Metal resource evaluation method and system based on geological exploration

Abstract

The invention belongs to the technical field of mineral resource evaluation, and provides a metal resource evaluation method and system based on geological exploration, and the method comprises the steps: obtaining and fusing the multi-source geological exploration data of a target region, and constructing a three-dimensional geological model and a prior knowledge base; based on the three-dimensional geologic model and the prior knowledge base, establishing and operating a numerical simulation model for describing the metallogenic process of the target area, and outputting a metallogenic parameter field; calibrating the metallogenic parameter field and the numerical simulation model by using multi-source geological exploration data, and generating an assimilated geological process model; generating a three-dimensional resource model based on the geological structure represented by the geological process model in combination with the multi-source geological exploration data, and calculating to obtain probability distribution of resource quantity; the sampling space of the economic parameters is dynamically determined according to the probability distribution of the resource quantity, cash flow simulation is conducted on the three-dimensional resource model through the sampling space, the probability distribution of the economic evaluation indexes is finally output, and the accuracy and reliability of metal resource evaluation are improved.

Description

Technical Field

[0001] This invention relates to the field of mineral resource evaluation technology, and in particular to a method and system for evaluating metal resources based on geological exploration. Background Technology

[0002] Mineral resource assessment is a core component of mineral exploration and development decision-making, and the accuracy of its results directly impacts investment risk, mine design, and resource utilization efficiency. Traditional metal resource assessment methods primarily rely on geostatistics (such as Kriging) to interpolate and extrapolate limited exploration data (such as borehole data) to estimate resource quantity and grade. These methods are essentially static statistical models based on spatial correlation, and have the following significant limitations: Traditional metal resource assessment methods typically treat ore bodies as static entities, and their estimation results rely excessively on existing exploration projects, resulting in inaccurate predictions for areas with low exploration levels or deep areas. Furthermore, they are mostly limited to providing deterministic values ​​or simple sensitivity analyses, making it difficult to systematically quantify the uncertainties across the entire chain from geological understanding to economic value. This leads to the inability to effectively transmit geological risks to economic assessments. At the same time, geological modeling, resource estimation, and economic evaluation are disconnected from each other, and economic parameters fail to be dynamically correlated with the uncertainties of the geological model.

[0003] Therefore, it is necessary to provide a method and system for assessing metal resources based on geological exploration to solve the above-mentioned technical problems. Summary of the Invention

[0004] To address the aforementioned technical issues, this invention provides a method and system for assessing metal resources based on geological exploration. By constructing an assessment system driven by geological process mechanisms, integrating multi-source data and knowledge, and quantifying uncertainty, it achieves an improvement in metal resource assessment from static description to dynamic prediction, and from deterministic estimation to probabilistic risk quantification. This provides more scientific and reliable technical support for mineral resource exploration deployment, asset valuation, and investment decisions.

[0005] This invention provides a method for assessing metallic resources based on geological exploration, the method comprising the following steps: Acquire and integrate multi-source geological exploration data of the target area to construct a three-dimensional geological model and a prior knowledge base; Based on the aforementioned three-dimensional geological model and prior knowledge base, a numerical simulation model describing the mineralization process in the target area is established and run, and the mineralization parameter field is output. Using the multi-source geological exploration data, the ore-forming parameter field and the numerical simulation model are calibrated to generate an assimilated geological process model; Based on the geological structure characterized by the geological process model, combined with the multi-source geological exploration data, a three-dimensional resource model is generated using geostatistical methods, and the probability distribution of resource quantity is calculated. The sampling space of economic parameters is dynamically determined based on the probability distribution of the resource quantity, and the cash flow simulation of the three-dimensional resource model is performed using the sampling space to finally output the probability distribution of economic evaluation indicators.

[0006] Preferably, the acquisition and fusion of multi-source geological exploration data of the target area to construct a three-dimensional geological model and prior knowledge base includes: Acquire multi-source geological exploration data of the target area and preprocess it to form a fused dataset; The fused dataset is input into a pre-trained feature extraction network, which outputs a comprehensive feature field. Based on geological prior rules and the comprehensive feature field, the prior knowledge base is generated through a knowledge graph construction method. Based on the geological constraints defined in the prior knowledge base, a three-dimensional geological interpretation of the comprehensive feature field is performed to construct a three-dimensional geological model of the target area.

[0007] Preferably, the step of establishing and running a numerical simulation model describing the mineralization process of the target area based on the three-dimensional geological model and prior knowledge base, and outputting a mineralization parameter field, includes: Based on the geometric structure and physical property parameters of the three-dimensional geological model, and combined with the mechanistic constraints defined in the prior knowledge base, an initial numerical simulation model describing the coupled thermal-fluid-mechanical-chemical mineralization process is constructed. Target observation data associated with the mineralization process are extracted from the multi-source geological exploration data. The target observation data are compared with the preliminary output results of the initial numerical simulation model. Based on the comparison results, the parameters of the initial numerical simulation model are iteratively optimized to generate a calibrated numerical simulation model. The numerical simulation model is run to dynamically simulate the mineralization process and output a mineralization parameter field, wherein the mineralization parameter field includes at least one of a temperature field, a pressure field, a chemical substance concentration field, and a fluid velocity field.

[0008] Preferably, the step of calibrating the ore-forming parameter field and the numerical simulation model using the multi-source geological exploration data to generate an assimilated geological process model includes: The numerical simulation model is run based on the simulation parameters to generate a corresponding simulated mineralization parameter field, and simulated observation data is extracted from the simulated mineralization parameter field. Calculate the mismatch between the target observation data and the simulated observation data, and update the simulation parameters using the mismatch. The generation of the simulated observation data and the updating of the simulation parameters are executed iteratively until the mismatch is less than a preset threshold. The corresponding simulation parameters and the optimized numerical simulation model are then defined together as the geological process model.

[0009] Preferably, the geological structure characterized by the geological process model, combined with the multi-source geological exploration data, is used to generate a three-dimensional resource model using geostatistical methods, and the probability distribution of resource quantity is calculated, including: From the geological process model, extract a structural model that characterizes the three-dimensional geological structure and lithofacies distribution in the final state after the mineralization period; Using the structural model as the training image required for geostatistical methods, and combining it with the metal precipitation data in the ore-forming parameter field, structural constraints and soft data constraints in three-dimensional space are defined. Using the sample grade data from the multi-source geological exploration data as hard data, under the constraints of the structure and soft data, geostatistical methods are applied to perform random simulations to generate multiple three-dimensional grade models with equal probability. The set of all the three-dimensional grade models is defined as the three-dimensional resource model, and the amount of metal resources in each model is counted to calculate the probability distribution of the resource amount.

[0010] Preferably, the step of dynamically determining the sampling space of economic parameters based on the probability distribution of the resource quantity, and using the sampling space to perform cash flow simulation on the three-dimensional resource model, ultimately outputting the probability distribution of economic evaluation indicators, includes: Based on the probability distribution of the resource quantity, the probability distribution of at least one economic parameter is determined through predefined economic-resource coupling rules; Random sampling is performed based on the probability distribution of the economic parameters to generate a set of economic parameter scenarios. Then, in combination with the three-dimensional grade models in the defined three-dimensional resource model, a full life-cycle cash flow simulation is performed. Summarize all the obtained cash flow simulation results, calculate and output the probability distribution of economic evaluation indicators.

[0011] Preferably, the probability distribution of economic evaluation indicators is calculated and output from all the summarized cash flow simulation results, including: Based on the coupled simulation results of each three-dimensional grade model and economic parameter scenario, a probabilistic statistical analysis is performed. The probabilistic statistical analysis involves using the kernel density estimation method to fit the probability density distribution of the net present value and calculating the empirical distribution of the internal rate of return. Based on the probability density distribution and the empirical distribution, calculate at least one of the risk indicators among the value at risk, conditional value at risk, and break-even probability. The final output is a structured probability report containing the probability density distribution, empirical distribution, and risk indicators.

[0012] This invention also provides a geological exploration-based metal resource assessment system for executing a geological exploration-based metal resource assessment method, the assessment system comprising: The fusion modeling module is used to acquire and fuse multi-source geological exploration data of the target area to construct a three-dimensional geological model and a prior knowledge base; The mineralization simulation module is used to establish and run a numerical simulation model describing the mineralization process of the target area based on the three-dimensional geological model and prior knowledge base, and output the mineralization parameter field. The calibration and assimilation module is used to calibrate the ore-forming parameter field and the numerical simulation model using the multi-source geological exploration data, and generate an assimilated geological process model. The resource distribution calculation module is used to generate a three-dimensional resource model based on the geological structure characterized by the geological process model, combined with the multi-source geological exploration data, using geostatistical methods, and to calculate the probability distribution of resource quantity. The economic assessment module is used to dynamically determine the sampling space of economic parameters based on the probability distribution of the resource quantity, and to use the sampling space to simulate cash flow in the three-dimensional resource model, and finally output the probability distribution of economic assessment indicators.

[0013] Compared with related technologies, the metal resource assessment method and system based on geological exploration provided by this invention has the following beneficial effects: This invention achieves a fundamental improvement from static description to dynamic prediction and from deterministic estimation to probabilistic risk quantification by constructing an evaluation system driven by geological process mechanisms, integrating multi-source data and knowledge, and quantifying uncertainty. Specifically, it enhances the scientific prediction capability for resources in exploration-blank areas by introducing four-dimensional numerical simulation of mineralization processes; it quantifies the uncertainty across the entire chain from geological understanding to economic value by adopting probabilistic resource modeling and dynamic coupling with economic parameters; it deeply integrates multi-source heterogeneous data and prior geological knowledge by utilizing multimodal deep learning and knowledge graph technology; and the final output is a structured probabilistic report containing indicators such as risk value, providing refined and reliable support for resource exploration and investment decisions based on risk trade-offs. Attached Figure Description

[0014] Figure 1 A flowchart illustrating a method for assessing metal resources based on geological exploration, provided by this invention; Figure 2 This is a schematic diagram of the module structure of a metal resource assessment system based on geological exploration provided by the present invention. Detailed Implementation

[0015] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the drawings, not all structures. Moreover, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0016] It should also be noted that, for ease of description, the accompanying drawings show only the parts relevant to the invention and not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. The process may correspond to a method, function, procedure, subroutine, subroutine, etc.

[0017] Example 1 This invention provides a method for assessing metal resources based on geological exploration, with reference to... Figure 1 As shown, the evaluation method includes the following steps: S1: Acquire and integrate multi-source geological exploration data of the target area to construct a three-dimensional geological model and prior knowledge base.

[0018] Specifically, step S1 includes the following steps: S11: Acquire multi-source geological exploration data of the target area and preprocess it to form a fused dataset.

[0019] In this embodiment, the specific implementation process of this step is as follows: First, multi-source geological exploration data were collected from the target area, including: 1:50,000 scale geological map vector data; high-precision ground magnetic survey data (point spacing of 50 meters); 1:50,000 scale soil geochemical survey data (sampling density of 1 point / square kilometer); and borehole data (including wellhead coordinates, inclination data, lithology logging, and chemical analysis grade data).

[0020] Subsequently, data preprocessing was performed: In ArcGIS 10.8 software, its "Projection Transformation" tool was used to convert the coordinate system of all spatial data to the UTMZone 50N coordinate system of the target area. For magnetic data, the "Diurnal Variation Correction" module of Geosoft Oasis Montaj software was used to correct the data by inputting local diurnal variation station data. For geochemical data, outliers were removed using the commonly used statistical method of "mean ± 3 standard deviations". For borehole data, the "Minimum Curvature Method" was used to calculate well tracks to ensure the accuracy of the sample's spatial location.

[0021] Finally, to form a fused dataset, the processed point data was interpolated into a grid using the "ordinary kriging method". The variogram model was selected as a spherical model based on the spatial characteristics of the data to generate a grid file with a resolution of 50 meters × 50 meters. This grid file was then imported into 3D geological modeling software (such as LeapfrogGeo) to form a unified fused dataset that can be used for 3D modeling.

[0022] S12: Input the fused dataset into the pre-trained feature extraction network and output the comprehensive feature field.

[0023] In this embodiment, the specific implementation process of this step is as follows: First, a feature extraction network is constructed, which is based on the U-Net architecture. The encoder part contains four downsampling stages, each of which uses two 3×3 convolutional layers (with the number of convolutional kernels being 64, 128, 256, and 512 respectively) and the ReLU activation function, followed by a 2×2 max pooling layer; the decoder part performs symmetrical upsampling.

[0024] This network is implemented in PyTorch 1.12. The pre-training process is as follows: The publicly available large-scale geological dataset "MineralNet" (containing 100,000 multispectral remote sensing geological annotation samples) is used as the training set. The classification task is the pre-training objective, and the Adam optimizer is used (initial learning rate set to 0.001, training for 100 epochs). After pre-training, it is transferred to this task. The fused dataset (grid data) obtained in S11 is resampled to a uniform size of 256×256 and normalized. The input region is used as the network's input. The network's output layer uses a sigmoid activation function to generate a comprehensive feature field of the same size as the input. Each pixel value in this feature field represents the comprehensive probability or feature intensity that the location belongs to a favorable mineralization area. The comprehensive feature field of the target region can be obtained through forward propagation.

[0025] S13: Based on geological prior rules and the comprehensive feature field, the prior knowledge base is generated through a knowledge graph construction method.

[0026] In this embodiment, the specific implementation process of this step is as follows: The knowledge graph was constructed using the Protégé ontology editing tool. First, the core geological entity classes (such as strata, rock mass, and faults) and their relationships (such as hasContactWith and controls) were manually defined.

[0027] Then, the mineralization patterns of typical deposits are extracted from the USGS deposit model database and manually entered in the form of IF-THEN rules, such as "IF the presence of intermediate-acidic rock mass AND the presence of ring structure THEN inferred to be a porphyry copper mineralization system".

[0028] Next, the comprehensive feature field obtained from S12 is used in the Python environment to perform unsupervised clustering using the KMeans algorithm of the Scikit-learn library (with the number of clusters set to 5), dividing the feature space into 5 geological units.

[0029] Finally, using the Neo4j graph database, the defined ontology and rules are used as nodes and relationships to create a graph, and the clustering results (geological unit labels for each grid cell) are stored as spatial attributes of the nodes. This generates a structured, queryable prior knowledge base containing geological rules and spatial distribution characteristics.

[0030] S14: Based on the geological constraints defined in the prior knowledge base, perform a three-dimensional geological interpretation of the comprehensive feature field and construct a three-dimensional geological model of the target area.

[0031] In this embodiment, the specific implementation process of this step is as follows: Three-dimensional geological modeling is carried out. First, the rules about stratigraphic sequences in the knowledge base generated by S13 (such as "strata A covers strata B") are converted into trend constraints for the modeling software.

[0032] Then, the fused datasets formed in S11 (such as geochemical anomalies and magnetic anomalies) and the comprehensive feature field obtained in S12 are used as input data for the "implicit modeler". The "Discrete Smoothing Interpolation (DSI)" algorithm is selected as the core modeling algorithm. During the calculation process, the constraints in the knowledge base, such as stratigraphic contact relationships and fault cutting relationships, are set as "boundary conditions" to force the model to satisfy the corresponding topological relationships when passing through these known geological control points.

[0033] By adjusting the tension factor and continuity threshold of the DSI algorithm, the generated 3D geological interface is made to both fit the input data and conform to geological laws. Finally, a 3D geological model composed of a triangular mesh, containing the spatial morphology and contact relationships of various geological bodies, is output.

[0034] S2: Based on the aforementioned three-dimensional geological model and prior knowledge base, establish and run a numerical simulation model describing the mineralization process in the target area, and output a mineralization parameter field.

[0035] Specifically, step S2 includes the following steps: S21: Based on the geometric structure and physical property parameters of the three-dimensional geological model, and combined with the mechanistic constraints defined in the prior knowledge base, construct an initial numerical simulation model describing the coupled thermal-fluid-mechanical-chemical mineralization process.

[0036] In this embodiment, the specific implementation process of this step is as follows: an initial model is constructed using a numerical simulation platform that supports multiphysics coupling.

[0037] First, the 3D geological model generated by S14 was imported into the platform in STL format. It was then discretized into a computational mesh using the mesh generation module, with a quality control standard set to a maximum mesh skewness of less than 0.8. In the material property definition module, physical property parameters were set according to the stratigraphic lithology zones: for granite bodies, a thermal conductivity of 2.9 W / (m·K) and a porosity of 0.08 were set; for fault zones, a permeability of 1×10⁻⁶ was set. -13 m² and porosity 0.15 are parameters derived from the standard value range of the corresponding lithology in the USGS rock physics database.

[0038] Next, a fully coupled thermo-fluid-mechanical-chemical (THMC) governing equation set was established: Darcy flow and Boussinesq approximation were enabled in the fluid flow module; convection-diffusion coupling was enabled in the heat transport module; and a precipitation kinetics term based on the Arrhenius equation was added to the chemical transport module, with the activation energy set to 65 kJ / mol (based on mineral deposit data in the S13 knowledge base). Based on the regional paleogeothermal gradient of 30°C / km, the initial temperature field was set to a linear distribution (15°C at the surface and 165°C at 5 km from the basement). Finally, a fully implicit solver was selected, with a relative tolerance set to 1 × 10⁻⁶. -5 The initial numerical model suitable for simulating mineralization processes was constructed.

[0039] S22: Extract target observation data associated with the mineralization process from the multi-source geological exploration data, compare the target observation data with the preliminary output results of the initial numerical simulation model, and iteratively optimize the parameters of the initial numerical simulation model based on the comparison results to generate a calibrated numerical simulation model.

[0040] In this embodiment, the specific implementation process of this step is as follows: The ensemble smoothing data assimilation algorithm (ES-MDA) was used for model calibration. First, the target observation dataset was extracted from the borehole data, including homogenization temperature data for 20 fluid inclusions (280±15°C) and sulfur isotope S values ​​for 15 fluid inclusions (-2.5‰ to +3.5‰). Algorithm parameters were set as follows: 4 iterations and a perturbation coefficient sequence. (Based on the geometric decrease criterion), generate a set of 100 initial parameters.

[0041] The implementation process is as follows: A main control program is written to implement the ES-MDA algorithm loop, batch-calling the numerical simulation platform to perform forward modeling calculations in each iteration; simulated temperature and component concentration values ​​are extracted at the observation points; the weighted root mean square error is calculated (temperature weight 0.7, component weight 0.3); the L-BFGS optimization algorithm is used to update the permeability field and heat source intensity parameters. Optimization is terminated when the mismatch reduction rate is less than 5% for two consecutive iterations, and the calibrated model parameter set is output.

[0042] S23: Run the numerical simulation model to dynamically simulate the mineralization process and output the mineralization parameter field, wherein the mineralization parameter field includes at least one of the temperature field, pressure field, chemical substance concentration field and fluid velocity field.

[0043] In this embodiment, the specific implementation process of this step is as follows: Set the transient simulation parameters: time range 0-5 Ma (covering the main mineralization period), step size 50,000 years (ensuring Coulomb number < 0.8). Run the calibrated numerical model and dynamically solve the THMC coupled equations. After the simulation is completed, output the three-dimensional parameter field data at the final time (5 Ma), including: temperature field (°C), pressure field (MPa), ore-forming element concentration field (ppm), and flow velocity field (m / s).

[0044] Output requirements: Data format should be VTK standard, including node coordinates, parameter values, and time step information; spatial resolution should be consistent with the geological model (50m grid); additional spatiotemporal evolution sequences of temperature and concentration fields should be output. The final generated parameter field file can be directly transferred to the subsequent resource modeling module via API interface, forming a complete analysis chain.

[0045] S3: Using the multi-source geological exploration data, calibrate the ore-forming parameter field and the numerical simulation model to generate an assimilated geological process model.

[0046] Specifically, step S3 includes the following steps: S31: Run the numerical simulation model based on the simulation parameters to generate the corresponding simulated mineralization parameter field, and extract simulated observation data from the simulated mineralization parameter field.

[0047] In this embodiment, the specific implementation process of this step is as follows: The S22-calibrated numerical simulation model was invoked via the COMSOLAPI interface. Simulation parameters were set to the best estimates for the current iteration, including permeability field distribution and heat source intensity parameters. A transient simulation was run with a time range of 0-5 Ma and an output time step of 0.1 Ma, generating simulation results for 50 time steps. After the simulation, the spatial coordinates of the 20 fluid inclusion sampling points and 15 sulfur isotope sampling points used in S22 were analyzed. The temperature and sulfur isotope values ​​at corresponding locations were extracted from the simulation results using cubic spline interpolation. The extracted data included the temperature evolution sequence and sulfur isotope composition sequence for each sampling point at each time step. Finally, a four-dimensional data table containing the sampling point number, spatial coordinates, time step, and corresponding parameter values ​​was generated and stored in HDF5 format as input data for subsequent mismatch calculations.

[0048] S32: Calculate the mismatch between the target observation data and the simulated observation data, and update the simulation parameters using the mismatch.

[0049] In this embodiment, the specific implementation process of this step is as follows: First, the simulated observation data generated in S31 and the actual observation data used in S22 are read. For temperature data, the root mean square error (RMSE) between the simulated temperature sequence and the actual measured temperature value at each sampling point is calculated. For sulfur isotope data, the mean absolute error (MAE) between the simulated and measured values ​​of sulfur isotopes is calculated. The total mismatch is obtained by weighted summation of the two error terms, with the weighting coefficients set to 0.7 for temperature error and 0.3 for sulfur isotope error. The simulation parameters are updated using the L-BFGS optimization algorithm. The `optimize.L-BFGS-B` function is called in the SciPy 1.8.0 library, with a maximum iteration count of 100 and a gradient tolerance of 1 × 10⁻⁶. -6 It automatically adjusts the distribution of permeability field and heat source intensity parameters by minimizing the total mismatch.

[0050] S33: Iteratively execute the generation of the simulated observation data and the update of the simulation parameters until the mismatch is less than a preset threshold, and define the corresponding simulation parameters and the optimized numerical simulation model together as the geological process model.

[0051] In this embodiment, the specific implementation process of this step is as follows: Establish a complete iterative optimization loop. Set the convergence threshold to a relative change rate of less than 2% for the total mismatch, and the maximum number of iterations to 20. In specific implementation, execute steps S31 and S32 sequentially in the While loop: first, run the numerical simulation under the current parameters and extract the observation data (S31); then calculate the current mismatch and update the parameters (S32). Record the mismatch value after each iteration. When the relative change rate of the mismatch is less than 2% for three consecutive iterations, it is considered converged.

[0052] If convergence is not achieved after reaching the maximum number of iterations, the parameter set with the smallest mismatch in the historical iterations is selected as the final result. After iteration, the final optimized permeability field distribution, heat source intensity parameters, and corresponding numerical simulation models are output, collectively defined as the assimilated geological process model. This model is saved in both COMSOL model file format (.mph) and parameter matrix file (.mat) to ensure the model's reproducibility and portability.

[0053] S4: Based on the geological structure characterized by the geological process model, and combined with the multi-source geological exploration data, a three-dimensional resource model is generated using geostatistical methods, and the probability distribution of resource quantity is calculated.

[0054] Specifically, step S4 includes the following steps: S41: Extract a structural model from the geological process model that represents the three-dimensional geological structure and lithofacies distribution in the final state after the mineralization period.

[0055] In this embodiment, the specific implementation process of this step is as follows: First, the geological process model generated by S33 (saved as a .mph file) was called through the API interface of COMSOL Multiphysics 6.0 software. This model contains the thermo-fluid-mechanical-chemical coupling results under the final state after the mineralization period (i.e., simulation time 5 Ma). In the COMSOL post-processing module, the "Export Geometry" function was selected to extract the boundary surfaces of three-dimensional geological structures (such as faults and stratigraphic interfaces) and lithofacies distributions (such as granite bodies and alteration zones), and the output format was an STL file.

[0056] Meanwhile, the "Grid Data Export" tool was used to export the physical property parameter field (such as lithological coding) in VTK format, with the grid resolution consistent with the original geological model (50m×50m×50m).

[0057] Next, the PyVista library was used in the Python 3.9 environment to read the VTK file, and different lithofacies units were separated by threshold filtering (e.g., retaining units with lithofacies coding values ​​in the range of 1-5). The MarchingCubes algorithm was then used to generate three-dimensional isosurfaces for each lithofacies.

[0058] Finally, the extracted structural model is stored in a hierarchical data format (HDF5), which includes geometric grids, lithofacies labels, and spatial coordinate attributes, to facilitate subsequent geostatistical analysis.

[0059] S42: Using the structural model as the training image required for geostatistical methods, and combining it with the metal precipitation data in the ore-forming parameter field, define structural constraints and soft data constraints in three-dimensional space.

[0060] In this embodiment, the specific implementation process of this step is as follows: First, the structural model (HDF5 format) extracted from S41 was imported into the geostatistics software SGeMS2.5 as the training image required for Multipoint Geostatistics (MPS). In the "Training Image Manager" of SGeMS, the grid size of the training image was set to 50m × 50m × 50m, and the geological structure trend was defined (e.g., the anisotropy ratio of lithofacies bodies was 1:2:1, and the corresponding major axis direction was the direction of the regional tectonic line).

[0061] Then, metal precipitation data (such as copper concentration field) are extracted from the ore-forming parameter field (VTK format) output by S23, and normalized to a normal value using MinMaxScaler from the Scikit-learn library in Python. The interval serves as a soft data constraint. Soft data is imported through the "Soft Probability" module of SGeMS and converted into a conditional probability field (e.g., areas with concentration values ​​above 100 ppm are assigned a mineralization probability of 0.8). Structural constraints are defined using geological patterns (such as lithofacies contact relationships) from the training images. The search template size is set to 5×5×5 grid cells using the "SNESIM" algorithm template in SGeMS, and rotation invariance is enabled to capture complex structures.

[0062] Finally, a constraint file (XML format) containing the structure training image and soft data probability field is generated for subsequent stochastic simulations.

[0063] S43: Using the sample grade data from the multi-source geological exploration data as hard data, under the structural constraints and soft data constraints, geostatistical methods are applied to perform random simulation to generate multiple three-dimensional grade models with equal probability.

[0064] In this embodiment, the specific implementation process of this step is as follows: First, borehole sample analysis grade data (e.g., copper grade, in ppm) are extracted from the S11 fusion dataset as hard data conditions. In the SGeMS software, the hard data, including sample coordinates (X, Y, Z) and grade values, is loaded using the "Data Import" tool, and the spatial distribution of the data is checked (e.g., variogram analysis to ensure the absence of systematic bias).

[0065] Next, a stochastic simulation was performed using the Sequential Gaussian Simulation (SGS) method: in the simulation parameter settings, a spherical model was selected as the variogram model (nuclear value 0.1, sill value 1.0, range 200 meters), and training images and soft data constraints generated by S42 were integrated. During the simulation, 100 equally probable realizations (i.e., random seed number from 1 to 100) were set, and the grid size of each realization was consistent with that of the training images (50-meter resolution).

[0066] During simulation, the SGS algorithm sequentially traverses each grid node, calculating the conditional distribution based on hard data, training image structure, and soft data probabilities, and randomly samples from it to generate grade values. Ultimately, it outputs 100 three-dimensional grade models (in GSLIB binary file format), each containing a grade estimate for each grid cell, representing the uncertainty in resource distribution.

[0067] S44: Define the set of all the three-dimensional grade models as the three-dimensional resource model, and count the amount of metal resources in each model to calculate the probability distribution of the resource amount.

[0068] In this embodiment, the specific implementation process of this step is as follows: First, the 100 three-dimensional grade models (GSLIB files) generated by S43 are collectively defined as a three-dimensional resource model. In the Python environment, the Pandas and NumPy libraries are used to read each grade model file, and the metal resource quantity of each model is calculated based on the grid cell volume (50m×50m×50m=125000 cubic meters) and rock density (default value 2.7 tons / cubic meter, which can be adjusted according to lithofacies).

[0069] The specific calculation is as follows: For each grid cell, the formula for calculating the metal content is:

[0070] In the formula Indicates the amount of metal. Indicates grade value, Represents the volume of a mesh cell. Indicates rock density, This represents the unit conversion factor from ppm to mass fraction.

[0071] Then, summing all grid cells yields the total resource quantity.

[0072] Next, statistical analysis was performed on the total resource quantities of the 100 implementations: the normal distribution probability density function was fitted using the `norm.fit` function from the SciPy library, and a cumulative distribution function (CDF) was generated. Simultaneously, descriptive statistics of the resource quantities (such as P10, P50, and P90 quantiles) were calculated, and histograms and kernel density estimation plots were plotted using Matplotlib. Finally, a resource quantity probability distribution report (PDF format) was output, including a list of resource quantities for each implementation, probability distribution curves, and key quantiles, providing input for economic evaluation.

[0073] S5: Dynamically determine the sampling space of economic parameters based on the probability distribution of the resource quantity, and use the sampling space to simulate cash flow in the three-dimensional resource model, and finally output the probability distribution of economic evaluation indicators.

[0074] Specifically, step S5 includes the following steps: S51: Based on the probability distribution of the resource quantity, determine the probability distribution of at least one economic parameter through a predefined economic-resource coupling rule.

[0075] In this embodiment, the specific implementation process of this step is as follows: First, in the Python 3.9 environment, read the resource quantity probability distribution report (PDF format) generated by S44 and extract the resource quantity values ​​(unit: tons) corresponding to the P10, P50, and P90 quantiles.

[0076] Subsequently, a parameter mapping relationship is established based on predefined economic-resource coupling rules: for mining cost parameters, a linear regression model (based on historical mine databases, such as SNLMetals & Mining) is used to establish the relationship between resource scale and unit mining cost. For example, when the resource quantity changes from P10 to P90, the unit mining cost is linearly adjusted proportionally within the range of $28-35 / ton. For the metal price parameters, a log-normal distribution was used for simulation. The mean and standard deviation were calculated based on the historical copper price data of the London Metal Exchange (LME) over the past 5 years, and the skewness characteristics were fitted by the Johnson-SU distribution.

[0077] All coupling rules are stored in YAML configuration files, containing variable relationships, data sources, and distribution types. Finally, Monte Carlo simulation is used to generate 10,000 random samples using the NumPy library, generating a corresponding probability distribution for each economic parameter (such as mining cost, metal price, and beneficiation recovery rate). The output is a probability distribution file in HDF5 format, containing the sampled values ​​of each parameter, statistical characteristics, and distribution goodness-of-fit indices.

[0078] S52: Random sampling is performed based on the probability distribution of the economic parameters to generate a set of economic parameter scenarios, and then the full life cycle cash flow simulation is performed by combining the three-dimensional quality models in the defined three-dimensional resource model.

[0079] In this embodiment, the specific implementation process of this step is as follows: First, read the probability distribution of each economic parameter from the HDF5 probability distribution file generated by S51, and use the Latin hypercube sampling method (LHS) to generate 1000 sets of economic parameter scenarios to ensure that the sampling space uniformly covers the distribution of each parameter.

[0080] Each scenario includes key parameters such as mining costs, metal prices, discount rate (set to 8%), and ore beneficiation recovery rate (default 85%±5%). Simultaneously, 100 three-dimensional grade models (GSLIB format) generated by S43 are loaded as input for geological uncertainty.

[0081] Cash flow simulation was implemented using a self-developed Python module, with the simulation period set at 20 years for the entire life cycle of the mine (including a 2-year construction period and an 18-year production period).

[0082] For each combination of economic parameter scenario and grade model (a total of 100,000 simulations), the following calculations are performed sequentially: Based on the spatial distribution characteristics of the grade model, the Lerchs-Grossmann algorithm is used to optimize the open-pit mining boundary and determine the annual mining sequence; the annual operating cost is calculated based on the mining cost parameters; the annual sales revenue is calculated based on the metal price parameters and the beneficiation recovery rate; finally, the net cash flow for each year is calculated using the discount rate, and the net present value (NPV) is obtained by summing them up.

[0083] All simulation results are stored in a Parquet file as a structured array, containing the input parameters for each simulation, annual cash flows, and the final NPV value.

[0084] S53: Summarize all the obtained cash flow simulation results, calculate and output the probability distribution of economic evaluation indicators.

[0085] In this embodiment, step S53 is performed by the following steps: S53a: Based on the coupled simulation results of each three-dimensional grade model and economic parameter scenario, a probabilistic statistical analysis is performed, wherein the probabilistic statistical analysis is to fit the probability density distribution of net present value using the kernel density estimation method and calculate the empirical distribution of internal rate of return.

[0086] In this embodiment, the specific implementation process of this step is as follows: First, the Pandas library was used in the Python environment to read the Parquet format simulation results file generated in step S52. This file contains 100,000 sets of simulation data (corresponding to combinations of 100 grade models and 1,000 economic parameter scenarios). The net present value data was preprocessed, including outlier removal (using the Tukey method, defining outliers as data below Q1 - 1.5 IQR or above Q3 + 1.5 IQR) and data standardization.

[0087] Next, the probability density distribution of NPV is fitted using the kernel density estimation method: the gaussian_kde function from the SciPy library is used, and the bandwidth is automatically calculated using the Silverman criterion, the formula of which is:

[0088] in The standard deviation of the sample is 1. This represents the sample size. By setting the number of grid points to 1000, a smooth probability density curve is generated within the NPV range.

[0089] For calculating the internal rate of return (IRR), a numerical iterative method (Newton-Raphson algorithm) is used for each cash flow series to find the discount rate that makes the NPV zero. The convergence condition for the iteration is set to a relative error of less than 1 × 10⁻⁶. 6 .

[0090] Finally, the cumulative probability of IRR is calculated using the empirical distribution function with a step size of 0.1%, generating a complete empirical distribution table. The goodness of fit of the distribution is then verified by the Kolmogorov-Smirnov test.

[0091] S53b: Based on the probability density distribution and the empirical distribution, calculate at least one of the following risk indicators: value at risk, conditional value at risk, and break-even probability.

[0092] In this embodiment, the specific implementation process of this step is as follows: First, based on the NPV probability density distribution obtained from S53a, the Value at Risk (VaR) is calculated at a 5% significance level. Specifically, the NPV simulation results are sorted in ascending order, and the NPV value corresponding to the 5th percentile is taken as VaR. 0.05 =NPV [n×0.05] , where m is the total number of simulations.

[0093] The conditional value at risk is calculated by taking the arithmetic mean of all NPV values ​​below VaR, using the following formula: ; For the break-even probability, the ratio of the number of simulations with an NPV greater than zero to the total number of simulations is calculated. .

[0094] Simultaneously, other key risk indicators are calculated, including downside standard deviation (considering only the standard deviation of negative NPV), skewness, and kurtosis, among other distribution shape indicators. All calculations are performed using NumPy array operations, employing a block processing strategy for large-scale data to ensure computational efficiency. The final result is a detailed report containing the values ​​of each risk indicator and explanations of the calculation process.

[0095] S53c: The final output is a structured probability report containing the probability density distribution, empirical distribution, and risk indicators.

[0096] In this embodiment, the specific implementation process of this step is as follows: First, a comprehensive report generator is created using Python's ReportLab and Matplotlib libraries. The report structure consists of three parts: The first part is the visualization of probability distribution, including the NPV kernel density estimation curve (with dual Y-axis to display probability density and cumulative probability), the IRR empirical distribution plot (with confidence intervals), and a three-dimensional scatter plot to show the interaction between the grade model and economic parameters. The second part is a summary table of risk indicators, which lists the values, confidence intervals and statistical significance of indicators such as VaR, CVaR and break-even probability in tabular form. The third part is the sensitivity analysis, which uses Tornado plots to show the degree of influence of each economic parameter on NPV.

[0097] The report outputs in both PDF and HTML formats. The HTML format supports interactive queries and utilizes the Plotly library to create dynamic charts (such as displaying specific values ​​on mouse hover). Simultaneously, a machine-readable JSON summary file is generated, containing structured data for all distribution parameters and risk indicators. Finally, version control records the report generation parameters to ensure reproducibility, and all output files are archived by project number and timestamp.

[0098] Example 2 This invention also provides a metal resource assessment system based on geological exploration, for implementing a metal resource assessment method based on geological exploration, with reference to... Figure 2 As shown, the evaluation system includes: The fusion modeling module 100 is used to acquire and fuse multi-source geological exploration data of the target area to construct a three-dimensional geological model and a prior knowledge base.

[0099] The mineralization simulation module 200 is used to establish and run a numerical simulation model describing the mineralization process of the target area based on the three-dimensional geological model and prior knowledge base, and output the mineralization parameter field.

[0100] The calibration and assimilation module 300 is used to calibrate the ore-forming parameter field and the numerical simulation model using the multi-source geological exploration data, and generate an assimilated geological process model.

[0101] The resource distribution calculation module 400 is used to generate a three-dimensional resource model based on the geological structure characterized by the geological process model, combined with the multi-source geological exploration data, using geostatistical methods, and to calculate the probability distribution of resource quantity.

[0102] The economic assessment module 500 is used to dynamically determine the sampling space of economic parameters based on the probability distribution of the resource quantity, and to use the sampling space to simulate cash flow of the three-dimensional resource model, and finally output the probability distribution of economic assessment indicators.

[0103] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0104] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0105] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A method for assessing metallic resources based on geological exploration, characterized in that, The evaluation method includes the following steps: Acquire and integrate multi-source geological exploration data of the target area to construct a three-dimensional geological model and a prior knowledge base; Based on the aforementioned three-dimensional geological model and prior knowledge base, a numerical simulation model describing the mineralization process in the target area is established and run, and the mineralization parameter field is output. Using the multi-source geological exploration data, the ore-forming parameter field and the numerical simulation model are calibrated to generate an assimilated geological process model; Based on the geological structure characterized by the geological process model, combined with the multi-source geological exploration data, a three-dimensional resource model is generated using geostatistical methods, and the probability distribution of resource quantity is calculated. The sampling space of economic parameters is dynamically determined based on the probability distribution of the resource quantity, and the cash flow simulation of the three-dimensional resource model is performed using the sampling space to finally output the probability distribution of economic evaluation indicators.

2. The method for assessing metallic resources based on geological exploration according to claim 1, characterized in that, The acquisition and fusion of multi-source geological exploration data of the target area to construct a three-dimensional geological model and prior knowledge base includes: Acquire multi-source geological exploration data of the target area and preprocess it to form a fused dataset; The fused dataset is input into a pre-trained feature extraction network, which outputs a comprehensive feature field. Based on geological prior rules and the comprehensive feature field, the prior knowledge base is generated through a knowledge graph construction method. Based on the geological constraints defined in the prior knowledge base, a three-dimensional geological interpretation of the comprehensive feature field is performed to construct a three-dimensional geological model of the target area.

3. The method for assessing metallic resources based on geological exploration according to claim 2, characterized in that, The numerical simulation model, based on the three-dimensional geological model and prior knowledge base, is established and run to describe the mineralization process in the target area, outputting a mineralization parameter field, including: Based on the geometric structure and physical property parameters of the three-dimensional geological model, and combined with the mechanistic constraints defined in the prior knowledge base, an initial numerical simulation model describing the coupled thermal-fluid-mechanical-chemical mineralization process is constructed. Target observation data associated with the mineralization process are extracted from the multi-source geological exploration data. The target observation data are compared with the preliminary output results of the initial numerical simulation model. Based on the comparison results, the parameters of the initial numerical simulation model are iteratively optimized to generate a calibrated numerical simulation model. The numerical simulation model is run to dynamically simulate the mineralization process and output a mineralization parameter field, wherein the mineralization parameter field includes at least one of a temperature field, a pressure field, a chemical substance concentration field, and a fluid velocity field.

4. The method for assessing metallic resources based on geological exploration according to claim 3, characterized in that, The step of calibrating the ore-forming parameter field and the numerical simulation model using the multi-source geological exploration data to generate an assimilated geological process model includes: The numerical simulation model is run based on the simulation parameters to generate a corresponding simulated mineralization parameter field, and simulated observation data is extracted from the simulated mineralization parameter field. Calculate the mismatch between the target observation data and the simulated observation data, and update the simulation parameters using the mismatch. The generation of the simulated observation data and the updating of the simulation parameters are executed iteratively until the mismatch is less than a preset threshold. The corresponding simulation parameters and the optimized numerical simulation model are then defined together as the geological process model.

5. The method for assessing metallic resources based on geological exploration according to claim 4, characterized in that, The geological structure characterized by the geological process model, combined with the multi-source geological exploration data, is used to generate a three-dimensional resource model using geostatistical methods, and the probability distribution of resource quantity is calculated, including: From the geological process model, extract a structural model that characterizes the three-dimensional geological structure and lithofacies distribution in the final state after the mineralization period; Using the structural model as the training image required for geostatistical methods, and combining it with the metal precipitation data in the ore-forming parameter field, structural constraints and soft data constraints in three-dimensional space are defined. Using the sample grade data from the multi-source geological exploration data as hard data, under the constraints of the structure and soft data, geostatistical methods are applied to perform random simulations to generate multiple three-dimensional grade models with equal probability. The set of all the three-dimensional grade models is defined as the three-dimensional resource model, and the amount of metal resources in each model is counted to calculate the probability distribution of the resource amount.

6. The method for assessing metallic resources based on geological exploration according to claim 5, characterized in that, The process of dynamically determining the sampling space of economic parameters based on the probability distribution of the resource quantity, and using the sampling space to simulate cash flow in the three-dimensional resource model, ultimately outputting the probability distribution of economic evaluation indicators, includes: Based on the probability distribution of the resource quantity, the probability distribution of at least one economic parameter is determined through predefined economic-resource coupling rules; Random sampling is performed based on the probability distribution of the economic parameters to generate a set of economic parameter scenarios. Then, in combination with the three-dimensional grade models in the defined three-dimensional resource model, a full life-cycle cash flow simulation is performed. Summarize all the obtained cash flow simulation results, calculate and output the probability distribution of economic evaluation indicators.

7. The method for assessing metallic resources based on geological exploration according to claim 6, characterized in that, The aggregated cash flow simulation results are used to calculate and output the probability distribution of economic evaluation indicators, including: Based on the coupled simulation results of each three-dimensional grade model and economic parameter scenario, a probabilistic statistical analysis is performed. The probabilistic statistical analysis involves using the kernel density estimation method to fit the probability density distribution of the net present value and calculating the empirical distribution of the internal rate of return. Based on the probability density distribution and the empirical distribution, calculate at least one of the risk indicators among the value at risk, conditional value at risk, and break-even probability. The final output is a structured probability report containing the probability density distribution, empirical distribution, and risk indicators.

8. A metal resource assessment system based on geological exploration, used to execute a metal resource assessment method based on geological exploration as described in any one of claims 1 to 7, characterized in that, The evaluation system includes: The fusion modeling module is used to acquire and fuse multi-source geological exploration data of the target area to construct a three-dimensional geological model and a prior knowledge base; The mineralization simulation module is used to establish and run a numerical simulation model describing the mineralization process of the target area based on the three-dimensional geological model and prior knowledge base, and output the mineralization parameter field. The calibration and assimilation module is used to calibrate the ore-forming parameter field and the numerical simulation model using the multi-source geological exploration data, and generate an assimilated geological process model. The resource distribution calculation module is used to generate a three-dimensional resource model based on the geological structure characterized by the geological process model, combined with the multi-source geological exploration data, using geostatistical methods, and to calculate the probability distribution of resource quantity. The economic assessment module is used to dynamically determine the sampling space of economic parameters based on the probability distribution of the resource quantity, and to use the sampling space to simulate cash flow in the three-dimensional resource model, and finally output the probability distribution of economic assessment indicators.

Images