geological exploration report text paragraph for which a text tracing reference has been established, the visual tracing module 500 reads the corresponding... and byte position range, and in the pre-established The inverted index queries for 2D geometric nodes and 3D interpolation voxels containing corresponding text source references; for the obtained 3D voxels, the visualization source module 500 processes each voxel... Different degrees of highlighting are applied to the spatial interpolation weights corresponding to the current text source, thereby enabling bidirectional collaborative interaction by locating 3D geological entities from the geological exploration report text.

[0171] S530, the visualization tracing module 500 performs adaptive transparency rendering calculations based on spatial weights, visually presenting the spatial interpolation weights of different data sources on the current interpolation voxel attributes. The visualization tracing module 500 traces the multi-source index array. Extract the first-side spatial interpolation weights corresponding to each side of the text. Interpolation weights with the second side space .

[0172] The visualization and tracing module 500 substitutes the extracted spatial interpolation weights into the set linear mapping formula to dynamically calculate the opacity coefficient of the highlighted background color of the corresponding jump paragraph in the document reader. The spatial interpolation weights on either side are set to... The corresponding opacity factor for the highlight background is The calculation formula is as follows: ; In the formula, This represents the opacity coefficient of the target highlight background color obtained through dynamic calculation. This represents the spatial interpolation weight parameters corresponding to the parsed text on this side. This indicates the preset upper limit threshold for opacity. This indicates the preset lower limit threshold for opacity.

[0173] The formula involves an upper limit threshold for opacity. Typically set between 0.8 and 1.0, this value characterizes the strong visual feedback corresponding to higher spatial interpolation weights; the lower limit threshold for opacity. The threshold is typically set between 0.1 and 0.3 to ensure that the bright background color still has basic, slight visibility even with extremely low interpolation weights. Those skilled in the art can adjust the threshold according to the color contrast parameters of the terminal display device.

[0174] when One side of When the value is null, the visualization tracing module 500 does not perform text positioning and background color rendering on that side, but still retains the original spatial interpolation weights on that side; for those with effective The highlight opacity of the text on the other side is still calculated according to the corresponding original spatial interpolation weights, and is not renormalized due to the lack of the text source on the other side, so that the highlight level can still reflect the actual spatial distance relationship between the current three-dimensional voxel and the effective source profile.

[0175] The visualization and tracing module 500 will calculate different opacity coefficients. The transparency channel value is injected into the front-end graphics renderer of the document reader to update the background highlight color of the target text paragraph. Through this adaptive rendering mechanism, the text paragraph corresponding to the profile closer to the picking voxel will be displayed with a dark highlight, while the text paragraph corresponding to the profile farther away will be displayed with a semi-transparent light highlight, thus intuitively restoring the weight decay distribution state inside the three-dimensional interpolation space on the two-dimensional text terminal.

[0176] To further illustrate the implementation process and technical effects of this invention, the following description is provided in conjunction with specific application scenarios and experimental data. The specific numerical values, scenario parameters, and comparative experiments described below are only used to explain the implementation principle of this invention and do not limit the scope of protection of this invention.

[0177] In one specific application embodiment, the work area to be processed includes two parallel two-dimensional vector geological profile maps (denoted as profiles) spaced 5050 meters apart. and cross section ), and the corresponding exploration report.

[0178] First, the knowledge graph construction module 100 performs natural language parsing on the geological exploration report. Three sets of standard stratigraphic entities are extracted: the overlying layer is siltstone (entity identifier). The middle layer is mudstone (physical identifier). The underlying layer is limestone (physical identifier). Based on the extracted geological relationships, a directed acyclic graph of textual semantics is constructed. The standard stratigraphic sequence was confirmed as follows: Simultaneously, the starting byte offset of siltstone in the standardized text stream file is extracted. .

[0179] During step S200, the topology parsing module 200 parses the two-dimensional vector geological profile. In the image, siltstone, mudstone, and limestone elements are fully presented, and their spatial contact relationships are shown. Completely identical. However, in the cross-section In the middle, due to local stratigraphic pinch-out, the siltstone planar element nodes With limestone planar primitive nodes Direct contact.

[0180] The topology resolution module 200 performs a joint topology comparison to retrieve and discover nodes. With nodes In the knowledge graph, it cannot be explained by interlacing and cutting relationships, and according to the standard sequence, there are unrepresented intermediate stratigraphic entities. (Mudstone). Therefore, the topology resolution module 200 cuts off the node. and The initial adjacent edges between them, and the coordinates of the midpoint of the contact boundary. Injection area attribute Missing stratigraphic topology placeholder nodes Forced reconstruction of the adjacency sequence is .

[0181] In the three-dimensional interpolation calculation step S400, the spatial interpolation module 400 performs cross-section calculations. Actual mudstone profile and cross-section Missing stratigraphic topology placeholder nodes Contour lofting is performed between these points to generate a three-dimensional set of mudstone voxels that gradually transitions from the actual thickness to zero area (pincerization).

[0182] Select an interpolation voxel within this 3D set Obtain its three-dimensional spatial coordinates and calculate the Euclidean distance from the voxel to the profile. Distance on the plane meters, to the cross-section Distance on the plane Meters. A minimum distance tolerance constant is set for the inverse distance weighting calculation. Meters, power parameter .

[0183] because and The spatial interpolation module 400 calculates the first side (section) according to the inverse distance weighting formula. ) and the second side (section) Spatial interpolation weights: First-side weights Second side weights .

[0184] Subsequently, the aforementioned weights are jointly encapsulated with the underlying text source tracing references to generate a multi-source source tracing index array. .

[0185] In the front-end rendering stage of step S500, when the user picks voxels in the 3D viewport... At that time, the visualization and tracing module 500 extracts the array and sets an upper limit threshold for the opacity of the highlighted background color. lower threshold .

[0186] For the cross section The corresponding exploration report paragraph, the document reader pointer jumps to Lock the corresponding text at that location and calculate its highlight opacity: .

[0187] In the corresponding front-end interface, due to voxels Closer to the cross-section in space (2020 meters), the corresponding text area is highlighted with a high opacity (approximately 75.4%); while for more distant sections... The corresponding text highlight transparency is approximately 44.6%. Through precise numerical mapping, a visual dimensionality reduction representation of the stratigraphic interpolation space confidence was achieved.

[0188] The technical features of the above embodiments can be combined in any way (as long as there is no contradiction in the combination of these technical features). For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described; these embodiments not explicitly written should also be considered to be within the scope of this specification.

Claims

1. A smart analysis and visualization system for geological exploration results, characterized in that, include: The knowledge graph construction module is used to construct a text semantic directed acyclic graph and a knowledge graph of geological entity relationships based on geological exploration report text. It extracts the document identifier and starting byte offset of each node in the corresponding standardized text stream file, generates text source references and encapsulates them as node attributes into the node attribute structure of the corresponding node. The topology parsing module is used to generate an initial spatial adjacency graph based on the input two-dimensional vector geological profile map, perform a topological comparison between the initial spatial adjacency graph and the geological entity relationship knowledge graph, and inject topological placeholder nodes at the missing strata locations to generate a final spatial geometric adjacency graph. The isomorphic matching module is used to calculate the subgraph isomorphic node mapping relationship between the text semantic directed acyclic graph and the final spatial geometric adjacency graph, and write the geological entity attributes and the text source reference into the attribute structure of the corresponding node according to the mapping relationship; The spatial interpolation module is used to perform interpolation operations on adjacent two-dimensional profiles that have completed attribute writing to generate a three-dimensional voxel set, calculate the spatial interpolation weight of the interpolated voxels relative to the adjacent two-dimensional profiles, read the text source references in the corresponding nodes of the adjacent two-dimensional profiles, and jointly encapsulate each text source reference with the corresponding spatial interpolation weight to generate a multi-source source index array. The visualization and tracing module is used to render the three-dimensional voxel set to generate a three-dimensional view. When receiving an interactive instruction for the interpolated voxel, it reads the multi-source tracing index array, locates and displays the corresponding text paragraph according to the document identifier and the starting byte offset, and displays the text paragraph with weights based on the spatial interpolation weight.

2. The intelligent analysis and visualization system for geological exploration results according to claim 1, characterized in that, The knowledge graph construction module divides the relationship edges corresponding to stratigraphic superposition relationships into a stratigraphic superposition relationship edge set, and divides the relationship edges corresponding to magmatic intrusion relationships, fault cutting relationships, or other relationships that cause existing geological bodies to intersect or cut into an intersecting and cutting relationship edge set. It sets an intersecting and cutting relationship flag with a first value for the relationship edges in the intersecting and cutting relationship edge set, and sets an intersecting and cutting relationship flag with a second value different from the first value for the relationship edges in the stratigraphic superposition relationship edge set. The knowledge graph construction module is also used to determine the starting character position number of the text mention corresponding to the target geological entity in the standardized text stream file, and to calculate the cumulative value of the byte length occupied by each character before the starting character position number according to the preset character encoding format, and to determine the obtained cumulative value as the starting byte offset.

3. The intelligent analysis and visualization system for geological exploration results according to claim 2, characterized in that, When generating the initial spatial adjacency graph, the topology parsing module extracts the endpoint coordinates of line elements in the two-dimensional vector geological profile. Based on the set spatial tolerance, it performs node capture and merging operations on the endpoints of line segments whose distance is less than the spatial tolerance. According to the connection relationship of the merged line segments, it performs boundary closure repair and extracts closed planar geological elements. Each closed planar geological element is established as a planar element node, and spatial adjacency edges are established according to the shared boundary between planar elements or the contact relationship formed within the spatial tolerance range, thereby generating the initial spatial adjacency graph.

4. The intelligent analysis and visualization system for geological exploration results according to claim 3, characterized in that, When performing the topology comparison, the topology parsing module traverses the pairs of adjacent planar primitive nodes in the initial spatial adjacency graph. Based on the relationship records of the geological entities corresponding to the adjacent planar primitive nodes in the geological entity relationship knowledge graph, it determines whether the direct spatial contact between the pairs of adjacent planar primitive nodes can be explained by the interpenetration and cutting relationship corresponding to the interpenetration and cutting relationship marker. When the direct spatial contact cannot be explained by the interpenetration and cutting relationship, the stratigraphic superposition relationship path between the geological entities corresponding to the adjacent planar primitive nodes is retrieved from the stratigraphic superposition relationship edge set of the geological entity relationship knowledge graph; When the stratigraphic overlay path contains at least one intermediate geological entity other than the geological entity corresponding to the adjacent planar primitive node, the at least one intermediate geological entity is identified as an unexpressed intermediate stratigraphic entity that needs topological compensation at the current direct spatial contact location.

5. The intelligent analysis and visualization system for geological exploration results according to claim 4, characterized in that, The topology parsing module establishes corresponding topology placeholder nodes respectively; The topological placeholder node is a virtual topological node that does not have actual planar geological elements. The node attribute structure includes at least the geological entity identifier corresponding to the unexpressed intermediate stratum entity, the virtual node flag, the geometric area attribute with a value of zero, and the geometric anchoring information. The geometric anchoring information is determined based on the contact boundary between the corresponding adjacent planar primitive node pairs before the topological placeholder node is injected. After establishing the topological placeholder nodes, the topology parsing module deletes the original direct spatial adjacency edges between the adjacent planar primitive node pairs, and connects each topological placeholder node in series with the adjacent planar primitive node pairs according to the geological entity arrangement order in the stratigraphic superposition path, so as to generate the final spatial geometric adjacency graph.

6. The intelligent analysis and visualization system for geological exploration results according to claim 1, characterized in that, The isomorphic matching module extracts the color channel values ​​of the primitive nodes in the two-dimensional vector geological profile as primitive fill color parameters, and calculates the square root of the sum of the squares of the differences between the primitive fill color parameters and the preset standard color values ​​in the color space as the color distance. When the color distance is less than or equal to a preset threshold, the standard geological entity identifier corresponding to the standard color value is added to the candidate geological entity set. The candidate geological entity set is used as a node mapping constraint to search for the largest common subgraph mapping that satisfies the edge relationship consistency constraint and completes the attribute writing.

7. The intelligent analysis and visualization system for geological exploration results according to claim 5, characterized in that, The spatial interpolation module determines the correspondence between the contours of similar geological entities in adjacent two-dimensional geological profiles based on the geological entity identifiers carried by the spatial geometric nodes in the adjacent two-dimensional geological profiles that have completed attribute writing. It extracts the source contour and target contour that establish the correspondence, performs contour lofting operation on the source contour and the target contour to establish a three-dimensional continuous boundary representation model, and performs three-dimensional spatial voxelization operation on the three-dimensional continuous boundary representation model to generate the three-dimensional voxel set. When the spatial geometric node corresponding to the source contour or target contour is the topological occupant node, the spatial interpolation module reads the geometric anchoring information of the topological occupant node and constructs a degraded source contour or degraded target contour with zero area based on the geometric anchoring information. This causes the geological entity contour on the other side, which is the same as the geological entity corresponding to the topological occupant node and has an actual area, to continuously shrink towards the degraded source contour or degraded target contour, so as to generate a pinch-out transition three-dimensional geometry in which the stratum thickness gradually decreases towards the position defined by the geometric anchoring information.

8. The intelligent analysis and visualization system for geological exploration results according to claim 1, characterized in that, The spatial interpolation module extracts the center point coordinates of the interpolation voxel and calculates the first Euclidean distance from the center point to the first adjacent two-dimensional geological profile projection reference plane and the second Euclidean distance to the second adjacent two-dimensional geological profile projection reference plane, respectively. When the first Euclidean distance or the second Euclidean distance is less than the preset minimum distance tolerance constant, the spatial interpolation weight corresponding to the side with a distance less than the minimum distance tolerance constant is determined to be 1, and the spatial interpolation weight corresponding to the other side is determined to be 0. When both the first Euclidean distance and the second Euclidean distance are greater than or equal to the minimum distance tolerance constant, the negative power exponent values ​​of the first Euclidean distance and the second Euclidean distance are calculated respectively, and the ratio of the negative power exponent value corresponding to any side to the sum of the negative power exponent values ​​of both sides is determined as the spatial interpolation weight corresponding to that side.

9. The intelligent analysis and visualization system for geological exploration results according to claim 1, characterized in that, The interaction command is a picking interaction command for the interpolation voxel. When the visualization tracing module receives the picking interaction command, it back-projects the click position in the screen coordinate system to the three-dimensional world coordinate system to generate a three-dimensional virtual picking ray, calculates the intersection state between the three-dimensional virtual picking ray and the bounding boxes of each voxel in the three-dimensional view, and determines the voxel at the intersection point closest to the virtual camera as the target interpolation voxel when there are intersecting voxels. The visualization tracing module reads the multi-source tracing index array of the target interpolation voxel and sends the document identifier, starting byte offset and text fragment length contained in the valid text tracing reference to the built-in document reader. The built-in document reader determines the corresponding standardized text stream file based on the document identifier, locates the starting position of the target text based on the starting byte offset, reads the target text segment based on the text segment length, performs forward and backward delimiter scanning with the target text segment as the center, and determines and displays the geological exploration report text paragraph corresponding to the target text segment.

10. The intelligent analysis and visualization system for geological exploration results according to claim 9, characterized in that, The visualization tracing module obtains the spatial interpolation weight corresponding to any valid text tracing reference in the multi-source tracing index array, calculates the difference between a preset upper opacity threshold and a preset lower opacity threshold, adds the product of the difference and the spatial interpolation weight to the lower opacity threshold, and obtains the target highlight background opacity coefficient corresponding to the text tracing reference. The target highlight background opacity coefficient is then provided as a transparency channel value to the graphics renderer of the built-in document reader to perform highlight background rendering on the corresponding geological exploration report text paragraph. The visualization tracing module is also used to listen for text selection or text positioning interaction events in the built-in document reader. After obtaining the document identifier and byte position range corresponding to the text paragraph selected by the user, it queries the three-dimensional interpolation voxels containing the text tracing references corresponding to the document identifier and byte position range according to the pre-established text tracing reference reverse index, and reads the spatial interpolation weights corresponding to the selected text paragraphs in the multi-source tracing index array of each queried three-dimensional interpolation voxel. The spatial interpolation weights are used to control the highlighting degree of the corresponding three-dimensional interpolation voxels in the three-dimensional view.

Images