Geological model dynamic construction method and system based on digital twinning

Abstract

The invention discloses a geologic model dynamic construction method and system based on digital twinning, and relates to the technical field of geologic modeling, and the method comprises the steps: carrying out the semantic interpretation of multi-source geologic data based on a domain ontology library, and generating an initial geologic model; representing the initial geologic model as a semantic topological graph; in response to updating of the exploration data, performing semantic mapping on the exploration data and the domain ontology library, performing topology diffusion analysis and geological law constraint in a semantic topological graph based on a mapping result, and determining a semantic topological influence domain to be updated; creating a virtual branch for the semantic topology influence domain, and copying the geologic volume metadata in the influence domain into the virtual branch; in the virtual branch, according to the exploration data and the existing data, reconstructing the attribute of the geologic volume element to obtain an updated virtual branch; and combining the change content after conflict resolution to a semantic topological graph to obtain an updated geological model. The invention provides a convenient and reliable dynamic geological modeling method.

Description

Technical Field

[0001] This invention relates to the technical field of geological modeling, and in particular to a method and system for dynamically constructing geological models based on digital twins. Background Technology

[0002] With the deepening application of information technology in geological engineering, constructing high-precision three-dimensional geological models has become a core challenge in realizing digital twins of the geological environment. Traditional three-dimensional geological modeling methods usually build models once based on static datasets, which exposes many limitations when faced with continuously acquired new exploration data (such as tunnel excavation and new boreholes).

[0003] Traditional methods often require reprocessing all historical data and reconstructing the entire model once new local data is found to differ from the original model. This process is computationally intensive and time-consuming, making it difficult to meet the needs of rapid, real-time model iteration in engineering projects and unable to effectively support real-time decision-making based on digital twins. Furthermore, model updates are usually one-time overwrite operations, lacking effective version management and isolation mechanisms. This makes the update process risky, as errors are difficult to roll back, and it is impossible to trace the content, basis, and decision-making process of each update, which is detrimental to collaborative work and quality control.

[0004] Furthermore, in the traditional modeling process, geological interpretation (semantic model) and preparing input data for numerical simulation (physical mechanism model) are two separate steps, requiring a lot of manual intervention for data conversion. This is not only inefficient but also prone to introducing errors, causing the physical simulation results to fail to truly reflect geological understanding, thus reducing the accuracy of the virtual-real interaction of digital twins.

[0005] Therefore, there is an urgent need in this field for a convenient and reliable dynamic geological modeling method. Summary of the Invention

[0006] The technical problem solved by this invention is that traditional geological model construction methods are not convenient and reliable enough.

[0007] To address the aforementioned technical problems, this invention provides the following technical solution: a method for dynamically constructing geological models based on digital twins, comprising:

[0008] Semantic interpretation of multi-source geological data is performed based on a domain ontology library to generate an initial geological model;

[0009] The initial geological model is represented as a semantic topological graph;

[0010] In response to the update of exploration data, the exploration data is semantically mapped to the domain ontology library. Based on the mapping results, topological diffusion analysis and geological law constraints are performed on the semantic topology graph to determine the semantic topological influence domain to be updated. The influence domain is a subgraph of the semantic topology graph.

[0011] Create a virtual branch for the semantic topological influence domain and copy the geological body metadata within the influence domain to the virtual branch;

[0012] Within the virtual branch, the attributes of the geological elements are reconstructed based on exploration data and existing data to obtain the updated virtual branch.

[0013] The updated virtual branches and semantic topology graph are subjected to change detection and conflict resolution.

[0014] The changes after conflict resolution are merged into the semantic topology graph to obtain an updated geological model.

[0015] Preferably, the nodes of the semantic topology graph are geological elements with globally unique semantic identifiers and geometric information, and the edges of the semantic topology graph are used to represent the topological relationships of the spatial relationships between geological elements.

[0016] Before performing semantic interpretation of multi-source geological data based on a domain ontology library to generate an initial geological model, the method also includes:

[0017] Geological concepts are extracted from geological standards and expert knowledge, and these geological concepts are defined as classes.

[0018] Establish hierarchical classification relationships between classes;

[0019] Define attributes for each class and the object relationships between classes;

[0020] Define logical axioms for constraining classes, attributes, and relationships. These logical axioms include constraints on mutual exclusion, domain, and range, resulting in the domain ontology library.

[0021] Within the virtual branch, the attributes of geological elements are reconstructed based on exploration data and existing data to obtain an updated virtual branch, including:

[0022] Within the virtual branch, based on exploration data and existing data, at least one of the geometric morphology and physical properties of the geological volume element is reconstructed to obtain the updated virtual branch.

[0023] Preferably, semantic interpretation of multi-source geological data is performed based on a domain ontology library to generate an initial geological model, including:

[0024] Data cleaning and spatial registration are performed on multi-source raw geological data, which include at least two of the following: borehole data, geophysical data, and remote sensing data.

[0025] Based on the domain ontology library, semantic annotation is performed on the cleaned and registered data;

[0026] Based on the semantically annotated data, continuous three-dimensional geological interfaces are generated through spatial interpolation algorithms or surface reconstruction algorithms, and three-dimensional geological elements are formed to create an initial geological model.

[0027] Preferably, the initial geological model is represented as a semantic topological graph, including:

[0028] Assign a globally unique semantic identifier to each geological volume element in the initial geological model;

[0029] Based on the geometric mesh of each geological volume element, the topological relationships between the volume elements are determined by a preset spatial analysis algorithm. The topological relationships include adjacency relationships, inclusion relationships, and fault contact relationships.

[0030] Using local geological elements as graph nodes, each node attribute includes at least a semantic identifier, a geometric information reference, and a semantic type. Topological relationships are used as graph edges, with edge attributes including relationship type. A semantic topological graph is instantiated and generated, and stored and managed using a graph database.

[0031] Preferably, in response to the update of exploration data, semantic mapping is performed between the exploration data and the domain ontology library. Based on the mapping results, topological diffusion analysis and geological regularity constraints are performed on the semantic topological graph to determine the semantic topological influence domain to be updated, including:

[0032] In response to updates to exploration data, the updated exploration data is semantically mapped to the domain ontology library;

[0033] Starting with the semantic topology graph nodes that are directly associated with the mapping results, a preset breadth-first search or depth-first search algorithm is used to traverse along the topology edges to obtain the initial set of affected geological elements.

[0034] The initial set is optimized by applying predefined geological spatial constraint rules and semantic logic rules to obtain the semantic topological influence domain;

[0035] Among them, the geological spatial constraint rules include distance filtering rules based on spatial buffers;

[0036] Semantic logic rules are predefined and stored in the domain ontology library. These rules include lithological sequence continuity rules and fault influence range rules.

[0037] Preferably, a virtual branch is created for the semantic topological influence domain, and the geological body metadata within the influence domain is copied to the virtual branch, including:

[0038] Establish an update session based on the updated exploration data;

[0039] Assign a unique branch identifier to this update session, initialize an empty graph structure as a container for the virtual branch, and create the virtual branch;

[0040] Copy all attribute values ​​of the nodes corresponding to the geological body metadata within the influence domain to generate a data copy that is completely independent of the source data in the semantic topology graph.

[0041] Populate the virtual branch with data copies and add a status flag to each data copy to indicate that it is in a state to be corrected;

[0042] Map the topological relationship edges between different geological elements in the semantic topological influence domain to the nodes corresponding to the data replicas within the virtual branch.

[0043] Preferably, within the virtual branch, the attributes of the geological volume elements are reconstructed based on exploration data and existing data to obtain an updated virtual branch, including:

[0044] Spatial registration and semantic consistency checks are performed on the exploration data and the existing data copied from the semantic topology graph within the virtual branch to form a multi-source dataset;

[0045] Using multi-source datasets as spatial constraints, surface reconstruction is performed through local interpolation algorithms to correct the geometric morphology of geological elements within virtual branches;

[0046] Based on multi-source datasets, geostatistical methods or physical information neural networks are used to correct the physical and mechanical parameters of geological elements within virtual branches, generating a heterogeneous physical parameter field.

[0047] The virtual branch is updated based on the heterogeneous physical parameter field.

[0048] Preferably, the updated virtual branches and semantic topology graph are subjected to change detection and conflict resolution, including:

[0049] By comparing the updated virtual branches with the semantic topology graph, the differences in geometric mesh, topological connection relationships, and physical attribute values ​​between the virtual branches and the semantic topology graph are identified, and a change set is generated.

[0050] Based on predefined conflict rules, detect semantic conflicts and data reliability conflicts in the change set;

[0051] For identified conflicts, a resolution strategy is executed. If the conflict is covered by a predefined resolution rule, it is automatically resolved. If the conflict cannot be automatically resolved, the human-computer interaction interface is triggered to receive expert decision instructions. The predefined resolution rules include at least one of the following: data source priority rule, data timestamp newer priority rule, and data with higher accuracy priority rule.

[0052] Preferably, the conflict-resolved changes are merged into the semantic topology map to obtain an updated geological model, including:

[0053] The changes after conflict resolution are merged into the semantic topology graph using database transactions to obtain an updated geological model.

[0054] In response to the update of the geological model, a version snapshot is generated. The version snapshot records the changeset, conflict resolution record, timestamp, operator, and metadata of the data source that triggered the update corresponding to this merge.

[0055] Archive the merged virtual branches.

[0056] On the other hand, the present invention also provides a dynamic geological model construction system based on digital twins, the system comprising:

[0057] The model generation module is used to perform semantic interpretation on multi-source geological data based on the domain ontology library and generate an initial geological model.

[0058] The model representation module is used to represent the initial geological model as a semantic topological graph;

[0059] The influence domain determination module is used to respond to the update of exploration data, perform semantic mapping between exploration data and domain ontology library, perform topological diffusion analysis and geological law constraints on semantic topology graph based on the mapping results, and determine the semantic topological influence domain to be updated. The influence domain is a subgraph of semantic topology graph.

[0060] The branch creation module is used to create virtual branches for semantic topological influence domains and copy the geological body metadata within the influence domain to the virtual branches;

[0061] The branch reconstruction module is used to reconstruct the attributes of geological elements within a virtual branch based on exploration data and existing data, thereby obtaining an updated virtual branch.

[0062] The conflict resolution module is used to perform change detection and conflict resolution on the updated virtual branches and semantic topology graph;

[0063] The model update module is used to merge the changes after conflict resolution into the semantic topology map to obtain an updated geological model.

[0064] The beneficial effects of this invention are as follows: By defining the semantic topological influence domain, the update scope is precisely locked within the local subgraph affected by new data, avoiding global reconstruction, greatly reducing computational overhead, improving update efficiency, and meeting the real-time requirements of digital twins; the update process is isolated through virtual branches, and combined with incremental changes and conflict resolution, the entire update process is safe, controllable, and has complete traceability; from initial construction to dynamic update, the entire process is based on a unified domain ontology library for semantic interpretation and mapping, ensuring the consistency of model semantics, completely eliminating semantic fragmentation, improving the accuracy and automation level of virtual-real mapping in digital twins, thereby achieving convenient and reliable dynamic geological modeling. Attached Figure Description

[0065] Figure 1 A schematic diagram of the basic process of the dynamic construction method of geological model based on digital twin provided in the embodiments of the present invention;

[0066] Figure 2 A schematic diagram illustrating the basic process of constructing a domain ontology library provided in this embodiment of the invention;

[0067] Figure 3 Provided for embodiments of the present invention Figure 1 A flowchart illustrating the sub-steps of S140 in the middle;

[0068] Figure 4 Provided for embodiments of the present invention Figure 1 A flowchart illustrating the sub-steps of S150. Detailed Implementation

[0069] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0070] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for dynamically constructing a geological model based on digital twins is provided, including steps S110 to S170:

[0071] S110, based on the domain ontology library, performs semantic interpretation of multi-source geological data to generate an initial geological model;

[0072] S120, representing the initial geological model as a semantic topological graph;

[0073] S130, in response to the update of exploration data, semantically map the exploration data to the domain ontology library, perform topological diffusion analysis and geological law constraints on the semantic topology graph based on the mapping results, and determine the semantic topological influence domain to be updated. The influence domain is a subgraph of the semantic topology graph.

[0074] Preferably, the nodes of the semantic topology graph are geological elements with globally unique semantic identifiers and geometric information, and the edges of the semantic topology graph are used to represent the topological relationships of the spatial relationships between geological elements.

[0075] S140, create a virtual branch for the semantic topological influence domain, and copy the geological body metadata within the influence domain to the virtual branch;

[0076] S150, within the virtual branch, the attributes of the geological elements are reconstructed based on exploration data and existing data to obtain the updated virtual branch;

[0077] Specifically, within the virtual branch, based on exploration data and existing data, at least one of the geometric morphology and physical properties of the geological volume element is reconstructed to obtain the updated virtual branch.

[0078] S160 performs change detection and conflict resolution on the updated virtual branches and semantic topology graph;

[0079] S170, the changes after conflict resolution are merged into the semantic topology map to obtain an updated geological model.

[0080] In the field of information technology, an ontology is a formalized specification used to describe concepts, attributes, and their interrelationships within a specific domain. In this scheme, it acts as a grammatical dictionary for geology, transforming geological concepts described in natural language, such as "granite" and "normal fault," into computer-understandable standard terms with clear hierarchical relationships, attribute definitions, and logical constraints. For example, the ontology can rigorously define "granite" as a subclass of "igneous rock" and possess attributes such as "silica content" and "compressive strength." This ensures that geological information from different sources and described by different individuals can be interpreted and integrated unambiguously, laying a semantically consistent foundation for the entire process.

[0081] With this standardized semantic system in place, the project proceeds to the initial model construction phase. This phase involves cleaning and spatially registering heterogeneous geological data from multiple sources, such as borehole and geophysical surveys. Then, automated semantic annotation is performed by querying an ontology database, transforming the chaotic raw data into information rich in standard semantics. Subsequently, algorithms such as spatial interpolation are used to generate continuous three-dimensional geological volumes from these semantically labeled data points, forming the initial geological model.

[0082] Topology is a branch of geometry that studies the properties of graphs that remain invariant under continuous deformation. In geological models, topological relationships specifically refer to the spatial connections between geological bodies, such as "adjacent," "containing," and "faulted." The proposed solution transforms the initial model into a graph, where each node is a geological volume element with a unique semantic ID and geometric information, and each edge represents a topological relationship between these volumes.

[0083] When new exploration data (such as new profiles revealed during tunnel excavation) is input, a dynamic update process is triggered. The new data is first semantically mapped through the ontology database, meaning it is interpreted as a standard geological event. Subsequently, it performs topological diffusion analysis starting from the location of this event in the semantic topology map. The system automatically traverses along topological edges (such as "adjacent" relationships) to find all potentially affected geological elements, forming a preliminary influence range.

[0084] However, simple geographic connectivity might be too broad. Therefore, the solution incorporates geological constraints for optimization. For example, it applies geological rules such as finite fault influence zones or the need for continuous lithological sequences to filter the initial scope, ultimately defining a minimal semantic topological influence domain. This influence domain is a submap of the original map, used to characterize which existing model regions are specifically affected by the new data.

[0085] To ensure the safety and controllability of updates to the main model, a virtual branch is introduced. The system creates an independent branch environment for the aforementioned influence domain and completely copies all data within the influence domain to this branch. This means that any modification experiments conducted within the branch will not affect the stability of the main model. Within this secure sandbox of the virtual branch, the system coordinates new and old data to reconstruct the geometry of geological elements (such as adjusting rock strata interfaces to fit newly exposed surfaces) and physical properties (such as recalculating the elastic modulus of rocks), thereby obtaining a model version updated within the branch that better reflects the latest exploration findings.

[0086] After modifications within a branch are completed, they need to be merged back into the main model. At this point, the system intelligently performs change detection and conflict resolution. It carefully compares the differences between the branch and the main model and identifies potential conflicts, such as logical contradictions between the lithology determined by new data and existing understandings of adjacent areas in the main model. For simple conflicts, the system can automatically handle them according to preset rules (such as prioritizing high-precision data); for complex conflicts, it submits them to domain experts for decision-making. Finally, all confirmed changes are safely merged back into the main semantic topology graph in a transactional manner (ensuring the atomicity of operations), generating a new version of the geological model and recording a detailed version snapshot, thus achieving complete traceability of the update process.

[0087] For example, during mine tunnel excavation, newly exposed rock strata reveal a fault location that deviates by 10 meters from the original model. Traditional methods require reconstructing the entire model, taking hours to days. This solution, however, can quickly locate the fault's influence area (potentially involving only a few surrounding rock strata elements), update only this local area within the virtual branch, and after conflict checking, merge the precise corrections back into the main model within minutes, generating the latest parameters for stability simulation. This significantly improves the real-time response capability of the digital twin system and the reliability of engineering decisions.

[0088] The construction of a domain ontology library is essentially about establishing a formalized semantic specification for the geological information world. Its purpose is to transform the qualitative knowledge of human experts and scattered industry standards into a computer-executable and unambiguous structured knowledge system. For example... Figure 2 As shown, for the construction of the domain ontology library, preferably, before S110, the method further includes S101~S104:

[0089] S101 extracts geological concepts from geological standards and expert knowledge, and defines geological concepts as classes;

[0090] This process involves identifying and defining core geological entities, such as "granite" and "reverse fault," from standards and expert knowledge, and defining these entities as "classes" in ontology. In information science, a "class" is an abstract representation of a set of instances sharing common attributes, providing a classification template for specific geological data instances.

[0091] S102, Establish hierarchical classification relationships between classes;

[0092] This step is based on the inherent classification system of geology. For example, it establishes that "granite" is a subclass of "acidic intrusive rock," and "acidic intrusive rock" is a subclass of "igneous rock." This is-a generalization-specialization relationship constitutes a hierarchical classification structure to support attribute inheritance, that is, subclasses automatically inherit the attributes and constraints of all their parent classes. This greatly enhances the structure and reasoning ability of the knowledge base.

[0093] S103 defines the attributes for each class and the object relationships between classes;

[0094] Attributes are used to describe the intrinsic characteristics of a class. For example, data attributes such as "silica content" and "compressive strength" can be defined for the "granite" class.

[0095] Object relations are used to characterize the interactions between class instances. For example, the "fault" relation is defined with the domain of the "fault" class and the value domain of the "rock strata" class. This allows the system to formally express complex geological statements such as "fault F1 faulted rock strata S1".

[0096] S104 defines logical axioms for constraining classes, attributes, and relationships. These logical axioms include constraints on mutual exclusion, domain, and range, resulting in the domain ontology library.

[0097] Logical axioms are constraints imposed on classes, attributes, and relations. For example, the axiom of "mutual exclusion" declares that the "sandstone" class and the "granite" class are disjoint, meaning that an instance cannot belong to both classes simultaneously. Domain and range constraints stipulate that a "discontinuity" relation can only be emanated by instances of the "fault" class and must point to instances of the "rock strata" class.

[0098] These axioms, based on descriptive logic, enable automated inference engines to detect logical inconsistencies in data input and perform consistency checks.

[0099] Step S110 aims to utilize this ontology library to perform semantic interpretation and integration of multi-source raw geological data, ultimately generating an initial geological model that can be dynamically updated subsequently. This process essentially elevates heterogeneous, raw observation data into a digital geological body with unified semantics and a clear geometric expression. Preferably, S110 includes sub-steps S111-S113:

[0100] S111, perform data cleaning and spatial registration on multi-source raw geological data, which includes at least two of the following: borehole data, geophysical data, and remote sensing data;

[0101] Multi-source geological data, such as borehole core logging that reveals subsurface point information, geophysical data that provides planar or volumetric geophysical responses (such as seismic profiles and resistivity tomography), and remote sensing data that reflects macroscopic surface features, exhibit significant differences in data format, accuracy, coordinate system, and acquisition scale.

[0102] Data cleaning tasks include removing obvious outliers, filling in missing key information, and standardizing data formats.

[0103] Spatial registration requires the use of coordinate transformation algorithms (such as affine transformation based on control points) or point cloud registration algorithms (such as the Iterative Closest Point (ICP) algorithm) to unify all data under the same spatial reference datum.

[0104] By using S111, systematic and random errors in the data source are eliminated, and accurate spatial correspondences are established, providing a reliable data foundation for subsequent spatial integration and modeling.

[0105] S112, based on the domain ontology library, performs semantic annotation on the cleaned and registered data;

[0106] The system matches and maps cleaned and registered data instances (such as a lithological description of a borehole, "grey moderately weathered granite") to predefined concepts in the domain ontology. For example, through natural language processing or rule matching, the description is mapped to the standard class Granite_ModerateWeathering in the ontology and assigned corresponding attribute values ​​(such as weathering_degree: "Moderate").

[0107] By using an ontology library as a semantic dictionary, unstructured or semi-structured raw data is transformed into structured instance data rich in semantic information, thereby solving the ambiguity and inconsistency in the use of terminology in data from different sources and realizing the semantic fusion of multi-source data.

[0108] S113, based on semantically annotated data, generates continuous three-dimensional geological interfaces through spatial interpolation algorithms or surface reconstruction algorithms, and encloses three-dimensional geological elements to form an initial geological model.

[0109] Spatial interpolation algorithms (such as Kriging) estimate the attribute values ​​of unsampled areas based on data points with spatial coordinates and semantic labels, generating a continuous geological attribute field or interface function.

[0110] Surface reconstruction algorithms (such as the Moving Cubes algorithm) can construct continuous three-dimensional geological interfaces (such as stratigraphic interfaces and fault planes) based on these interface functions. Finally, these interfaces enclose closed three-dimensional geological elements, each of which is associated with the semantic information labeled in step S112, so as to infer the continuous distribution pattern of geological bodies in three-dimensional space from sparse sampling points, thereby constructing an initial three-dimensional geological model that has both geometric shape and rich semantic information.

[0111] For example, in oil and gas exploration in a sedimentary basin, it is necessary to integrate data from hundreds of boreholes with 2D seismic profiles. In stage S111, the depth benchmarks of all boreholes and the coordinate system of the seismic profiles are calibrated and registered. In stage S112, using the constructed sedimentary stratigraphic library, the description of "grayish-black mudstone containing organic matter" in the boreholes is automatically labeled as the standard class Shale_SourceRock, and the `organic_content: "High"` attribute is added to it. Simultaneously, the seismic reflection phase axis interpretation results are mapped to "stratigraphic interface" instances. In stage S113, using the labeled borehole points and seismic interpretation interfaces as constraints, the co-kriging interpolation method is used to generate the 3D spatial interfaces of the main reservoirs and caprocks within the basin, ultimately constructing an initial basin geological model containing semantic voxels such as Sandstone_Reservoir and Shale_Seal. This model not only reproduces the geometric structure of the strata but also directly identifies the spatial distribution of source, reservoir, and caprocks, providing a semantically clear and geometrically accurate initial state for subsequent dynamic updates based on new borehole data.

[0112] The core of step S120 lies in transforming the geometric model into a computational representation form that is easier to perform dynamic relationship analysis and incremental updates, namely a semantic topological graph. This transformation is a key data abstraction step for achieving accurate local updates in this invention. Preferably, S120 includes sub-steps S121~S123:

[0113] S121 assigns a globally unique semantic identifier to each geological volume element in the initial geological model;

[0114] Establish a unique and identifiable identity for geological entities in the digital space. This identifier is different from the row ID inside the database. It is a semantically meaningful logical pointer that is permanently immutable in the global scope of the system. It is usually designed using a Uniform Resource Identifier (URI) scheme, such as urn:geology:model:boundary_fault:001, to provide a stable and reliable index foundation for subsequent graph-based topology management and version tracking.

[0115] S122, based on the geometric mesh of each geological volume element, determines the topological relationship between volume elements through a preset spatial analysis algorithm. The topological relationship includes the relationship between adjacencies, the relationship of inclusion, and the relationship of fault contact.

[0116] Spatial analysis algorithms require precise calculation of the spatial relationships between 3D voxel meshes. Adjacency relationships are determined by whether the triangular mesh faces of two voxels share edges or vertices. Inclusion relationships are determined by using ray casting or parity rule algorithms to determine whether the spatial position of one voxel is completely within another voxel. Fault contact relationships require a comprehensive logical judgment that combines geometric calculations (such as mesh boundary distances) with the semantic type of the voxels (such as one being labeled as a "fault"), so as to automatically deduce the invariant connectivity relationships between entities from the spatial position of geometric data based on computational geometry and topology.

[0117] S123 uses local geological elements as graph nodes, with node attributes including at least semantic identifiers, geometric information references, and semantic types. Topological relationships are used as graph edges, with edge attributes including relationship types. A semantic topological graph is instantiated and generated, and stored and managed using a graph database.

[0118] Each geological element and its attributes are abstracted as a node in graph theory. Node attributes include at least its unique semantic identifier, a reference pointer to its detailed geometric grid data, and a semantic type inherited from the ontology library. The various topological relationships calculated in step S122 are concretized as edges connecting these nodes. The edge attributes need to explicitly record the relationship type. Finally, this graph structure composed of nodes and edges is stored in a dedicated graph database (such as Neo4j).

[0119] For example, in an advanced geological prediction model for a tunnel project, the initial model already includes geological elements such as "granite mass," "fault F1," and "fracture zone." In S121, the system assigns them unique IDs, such as GeoUnit_Granite_Mass and GeoUnit_Fault_F1. In S122, the spatial analysis algorithm calculates that the mesh of "fault F1" and the mesh of "granite mass" have boundary contact, thus establishing a "fault contact" relationship. At the same time, it calculates that the geometry of "fracture zone" is completely located inside "granite mass," establishing an "inclusion" relationship. In S123, all this information is constructed into a semantic topology graph and stored in the graph database.

[0120] When new data revealed during tunnel excavation indicates that the attitude of fault F1 needs correction, the system can immediately and accurately locate all rock strata elements (i.e., influence domains) that have a "fault contact" relationship with F1 by querying the graph, without having to traverse the entire model. This minimizes the update scope and improves efficiency. This graph-based relational representation enables updates from a static geometric model to a dynamic relational network.

[0121] The core of step S130 lies in the system's ability to automatically and accurately define the model scope requiring modification, i.e., the semantic topological influence domain, when new exploration data is input, thereby avoiding inefficient full updates to the entire model. Preferably, S130 includes sub-steps S131~S133:

[0122] S131, in response to the update of exploration data, semantically map the updated exploration data with the domain ontology library;

[0123] By using a pre-built domain ontology library as a standardized dictionary, raw exploration data that may have descriptive ambiguity (such as the text description "grey strongly weathered granite" in borehole logging) is transformed into unambiguous standard semantic events within the system.

[0124] Based on ontology matching and concept recognition, we ensure that new data can be integrated into a unified semantic framework, providing consistent input for subsequent analysis.

[0125] S132, starting from the semantic topology graph node directly associated with the mapping result, the preset breadth-first search or depth-first search algorithm is used to traverse along the topology edge to obtain the initial set of affected geological elements.

[0126] The system starts with one or more nodes that are directly associated with a new data event in the semantic topology graph, and uses graph traversal algorithms such as breadth-first search or depth-first search to traverse along the pre-established topological relationship edges (such as "adjacent" or "fault contact").

[0127] This process collects all nodes reachable on the topological network to form an initial set of affected geological elements, transforming the spatial connectivity problem of the geological model into a problem of solving the connected components of the graph structure, thus achieving rapid and automatic preliminary identification of the scope of influence.

[0128] S133, the initial set is optimized by applying predefined geological spatial constraint rules and semantic logic rules to obtain the semantic topological influence domain; wherein, the geological spatial constraint rules include distance filtering rules based on spatial buffers; the semantic logic rules are predefined and stored in the domain ontology library, and the semantic logic rules include lithological sequence continuity rules and fault influence range rules.

[0129] However, connectivity judgments based solely on geometric topology may result in overfitting with an excessively large range. Therefore, step S133 introduces a rule optimization mechanism based on geological knowledge.

[0130] This step applies two types of rules to filter the initial set: one is geological spatial constraint rules, such as establishing a spatial buffer (e.g., 50 meters on both sides) for a newly identified fault, filtering out voxels in the initial set that are outside the buffer and have no direct topological connectivity; the other is semantic logic rules called from the domain ontology library, such as the "lithological sequence continuity rule", which determines whether newly discovered "sandstone" lithological data may affect distant "granite" voxels that have no topological connectivity, thereby performing semantic-level filtering.

[0131] By encoding professional geological knowledge into computable constraints, and combining spatial analysis with logical reasoning, purely geometric topological results are optimized into precise semantic topological influence domains that conform to geological understanding.

[0132] For example, during the excavation of a mine roadway, a new small fault was discovered at the exposed face.

[0133] In stage S131, the fault data is semantically mapped to standard events. In stage S132, the system performs a topological traversal starting from the node corresponding to the fault location in the graph, initially collecting dozens of rock strata directly or indirectly adjacent to it. In stage S133, the system first applies the fault influence range rule, setting a limited distance threshold (e.g., 20 meters) to filter out distant voxels. Then, it applies the lithology sequence rule to determine that the fault only affects the local sedimentary rock sequence and not unconnected rock masses within distant hydrogeological units. After these two optimization steps, the final influence domain is precisely limited to several key voxels within a specific lithology and range around the fault, providing a precise target range for subsequent minimization updates.

[0134] The core of step S140 is to create a logically isolated, independently operable virtual working environment for the affected domain. For example... Figure 3 As shown, preferably, S140 includes sub-steps S141 to S145:

[0135] S141, Establish an update session based on the updated exploration data;

[0136] Create a separate management unit with a complete context for the entire update lifecycle triggered by specific exploration data. This session will record all metadata of this update, such as the trigger source, operation time, user ID, etc., and provide a transactional management framework for the entire update process, ensuring that all subsequent operations can be tracked and controlled as a logical whole.

[0137] S142, Assign a unique branch identifier to this update session, initialize an empty graph structure as a container for the virtual branch, and create the virtual branch;

[0138] The system assigns a globally unique identifier to the current update session and initializes an empty graph structure as a container for the virtual branch. This virtual branch is conceptually parallel to the semantic topology graph of the trunk, but physically it is an independent data structure.

[0139] By creating independent namespaces and data environments, development is isolated from the main model. This means that any exploratory modifications made within the branch will not affect the stability and consistency of the main model (i.e., the authoritative production version), providing a safe sandbox environment for high-risk geological model editing.

[0140] S143, copy all attribute values ​​of the nodes corresponding to the geological body metadata in the influence domain to generate a data copy that is completely independent of the source data in the semantic topology graph;

[0141] Deep copying of data within the affected domain is not simply about creating pointers or references. Instead, it involves completely copying all attribute values ​​(including semantic identifiers, geometric mesh references, physical properties, etc.) of the affected nodes in the semantic topology graph, generating a completely independent data copy. This ensures that data operations within the virtual branch are completely decoupled from the main data. Any modifications, calculations, or even erroneous operations within the branch will not directly pollute or damage the main data, thus guaranteeing the security and integrity of the main data from the bottom layer.

[0142] S144, populate the virtual branch with data copies and add a status flag to each data copy to indicate that it is in a state to be corrected;

[0143] The system populates the created data copies into the initialized virtual branch container and adds a specific status tag (such as status: "pending revision") to each copy. This tag is a metadata label used to identify that these data objects are in a "pending revision" state, providing a basis for status management in the subsequent update process. The system can use this tag to identify which objects are the targets that need to be processed, thereby optimizing the processing logic and providing a clearer context for conflict detection.

[0144] S145 maps the topological relationship edges between geological elements in the semantic topological influence domain to the nodes corresponding to the data replicas within the virtual branch.

[0145] The system accurately maps all topological relationship edges between geological elements in the semantic topological influence domain subgraph to newly created data replica nodes within the virtual branch. This maintains the integrity of the spatial logical relationships between geological bodies and ensures that, in the isolated branch environment, these replica nodes still maintain topological constraints such as "adjacency" and "containment" consistent with the real world. As a result, a self-consistent and complete subgraph in terms of topological relationships is formed within the branch, providing a correct relational basis for subsequent collaborative reconstruction calculations within the branch.

[0146] Step S150 is used to collaboratively reconstruct the geometric morphology and physical properties of geological elements by integrating newly acquired exploration data with existing historical data within the branch, in order to generate a model version that better aligns with the latest geological understanding. This process essentially integrates new observational evidence into the existing knowledge system and enables the model to evolve autonomously. Figure 4 As shown, preferably, S150 includes sub-steps S151 to S154:

[0147] S151, Spatial registration and semantic consistency checks are performed on the exploration data and the existing data copied from the semantic topology graph within the virtual branch to form a multi-source dataset;

[0148] The new exploration data that triggered this update (such as point clouds of tunnel exposure surfaces) will be spatially aligned and semantically consistent with the existing data copied from the main semantic topology map within the virtual branch (such as borehole data and geophysical interpretation results from the initial model).

[0149] Spatial registration ensures that new and old data are in the same coordinate system through algorithms such as Iterative Closest Point (ICP); semantic consistency checks, based on the domain ontology library, verify the logical compatibility of data at the semantic level, such as lithological classification, to eliminate systematic differences in space and semantics between multi-source data, forming an internally consistent and complementary fusion dataset, providing high-quality constraints for subsequent reconstruction calculations.

[0150] S152 utilizes multi-source datasets as spatial constraints and performs surface reconstruction through local interpolation algorithms to correct the geometric morphology of geological elements within virtual branches.

[0151] This step uses the multi-source dataset formed in step S151 as a spatial constraint and corrects the geometric interface of the geological elements within the virtual branch through local interpolation algorithms (such as moving least squares MLS) or machine learning-based surface reconstruction algorithms.

[0152] For example, using newly revealed point cloud data as a strong constraint, the shape and position of nearby rock strata interfaces can be adjusted to better match actual observations. The core of using scattered data to locally optimize continuous geometric surfaces lies in achieving accurate fitting of local geometric morphology while maintaining the overall structural stability of the model.

[0153] S153, based on multi-source datasets, uses geostatistical methods or physical information neural networks to correct the physical and mechanical parameters of geological elements within virtual branches, generating a heterogeneous physical parameter field.

[0154] This allows for the assignment of more precise physical and mechanical parameters to geological elements and the generation of spatially heterogeneous parameter fields.

[0155] This step is based on a coordinated multi-source dataset and is performed using geostatistical methods such as co-kriging interpolation to analyze spatial variability using variograms or physical information neural networks (PINN).

[0156] Geostatistical methods focus on mining spatial correlation patterns from data and performing interpolation predictions; while physical information neural networks embed mechanical equations such as Hooke's law as constraints into the neural network training process, using a combination of data-driven and physical law-driven approaches to invert rock mass parameters.

[0157] By using spatial statistics or physical laws to constrain discrete point observation data, continuous distributed field data can be deduced, thereby completely solving the problem of mechanistic model distortion caused by the parameter homogenization assumption in traditional methods.

[0158] S154 updates the virtual branch based on the non-homogeneous physical parameter field.

[0159] The system assigns the new geometric shape and heterogeneous physical parameter field calculated in steps S152 and S153 to the corresponding geological body metadata copy within the virtual branch, thereby completing the update of the model within the branch. This completes the persistence of data from the calculation process to the model state, transforming the virtual branch from an initial data copy container into a new candidate model version after data assimilation and algorithm optimization.

[0160] Step S160 is used to systematically evaluate the differences between the branch modifications and the main model, and resolve any contradictions. This process is a key quality control step to ensure the geological rationality and data reliability of the update results. Preferably, S160 includes sub-steps S161-S163:

[0161] S161, compare the updated virtual branch with the semantic topology graph, identify the differences in geometric mesh, topological connection relationship and physical attribute value between the virtual branch and the semantic topology graph, and generate a change set;

[0162] A comprehensive comparison is made between the updated virtual branches and the semantic topology graph used as a benchmark.

[0163] This process requires precise identification of differences at three levels: at the geometric level, morphological changes are quantified by calculating the Euclidean distance of mesh vertex coordinates or using the Hausdorff distance algorithm; at the topological level, additions, deletions, and modifications to topological relationships are detected, such as changes in "adjacent" relationships becoming "separated" due to geometric modifications; and at the attribute level, numerical differences in physical and mechanical parameters are compared. All these differences are captured by the system and structurally recorded as a "change set," which is essentially a detailed list of incremental modifications, precisely describing all changes from the old version to the new version. This provides complete and objective factual evidence for subsequent conflict analysis based on version difference algorithms.

[0164] S162, based on predefined conflict rules, detects semantic conflicts and data reliability conflicts in the change set;

[0165] The system scans the change set according to predefined logical rules, mainly identifying two types of conflicts: First, semantic conflicts, where the modification result contradicts the geological logic defined in the domain ontology library. For example, in a virtual branch, the lithology of a voxel is changed from "granite" to "sandstone," but the voxel still maintains the topological relationship of "the same rock layer sequence" with multiple "granite" voxels in the main model, which violates the geological law of lithological sequence continuity. Second, data reliability conflicts, where the new and old data have inconsistent judgments on the same attribute (e.g., the newly measured wave velocity is 4500 m / s while the original value is 4000 m / s). The system will assess its relative reliability based on metadata such as the authority of the data source, acquisition accuracy, instrument error, and timestamp, and mark the challenge of low-reliability data to high-reliability data.

[0166] S163. For the identified conflict, execute the resolution strategy. If the conflict is covered by the predefined resolution rules, it will be automatically resolved. If the conflict cannot be automatically resolved, the human-computer interaction interface will be triggered to receive expert decision instructions. The predefined resolution rules include at least one of the following: data source priority rules, data timestamp newer priority rules, and data with higher accuracy priority rules.

[0167] For conflicts with clear rules and low uncertainty, the system uses predefined automatic resolution rules. For example, the "higher data accuracy priority rule" assigns higher weight to laboratory test results than field logging results; the "newest data timestamp priority rule" assumes that the latest observation data better reflects the current state. However, for complex conflicts involving significant changes in geological understanding, those not covered by existing rules, or those with high uncertainty (such as fundamental disagreements on the interpretation of key fault properties), the system will pause the automatic process and trigger a human-computer interaction interface. This interface clearly presents the contextual information of the conflict (such as conflict location, type, comparison of old and new data, and reliability assessment) to domain experts, who then use their experience to make a judgment and input the final decision instruction.

[0168] For example, during deep development in a mining area, a new tunnel revealed a rock stratum with a dip angle 15 degrees steeper than the original model. In stage S161, the system compared the change in the rock stratum's geometry and the resulting change in its topological relationship with the overlying strata. In stage S162, the conflict detection module found that this geometric modification caused a significant discrepancy between the rock stratum's thickness in the branch and the thickness inferred from borehole data on adjacent exploration lines (data reliability conflict). In stage S163, the system first attempted to resolve the conflict automatically: since the tunnel exposure data is direct measurement data with higher accuracy than the indirect inference from boreholes, the "higher accuracy data priority rule" automatically took effect, adopting the new dip angle. However, the system simultaneously detected that the dip angle change might indicate the existence of an unknown fault, which was a newly generated semantic conflict, exceeding the processing scope of the automatic rules. Therefore, the system submitted this conflict to geological experts, who, after comprehensive analysis, confirmed the existence of a small fault and decided to add the fault to the model.

[0169] Step S170 marks the final stage of this dynamic update cycle, used to securely and reliably integrate the rigorously verified and coordinated changes into the backbone model and establish a complete update audit trail. Preferably, S170 includes sub-steps S171-S173:

[0170] S171, merge the conflict-resolved changes into the semantic topology graph using a database transaction to obtain an updated geological model;

[0171] The confirmed change set after conflict resolution is committed and applied to the main semantic topology graph via database transactions.

[0172] Database transactions possess the characteristics of atomicity, consistency, isolation, and durability. In this scenario, atomicity ensures that the merge operation is either fully executed or completely rolled back in the event of any failure, thereby absolutely guaranteeing the consistency of the backbone model before and after the update and avoiding data corruption or logical contradictions caused by partial update success and partial failure.

[0173] By using transaction mechanisms, a series of complex add, delete, and modify operations are encapsulated into an indivisible unit, thereby ensuring that the backbone data is in a logically complete and stable state at all times.

[0174] S172, in response to the update of the geological model, a version snapshot is generated. The version snapshot records the changeset, conflict resolution record, timestamp, operator, and metadata of the data source that triggered the update corresponding to this merge.

[0175] In response to a successful update of the geological model, the system automatically creates a lightweight version snapshot object. This snapshot is not a redundant backup of the complete model data, but rather a precise record of the incremental changes made by this merge relative to the previous version. Its content typically includes: the set of changes involved in this merge, i.e., a detailed list of all geometric, topological, and attribute modifications; a complete record of conflict resolution, including the content of the conflict, the resolution strategy, and the decision basis; and key metadata, such as timestamps, operator identity, and the original exploration data source identifier that triggered this update.

[0176] Based on the incremental storage concept of version control systems, this approach records the differences between versions instead of the full dataset, thus greatly reducing the storage overhead of version management while ensuring complete traceability.

[0177] S173 archives the merged virtual branches.

[0178] After confirming the successful merge and generating the version snapshot, the system moves the virtual branch used in this update to the archive storage area and marks its status as "merged". The archive branch and its internal data are preserved for a long time as historical evidence of the update process to achieve full lifecycle management of the update process. The archived branch provides an immutable original data foundation for subsequent model auditing, historical status backtracking, and even the evaluation and optimization of update strategies.

[0179] This application's embodiments, by defining the semantic topological influence domain, precisely lock the update scope within the local subgraph affected by new data, avoiding global reconstruction, greatly reducing computational overhead, improving update efficiency, and meeting the real-time requirements of digital twins. Virtual branches isolate the update process, and combined with incremental changes and conflict resolution, make the entire update process safe, controllable, and fully traceable. From initial construction to dynamic updates, semantic interpretation and mapping are based entirely on a unified domain ontology library, ensuring model semantic consistency, completely eliminating semantic fragmentation, and improving the accuracy and automation level of virtual-real mapping in digital twins, thereby achieving convenient and reliable dynamic geological modeling.

[0180] On the other hand, a dynamic geological model construction system based on digital twins is also provided, which includes:

[0181] The model generation module is used to perform semantic interpretation on multi-source geological data based on the domain ontology library and generate an initial geological model.

[0182] The model representation module is used to represent the initial geological model as a semantic topological graph;

[0183] The influence domain determination module is used to respond to the update of exploration data, perform semantic mapping between exploration data and domain ontology library, perform topological diffusion analysis and geological law constraints on semantic topology graph based on the mapping results, and determine the semantic topological influence domain to be updated. The influence domain is a subgraph of semantic topology graph.

[0184] The branch creation module is used to create virtual branches for semantic topological influence domains and copy the geological body metadata within the influence domain to the virtual branches;

[0185] The branch reconstruction module is used to reconstruct the attributes of geological elements within a virtual branch based on exploration data and existing data, thereby obtaining an updated virtual branch.

[0186] The conflict resolution module is used to perform change detection and conflict resolution on the updated virtual branches and semantic topology graph;

[0187] The model update module is used to merge the changes after conflict resolution into the semantic topology map to obtain an updated geological model.

[0188] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0189] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for dynamically constructing geological models based on digital twins, characterized in that, include: Semantic interpretation of multi-source geological data is performed based on a domain ontology library to generate an initial geological model; The initial geological model is represented as a semantic topological graph; In response to the update of exploration data, the exploration data is semantically mapped to the domain ontology library. Based on the mapping results, topological diffusion analysis and geological law constraints are performed on the semantic topology graph to determine the semantic topological influence domain to be updated. The influence domain is a subgraph of the semantic topology graph. Create a virtual branch for the semantic topological influence domain and copy the geological body metadata within the influence domain to the virtual branch; Within the virtual branch, the attributes of the geological elements are reconstructed based on exploration data and existing data to obtain the updated virtual branch. The updated virtual branches and semantic topology graph are subjected to change detection and conflict resolution. The changes after conflict resolution are merged into the semantic topology graph to obtain an updated geological model.

2. The method as described in claim 1, characterized in that, The nodes of the semantic topology graph are geological elements with globally unique semantic identifiers and geometric information, and the edges of the semantic topology graph are used to represent the topological relationships of the spatial relationships between geological elements. Before performing semantic interpretation of multi-source geological data based on a domain ontology library to generate an initial geological model, the method further includes: Geological concepts are extracted from geological standards and expert knowledge, and these geological concepts are defined as classes. Establish a hierarchical classification relationship between the classes; Define attributes for each class and the object relationships between classes; Define logical axioms for constraining classes, attributes, and relationships, including constraints on mutual exclusion, domain, and range, to obtain the domain ontology library; Within the virtual branch, the attributes of geological elements are reconstructed based on exploration data and existing data to obtain an updated virtual branch, including: Within the virtual branch, based on exploration data and existing data, at least one of the geometric morphology and physical properties of the geological volume element is reconstructed to obtain the updated virtual branch.

3. The method as described in claim 2, characterized in that, The process of semantically interpreting multi-source geological data based on a domain ontology library to generate an initial geological model includes: Data cleaning and spatial registration are performed on multi-source raw geological data, which includes at least two of the following: borehole data, geophysical data, and remote sensing data. Based on the domain ontology library, semantic annotation is performed on the cleaned and registered data; Based on the semantically annotated data, a continuous three-dimensional geological interface is generated using a spatial interpolation algorithm or a surface reconstruction algorithm, and these interfaces are used to form three-dimensional geological elements, thus creating the initial geological model.

4. The method as described in claim 3, characterized in that, The process of representing the initial geological model as a semantic topological graph includes: Assign a globally unique semantic identifier to each geological volume element in the initial geological model; Based on the geometric mesh of each geological volume element, the topological relationships between the volume elements are determined by a preset spatial analysis algorithm. The topological relationships include adjacency relationships, inclusion relationships, and fault contact relationships. Using each geological element as a graph node, with node attributes including at least a semantic identifier, geometric information reference, and semantic type, and the topological relationship as graph edges, with edge attributes including relationship type, the semantic topological graph is instantiated and generated, and stored and managed using a graph database.

5. The method as described in claim 4, characterized in that, In response to the update of exploration data, the exploration data is semantically mapped to the domain ontology library. Based on the mapping results, topological diffusion analysis and geological regularity constraints are performed on the semantic topological graph to determine the semantic topological influence domain to be updated, including: In response to updates to exploration data, the updated exploration data is semantically mapped to the domain ontology library; Starting with the semantic topology graph nodes that are directly associated with the mapping results, a preset breadth-first search or depth-first search algorithm is used to traverse along the topology edges to obtain the initial set of affected geological elements. The initial set is optimized by applying predefined geological spatial constraint rules and semantic logic rules to obtain the semantic topological influence domain; The geological spatial constraint rules include distance filtering rules based on spatial buffers; The semantic logic rules are predefined and stored in the domain ontology library. The semantic logic rules include lithological sequence continuity rules and fault influence range rules.

6. The method as described in claim 5, characterized in that, The process of creating a virtual branch for the semantic topological influence domain and copying the geological body metadata within the influence domain to the virtual branch includes: Establish an update session based on the updated exploration data; Assign a unique branch identifier to this update session and initialize an empty graph structure as a container for the virtual branch to create the virtual branch; Copy all attribute values ​​of the nodes corresponding to the geological body metadata within the influence domain to generate a data copy that is completely independent of the source data in the semantic topology graph. The data copies are populated into the virtual branch, and a status flag indicating that each data copy is in a state to be corrected is added; The topological relationship edges between geological elements in the semantic topological influence domain are mapped to the nodes corresponding to the data replicas within the virtual branch.

7. The method as described in claim 6, characterized in that, Within the virtual branch, the attributes of geological elements are reconstructed based on exploration data and existing data to obtain an updated virtual branch, including: Spatial registration and semantic consistency checks are performed on the exploration data and the existing data copied from the semantic topology graph within the virtual branch to form a multi-source dataset; Using the multi-source dataset as a spatial constraint, surface reconstruction is performed through a local interpolation algorithm to correct the geometric morphology of geological elements within the virtual branch; Based on the multi-source dataset, geostatistical methods or physical information neural networks are used to correct the physical and mechanical parameters of geological elements within the virtual branch, generating a heterogeneous physical parameter field. The virtual branch is updated based on the heterogeneous physical parameter field.

8. The method as described in claim 7, characterized in that, The process of performing change detection and conflict resolution on the updated virtual branches and semantic topology graph includes: By comparing the updated virtual branch with the semantic topology graph, the differences in geometric mesh, topological connection relationship and physical attribute value between the virtual branch and the semantic topology graph are identified, and a change set is generated. Based on predefined conflict rules, semantic conflicts and data reliability conflicts are detected in the change set; For the identified conflicts, a resolution strategy is executed. If the conflict is covered by a predefined resolution rule, it is automatically resolved. If the conflict cannot be automatically resolved, the human-computer interaction interface is triggered to receive expert decision instructions. The predefined resolution rule includes at least one of the following: data source priority rule, data timestamp newer priority rule, and data with higher precision priority rule.

9. The method as described in claim 8, characterized in that, The process of merging the conflict-resolved changes into the semantic topology graph to obtain an updated geological model includes: The changes after conflict resolution are merged into the semantic topology graph using database transactions to obtain an updated geological model. In response to the update of the geological model, a version snapshot is generated. The version snapshot records the changes set, conflict resolution records, timestamps, operators, and metadata of the data source that triggered the update corresponding to this merge. Archive the merged virtual branches.

10. A dynamic geological model construction system based on digital twins, characterized in that, The system includes: The model generation module is used to perform semantic interpretation on multi-source geological data based on the domain ontology library and generate an initial geological model. The model representation module is used to represent the initial geological model as a semantic topological graph; The influence domain determination module is used to respond to the update of exploration data, perform semantic mapping between exploration data and domain ontology library, perform topological diffusion analysis and geological law constraints on semantic topology graph based on the mapping results, and determine the semantic topological influence domain to be updated. The influence domain is a subgraph of semantic topology graph. The branch creation module is used to create virtual branches for semantic topological influence domains and copy the geological body metadata within the influence domain to the virtual branches; The branch reconstruction module is used to reconstruct the attributes of geological elements within a virtual branch based on exploration data and existing data, thereby obtaining an updated virtual branch. The conflict resolution module is used to perform change detection and conflict resolution on the updated virtual branches and semantic topology graph; The model update module is used to merge the changes after conflict resolution into the semantic topology map to obtain an updated geological model.

Images