A three-dimensional fluorite ore exploration modeling and analysis system

Abstract

The present application relates to geological exploration and three-dimensional modeling technical field, disclose a kind of fluorite ore three-dimensional exploration modeling and analysis system, the system includes: data preprocessing and metallogenic period definition module, space-time geological structure tensor field construction and dynamic updating module, dynamic geodesic distance field calculation module, and multi-stage feedback driven level set modeling module.The present application is evolved by executing multi-stage time sequence, the model boundary of previous geological period evolution convergence is used as the starting boundary of current period, and the inheritance type mineralization is simulated.In evolution, the system also performs dynamic feedback: narrow band area near the current evolution surface is identified, the tensor field module is instructed to update locally in the narrow band, and the geodesic distance field is recalculated to guide the next evolution.The present application can also generate a three-dimensional boundary probability volume by applying a disturbance to the input parameters and repeating the modeling to quantify the uncertainty of the model.

Description

A 3D exploration modeling and analysis system for fluorite deposits Technical Field

[0001] This invention relates to the field of geological exploration and 3D modeling technology, specifically a 3D exploration modeling and analysis system for fluorite deposits. Background Technology

[0002] Three-dimensional geological modeling is an indispensable core technology in current mineral exploration, reserve estimation, and mine planning. Existing modeling processes typically rely on integrating multi-source heterogeneous geological information such as borehole data, geological maps, and geophysical inversion data to construct the morphology and occurrence of ore bodies in three-dimensional space.

[0003] In constructing orebody models, implicit modeling techniques, particularly those based on radial basis functions (RBF), level sets, or geostatistics, are widely used because they can quickly and automatically generate continuous three-dimensional geological surfaces from discrete exploration data points. These methods automate the modeling process to some extent and often attempt to incorporate geological constraints such as geophysical inversion data volumes to improve the model's predictive accuracy in areas with sparse borehole data.

[0004] However, existing 3D modeling techniques still have significant limitations when dealing with deposits with complex geological structures and multiple mineralization processes (such as fluorite deposits). Most modeling methods are essentially static, performing geometric interpolation only on the current final morphology of the ore body, ignoring the dynamic evolution of the deposit as a result of multiple phases of geological tectonic activity. This leads to insufficient geological genetic plausibility in the models. Furthermore, many implicit algorithms tend to assume isotropy during spatial interpolation, making it difficult to accurately depict the morphology of ore bodies strongly controlled by faults, folds, and other structures (i.e., anisotropy). This results in model boundaries that may unreasonably cross ore-controlling structures, contradicting actual mineralization patterns. In addition, existing modeling workflows are often one-stop processes, lacking a dynamic feedback mechanism to adaptively adjust local geological constraints based on boundary evolution during the modeling process. This leads to low computational efficiency and insufficient depiction of boundary details. Ultimately, these methods typically provide only a single, deterministic model result, failing to quantify the model uncertainties caused by data sparsity and parameter selection. This introduces hidden risks to subsequent resource estimation and exploration decisions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a three-dimensional exploration modeling and analysis system for fluorite deposits. This system solves the problems of existing three-dimensional geological modeling methods when dealing with complex deposits that are multi-phase and tectonically controlled. These problems stem from the static, isotropic, and lacking dynamic feedback nature of the models, resulting in models that do not conform to geological genesis, have low boundary accuracy, and are unable to quantify uncertainties.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a three-dimensional exploration modeling and analysis system for fluorite deposits, comprising:

[0007] The data preprocessing and metallogenic phase definition module is used to receive multi-source heterogeneous geological data and assign metallogenic tectonic elements to multiple metallogenic phase sets with chronological order.

[0008] The spatiotemporal geological structure tensor field construction and dynamic update module constructs an initial structural tensor field at the nodes of the three-dimensional spatial grid for each geological period based on the set of mineralization periods assigned by the data preprocessing and mineralization period definition module.

[0009] The dynamic geodesic distance field calculation module uses the structural tensor field constructed by the spatiotemporal geological structure tensor field construction and dynamic update module as a metric to calculate the geodesic distance field.

[0010] And a multi-stage feedback-driven level set modeling module, which, based on the geodesic distance field calculated by the dynamic geodesic distance field calculation module, drives an evolving level set function to zero level surface to express the ore body boundary, and performs multi-stage temporal evolution, taking the model boundary after the convergence of the previous geological period evolution as the starting boundary of the current geological period evolution.

[0011] In one specific implementation, the data preprocessing and mineralization period definition module is further used to register the received multi-source heterogeneous geological data (e.g., borehole data, geological map data, and 3D geophysical data volumes) to the same engineering coordinate system through a unified affine transformation. This step provides a unified spatial reference for the assignment of structural elements and the subsequent construction of the spatiotemporal geological structure tensor field on a 3D spatial grid, ensuring the consistency of spatial data.

[0012] In a preferred embodiment, the data preprocessing and mineralization period definition module assigns the structural elements within a human-computer interactive environment that integrates a 3D visualization window. Users can select the geometry representing the structural element and assign it to the target geological period set. Simultaneously, the system provides immediate visual feedback in the 3D visualization window by updating rendering attributes (such as changing color or transparency), improving the intuitiveness and accuracy of the assignment.

[0013] Furthermore, the core principle of the spatiotemporal geological structure tensor field construction and dynamic update module in constructing the initial structural tensor field is the fusion of multi-source information. Specifically, this module is used to perform a weighted linear combination of the tectonic source tensor calculated based on geological structural data (such as fault attitude) and the geophysical source tensor calculated based on the gradient distribution of three-dimensional geophysical data volumes (such as gravity or magnetic data). In this way, the generated structural tensor field can simultaneously reflect the macroscopic tectonic distribution and local physical property changes, providing a basis for subsequent anisotropic distance calculations.

[0014] The dynamic geodesic distance field calculation module is specifically implemented by solving an anisotropic equation to calculate the geodesic distance field. A key technical feature of this calculation is that the inverse matrix of the structure tensor field constructed in the previous step is used as a metric to define local distances in space. This anisotropic metric allows distance calculations to propagate faster along the dominant direction of geological structures (such as veins) and slower in the perpendicular direction, thus enabling the calculated distance field to accurately reflect the anisotropy of the geological structure.

[0015] In the multi-stage feedback-driven level set modeling module, the evolution of the level set function is controlled by a composite velocity function. This composite velocity function preferably includes at least two components: a data-driven component determined by the geodesic distance field, used to attract the evolving surface to the ore body boundary (i.e., the position with the largest gradient in the distance field); and a curvature velocity component used to keep the boundary smooth, so as to avoid unreasonable sharp or jagged shapes in the model.

[0016] A core innovation of this invention lies in the fact that the multi-stage feedback-driven level set modeling module is also used to execute a dynamic feedback loop during the evolution process to achieve adaptive focusing of the computation. This process specifically includes: first, identifying a narrow band region near the current evolution surface (i.e., the zero level plane); then, instructing the spatiotemporal geological structure tensor field construction and dynamic update module to perform local updates only within this narrow band region to generate an updated structure tensor field; finally, calling the dynamic geodesic distance field calculation module to recalculate the geodesic distance field within this narrow band region using the updated structure tensor field, and using this high-precision distance field to guide the next step of the level set function's evolution.

[0017] In the aforementioned local update step of dynamic feedback, the spatiotemporal geological structure tensor field construction and dynamic update module can use more refined parameters than those used during initial construction when recalculating the structure tensor field within a narrow band region. For example, it can use smaller spatial smoothing range parameters or higher-resolution geophysical inversion results. This strategy can significantly improve the model accuracy in key boundary regions without sacrificing overall computational efficiency.

[0018] Furthermore, the system provided by this invention may also include a model post-processing and uncertainty analysis module. One function of this module is to transform the finally converged implicit level set function (a scalar field) into an explicit 3D mesh surface model that can be used for 3D visualization and geological analysis after all geological evolutions have been completed, using an isosurface extraction algorithm (such as the moving cube method).

[0019] Another important function of the model post-processing and uncertainty analysis module is to perform uncertainty quantification. This is achieved by applying a set of random perturbations to the system's key input parameters (such as the weights of the source tensor and the confidence level of the geophysical data) and repeatedly executing the entire modeling process (from tensor field construction to level set evolution), thereby generating a set of model implementations. Finally, the module statistically calculates a three-dimensional boundary probability volume based on this. Each voxel value in this probability volume represents the probability that the orebody boundary will appear at that location, thus achieving an intuitive quantification of the uncertainty of the orebody boundary location and providing a risk assessment basis for subsequent exploration decisions.

[0020] This invention provides a three-dimensional exploration modeling and analysis system for fluorite deposits. It has the following beneficial effects:

[0021] 1. This invention employs a multi-stage temporal evolution modeling process, using the converged boundary of the model from the previous geological phase as the starting boundary of the current phase. Simultaneously, it utilizes the spatiotemporal geological structure tensor field and the dynamic geodesic distance field to control the evolution direction. This approach simulates the inherited evolution of ore bodies across different geological tectonic phases, resulting in a final 3D ore body model that more closely resembles real mineralization patterns, significantly improving the model's geological rationality and morphological accuracy.

[0022] 2. This invention employs a multi-stage feedback-driven level set modeling module that performs dynamic feedback during evolution. This module automatically identifies narrow band regions near the current evolving surface and instructs the spatiotemporal geological structure tensor field construction and dynamic update module to perform local updates only within these narrow band regions. This adaptive focusing mechanism avoids global recalculation of the entire 3D spatial mesh, significantly improving computational efficiency while concentrating computational resources on the most critical ore body boundary regions, thereby enhancing the local accuracy and detail rendering capability of the model boundaries.

[0023] 3. By setting up a model post-processing and uncertainty analysis module, after modeling is completed, a three-dimensional boundary probability volume can be statistically calculated by applying perturbations to the input parameters and repeatedly executing the entire modeling process. This probability volume intuitively quantifies the uncertainty of the ore body boundary location, overcoming the deficiency that a single deterministic model cannot reflect risk, and providing a more scientific and reliable decision-making basis for subsequent resource estimation and exploration borehole layout. Attached Figure Description

[0024] Figure 1 is a diagram of the overall functional modules of the present invention;

[0025] Figure 2 is a flowchart of the multi-stage temporal evolution of the three-dimensional exploration modeling of the present invention;

[0026] Figure 3 is a detailed flowchart of the dynamic feedback closed loop of the present invention;

[0027] Figure 4 is a flowchart of the post-processing and uncertainty quantification of the present invention. Detailed Implementation

[0028] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Referring to Figure 1, the present invention provides a three-dimensional exploration modeling and analysis system for fluorite deposits, which achieves accurate modeling of complex ore bodies through the collaborative work of a series of functional modules.

[0030] The system includes a data preprocessing and metallogenic phase definition module. This module receives multi-source, heterogeneous geological data, such as borehole data, geological map data, and 3D geophysical data volumes. A key function is that this module allows users to assign input mineralization-related structural elements, such as faults or fractures, to multiple sets of metallogenic phases with a chronological order, based on prior geological knowledge. Among them An index representing geological periods.

[0031] The system further includes a module for constructing and dynamically updating the spatiotemporal geological structure tensor field. This module is based on the set of periods defined by the aforementioned modules. For each geological period At each node of the three-dimensional spatial grid At this point, construct an initial structure tensor field. The tensor field is a three-dimensional matrix whose mathematical form represents the local spatial connectivity and directionality within a given geological period. Furthermore, the spatiotemporal geological structure tensor field construction and dynamic update module has a dynamic update interface that can receive external commands to locally recalculate the tensor field within a specified region.

[0032] The system also includes a dynamic geodesic distance field calculation module. This module utilizes the spatiotemporal geological structure tensor field construction and dynamic update module to generate the structural tensor field. As a metric for defining spatial distance, the geodesic distance field is calculated by solving the anisotropic equation. .

[0033] The system further includes a multi-stage feedback-driven level set modeling module, which serves as the core execution engine of the entire system. The multi-stage feedback-driven level set modeling module operates within an evolutionary function. The ore body boundary is implicitly expressed on the zero-level surface. Its evolution process is decomposed into multiple stages, each corresponding to a geological period. In a crucial step, the first The evolution of the stages will inherit the first stage. Model after stage evolution convergence The boundary is taken as its starting boundary.

[0034] During the evolution process, this level set modeling module continuously identifies the current evolving surface. The nearby narrowband region. Then, through a feedback path, it instructs the ST-GSTF building blocks to perform local updates only within this narrowband region, generating a more accurate tensor field. Subsequently, the dynamic geodesic distance field calculation module was invoked again to utilize the updated... Recalculate the geodesic distance field within this narrow band region. This geodesic distance field contains updated information. It is then used to guide the next step of the evolution of the level set function, thus forming a dynamic feedback loop of evolution-identification-update-recalculation-re-evolution.

[0035] Finally, the system includes a model post-processing and uncertainty analysis module. The multi-stage feedback-driven level set modeling module completes the evolution across all phases and outputs the final deterministic orebody model. Subsequently, the model post-processing and uncertainty analysis module performs geometric morphology analysis and resource estimation on the final model. Simultaneously, by applying perturbations to the input data and repeatedly executing the entire modeling process, the module generates a set of model implementations and statistically calculates a three-dimensional boundary probability volume to quantify the uncertainty of the ore body boundary location.

[0036] A specific implementation of the data preprocessing and mineralization period definition module first performs input and spatial registration of multi-source heterogeneous data. This process aims to integrate various types of geological exploration data from different sources and with different structures into a unified data environment that can be directly used by subsequent modules.

[0037] The first type of data processed in this step is borehole data. Borehole data is typically provided in multiple tabular files, including a borehole file recording the borehole location, a slewing file recording changes in the borehole trajectory, and an attribute file recording lithology, grade, or structural attitude at different depth intervals along the trajectory. The system first reads the borehole file to obtain the initial three-dimensional coordinates of each borehole. Next, based on the depth, dip, and azimuth records in the slewing file, the system calculates the borehole trajectory using trajectory calculation methods such as the minimum curvature method. This calculation process converts the slewing data into a series of three-dimensional spatial points distributed along the borehole trajectory. Finally, the system precisely maps the interval data from the attribute file (e.g., ore from depth 100 meters to 102 meters) onto this three-dimensional spatial point sequence, generating a discrete set of points with geological attributes (such as grade values, lithology codes, and structural attitude normals), providing a data source for subsequent calculations.

[0038] The second type of data processed in this step is geological maps and structural data. This type of data typically originates from digitized geological maps or exploration line profiles, stored in vector file formats (such as DXF files). The system parses these files, extracting linear or planar geometric information representing structural elements such as faults and lithological boundaries. For linear elements, the system converts them into a set of ordered three-dimensional spatial points. For discrete measurement points representing structural attitudes, the system directly extracts their spatial coordinates and corresponding normal vectors. For large faults that need to be represented as continuous surfaces, the system can generate triangular mesh patches based on the input control points and boundary lines, thereby obtaining a set of discrete points covering the structural surface. and its corresponding normal vector These structured construction data are used to build the structure tensor field. Direct input.

[0039] The third type of data processed in this step is 3D geophysical data volume. This type of data is usually the product of geophysical inversion and is stored in the form of regular or irregular 3D grids, with each grid node... A data volume is formed by corresponding physical attribute values. The system reads the grid definition (including origin coordinates, grid spacing, rotation angle) and attribute values ​​of each node of the data volume, and loads them into memory in preparation for coordinate registration.

[0040] After all the above data has undergone its respective structured processing, it will enter the spatial registration process. The system first defines a unified engineering coordinate system. For data from each source, if its original coordinate system is inconsistent with the engineering coordinate system, the system will apply an affine transformation matrix to transform it to the engineering coordinate system. This transformation can be expressed as:

[0041] ;

[0042] in, It is the three-dimensional coordinate vector of the data point in the original coordinate system. It is its new coordinate vector in the unified engineering coordinate system. It is a 3×3 transformation matrix used to handle rotation, scaling, and shearing. It is a three-dimensional translation vector. By uniformly applying this transformation to the grid framework of all borehole points, structural data points, and geophysical data volumes, the precise spatial alignment and consistency of all data are ensured, laying the foundation for subsequent multi-source data fusion modeling.

[0043] After completing data input and registration, the data preprocessing and metallogenic period definition module then executes an interactive metallogenic period assignment method based on geological priors. The purpose of this method is to transform the geological experts' understanding of the regional metallogenic history into a data structure with time-series attributes that can be processed by computers, providing the basic input for subsequent multi-stage time-series modeling.

[0044] This method is implemented in a human-computer interaction environment that integrates a 3D visualization window, a list of data objects, and a period management panel. The system first renders all spatially registered structural element geometries, such as fault planes and fracture networks, in the 3D visualization window with a uniform default visual style (e.g., uniform colors and transparency).

[0045] Users define geological periods through the period management panel. This panel provides functions for creating, naming, and sorting geological periods. Users can create periods with clear geological meaning and chronological order, such as Period 1: main mineralization period faults and Period 2: later-stage cross-cutting fractures, based on mineralization patterns. The system internally associates these periods with integer indices. Perform a unique mapping and maintain their order.

[0046] Once the assignment process begins, users can select one or more construction elements to be assigned in two ways. The first way is to directly select the geometry representing a specific construction in the 3D visualization window using interactive methods such as mouse clicks; the selected geometry will be highlighted. The second way is to select the corresponding construction element entry by name or attribute from the data object list; this operation will simultaneously highlight the corresponding geometry in the 3D visualization window.

[0047] After selecting one or more structural elements, the user then selects a target geological period in the period management panel, such as Period 1: Main Metallogenic Fault. The user then activates an assignment command. Upon receiving this command, the system performs two core operations. First, at the data level, the system indexes the identifiers of all selected structural elements to the target period. Related construction sets This means that the index for that period has been written into or associated with the data structure of these structural elements. .

[0048] Secondly, at the visual level, the system immediately updates the rendering attributes of the assigned construction elements in the 3D visualization window. For example, all assigned elements... The structural elements will be uniformly changed to the first preset color (such as red), and the assigned time will be changed accordingly. The element is then changed to a second preset color (such as blue). This instant visual feedback provides users with clear and unambiguous confirmation, allowing them to intuitively check and verify the correctness of the assigned operation.

[0049] The user repeatedly performs the operations of selecting structures, selecting geological periods, and executing assignments until all key tectonic elements related to mineralization are categorized into the corresponding geological period sets. In this step, the originally disordered constructed data is organized into a dataset with a defined time series. This provides a crucial time-dimensional input for the subsequent construction of the spatiotemporal geological structure tensor field.

[0050] The spatiotemporal geological structure tensor field construction and dynamic update module first performs the initialization of the three-dimensional mesh. The purpose of this step is to establish a discretized three-dimensional computational space for the subsequent calculation, storage, and evolution of all field quantities (including the structural tensor field, geodesic distance field, and level set function).

[0051] This process begins by determining a three-dimensional computational domain boundary based on all spatially registered data points. The system traverses all borehole data points, structural feature geometric points, and all nodes of the geophysical data volume to calculate the minimum axis-aligned bounding box that completely encloses these data points. To avoid potential boundary effects in subsequent calculations, the system expands this minimum bounding box along each coordinate axis by a preset scale or absolute distance to form the final computational domain.

[0052] Next, the computational domain is discretized into a regular Cartesian coordinate grid. This grid is defined by its origin coordinates in a unified engineering coordinate system. Three mutually orthogonal coordinate axes, and the number of nodes along each axis. The resolution of the mesh is uniquely determined by the node spacing along each axis. These spacings are defined as user-configurable parameters, and their selection directly affects the level of detail and computational resource requirements of the final model.

[0053] Any node in the grid can be accessed via its integer index triplet. To identify, among which This index corresponds to the spatial coordinates of the node. There is a clear mapping relationship between them, and the formula is:

[0054] ;

[0055] in, Represents the spatial coordinates of the nodes. , , Integer index, , , Indicates the coordinates of the grid origin. express axial grid spacing, express axial grid spacing, express Axis grid spacing.

[0056] This discrete grid structure forms the foundational data container for all subsequent calculations in this invention. During the initialization phase, the system initializes each node in the grid. Allocate storage space to store the structure tensor that will be calculated later. The nine components, geodesic distance scalar value and the scalar value of the level set function After initialization, the system obtains a ready-to-use three-dimensional discrete computing environment capable of handling multi-physics quantities.

[0057] After the 3D mesh initialization is complete, the spatiotemporal geological structure tensor field construction and dynamic update module then executes the phased tensor field initialization construction method. This method is applied to each geological phase defined in the data preprocessing module. Independently at all nodes of the 3D mesh Above, compute and store an initial structure tensor. .

[0058] This construction process integrates information from two different sources. The first source is geological structural data. For the current processing period... All structural elements (sets) within The system at each grid node Calculate a source tensor at point This calculation is performed on the period. The influence of all structural elements within the surface is spatially weighted and aggregated. Specifically, the contribution of each structural patch to its neighboring spatial points is determined by its normal vector. The decision is made, and its influence weight varies with the grid node. The decay occurs with increasing distance. This method allows the tensor field to... It can encode well-defined anisotropic directions as defined by geological structures.

[0059] The second source is three-dimensional geophysical data volumes. The system analyzes geophysical data volumes. In each grid node The gradient distribution within a local neighborhood is used to calculate a geophysical source tensor. This process typically involves the data body gradient field Local structural tensor analysis is performed to capture the potential geological structural orientation implied by the boundaries of geophysical anomalies. Finally, the system obtains the period by performing a weighted linear combination of the tensors from the two sources mentioned above. initial structure tensor field The combination method is as follows:

[0060] ;

[0061] in, and It is a user-configurable, non-negative weighting coefficient used to adjust the relative importance of hard geological data and soft geophysical data when defining the final structure tensor field.

[0062] This construction process targets all geological phases. Execution is performed sequentially, ultimately generating a set of static initial tensor fields that characterize the tectonic environments of different geological periods. This set of tensor fields provides a fundamental spatial metric for subsequent geodesic distance calculations and multi-stage evolution modeling.

[0063] Referring to Figure 3, a local dynamic update mechanism for the tensor field based on the evolutionary state is illustrated. This mechanism is implemented in the spatiotemporal geological structure tensor field construction and dynamic update module. Its purpose is to concentrate computational resources during the modeling process on the boundary region of the forming ore body, and to ensure the high fidelity of the final model morphology by dynamically improving the accuracy of the structural tensor field in this region.

[0064] The tensor field local dynamic update mechanism is built upon a feedback loop driven by a level set evolution module. This mechanism applies to any phase of the multi-stage level set evolution. During the iteration process, the tensor field local dynamic update mechanism first receives the current time from the level set evolution module. level set function .

[0065] Based on the received level set function, the spatiotemporal geological structure tensor field construction and dynamic update module performs a narrowband region identification operation. Narrowband region Defined as the current evolution surface (i.e., the zero level set) A finite-width three-dimensional spatial region near ( ). Mathematically, this region consists of all mesh nodes that satisfy the following conditions. constitute:

[0066] ;

[0067] in, It is a preset positive number representing half the width of the narrow band. This narrow band precisely defines the location of the current orebody boundary and the adjacent space it will evolve into.

[0068] After identifying the narrow band region, the spatiotemporal geological structure tensor field construction and dynamic update module is triggered to execute a local, high-precision tensor field recalculation task. This recalculation task is only performed on the grid nodes within the narrow band region, while the tensor values ​​of nodes outside the region remain unchanged. This recalculation process is similar in workflow to the initialization construction method, but there are fundamental differences in the use of calculation parameters and data sources, aiming to improve accuracy.

[0069] Specifically, when recalculating the source tensor, the system will use a smaller spatial scale parameter than during initialization (i.e., the updated scale parameter value is smaller than the initial value). Reducing this parameter means reducing the spatial smoothing range of the construction effects, allowing the tensor field to respond more sensitively to local, subtle construction changes, thereby forming sharper anisotropic features near the boundary.

[0070] Recalculate the geophysical source tensor At the same time, the system also uses a smaller smoothing scale. (Right now The system can use geophysical source tensors to calculate the gradient of geophysical data, or it can call upon a higher-resolution local geophysical inversion result that was not used during the initialization phase due to computational costs. This allows the geophysical source tensor to reflect the more dramatic physical property variations that typically exist near the orebody boundaries.

[0071] After completing the local recalculation, the spatiotemporal geological structure tensor field construction and dynamic update module generates a new and updated structural tensor field. The field is in a narrow band area. The inner part consists of newly calculated high-precision tensor values, while the outer part is consistent with the previous tensor field. same.

[0072] Finally, the spatiotemporal geological structure tensor field construction and dynamic update module will use this locally updated tensor field. The output is then passed to the dynamic geodesic distance field calculation module. Based on this more accurate metric, the distance field module recalculates the geodesic distances within the narrow band region, guiding the level set function to evolve in the next time step to better reflect geological realities. By continuously executing the cycle of identifying narrow bands, local recalculation, and output update at each evolution stage, this mechanism constructs an adaptive, boundary-focused modeling process, as shown in Figure 3, a closed-loop path, thereby achieving accurate characterization of complex orebody boundary morphologies.

[0073] The dynamic geodesic distance field calculation module aims to establish a mathematical model for the structure tensor field. It is transformed into a metric of anisotropic distance. This invention does not use Euclidean distance, but instead... inverse matrix The Riemannian metric tensor, used to define local spatial metrics, reduces costs along structurally advantageous directions. Based on this, the geodesic distance field to be solved... Anisotropic equations satisfying the following form:

[0074] ;

[0075] in, It is its spatial gradient. The boundary conditions are set as the set of all known mineralization point source points. The distance at that location is zero. Solving this equation yields a distance field that reflects the anisotropic characteristics of geology.

[0076] The dynamic geodesic distance field calculation module then executes a numerical implementation step based on the Fast Marching Method (FMM) to efficiently solve the established anisotropic equations. The purpose of this numerical implementation is to transform the continuous partial differential equations into a stable and efficient computational process on a three-dimensional discrete grid, thereby obtaining the values ​​of each grid node. Geodesic distance at the location .

[0077] This numerical calculation employs an ordered frontier propagation algorithm. The system provides each node in the 3D mesh with... Maintain a state, which can be one of the following three: Known, representing the geodesic distance of the node. It has been determined; the critical (Trial) indicates that the node is located within the narrow band of the propagation front, and its... The value is a temporary, undetermined estimate; or unknown (Far), indicating that the node has not yet been reached by the propagation front.

[0078] The system uses a priority queue data structure, typically implemented as a min-heap, to store all nodes in a critical state. This priority queue is based on the node's current geodesic distance. Sort the data to ensure that the top of the queue is always the critical node with the minimum temporary distance value.

[0079] At the start of the calculation, the system is initialized. This includes the set of all source points representing known mineralization points. Nodes within It is set to a known state, and its geodesic distance is set to... All other non-source nodes are set to an unknown state, and their distance values ​​are initialized to infinity. Subsequently, all unknown nodes that are directly adjacent to known nodes are transitioned to a critical state, and the system calculates an initial temporary distance value for them and inserts them (and their temporary distances) into a priority queue.

[0080] After initialization, the system enters an iterative calculation loop. In each iteration, the system extracts (Extract-Min) the node with the minimum temporary distance value from the priority queue, denoted as . This node The state is permanently set to known, and its current distance value is... It has been confirmed as the final solution.

[0081] Next, the system traverses. All neighboring nodes that are not yet in a known state For each such neighbor node The system executes an UpwindUpdate operation. The core of this operation lies in using the discretized form of the established anisotropic equations to... final distance value as well as Use the distance values ​​of other known neighbors to recalculate. A new temporary distance .

[0082] The local solution process uses an upstream finite difference scheme to approximate the gradient. Furthermore, the computation of this format explicitly and locally includes the calculation at the nodes. inverse structure tensor at the location The component of this ensures that the distance calculation is anisotropic, and its propagation speed and direction are strictly controlled by the local geological structure tensor, thereby achieving the effect of propagation along the dominant path.

[0083] After calculating the new temporary distance Then, the system compared it with... Current stored distance value Compare. If the new temporary distance Smaller, the system will be updated. Distance value .like If the original state is unknown, then update its state to critical and insert it into the priority queue. If If the original state is critical, the system updates its position in the priority queue (Decrease-Key operation) to reflect its new, smaller distance value.

[0084] This iterative loop continues, repeatedly extracting the node with the minimum distance from the priority queue and setting it as known, then updating its neighboring nodes, until the priority queue is empty. At the end of the loop, a final geodesic distance field has been calculated for all reachable grid nodes. This forms a complete distance field for use by the level set evolution module.

[0085] The function of the dynamic geodesic distance field calculation module is to respond to dynamic calling and update commands. In each geological period... In the initial stage, the dynamic geodesic distance field calculation module is called for the first time to receive the initial structure tensor field. It also calculates and outputs the initial geodesic distance field covering the entire grid. The crucial invocation occurs within the dynamic feedback loop of the level set evolution. This is because the spatiotemporal geological structure tensor field construction and dynamic update module is based on narrowband regions. A locally updated tensor field was generated. Subsequently, the dynamic geodesic distance field calculation module was invoked again. The dynamic geodesic distance field calculation module received... As new input, and by re-executing the fast travel method, a completely new geodesic distance field reflecting higher-precision tectonic information is calculated. This is then returned to the level set modeling module to ensure that the distance field and the structure tensor field are updated synchronously.

[0086] The multi-stage feedback-driven level set modeling module uses implicit level set functions. This is used to characterize the orebody boundary. The orebody boundary is defined as a zero-level set. Inside the ore body ,external In the initial stage of multi-stage evolution The function is initialized. The system utilizes a known set of mineralization points. As a seed, calculate all grid nodes. The Euclidean distance to the seed boundary is used to generate an initial signed distance function. .

[0087] Referring to Figure 2, which is a schematic diagram of the multi-stage temporal evolution modeling process of this invention, the temporal inheritance mechanism of the multi-stage evolution described in detail in this section is the core technical step for simulating the geological mineralization process shown in Figure 3. The purpose of this mechanism is to ensure that each geological period... The modeling is all done in the previous geological period. This was done based on the established ore body model, thus achieving the correct superposition of geological events in time and space.

[0088] In the multi-stage feedback-driven level set modeling module, when targeting the first... geological periods (of which) The level set evolution process will terminate the iterative calculation for that period after satisfying a preset convergence criterion (e.g., the position change of the zero level set is less than a very small threshold within a certain number of time steps). At this point, the system obtains a representation of the 1st level set on the 3D mesh. The convergent level set function of the final orebody morphology of each period is denoted as The zero level set of this function That is, the first Final orebody boundary of the stage .

[0089] When the system begins executing the... When modeling tasks for individual geological periods, it does not use the initial mineralization source point as the basis. Instead of the previous initialization method, the system will initialize the level set function that converged in the previous stage. As a complete field containing geometric and topological information, it is directly designated as the current field. The initial function for the stage evolution. This inheritance operation is executed as follows:

[0090] ;

[0091] in, It is the first Stage in time The level set function at time.

[0092] By performing this inheritance operation, the first... The ore body model formed in stages (including its internal regions) and external areas The complex geometry and topology of the object are transmitted completely and without distortion to the next level. The starting point of the phase.

[0093] Subsequently, the multi-stage feedback-driven level set modeling module will call the level set modeling module with the first stage feedback. Phase geological structure collection Corresponding structure tensor field and geodesic distance field Or in a dynamic feedback closed loop The generated updated geodesic distance field The system is based on these first... From the field quantity of each period, a completely new evolution velocity field is calculated. This new velocity field It will act on the inherited initial boundary. This allows it to begin a new evolution. As shown in Figure 3, this mechanism enables the formation of the ore body in the first phase. On the boundary, regeneration occurs due to the second phase of tectonic activity. The controlled secondary ore bodies thus realistically simulate the geological process of multi-stage superimposed mineralization.

[0094] Multi-stage feedback-driven level set modeling module calculates composite velocity function To control the level set equation The evolution of the velocity function. Designed as a weighted sum of data-driven and smoothing terms:

[0095] ;

[0096] in, It is a data-driven velocity component, derived from the geodesic distance field. The decision was made to guide the evolutionary curve to expand towards the reachable region; It is the curvature velocity component, used to maintain boundary smoothness, and is related to the average curvature of the current boundary. Proportional. The system calculates at each time step. And update Drive model evolution For data-driven item weights, This represents the weight of the curvature term.

[0097] Referring to Figure 3, which illustrates in detail the collaborative workflow of the dynamic feedback closed loop of this invention, a multi-stage feedback-driven level set modeling module acts as the central coordinator. This workflow ensures real-time, synchronous, and high-precision collaborative evolution among the geological structure tensor field, the geodesic distance field, and the level set function.

[0098] This collaborative workflow is in Each time step of the evolution of the period level set (or several preset time steps) are triggered.

[0099] The multi-stage feedback-driven level set modeling module first bases the current level set function state... Identify the narrow band region where the evolutionary frontier is located. That is, all that satisfy Grid nodes The collection. This narrow band region. It defines the active area where current computing resources need to be concentrated.

[0100] Subsequently, the level set modeling module sends a local update request to the spatiotemporal geological structure tensor field construction and dynamic update module, and updates this narrow band region. The spatial extent is passed as a parameter. Upon receiving this request, the tensor field construction and dynamic update module immediately executes a local dynamic update mechanism: it only applies to narrow-band regions. Within the grid nodes, call higher-precision calculation parameters (such as a smaller smoothing scale). and Alternatively, a higher resolution data source can be used to recalculate its structure source tensor and geophysical source tensor. The product of this operation is an updated structure tensor field with higher accuracy over a narrow bandgap region. .

[0101] The updated metric tensor field Upon completion, the spatiotemporal geological structure tensor field construction and dynamic update module does not directly return it to the multi-stage feedback-driven level set modeling module, but instead passes it to the dynamic geodesic distance field calculation module.

[0102] The dynamic geodesic distance field calculation module receives this updated metric tensor field. Then, immediately perform the range field update operation. This is based on the improved accuracy in the narrowband region. Recalculate a completely new geodesic distance field covering the entire computational domain. .

[0103] Finally, this distance field, which has just been calculated and reflects the latest high-precision structural information, This is returned to the level set modeling module. The level set modeling module receives... Then, immediately use it to update the composite velocity function. In particular, updating its data-driven components The level set modeling module then uses this velocity field based on the latest high-precision information. By solving the level set equation, the level set function for the next time step is calculated and updated. .

[0104] This from Departure, Passing through Identification, Local recalculation Global recalculation Update, finally back The complete closed loop, in the first This process is repeated iteratively before the convergence of each evolutionary phase. This collaborative workflow ensures that the model's evolutionary path is always guided by the most accurate local geological features, thereby achieving high-fidelity dynamic simulation of complex orebody morphologies.

[0105] Multi-stage feedback-driven level set modeling module in the current geological period Each evolutionary time step After each update, a convergence criterion and termination condition check will be performed. The purpose of this check is to determine whether the orebody model morphology of the current period has reached stability, or whether the evolution should be terminated due to other conditions.

[0106] The system primarily monitors the following conditions:

[0107] The first type is the numerical convergence criterion. This criterion is used to measure the evolutionary front (i.e., the zero-level set). The magnitude of the movement. The system at each time step. Next, the current level set function will be calculated. With the function of the previous time step The difference between them. In a specific implementation, the system calculates in the narrowband region. The maximum absolute difference of function values ​​across all grid nodes. This maximum difference is considered acceptable when it remains less than a user-preset, minimal positive tolerance across multiple consecutive time steps. At this point, the system considers that the evolutionary boundary no longer undergoes meaningful changes and has reached a numerically stable state.

[0108] The second type is the data-driven saturation criterion. This criterion is related to the defined velocity function. Closely related. The main driving force of evolution. Based on the updated Defined. When the evolutionary frontier When it has expanded and fully entered the unfavorable mineralization region defined by the distance field (e.g., the geodesic distance field of all points on the front), All are greater than the preset distance threshold. Data is the main driving force. The value will become zero or negative across the entire frontier. At this point, the model evolution has reached physical saturation, the driving force disappears, and the evolution automatically stops.

[0109] The third category is resource constraints. This is a safeguard mechanism designed to prevent the computation process from getting stuck in an infinite loop or taking too long. The system maintains a self-regulating schedule. Iteration time step counter since the start .once Exceeded the user-defined maximum number of iterations. Regardless of whether the first two criteria are met, the system will forcibly terminate the evolution of the current period.

[0110] When any of the above conditions (numerical convergence, data saturation, or reaching the maximum number of iterations) are met, the system will stop the process. The evolution process of each period. At this point, the level set function stored on the grid... Marked as the final convergence result for this period Final convergence result It will be retained by the system and used as the next geological period. Initial function at the start of evolution .

[0111] Referring to Figure 4, in all geological periods After the evolution is complete, the model post-processing and uncertainty analysis module processes the final level set function. Post-processing is then performed. First, the system uses an isosurface extraction algorithm (such as the traveling cube algorithm) to extract the implicit final level set function. Transformed into an explicit three-dimensional mesh surface composed of triangular facets. Based on this explicit model, the system can calculate the total surface area and total volume of the ore body. Secondly, the model post-processing and uncertainty analysis module provides a data interface to export the generated 3D mesh model of the ore body into standard file formats (such as DXF and STL). These files can be imported by mainstream mining software (such as Surpac and Datamine) as hard boundary constraints in resource estimation (such as kriging interpolation).

[0112] Another function of the model post-processing and uncertainty analysis module is to provide a method for quantifying boundary uncertainties based on Monte Carlo simulation.

[0113] First, the system defines a set of key parameters (such as tensor field combination weights). Speed ​​weight (etc.) and their probability distribution.

[0114] The system then executes a loop, randomly selecting a set of parameter values ​​from these distributions in each loop and running the multi-stage modeling process once. This process will generate different final models. .

[0115] Finally, the model post-processing and uncertainty analysis module performs statistical analysis on different final models and calculates the results for each grid node. In this simulation, it was determined to be located inside the ore body. ) three-dimensional probability field Three-dimensional probability field The location uncertainty of the model boundary is intuitively quantified (the region with a value close to 0.5 is the high uncertainty region).

Claims

1. A three-dimensional exploration modeling and analysis system for fluorite deposits, characterized in that, This includes a data preprocessing and metallogenic phase definition module, which receives multi-source heterogeneous geological data and assigns metallogenic tectonic elements to a set of multiple metallogenic phases with a chronological order. The spatiotemporal geological structure tensor field construction and dynamic update module constructs an initial structural tensor field at the nodes of the three-dimensional spatial grid for each geological period based on the set of mineralization periods assigned by the data preprocessing and mineralization period definition module. The dynamic geodesic distance field calculation module uses the structural tensor field constructed by the spatiotemporal geological structure tensor field construction and dynamic update module as a metric to calculate the geodesic distance field. The multi-stage feedback-driven level set modeling module drives an evolving level set function to express the ore body boundary on the zero level surface based on the geodesic distance field calculated by the dynamic geodesic distance field calculation module, and performs multi-stage temporal evolution, taking the model boundary after the convergence of the previous geological period evolution as the starting boundary of the current geological period evolution.

2. The fluorite ore three-dimensional exploration modeling and analysis system according to claim 1, characterized in that, The data preprocessing and mineralization period definition module is also used to: register the received multi-source heterogeneous geological data to the same engineering coordinate system through a unified affine transformation, so as to provide a unified spatial reference for the assignment of the structural elements and the subsequent construction of the spatiotemporal geological structure tensor field on the three-dimensional spatial grid.

3. The fluorite ore three-dimensional exploration modeling and analysis system according to claim 1, characterized in that, When assigning the structural elements, the data preprocessing and mineralization period definition module is specifically used to: allow the user to select the geometry representing the structural element in a human-computer interaction environment that integrates a three-dimensional visualization window, and assign it to the target geological period set; at the same time, provide real-time visual feedback by updating the rendering attributes in the three-dimensional visualization window.

4. The fluorite ore three-dimensional exploration modeling and analysis system according to claim 1, characterized in that, The spatiotemporal geological structure tensor field construction and dynamic update module is specifically used to construct the initial structure tensor field by performing a weighted linear combination of the tectonic source tensor calculated based on geological structure data and the geophysical source tensor calculated based on the gradient distribution of three-dimensional geophysical data volume.

5. The fluorite ore three-dimensional exploration modeling and analysis system according to claim 1, characterized in that, The dynamic geodesic distance field calculation module is specifically used to calculate the geodesic distance field by solving the anisotropic equation, wherein the inverse matrix of the structure tensor field is used as a metric to define the local spatial distance.

6. The fluorite ore three-dimensional exploration modeling and analysis system according to claim 1, characterized in that, The multi-stage feedback-driven level set modeling module is specifically used to: control the evolution of the level set function through a composite velocity function, the composite velocity function including a data-driven component determined by the geodesic distance field and a curvature velocity component for maintaining boundary smoothness.

7. The fluorite ore three-dimensional exploration modeling and analysis system according to claim 1, characterized in that, The spatiotemporal geological structure tensor field construction and dynamic update module is also used to: when receiving a local update command, recalculate the structure tensor field in the narrow band region using a smaller spatial smoothing range parameter or a higher resolution geophysical inversion result than during initial construction.

8. The fluorite ore three-dimensional exploration modeling and analysis system according to claim 1, characterized in that, The multi-stage feedback-driven level set modeling module is also used to perform dynamic feedback during the evolution process, specifically including: identifying narrow band regions near the current evolution surface; instructing the spatiotemporal geological structure tensor field construction and dynamic update module to perform local updates only within the narrow band regions, generating an updated structure tensor field; and calling the dynamic geodesic distance field calculation module to recalculate the geodesic distance field using the updated structure tensor field, guiding the next step of the evolution of the level set function.

9. The fluorite ore three-dimensional exploration modeling and analysis system according to claim 1, characterized in that, It also includes a model post-processing and uncertainty analysis module, which is used to transform the final implicit level set function into an explicit three-dimensional mesh surface model after all geological evolutions have been completed, using an isosurface extraction algorithm.

10. A three-dimensional exploration modeling and analysis system for fluorite deposits according to claim 9, characterized in that, The model post-processing and uncertainty analysis module also includes: generating a set of model implementations by perturbing the input parameters and repeatedly executing the entire modeling process, and statistically calculating a three-dimensional boundary probability volume based on this, which is used to quantify the uncertainty of the ore body boundary position.

Images