Ore deposit three-dimensional geologic model intelligent prospecting prediction method and system, terminal and medium

Abstract

The invention relates to the field of geological exploration, in particular to an intelligent prospecting prediction method and system for an ore deposit three-dimensional geological model, a terminal and a medium. The method comprises the steps of obtaining multi-source geological data of a target area to construct an ore deposit three-dimensional geological model, inputting the ore deposit three-dimensional geological model into a trained intelligent prospecting prediction model for ore-forming potential analysis, and optimizing the model or generating an intelligent prospecting prediction scheme according to a predicted resource quantity confidence degree condition; when the model is constructed, three-dimensional inversion calculation, element anomaly field construction and the like are carried out, the model can be optimized through transfer learning, exploration data can be accessed in real time to realize dynamic updating, and a multi-target optimization model is established to output an exploration scheme; the invention also relates to a corresponding system, a terminal and a storage medium. The method achieves the technical effects of improving the accuracy and efficiency of prospecting prediction, dynamically optimizing the model according to the actual situation, reasonably planning the exploration scheme, and reducing the exploration cost and risk.

Description

Technical Field

[0001] This application relates to the field of geological exploration, and in particular to intelligent mineral exploration prediction methods, systems, terminals and media using three-dimensional geological models of mineral deposits. Background Technology

[0002] With the continued growth in global demand for mineral resources and the decreasing availability of easily accessible shallow mineral resources, mineral exploration is gradually shifting towards areas with complex geological conditions, such as deep layers and overburden. This presents greater challenges to mineral exploration, requiring more precise and efficient prospecting methods. Exploration of mineral resources in deep and complex geological conditions is of great significance for ensuring a stable supply of mineral resources in my country. It can expand the sources of mineral resources and meet the ever-growing needs of economic development. At the same time, it also plays a positive role in promoting the development of geological science and technological progress.

[0003] Traditional methods for predicting mineral deposits mainly fall into two categories. One relies on the experience and judgment of geological experts, who use their years of accumulated knowledge and practical experience to observe and analyze geological phenomena, thereby inferring potential target ore-bearing geological bodies. The other method is based on the analysis of two-dimensional geological maps, studying information such as geological structures and stratigraphic distribution on these maps to predict mineral deposits. These methods have provided guidance for mineral exploration to a certain extent and have played a significant role in past mineral development.

[0004] However, traditional mineral deposit prediction methods have many shortcomings. One prominent issue is the inaccurate characterization of complex geological structures. Due to the diversity and concealment of these structures, traditional methods struggle to accurately depict their true form and characteristics. Furthermore, the low efficiency of multi-source geological data fusion prevents the effective integration of various data types, thus underutilizing the data's value. Moreover, the insufficient accuracy in predicting the three-dimensional spatial morphology of ore bodies fails to accurately predict the distribution and morphology of target ore-bearing geological bodies in three-dimensional space, thereby affecting the accuracy and success rate of mineral exploration. Summary of the Invention

[0005] Firstly, in order to improve the efficiency and success rate of mineral exploration, this application provides an intelligent mineral exploration prediction method using a three-dimensional geological model of mineral deposits.

[0006] The intelligent mineral exploration and prediction method using a three-dimensional geological model of mineral deposits provided in this application adopts the following technical solution: Intelligent mineral exploration and prediction methods using three-dimensional geological models of mineral deposits include: Acquire multi-source geological data for the target area, including three-dimensional geological structure data, geophysical exploration data, geochemical element data, exploration engineering chemical analysis data, and known ore body characteristic parameters; A three-dimensional geological model of the ore deposit is constructed based on multi-source geological data. The three-dimensional geological model of the ore deposit includes a stratigraphic distribution model, a structural network model, a mineralization spatial model, a geophysical data model, and a geochemical data model. The three-dimensional geological model of the ore deposit is input into the trained intelligent mineral exploration prediction model, which integrates geological genesis rules and data-driven features through machine learning algorithms. Perform mineralization potential analysis: Output a three-dimensional mineralization probability distribution map and resource quantity prediction range based on the intelligent mineral exploration prediction model; If the confidence level of the predicted resource quantity is lower than the preset threshold, optimize the geological boundary parameters and ore body connection relationships of the three-dimensional geological model of the deposit. If the predicted resource quantity meets the preset confidence standard, an intelligent mineral exploration prediction scheme is generated, which includes the location of mineralized target areas and exploration suggestions.

[0007] By adopting the above technical solutions, integrating three-dimensional geological structures, geophysical fields, geochemical anomalies, and ore body characteristic parameters, the limitations of traditional two-dimensional planar prospecting are overcome, constructing a three-dimensional spatial model that more closely resembles the real geological environment, and providing a three-dimensional data foundation for the analysis of metallogenic regularities. By combining geological genetic rules with data-driven features, machine learning algorithms are used to automatically uncover hidden metallogenic control factors, avoiding the subjectivity of single empirical models and the blindness of data models, significantly improving the accuracy of metallogenic prediction under complex geological conditions. By triggering model parameter iteration through confidence threshold judgment, adaptive adjustments to the geological boundaries and ore body connection relationships are achieved, so that the prediction results continuously approach the actual ore body distribution as exploration data accumulates, reducing the blindness of drilling projects and the waste of resources.

[0008] Preferably, the step of constructing a three-dimensional geological model of the ore deposit includes: Three-dimensional inversion calculations are performed on geophysical exploration data to generate three-dimensional distributions of density, magnetic susceptibility, resistivity, and polarizability. A three-dimensional elemental anomaly field was constructed using geochemical elemental data and the Crick interpolation algorithm. By integrating geological structure data, geophysical and geochemical fields, and chemical analysis, a continuous geological interface is generated using implicit modeling techniques.

[0009] By adopting the above technical solutions, physical property parameters such as density, magnetic susceptibility, resistivity, and polarizability are transformed into continuously distributed three-dimensional physical property volumes, intuitively presenting the differences in physical properties of underground media and providing qualitative basis for identifying geological structures such as faults and lithological interfaces. By utilizing the spatial autocorrelation of geochemical data, a high-precision model of elemental concentration anomalies is generated, effectively delineating areas enriched by ore-forming elements and narrowing the scope of prospecting targets. By integrating multi-source data to construct smooth and continuous geological interfaces (such as stratigraphic interfaces and fault planes), the discontinuity problem of traditional explicit modeling in complex tectonic areas is solved, improving the spatial interpolation accuracy and geological rationality of three-dimensional geological models.

[0010] Preferred options also include: Obtain the data set of ore deposit models in historical exploration areas and the corresponding drilling, trenching, pitting, and shallow well verification results; Convolutional neural networks are used to extract the spatial morphological features of ore deposit models, and hybrid training samples are constructed by combining geological genesis knowledge. The regional adaptability of the intelligent mineral exploration prediction model is optimized through transfer learning algorithms.

[0011] By adopting the above technical solution, and combining the spatial morphological features of ore deposits (such as ore body occurrence, scale, and zonation) extracted by convolutional neural networks with geological genetic knowledge (such as ore-forming fluid migration paths and ore-controlling structural patterns), a composite training sample containing data features and prior knowledge is constructed to avoid the "black box" defect of purely data-driven models. By utilizing mature model parameters from historical exploration areas, the geological conditions of new target areas can be quickly adapted through transfer learning, reducing the dependence on massive labeled data in new areas. This is especially suitable for rapid modeling and prediction in areas with low exploration levels and scarce data.

[0012] Preferably, the method further includes: Real-time access to exploration engineering data, including borehole core data, well geophysical data, depth measurement data, and tunnel logging data; A dynamic update mechanism is established so that when the deviation between new exploration data and model predictions exceeds the tolerance range, the automatic iterative optimization of the three-dimensional geological model of the deposit is triggered.

[0013] By adopting the above technical solutions, dynamic data such as borehole cores, well geophysical exploration, and tunnel logging are integrated to construct a real-time linkage mechanism of "data acquisition - model update - prediction feedback," ensuring that the model always reflects the latest exploration results. By setting a deviation tolerance threshold, the system automatically identifies contradictions between model predictions and actual exploration data (such as a mismatch between the borehole mineralization location and the predicted mineralization probability), accurately locates the model error area, and triggers local updates, avoiding the lag of manual intervention and maintaining the timeliness and reliability of prediction results.

[0014] Preferably, the step of optimizing the three-dimensional geological model of the ore deposit includes: Refer to geophysical anomaly gradient zones to adjust tectonic surface attitude parameters; Recalculate the spatial connectivity index of the mineralized domain; Increase virtual borehole constraints for controllable exploration projects.

[0015] By adopting the above technical solutions, the structural parameters of the tectonic surface are constrained by geophysical anomaly gradient zones (such as gravity gradient zones and magnetic anomaly abrupt change zones), making the structural elements such as faults and folds in the model consistent with the measured geophysical data, thereby improving the accuracy of structural ore-controlling analysis. By calculating spatial connectivity indices (such as the continuity and bifurcation and merging characteristics of ore bodies in three-dimensional space), the spatial distribution model of mineralized bodies is optimized, avoiding resource estimation deviations caused by structural misconnections or fragmentation of mineralized domains. By adding virtual boreholes (simulating exploration projects in key locations that have not yet been constructed), the spatial constraints of the model are enhanced without the addition of new drilling data, which is especially suitable for refining the model in areas with sparse boreholes in the early stages of exploration, providing pre-research support for the deployment of exploration projects.

[0016] Preferably, the step of generating the intelligent mineral exploration prediction scheme includes: Extract the three-dimensional spatial coordinates and geometric parameters of high-probability mineralization units; Calculate the probability distribution function of resource quantity in each predicted target area; Generate an optimal exploration deployment scheme that includes a 3D navigation path.

[0017] By adopting the above technical solutions, the precise three-dimensional coordinates (X / Y / Z) and geometric parameters (such as ore body dip angle, depth, and thickness) of high-probability ore-forming units are extracted, providing direct target area coordinates for drilling engineering design and avoiding the spatial ambiguity problem of traditional two-dimensional planar maps. Based on Monte Carlo simulation and other statistical methods, a resource quantity probability distribution (such as tonnage-grade joint distribution) is generated to replace single numerical prediction, providing a quantitative decision-making basis including risk probability for mine planning. Combining the spatial distribution of ore bodies with the difficulty of engineering construction, a three-dimensional navigation scheme including the optimal borehole trajectory and tunnel excavation direction is generated, reducing the construction risk of exploration equipment in complex terrain and improving drilling efficiency.

[0018] Preferred options also include: Establish a multi-objective optimization model to simultaneously optimize exploration costs, prediction accuracy, and engineering risk indicators; The Monte Carlo algorithm is used to simulate and output exploration engineering combination schemes that meet multiple constraints.

[0019] By adopting the above technical solutions, multiple dimensions such as exploration cost (drilling workload, equipment investment), prediction accuracy (mineralization probability threshold, resource error range), and engineering risk (construction safety in fault fracture zones, groundwater inrush risk) are considered simultaneously. This avoids cost overruns caused by solely pursuing accuracy or construction risks caused by oversimplification. Through thousands of simulations of the implementation effects of different exploration engineering combinations, a Pareto optimal solution set that satisfies multiple constraints is output, providing decision-makers with diverse options that include cost-accuracy-risk trade-offs. This is especially suitable for the overall exploration planning of large mineral clusters.

[0020] Secondly, in order to improve the efficiency and success rate of mineral exploration, this application provides an intelligent mineral exploration prediction system based on a three-dimensional geological model of mineral deposits.

[0021] The intelligent mineral exploration and prediction system based on a three-dimensional geological model of mineral deposits provided in this application includes, The data acquisition module is used to acquire and standardize multi-source geological data; The model building module is used to generate three-dimensional geological models of mineral deposits that integrate geological, geophysical, and geochemical features; The intelligent analysis module deploys the trained mineral exploration prediction model and performs three-dimensional mineralization potential analysis. The dynamic optimization module is used to adjust model parameters based on prediction confidence feedback; The decision output module generates an intelligent mineral exploration prediction scheme that includes three-dimensional spatial positioning information.

[0022] By adopting the above technical solutions, the data acquisition module integrates multi-source heterogeneous data (such as CAD geological maps, GIS spatial data, and MapGIS graphic files) through standardized data interfaces, solving the problems of inconsistent data formats and chaotic coordinate systems in traditional mineral exploration, and improving data preprocessing efficiency by more than 50%. The model building module integrates professional tools such as geophysical inversion, geostatistical interpolation, and implicit modeling to achieve automated generation of 3D geological models of ore deposits from raw data, reducing the tedious operations and subjective errors of manual modeling. The intelligent analysis module deploys deep learning models based on high-performance computing clusters, supporting rapid calculation of 3D meshes with millions of nodes, reducing the time for mineral potential analysis from several days to several hours using traditional methods. The dynamic optimization module monitors the model's prediction effect in real time through a confidence feedback mechanism, automatically identifying "data blind spots" requiring intensive exploration, shifting exploration engineering deployment from experience-driven to data-driven. The decision output module generates an integrated delivery result including a 3D visualization model, resource prediction scheme, and exploration engineering CAD drawings, directly connecting to mine design to achieve seamless transformation and application of mineral exploration results.

[0023] Thirdly, this application provides a smart terminal, which adopts the following technical solution: A smart terminal includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed to perform the aforementioned intelligent mineral exploration prediction method based on a three-dimensional geological model of a mineral deposit.

[0024] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium storing a computer program that can be loaded by a processor and executed by the above-mentioned intelligent mineral exploration prediction method based on a three-dimensional geological model of a mineral deposit.

[0025] In summary, this application includes at least one of the following beneficial technical effects: 1. By integrating three-dimensional geological structures, geophysical fields, geochemical anomalies, and ore body characteristic parameters, this method overcomes the limitations of traditional two-dimensional planar prospecting, constructing a three-dimensional spatial model that more closely resembles the real geological environment, providing a three-dimensional data foundation for metallogenic regularity analysis. It combines geological genetic rules with data-driven features, automatically uncovering hidden metallogenic control factors through machine learning algorithms, avoiding the subjectivity of single empirical models and the blindness of data models, significantly improving the accuracy of metallogenic prediction under complex geological conditions. Furthermore, by triggering model parameter iteration through confidence threshold judgment, it achieves adaptive adjustment of geological boundaries and ore body connectivity, ensuring that prediction results continuously approach the actual ore body distribution as exploration data accumulates, reducing the blindness and resource waste in drilling projects. 2. Physical properties such as density, magnetic susceptibility, and polarizability are transformed into continuously distributed three-dimensional physical property volumes, intuitively presenting the differences in physical properties of underground media and providing qualitative basis for identifying geological structures such as faults and lithological interfaces; by utilizing the spatial autocorrelation of geochemical data, a high-precision model of elemental concentration anomalies is generated, effectively delineating areas enriched by ore-forming elements and narrowing the scope of prospecting targets; by integrating multi-source data to construct smooth and continuous geological interfaces (such as stratigraphic interfaces and fault planes), the discontinuity problem of traditional explicit modeling in complex tectonic areas is solved, improving the spatial interpolation accuracy and geological rationality of three-dimensional geological models; 3. By combining the spatial morphological features of ore deposits extracted by convolutional neural networks (such as ore body occurrence, scale, and zonation) with geological genetic knowledge (such as ore-forming fluid migration paths and ore-controlling structural patterns), a composite training sample containing data features and prior knowledge is constructed to avoid the "black box" defects of purely data-driven models. By utilizing mature model parameters from historical exploration areas, the geological conditions of new target areas can be quickly adapted through transfer learning, reducing the dependence on massive amounts of labeled data in new areas. This is especially suitable for rapid modeling and prediction in areas with low exploration levels and scarce data. Attached Figure Description

[0026] Figure 1 This is a flowchart of the intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to an embodiment of this application, mainly showing steps S100-S420.

[0027] Figure 2 This is a partial flowchart of the intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to an embodiment of this application, mainly showing steps SA1-SA3.

[0028] Figure 3 This is a partial flowchart of the intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to an embodiment of this application, mainly showing steps SB1-SB5.

[0029] Figure 4 This is a partial flowchart of the intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to an embodiment of this application, mainly showing steps SC1-SC3.

[0030] Figure 5 This is a partial flowchart of the intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to an embodiment of this application, mainly showing steps SD1-SD2. Detailed Implementation

[0031] The present application will be further described in detail below with reference to all the accompanying drawings.

[0032] This application discloses an intelligent mineral exploration and prediction method using a three-dimensional geological model of mineral deposits. (Refer to...) Figures 1-5 Intelligent mineral exploration and prediction methods using three-dimensional geological models of mineral deposits include: S100: Acquire multi-source geological data for the target area, including three-dimensional geological structure data, geophysical exploration data, geochemical element data, exploration engineering chemical analysis data, and known ore body characteristic parameters.

[0033] Specifically, stratigraphic data, fault distribution, and fold morphology are obtained from regional geological surveys. Combined with surface outcrop observations, geological profile measurements, and existing two-dimensional geological maps, an initial three-dimensional geological framework is constructed using three-dimensional modeling techniques (such as profile-based interpolation or geostatistical methods). Raw data obtained from geophysical methods such as gravity, magnetics, electrical resistivity, and seismic exploration (e.g., gravity anomalies, total magnetic field strength, polarizability and resistivity sounding curves, seismic reflection waves) are integrated and unified to the geographic coordinate system of the target area through data preprocessing (denoising, topographic correction, coordinate transformation). Geochemical analysis data (e.g., Au, Cu, Pb) from surface soil, rocks, and stream sediments are collected. The data includes the content of mineralization and the spatial coordinates of sampling points to ensure that the data covers the key metallogenic belts and potential anomaly areas of the target area; core samples and grooved samples are collected in exploration projects such as boreholes and tunnels for chemical analysis to obtain data such as ore body grade; the spatial location (three-dimensional coordinates), morphological parameters (strike, dip angle, thickness, depth), ore grade and mineralization type of the proven ore bodies are compiled as key reference data for model training and verification.

[0034] All data undergoes standardization processing, unifying data formats (such as XYZ coordinates and attribute tables) and spatial reference systems (such as the CGCS2000 coordinate system and Gaussian projection) to ensure spatial consistency and fusionability of multi-source data.

[0035] S200: Construct a three-dimensional geological model of the ore deposit based on multi-source geological data. The three-dimensional geological model of the ore deposit includes a stratigraphic distribution model, a structural network model, a mineralization spatial model, a geophysical data model, and a geochemical data model.

[0036] Specifically, the steps for constructing a three-dimensional geological model of a mineral deposit include: S210: Perform three-dimensional inversion calculations on geophysical exploration data to generate a three-dimensional distribution of density, magnetic susceptibility, resistivity, and polarizability; Specifically, a three-dimensional inversion is performed on the preprocessed geophysical data (such as gravity anomalies, magnetic data, polarizability and resistivity sounding data, and wave velocity data). The inversion results are used to constrain the morphology of geological interfaces and cannot be directly modified. Gravity and magnetic data: Using inversion algorithms based on density and magnetic susceptibility differences (such as equivalent source method and particle swarm optimization inversion), the gravity and magnetic field anomalies observed on the surface are converted into density and magnetic susceptibility distributions in the three-dimensional underground space, reflecting the differences in physical properties of strata, rock masses and ore bodies.

[0037] Electrical resistivity data: By inverting three-dimensional polarizability and resistivity (such as finite element or finite difference methods), a three-dimensional distribution of underground polarizability and resistivity is constructed to identify high-resistivity or low-resistivity high-polarization anomalies (such as ore bodies, water-bearing structures, and fault zones).

[0038] The inversion process improves accuracy by setting constraints (such as known geological interfaces and prior ranges of physical property parameters), and finally generates a continuous and smooth three-dimensional distribution of density, magnetic susceptibility, polarizability and resistivity.

[0039] S220: Construct a three-dimensional elemental anomaly field from geochemical element data using the Crick interpolation algorithm; Specifically, for geochemical element data (such as the element content at each sampling point), the Crick interpolation algorithm is used to construct a three-dimensional elemental concentration field at the Earth's surface: Variation function analysis: Calculate the spatial variability of element content, fit the semivariogram, and determine the spatial autocorrelation of element distribution (such as nugget effect, range, sill value).

[0040] 3D mesh generation: The target area is divided into regular 3D meshes (such as 10m×10m×10m cube units), with each mesh node as the interpolation target.

[0041] Crick interpolation calculation: Using the element content and variation function model of the surrounding effective sampling points, the element concentration estimate and estimation error of each grid node are calculated to generate a continuous three-dimensional element anomaly field, which intuitively shows the spatial enrichment pattern of ore-forming elements (such as high concentration anomaly areas).

[0042] S230: Integrates geological structure data, geophysical, geochemical fields, and chemical analysis, and uses implicit modeling techniques to generate continuous geological interfaces.

[0043] Specifically, implicit modeling techniques (such as level set methods and distance field modeling) are used to integrate geological structure data, geophysical and geochemical fields, and chemical analysis to construct continuous three-dimensional geological interfaces. Based on three-dimensional geological structure data, combined with physical property distribution (such as density differences to distinguish different strata) and elemental anomaly fields (such as background values ​​of specific elements in sedimentary strata), the stratigraphic interfaces are fitted by implicit functions to ensure that the strata are continuous in three-dimensional space and conform to the laws of geological sedimentation (such as layered distribution and angular unconformity contact).

[0044] Based on spatial attitude data (strike, dip, and extension length) of faults, folds, and other structures, combined with geophysical anomaly gradient zones (such as gravity gradient zones indicating fault locations and magnetic anomaly abrupt change zones indicating lithological contact zones), a complete structural network is constructed by generating structural interfaces such as fault planes and fold axis planes through implicit surface modeling.

[0045] Constrained by known ore body characteristics, and combined with elemental anomaly fields (greater than the main boundary position or specified value) and physical property distribution (such as the resistivity difference between the ore body and the surrounding rock), the spatial range of the mineralization domain is delineated through implicit functions, reflecting the continuity and zonation of the mineralization body (such as primary mineralization zone and oxidation zone).

[0046] S210 and S220 transform the raw geophysical and geochemical data into three-dimensional spatial distribution models (volume models), respectively. S230 then integrates these volume models, point / line structure data, and other analysis results to generate the final three-dimensional surface model (surface model / solid model) describing the boundaries of geological entities. Together, they form a complete geological model framework (strata, structure, mineralization). The geophysical and geochemical models themselves are also important components of the final three-dimensional geological model of the deposit (geophysical data model, geochemical data model).

[0047] S300: Input the three-dimensional geological model of the ore deposit into the trained intelligent mineral exploration prediction model, which integrates geological genesis rules and data-driven features through machine learning algorithms.

[0048] Specifically, model construction: An intelligent mineral exploration prediction model is built using machine learning algorithms (such as convolutional neural networks, random forests, or graph neural networks). This model integrates geological genesis rules with data-driven features. Geological genesis rules: Transform expert knowledge (such as prospecting indicators such as "high probability of mineralization at fault intersections" and "skarn-type ore bodies are easily formed at igneous rock contact zones") into prior constraints of the model. For example, introduce artificially designed features such as structural density and distance from igneous rocks into the model input layer.

[0049] Data-driven features: implicit features are automatically extracted from the three-dimensional geological model of the ore deposit through machine learning, such as the curvature of the stratigraphic contact zone, the connectivity of the structural network, and the gradient changes of the elemental anomaly field.

[0050] Model training: Using a dataset of ore deposit models from historical exploration areas (including three-dimensional geological structure, physical properties, elemental fields, and known ore body locations) as training samples, and labeling the probability of ore body existence, the model is trained through supervised learning to identify key spatial patterns related to mineralization.

[0051] Model input: The three-dimensional geological model of the target area (strata distribution, structural network, mineralization spatial model and physical properties, element field data) is converted into a grid-like or voxelized numerical matrix and input into the trained intelligent mineral exploration prediction model as the basic data for mineralization potential analysis.

[0052] S400: Perform mineralization potential analysis: Based on the intelligent mineral exploration prediction model, output a three-dimensional mineralization probability distribution map and resource quantity prediction range.

[0053] Specifically, the three-dimensional mineralization probability calculation: The intelligent mineral exploration prediction model performs grid-by-grid calculations on the input three-dimensional geological model of the mineral deposit, outputting the mineralization probability value (between 0 and 1) for each three-dimensional spatial grid node, and generates a three-dimensional mineralization probability distribution map through interpolation or visualization technology, intuitively displaying high-probability mineralization areas (such as areas with a probability ≥ 0.7 are highlighted in red).

[0054] Resource prediction interval estimation: For high-probability mineralization areas, the resource prediction interval is estimated by combining the geometric parameters of the ore body (such as thickness and depth) in the mineralization spatial model, the average grade of the elemental anomaly field, and the known resource quantity statistical law of the ore body, using the volume-grade method or statistical simulation method (such as Monte Carlo algorithm simulation).

[0055] S410: If the confidence level of the predicted resource quantity is lower than the preset threshold, optimize the geological boundary parameters and ore body connection relationships of the three-dimensional geological model of the deposit.

[0056] Specifically, it includes three steps, among which the collaborative optimization of key structural parameters involves systematically checking whether the attitude parameters (strike and dip angle) and their spatial combination relationships of the main structural surfaces (faults, fold axial surfaces) in the model are consistent with the regional geological patterns and the local structural features revealed by adjacent exploration projects.

[0057] Based on geological rationality (such as structural style matching, conjugate fault angle, and fold geometry coordination), the attitude parameters of relevant structural surfaces are adjusted in a coordinated manner to ensure that the topological relationship of the structural network in three-dimensional space is reasonable and consistent with geological understanding.

[0058] Geological reassessment of mineralized domain connectivity: Calculation of spatial connectivity index of mineralized domain using graph theory or spatial algorithms; Prioritize the reassessment of mineralization domain boundaries in the model caused by tectonic separation or data intervals based on geological evidence (such as alteration zoning continuity, mineralization patterns, and tectonic ore-controlling laws). For mineralized domains that are geologically considered to be continuous but are separated by the model, their connectivity is merged or adjusted based on geological logic.

[0059] Introducing virtual drilling to fill data gaps: In sparsely explored areas, virtual boreholes should be rationally arranged based on model geological concepts and regional metallogenic regularities; Define the expected lithological sequence, key geological interface locations, and mineralization information characteristics, and incorporate them as "conceptual constraints" into the modeling process to optimize the geological interface morphology and mineralization domain distribution in blank areas.

[0060] S420: If the predicted resource quantity meets the preset information standard, generate an intelligent mineral exploration prediction scheme that includes the location of mineralized target areas and exploration suggestions.

[0061] The steps to generate an intelligent mineral exploration prediction scheme include: SC1: Extract the three-dimensional spatial coordinates and geometric parameters of high-probability mineralization units; SC2: Calculate the probability distribution function of resource quantity in each predicted target area; SC3: Generates an optimal exploration deployment scheme that includes a 3D navigation path.

[0062] Specifically, extract high-probability mineralization units: In the three-dimensional mineralization probability distribution map, high-probability mineralization units are delineated by spatial threshold screening (e.g., mineralization probability ≥ 0.8), and the three-dimensional spatial coordinates (X / Y / Z range) and geometric parameters (e.g., ore body strike, dip angle, average thickness, and extension direction) of each unit are extracted to form an accurate three-dimensional target area location list.

[0063] Calculate the probability distribution function of resource quantity: For each predicted target area, the Monte Carlo algorithm is used to generate a probability distribution function of resource quantity by combining the element grade distribution, ore body geometric parameters, and statistical models (such as log-normal distribution and truncated Gaussian model) in the mineralization spatial model. For example, it outputs quantitative results such as "the probability of copper resource quantity ≥ 50,000 tons is 90%, and the probability of ≥ 100,000 tons is 60%", providing a scientific basis for resource assessment that includes risk probability.

[0064] Generate the optimal exploration deployment plan: 3D navigation path planning: Combining the spatial location of high-probability mineralized units, surface topography, underground structures (such as the distribution of fault fracture zones), and engineering construction difficulty, 3D path optimization algorithms (such as Dijkstra's algorithm and A* algorithm) are used to design the optimal exploration engineering path. For example, it can plan the shortest borehole trajectory for drilling projects, avoiding high-risk fault zones, and provide reasonable suggestions for tunnel excavation.

[0065] Visualization of the plan: The exploration deployment plan is presented in a three-dimensional visualization form, marking key engineering parameters such as borehole location, expected depth, and tunnel orientation, forming an optimal exploration plan that includes a three-dimensional navigation path, directly guiding field exploration and construction.

[0066] Other methods include: SA1: Obtain the data set of ore deposit models in historical exploration areas and the corresponding drilling, trenching, pitting and shallow well verification results; SA2: Convolutional neural networks are used to extract the spatial morphological features of the ore deposit model, and combined with geological genesis knowledge to construct hybrid training samples; SA3: Optimizes the regional adaptability of the intelligent mineral exploration prediction model through transfer learning algorithms.

[0067] Specifically, obtain historical datasets: collect complete ore deposit models (including three-dimensional geological structure, physical properties, elemental fields, and verified ore body locations) from multiple historical exploration areas to form a training dataset containing successful and unsuccessful cases.

[0068] Constructing mixed training samples: Spatial morphology feature extraction: Convolutional neural networks (CNN) are used to process the three-dimensional mesh data of historical ore deposit models to automatically extract the spatial morphology features of the ore body (such as ore body length, dip angle, branching morphology, and intersection with structural surfaces).

[0069] Integrating geological genesis knowledge: Geological experts label each historical case with geological genesis attributes (such as "controlled by NE-trending faults" and "related to granite intrusion"). This prior knowledge is combined with numerical features extracted by CNN to form a hybrid training sample that includes data-driven features and geological rules.

[0070] Transfer learning optimizes regional adaptability: Pre-trained model: First, the intelligent mineral exploration prediction model is pre-trained on historical datasets to enable it to master general mineralization rules.

[0071] Target area fine-tuning: For new target areas, the pre-trained model is fine-tuned using a small amount of local exploration data (such as a few boreholes or local geophysical anomalies). Through transfer learning techniques, historical experience is transferred to the new area, which can quickly adapt to the unique geological conditions of the target area (such as different structural styles and combinations of ore-forming elements) and reduce the dependence on a large amount of local labeled data.

[0072] The 3D geological model of the mineral deposit can also be automatically iterated, and the specific steps include: SB1: Real-time access to exploration engineering data, including borehole core data, well geophysical data, and tunnel logging data.

[0073] SB2: Establish a dynamic update mechanism. When the deviation between new exploration data and model predictions exceeds the tolerance range, the automatic iterative optimization of the three-dimensional geological model of the deposit is triggered.

[0074] Specifically, real-time access to exploration project data: Establish a data interface to receive newly generated exploration data in real time, including: Borehole core data: borehole coordinates, layer depth, lithological description, ore grade, etc.

[0075] Borehole geophysical data: Curves showing the changes in resistivity, polarizability, temperature, radioactivity, and acoustic velocity as a function of depth, measured in the borehole.

[0076] Tunnel logging data: Records of surrounding rock lithology, fault exposure locations, and mineralization zones during tunnel excavation.

[0077] All data is imported into the system in real time, and the system automatically performs format conversion and spatial coordinate correction to ensure that the coordinate system is consistent with the existing three-dimensional geological model of the ore deposit.

[0078] Dynamic update trigger mechanism: Set a data deviation tolerance (such as the threshold for the difference between the predicted value of the mineralization probability at the borehole mineralization location and the actual mineralization situation, or the threshold for the relative deviation between the resistivity anomaly in the new well logging and the model prediction value).

[0079] When the deviation between newly acquired exploration data and model predictions exceeds the tolerance range (for example, a borehole actually exposes an ore body in an area with a predicted mineralization probability of 0.3), the system automatically identifies the area as a model error zone, triggers the automatic iterative optimization process of the 3D geological model of the deposit, and recalculates the geological interface, physical property distribution, and mineralization domain range of the area to ensure that the model reflects the latest exploration results in real time.

[0080] The steps involved in optimizing the three-dimensional geological model of the ore deposit include: SB3: Referencing geophysical anomaly gradient zones to adjust tectonic surface attitude parameters; SB4: Recalculate the spatial connectivity index of the mineralized domain; SB5: Add virtual borehole constraints for controllable exploration projects.

[0081] Specifically, areas with significant changes in physical properties (such as faults and lithological boundaries) are identified from geophysical data such as gravity and magnetic methods. Based on this, parameters such as the strike and dip angle of faults and folds in the model are adjusted. At the same time, surface geological observations and structural information revealed by boreholes are combined to ensure that the adjusted structures conform to both geophysical anomalies and actual geological conditions.

[0082] The mineralized region is divided into grid cells. The spatial continuity of the mineralized body is determined based on whether adjacent cells belong to the mineralized domain and are separated by structures (such as faults). If the fault is a ore-guiding structure (allowing ore-forming fluids to pass through), the mineralization on both sides is preserved; if it is a barrier structure (preventing mineralization diffusion), the connection is broken. In this way, the problem of "false connections" or "false breaks" in mineralized bodies caused by structural model errors or insufficient data is corrected, and the continuity of mineralization distribution is more accurately reflected.

[0083] In key areas with limited exploration data (such as structural intersections, anomaly centers, or borehole gaps), virtual boreholes are designed. The depth, lithology, grade, and other attributes of the virtual boreholes are reasonably set based on surrounding actual data and model predictions. These virtual data are used as new constraints and integrated into the model adjustment process to supplement geological information in data-blind areas, reduce model uncertainty caused by insufficient boreholes, and make the morphology of geological interfaces and mineralized zones more accurate.

[0084] Also includes: SD1: Establish a multi-objective optimization model to simultaneously optimize exploration costs, prediction accuracy, and engineering risk indicators; SD2: Simulates and outputs exploration engineering combination schemes that satisfy multiple constraints using the Monte Carlo algorithm. Specifically, establish a multi-objective optimization model: Define three core optimization objectives: Exploration costs: Minimize drilling workload, equipment investment, and construction period.

[0085] Prediction accuracy: Maximize the accuracy of mineralization probability prediction and resource estimation.

[0086] Engineering risks: Minimize safety risks during construction (such as water inrush in fault zones and rock burst risks).

[0087] Construct a mathematical model to transform each objective into quantifiable indicators (such as cost in tens of thousands of yuan, accuracy in the rate of agreement between prediction and actual mineralization, and risk in the statistical value of historical accident probability), and set constraints (such as the total exploration budget not exceeding 5 million yuan and the maximum depth of a single borehole not exceeding 1000m).

[0088] Monte Carlo algorithm simulates and generates combinatorial schemes: Using Monte Carlo simulation technology, thousands of exploration engineering combination schemes (such as combinations of different numbers, locations and depths of boreholes) are randomly generated, with each scheme corresponding to a set of cost, accuracy and risk indicators.

[0089] All options are screened, and those that meet the constraints are retained. Pareto optimality analysis is then used to select the set of non-dominated options that achieve the best balance between cost, accuracy, and risk (i.e., options that cannot improve any metric without sacrificing other metrics).

[0090] The final output consists of 10-30 optimal exploration engineering combination schemes, each accompanied by a detailed index comparison table and a three-dimensional deployment diagram, providing decision-makers with diverse options that take into account economy, scientificity and safety.

[0091] The implementation principle of the intelligent mineral exploration prediction method, system, terminal, and medium based on the three-dimensional geological model of mineral deposits in this application is as follows: It integrates three-dimensional geological structure, geophysical field, geochemical anomalies, and ore body characteristic parameters, breaking through the limitations of traditional two-dimensional planar mineral exploration, and constructing a three-dimensional spatial model that more closely resembles the real geological environment, providing a three-dimensional data foundation for mineralization regularity analysis; it combines geological genetic rules with data-driven features, automatically mining hidden mineralization control factors through machine learning algorithms, avoiding the subjectivity of single empirical models and the blindness of data models, significantly improving the accuracy of mineralization prediction under complex geological conditions; it triggers model parameter iteration by judging confidence thresholds, achieving adaptive adjustment of geological boundaries and ore body connection relationships, so that the prediction results continuously approach the actual ore body distribution as exploration data accumulates, reducing the blindness and resource waste of drilling projects.

[0092] This application discloses an intelligent mineral exploration and prediction system based on a three-dimensional geological model of mineral deposits.

[0093] A three-dimensional geological model-based intelligent mineral exploration and prediction system for mineral deposits, including: The data acquisition module is used to acquire and standardize multi-source geological data; The model building module is used to generate three-dimensional geological models of mineral deposits that integrate geological, geophysical, and geochemical features; The intelligent analysis module deploys the trained mineral exploration prediction model and performs three-dimensional mineralization potential analysis. The dynamic optimization module is used to adjust model parameters based on prediction confidence feedback; The decision output module generates an intelligent mineral exploration prediction scheme that includes three-dimensional spatial positioning information. The exploration planning module is used to establish a multi-objective optimization model to simultaneously optimize exploration costs, prediction accuracy, and engineering risk indicators. The Monte Carlo algorithm is used to simulate and output exploration engineering combination schemes that meet multiple constraints.

[0094] This application provides a smart terminal.

[0095] A smart terminal includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed to perform the aforementioned intelligent mineral exploration prediction method based on a three-dimensional geological model of a mineral deposit.

[0096] This application provides a computer-readable storage medium. A computer-readable storage medium storing a computer program that can be loaded by a processor and executed by the above-mentioned intelligent mineral exploration prediction method based on a three-dimensional geological model of a mineral deposit.

[0097] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A three-dimensional geological model-based intelligent mineral exploration and prediction method for mineral deposits, characterized by: include: Acquire multi-source geological data for the target area, including three-dimensional geological structure data, geophysical exploration data, geochemical element data, exploration engineering chemical analysis data, and known ore body characteristic parameters; A three-dimensional geological model of the ore deposit is constructed based on multi-source geological data. The three-dimensional geological model of the ore deposit includes a stratigraphic distribution model, a structural network model, a mineralization spatial model, a geophysical data model, and a geochemical data model. The three-dimensional geological model of the ore deposit is input into the trained intelligent mineral exploration prediction model, which integrates geological genesis rules and data-driven features through machine learning algorithms. Perform mineralization potential analysis: Output a three-dimensional mineralization probability distribution map and resource quantity prediction range based on the intelligent mineral exploration prediction model; If the confidence level of the predicted resource quantity is lower than the preset threshold, optimize the geological boundary parameters and ore body connection relationships of the three-dimensional geological model of the deposit. If the predicted resource quantity meets the preset confidence standard, an intelligent mineral exploration prediction scheme is generated, which includes the location of mineralized target areas and exploration suggestions.

2. The intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to claim 1, characterized in that: The steps for constructing a three-dimensional geological model of the ore deposit include: Three-dimensional inversion calculations are performed on geophysical exploration data to generate three-dimensional distributions of density, magnetic susceptibility, resistivity, and polarizability. A three-dimensional elemental anomaly field was constructed using geochemical elemental data and the Crick interpolation algorithm. By integrating geological structure data, geophysical and geochemical fields, and chemical analysis, a continuous geological interface is generated using implicit modeling techniques.

3. The intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to claim 1, characterized in that: Also includes: Obtain the data set of ore deposit models in historical exploration areas and the corresponding drilling, trenching, pitting, and shallow well verification results; Convolutional neural networks are used to extract the spatial morphological features of ore deposit models, and hybrid training samples are constructed by combining geological genesis knowledge. The regional adaptability of the intelligent mineral exploration prediction model is optimized through transfer learning algorithms.

4. The intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to claim 1, characterized in that: The method further includes: Real-time access to exploration engineering data, including borehole core data, well geophysical data, depth measurement data, and tunnel logging data; A dynamic update mechanism is established so that when the deviation between new exploration data and model predictions exceeds the tolerance range, the automatic iterative optimization of the three-dimensional geological model of the deposit is triggered.

5. The intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to claim 1, characterized in that: The steps for optimizing the three-dimensional geological model of the ore deposit include: Refer to geophysical anomaly gradient zones to adjust tectonic surface attitude parameters; Recalculate the spatial connectivity index of the mineralized domain; Increase virtual borehole constraints for controllable exploration projects.

6. The intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to claim 1, characterized in that: The steps for generating the intelligent mineral exploration prediction scheme include: Extract the three-dimensional spatial coordinates and geometric parameters of high-probability mineralization units; Calculate the probability distribution function of resource quantity in each predicted target area; Generate an optimal exploration deployment scheme that includes a 3D navigation path.

7. The intelligent mineral exploration and prediction method using a three-dimensional geological model of a mineral deposit according to claim 1, characterized in that: Also includes: Establish a multi-objective optimization model to simultaneously optimize exploration costs, prediction accuracy, and engineering risk indicators; The Monte Carlo algorithm is used to simulate and output exploration engineering combination schemes that meet multiple constraints.

8. A three-dimensional geological model-based intelligent mineral exploration and prediction system, characterized by: include, The data acquisition module is used to acquire and standardize multi-source geological data; The model building module is used to generate three-dimensional geological models of mineral deposits that integrate geological, geophysical, and geochemical features; The intelligent analysis module deploys the trained mineral exploration prediction model and performs three-dimensional mineralization potential analysis. The dynamic optimization module is used to adjust model parameters based on prediction confidence feedback; The decision output module generates an intelligent mineral exploration prediction scheme that includes three-dimensional spatial positioning information.

9. A smart terminal, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any one of claims 1 to 7 for intelligent mineral exploration and prediction using a three-dimensional model of a mineral deposit.

10. A computer-readable storage medium, characterized in that, The system stores a computer program capable of being loaded by a processor and executing the intelligent mineral exploration prediction method based on a three-dimensional model of a mineral deposit as described in any one of claims 1 to 7.

Images