A deep web proxy-based mineral resource prediction method and device

Abstract

The application discloses a mineral resource prediction method and device based on a deep network agent, and comprises the following steps: surveying a deposit through an exploration component, obtaining exploration data of the deposit, and constructing a mineralization numerical calculation model of the deposit based on the exploration data; outputting a mineralization condition parameter calculation result through the mineralization numerical calculation model, and training a deep neural network model; after resampling model parameters, performing forward calculation on the mineralization condition parameters through the trained deep neural network agent model; generating a data set according to the forward calculation data combined with the exploration data, training a machine learning model based on the data set, and obtaining a trained machine learning model; outputting a preliminary prediction result through the trained machine learning model, detecting the preliminary prediction result through a prediction index, and outputting a prediction result. The mineralization data is obtained through the forward calculation, the most possible mineralization condition and mineralization situation are deduced through the mineralization data, and the prediction accuracy of the prediction model is improved.

Description

Technical Field

[0001] This invention relates to the field of mineral exploration technology, specifically to a mineral resource prediction method and apparatus based on deep network proxy. Background Technology

[0002] Current mineral resource exploration and prediction primarily rely on real-time monitoring of the geophysical properties within the deposit area. This data, combined with historical mineralization data, is used to predict mineralization based on changes in these properties, thus providing prospecting recommendations. Existing prediction models mainly set mineralization simulation and calculation conditions based on physical property parameters. However, due to the complexity of geological physical parameters within the deposit area and their significant impact on the prediction models, the accuracy of predictions is reduced, leading to uncertainty in mineralization predictions for the deposit area. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of existing technologies. This invention provides a mineral resource prediction method and apparatus based on deep network proxy, which calculates mineralization data within a mineral deposit area through forward modeling. By inversely extrapolating the most likely mineralization conditions and mineralization status from the mineralization data, the prediction accuracy of the prediction model is improved, thereby achieving accurate prediction of the mineralization status of the mineral deposit area.

[0004] This invention provides a mineral resource prediction method based on deep web proxy, the prediction method comprising:

[0005] S11: The ore deposit is surveyed using the exploration component to obtain exploration data of the ore deposit, and model parameters are extracted based on the exploration data;

[0006] Several index parameters are extracted from the exploration data, and a metallogenic numerical calculation model of the deposit is constructed based on these parameters.

[0007] S12: Output the calculation results of ore-forming conditions through the ore-forming numerical calculation model, and train the deep neural network model based on the calculation results of ore-forming condition parameters;

[0008] After resampling the model parameters, the model parameters are forward-modeled using a trained deep neural network surrogate model to obtain forward-modeling data;

[0009] S13: Generate a dataset based on the forward modeling data and the exploration data, and train the machine learning model based on the dataset to obtain the trained machine learning model;

[0010] S14: Output the prediction result through the trained machine learning model, and check whether the prediction index of the prediction result has reached the maximum value. If not, return to step S12; if yes, proceed to step S15.

[0011] S15: Output the prediction results.

[0012] Furthermore, the step of extracting several index parameters from the exploration data and constructing a metallogenic numerical calculation model for the deposit based on these parameters includes:

[0013] The physical properties of the ore deposit are extracted from the exploration data;

[0014] Construct a constitutive model of the ore deposit based on the exploration data;

[0015] The conditional parameters of the mineralization numerical calculation model are set based on the exploration data;

[0016] A numerical calculation model for the mineralization of the deposit is constructed by combining the physical property parameters, the constitutive model, and the conditional parameters.

[0017] Furthermore, the conditional parameters for setting the mineralization numerical calculation model based on exploration data include:

[0018] Based on the numerical changes of various physical parameters within the deposit area, the magnitude of the change in each physical parameter is obtained and calculated. Then, stress loading conditions and boundary constraints for mineralization simulation are set in conjunction with relevant research data on the mineralization background and model.

[0019] Furthermore, the step of outputting the calculation results of ore-forming condition parameters through the ore-forming numerical calculation model, and training the deep neural network model based on the calculation results of the ore-forming condition parameters, includes:

[0020] Sensitivity analysis of simulation parameters to obtain sensitive simulation parameters;

[0021] Sensitive simulation parameters are resampled, and simulation results are output based on high-fidelity numerical simulation.

[0022] A deep neural network model is trained based on the simulation results.

[0023] Furthermore, the sensitivity analysis of the simulation parameters, obtaining sensitive simulation parameters includes:

[0024] Sensitivity analysis of simulation parameters was performed using the COMSOL Multiphysics finite element solver to extract sensitive parameters.

[0025] Furthermore, the high-fidelity numerical simulation output includes:

[0026] Several sensitive parameters are retained from several physical property parameters, and several sensitive parameters are resampled;

[0027] The metallogenic numerical calculation model performs calculations on the resampled sensitive parameter data samples, and outputs the corresponding calculation results in the metallogenic numerical calculation model based on the sensitive parameters.

[0028] Perform high-fidelity numerical simulations on sensitive simulation parameters and output the simulation results.

[0029] Furthermore, the process of outputting the calculation results of ore-forming condition parameters through the ore-forming numerical calculation model, and training the deep neural network model based on the calculation results of the ore-forming condition parameters, also includes:

[0030] A training set and a test set are generated based on the simulation results, and the deep neural network is trained based on the training set and the test set.

[0031] Furthermore, the step of generating a dataset based on forward modeling data and the exploration data, and training a machine learning model based on the dataset to obtain the trained machine learning model includes:

[0032] We propose to use the random forest algorithm to split the training data into training and testing sets, train the machine learning model and optimize its hyperparameters to obtain the trained machine learning model.

[0033] Furthermore, the step of outputting prediction results through the trained machine learning model and detecting whether the prediction index of the prediction results has reached its maximum value includes:

[0034] The trained machine learning model is used to fuse the mineralization prediction variable information of unknown mineralized data points, calculate the posterior probability of mineralization, and output the preliminary prediction results.

[0035] The counting results are calculated based on the parameters of the machine learning model, and it is then determined whether the predictive index of the preliminary prediction result has reached its maximum value.

[0036] The present invention also provides a mineral resource prediction device based on deep web proxy, the prediction device being used to execute the prediction method described above;

[0037] Exploration module: Used to survey the ore deposit using exploration components, obtain exploration data of the ore deposit, and extract model parameters based on the exploration data;

[0038] The calculation model construction module is used to extract several index parameters from the exploration data and construct the metallogenic numerical calculation model of the deposit based on the several index parameters.

[0039] Training module: Used to output the calculation results of mineralization condition parameters through the mineralization numerical calculation model, and to train the deep neural network model based on the calculation results of the mineralization condition parameters;

[0040] Forward modeling module: After resampling the model parameters, the module performs forward modeling calculations on the model parameters using a trained deep neural network surrogate model to obtain forward modeling data;

[0041] Machine learning module: Generates a dataset based on the forward modeling data and the exploration data, trains the machine learning model based on the dataset, and obtains the trained machine learning model;

[0042] Judgment module: used to output prediction results through the trained machine learning model and detect whether the prediction index of the prediction results has reached the maximum value;

[0043] Output module: Used to output the prediction results.

[0044] This invention provides a mineral resource prediction method and apparatus based on deep neural network proxies. It extracts physical parameters, constitutive models, calculation conditions, and initial operating conditions of a mineral deposit area from exploration data. By combining physics, data, and models, it improves the accuracy of exploration data processing for mineral deposit areas. Based on the model parameters obtained from the exploration data processing, a deep neural network model is used to perform forward modeling calculations on the model parameters, outputting preliminary predictions of mineralization data for the mineral deposit area. Then, based on the preliminary predictions, a machine learning model is used for inverse calculations to extrapolate mineralization data within the mineral deposit area. By extrapolating the most probable mineralization conditions and mineralization characteristics from the mineralization data, accurate prediction of the mineralization characteristics of the mineral deposit area is achieved. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart of the mineral resource prediction method based on deep web proxy in an embodiment of the present invention;

[0047] Figure 2 This is a flowchart illustrating the ore-forming numerical calculation model in an embodiment of the present invention;

[0048] Figure 3 This is a schematic diagram of the training process of the deep neural network model in an embodiment of the present invention.

[0049] Figure 4 This is a schematic diagram of a mineral resource prediction device based on deep web proxy in an embodiment of the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Example 1:

[0052] Figure 1 A flowchart of a mineral resource prediction method based on deep web proxy in an embodiment of the present invention is shown. The prediction method includes:

[0053] S11: The ore deposit is surveyed using the exploration component to obtain exploration data of the ore deposit, and model parameters are extracted based on the exploration data;

[0054] Several index parameters are extracted from the exploration data, and a metallogenic numerical calculation model of the deposit is constructed based on these parameters.

[0055] The geological exploration of the deposit is carried out by exploration components, drilling operations are carried out at the corresponding locations in the mining area by drilling rigs, and geological sampling and analysis are conducted at the drilling locations. Based on the differences in the physical properties of various rocks and ores, such as density, magnetism, electrical properties, elasticity, and radioactivity, different physical methods and geophysical instruments are used to detect the changes in geological properties at different locations and depths within the mining area.

[0056] Based on the exploration data, physical property parameters of the ore deposit area are extracted, and structural models are generated based on the exploration data of the ore deposit area. Boundary conditions and initial conditions for mineralization calculations are analyzed based on the exploration data of the ore deposit area, so as to generate a mineralization numerical calculation model of the ore deposit area based on the exploration data.

[0057] S12: Output the calculation results of ore-forming condition parameters through the ore-forming numerical calculation model, and train the deep neural network model based on the calculation results of the ore-forming condition parameters;

[0058] After resampling the model parameters, the model parameters are forward-modeled using a mineralization numerical simulation model to obtain various types of parameters, such as stress field parameters (including the first principal stress, second principal stress, third principal stress, first principal strain, second principal strain, third principal strain, etc.), temperature field parameters (including temperature, gradient, etc.), fluid field parameters (including Darcy velocity field, flow rate, fluid potential, etc.), and chemical field (lead ion concentration, lead sulfide concentration, etc.).

[0059] Under the constraints of reasonable values ​​for model parameters, boundary conditions, and initial conditions, the model parameters are resampled at high density and fine intervals. By automatically reading a combination of resampled parameters and combining it with a deep neural network surrogate model, the gravity field, thermal field, flow field, and chemical field of the ore deposit area are solved quickly to obtain the ore-forming condition parameter values ​​corresponding to the input parameters. This completes a forward modeling calculation of the ore-forming condition parameters and yields forward modeling data.

[0060] S13: Generate a dataset based on the forward modeling data and the exploration data, and train the machine learning model based on the dataset to obtain the trained machine learning model.

[0061] The calculation results of each mineralization condition parameter are obtained by a deep neural network proxy model. A dataset is generated by combining the exploration data in the mineral deposit area. The machine learning model is trained on the dataset to optimize the accuracy of the machine learning model in predicting mineralization in the mineral deposit area.

[0062] Other data with unknown mineral content will be used as the data to be predicted. A random forest algorithm will be used to split the data for training the machine learning model into training and testing sets, and then the machine learning model will be trained and its hyperparameters optimized to obtain the trained machine learning model.

[0063] S14: Output preliminary prediction results through the trained machine learning model, and check whether the prediction index of the preliminary prediction results has reached the maximum value. If not, return to step S12; if yes, proceed to step S15.

[0064] By constructing models and training machine learning models for each mineralization condition parameter within the ore deposit area, and by combining the computational results of deep neural network surrogate models to inversely calculate different mineralization condition parameters, a mineralization prediction model with high prediction accuracy is obtained.

[0065] S15: Output the prediction results.

[0066] Through particle swarm intelligent optimization iterative inversion calculation, the particle swarm is initialized with discretized resampled simulated parameters. Based on different simulated parameters, i.e. different mineralization condition parameters, the machine learning model is repeatedly trained and constructed. Based on continuous iterative optimization calculation, the next parameter combination value is calculated for forward and inverse calculation until the model with the maximum prediction accuracy index is obtained.

[0067] Furthermore, by combining historical data of the ore deposit area, factors such as the metallogenic background, exploration level, and exploration conditions of the ore deposit area are obtained, the predicted target areas are optimized and ranked, the prospecting prospects and exploration risks are evaluated, and prospecting suggestions are proposed.

[0068] Example 2:

[0069] Figure 2 The flowchart of the ore-forming numerical calculation model in this embodiment of the invention is shown. The geological exploration of the ore deposit is carried out by the exploration component, and the drilling operation is carried out at the corresponding location in the mining area by the drilling rig. The geological sampling and analysis of the mining area at the drilling location is carried out. Based on the differences in the physical properties of various rocks and ores, such as density, magnetism, electrical properties, elasticity, and radioactivity, different physical methods and geophysical instruments are used to detect the changes in geological properties at different locations and depths in the mining area.

[0070] Several index parameters are extracted from the exploration data, and a metallogenic numerical calculation model of the deposit is constructed based on these parameters.

[0071] Specifically, the construction process of the ore-forming numerical calculation model includes:

[0072] The process of extracting several index parameters from exploration data and constructing a metallogenic numerical calculation model for the ore deposit based on these parameters includes:

[0073] S111: Extract the physical properties of the ore deposit from the exploration data.

[0074] By comprehensively analyzing the exploration data of the deposit area, such as the deposit profile and boreholes, a shallow geometric model of the deposit is constructed. That is, based on the exploration data of the deposit area, a geometric model of the geological features of the deposit surface is constructed.

[0075] Historical data of the mineral deposit area were queried, and physical properties such as gravity field, magnetic field, and electric field of the mineral deposit area were measured. Based on the physical characteristics of gravity field, magnetic field, and electric field in the historical data, a geometric model of the deep geological structure of the mineral deposit area was constructed.

[0076] Based on historical data of the ore deposit area and exploration data of the ore deposit area, a reconstruction model of the ore deposit area is constructed, which combines the shallow geometric model, the deep geological structure geometric model, and the reconstruction model with the initial geometric model of the ore deposit area.

[0077] Furthermore, by acquiring the physical characteristics of the geological features of the mining area and combining them with historical data of the deposit area to construct an initial geometric model of the deposit area, the structural characteristics of the deposit area can be accurately reflected.

[0078] S112: Construct a constitutive model of the ore deposit based on the exploration data.

[0079] Based on the exploration data, the metallogenic mechanism of the deposit area is summarized. The mineralization processes such as ore-controlling structures, magma, and hydrothermal fluids in the deposit area are analyzed through the exploration data. By querying historical data, a constitutive model of force-heat-fluid multi-field coupling numerical simulation describing the proposed ore deposit is established based on the dynamic models of force, heat, and flow physical fields and their coupling methods.

[0080] In this embodiment, the dynamic models of the three physical fields of force, heat, and flow are coupled using multiphysics simulation software to obtain the constitutive model of the force-heat-flow multifield coupled numerical simulation of the ore deposit in the ore deposit area.

[0081] S113: Set the conditional parameters of the mineralization numerical calculation model based on the exploration data;

[0082] Based on the exploration data of the deposit area, the range of rock and ore physical properties and fluid parameters is obtained. Based on the geological physical properties and fluid parameters of the deposit area, the metallogenic model of the deposit area is analyzed. That is, based on the numerical changes of each physical property parameter in the deposit area, the magnitude of the numerical change of each physical property parameter is obtained and the magnitude change value of each physical property parameter is calculated. Combined with the relevant research data on the metallogenic background and model, the stress loading conditions and boundary constraints of the metallogenic simulation are set.

[0083] Furthermore, the physical properties include: density, porosity, permeability, thermal conductivity, Poisson's ratio, fluid viscosity, fluid temperature, fluid pressure, etc. By analyzing the physical properties of the geological area of ​​the ore deposit, accurate prediction results of the geological area of ​​the ore deposit can be obtained.

[0084] By analyzing the physical properties and fluid parameters, the mineralization background and mineralization mode of the deposit area are determined, and the stress loading conditions and boundary constraints for mineralization simulation are proposed, providing a basis for determining model parameters, boundary conditions and initial conditions in the numerical simulation of the mineralization process.

[0085] S114: Construct a numerical calculation model for the mineralization of the deposit by combining the physical property parameters, the constitutive model, and the condition parameters.

[0086] Based on the above geometric model, as well as the model parameters, boundary conditions and initial conditions, an initial numerical calculation model is established. Multiphysics numerical simulation software is used to perform numerical simulation of the force-heat-fluid multi-field coupled mineralization dynamics process of the proposed ore deposit, and the initial force-heat-fluid multi-field coupled numerical calculation model is obtained.

[0087] The feasibility and rationality of the force-heat-fluid multi-field coupled numerical calculation model are analyzed, that is, multi-case calculation tests are performed based on multiple sets of test data to complete the debugging of the force-heat-fluid multi-field coupled numerical calculation model and obtain the mineralization numerical calculation model.

[0088] Furthermore, metallogenic history data of approximate deposits are extracted from the database, and geological physical property parameters and fluid parameters of the approximate deposits are extracted from the metallogenic history data of the approximate deposits. The geological physical property parameters and fluid parameters of the approximate deposits are then input into the metallogenic numerical calculation model of the deposit for calculation and verification. Based on the calculation results of the metallogenic numerical calculation model, the metallogenic numerical values ​​of the approximate deposits are compared. If the error value between the two is less than 0.3, then the metallogenic numerical calculation model meets the metallogenic simulation calculation requirements of the deposit area and improves the accuracy of the metallogenic numerical calculation model.

[0089] After resampling the model parameters, the model parameters are forward-modeled using a trained deep neural network surrogate model to obtain forward-modeled data.

[0090] The model parameters of the ore-forming numerical calculation model are calculated by using a deep neural network. The accuracy of the calculation of the ore-forming condition parameters in the ore deposit area is verified, thereby obtaining the forward calculation data.

[0091] Specifically, Figure 3 The diagram illustrates the training process of a deep neural network model in an embodiment of the present invention. The step of outputting the calculation results of mineralization condition parameters through the mineralization numerical calculation model and training the deep neural network model based on the calculation results of the mineralization condition parameters includes:

[0092] S121: Sensitivity analysis of simulation parameters.

[0093] Using the COMSOL Multiphysics finite element solver, sensitivity functions for force, heat, fluid, and chemical fields were defined. The geometric model and physics interface of the ore deposit region were configured within the COMSOL Multiphysics finite element solver. Based on the nature and scale of the problem, appropriate solvers and iterators were selected, and physics parameters for the geometric model of the ore deposit region, such as boundary conditions and source terms, were set. Finite element simulation calculations were then performed after setting parameters such as solution time and step size.

[0094] Furthermore, COMSOL Multiphysics is COMSOL's flagship product, a powerful multiphysics simulation software. Based on the finite element method, it simulates real physical phenomena by solving partial differential equations (single field) or systems of partial differential equations (multiple fields).

[0095] S122: Obtain sensitive parameters.

[0096] Sensitivity analysis of simulation parameters is performed using the COMSOL Multiphysics finite element solver to extract sensitive parameters. By defining objective functions such as mean stress and temperature range, sensitivity analysis is conducted on model parameters, boundary conditions, and initial conditions to identify simulation parameters (i.e., parameters of the mathematical-physical equations, such as elastic modulus E, density ρ, porosity ε, Poisson's ratio λ, thermal conductivity kp, constant-pressure heat capacity Cp, etc.) that are highly sensitive to the simulation results. A forward sensitivity calculation method is used to calculate the differential of the scalar objective function with respect to a set of specified control variables using the COMSOL Multiphysics finite element solver. The formula is as follows: where the objective function is Q(ζ), which is usually not an explicit expression of the single control variable ζ (parameters of the mathematical-physical equations). Instead, Q(ζ) is a function of the solution variable u, and u is an implicit expression of the control variable ζ (parameters of the mathematical-physical equations). The partial differential equation of the physical field is expressed as L(u(ζ),ζ)=0.

[0097]

[0098] Where ζ is the control variable, u is the solution variable, u(ζ) is the implicit expression of the control variable ζ, and Q(ζ) is the objective function; is the explicit partial derivative of Q with respect to ζ, d is the differential symbol, and dζ is the differential of ζ. It is the partial derivative of Q with respect to u. It is the partial derivative of L with respect to ζ. It is the partial derivative of u with respect to ζ.

[0099] S123: Resample the sensitive parameters.

[0100] After inputting several physical property parameters of the ore deposit region into the COMSOL Multiphysics finite element solver for sensitivity analysis, several sensitive parameters are retained among the several physical property parameters, and resampling operations are performed on the several sensitive parameters at equal intervals.

[0101] For example, numerical samples of each sensitive parameter are taken at preset time intervals, and the computational data samples of the sensitive parameters are obtained based on the resampling.

[0102] S124: Input the resampled sensitive parameters into the ore-forming numerical calculation model.

[0103] The ore-forming numerical calculation model performs calculations on the resampled sensitive parameter data samples, and outputs the corresponding calculation results in the ore-forming numerical calculation model based on the sensitive parameters.

[0104] S125: Performs high-fidelity numerical simulation on sensitive simulation parameters and outputs the simulation results.

[0105] Using MATLAB scripts, the system automatically resamples and extracts values ​​for highly sensitive numerical simulation parameters, and controls COMSOL Multiphysics to solve the problem. It outputs ore-forming condition parameters (i.e., high-fidelity simulation data) in batches, such as stress, strain, fluid flux, temperature, and ore-forming concentration, and saves the input parameters and output results as a CSV file.

[0106] S126: Generate a training set and a test set based on the simulation results, and train the deep neural network based on the training set and the test set.

[0107] The system automatically reads the CSV data files of the high-fidelity numerical simulation results in batches, merges them into a dataset, and divides the dataset into a training set and a test set. The deep neural network is trained based on the training set, and the trained deep neural network model is tested and verified through the test set to ensure the computational accuracy of the deep neural network model.

[0108] First, the numerical simulation input parameters, spatial coordinates (x, y) of the simulation unit, and time (t) in the automatically batch-read dataset are used as feature variables. The simulation results are used as prediction variables. The dataset is divided into training set and test set according to a certain ratio. The deep neural network model is trained on the training set and tested on the test set. High-precision and fast deep neural network surrogate models for solving the force field, thermal field, flow field, and chemical field in the multi-field coupled numerical model of mineralization dynamics are constructed respectively.

[0109] The simulation input parameters include parameters of the mathematical-physical equations, such as elastic modulus E, density ρ, porosity ε, Poisson's ratio λ, thermal conductivity kp, and constant-pressure heat capacity Cp.

[0110] The simulation results include parameters such as principal strain (Ep), principal stress (Sp), equivalent stress (Mises), and shear stress (Evol) in the force field, temperature (T) in the thermal field, and fluid velocity (U) and flux (Q) in the flow field.

[0111] The simulation condition parameters in CSV format from step S12 are read, and the exploration results are linked with the simulation condition parameters through spatial overlay. The simulation condition parameters and exploration data are then compiled into a dataset for a machine learning model. Based on the machine learning model, data fusion is performed on the simulation condition parameters and exploration data. The binary variables indicating whether minerals are present or not in the dataset are used as target variables, and the data labeled as mineral-present and mineral-free are used as training data for the machine learning model.

[0112] Other data with unknown mineral content will be used as the data to be predicted. A random forest algorithm will be used to split the data for training the machine learning model into training and testing sets, and then the machine learning model will be trained and its hyperparameters optimized to obtain the trained machine learning model.

[0113] The trained machine learning model is used to fuse the mineralization prediction variable information of the above-mentioned unknown mineralized data points, calculate the posterior probability of mineralization, and output the preliminary prediction results.

[0114] Based on the number of mineralization condition parameters recorded in the CSV data table, the mineralization condition parameters in the CSV data table are sequentially input into the machine learning model. The machine learning model uses the input mineralization condition parameters as variables to output the mineralization probability of the deposit area corresponding to the mineralization condition parameters, and counts the parameter operations of the machine learning model.

[0115] Based on the parameter calculation and counting results of the machine learning model, and by detecting whether the prediction index of the preliminary prediction result has reached its maximum value, that is, whether the machine learning model has completed the mineralization prediction calculation for all mineralization condition parameters of the CSV data table.

[0116] If not, return to step S12 and construct a mineralization prediction model for the deposit region based on the next mineralization condition parameter data, thereby ensuring that the machine learning model can complete the mineralization prediction of each mineralization condition parameter in the deposit region, and thus realize the inversion calculation of the mineralization conditions of the deposit region.

[0117] Through particle swarm intelligent optimization iterative inversion calculation, the particle swarm is initialized with discretized resampled simulation parameters, and the model construction and calculation in steps S12 to S14 are repeated, so that the various mineralization condition parameters of the ore deposit area can complete the prediction of mineralization calculation in the ore deposit area.

[0118] Furthermore, by continuously performing iterative optimization calculations and forward and inverse calculations of the next parameter combination values ​​until the model with the maximum prediction accuracy index value is obtained, the accuracy of the mineralization prediction of the machine learning model can be effectively improved, thereby generating accurate mineral exploration suggestions for the mineral deposit area.

[0119] This invention provides a mineral resource prediction method based on deep network proxy. By performing forward modeling and inverse modeling verification of mineralization parameters in a mineral deposit area, the method accurately predicts the mineralization status of the area. Forward modeling is used to calculate mineralization data within the deposit area. The most probable mineralization conditions and mineralization status are then inversely derived from the mineralization data, thereby achieving accurate prediction of the mineralization status of the deposit area.

[0120] Example 3:

[0121] Figure 4 This diagram illustrates a mineral resource prediction device based on a deep web proxy, as shown in an embodiment of the present invention. The deep web proxy-based mineral resource prediction device includes:

[0122] Exploration module 10: Used to survey the ore deposit through exploration components, obtain exploration data of the ore deposit, and extract model parameters based on the exploration data.

[0123] Several index parameters are extracted from the exploration data, and a metallogenic numerical calculation model of the deposit is constructed based on these parameters.

[0124] The geological exploration of the deposit is carried out by exploration components, drilling operations are carried out at the corresponding locations in the mining area by drilling rigs, and geological sampling and analysis are conducted at the drilling locations. Based on the differences in the physical properties of various rocks and ores, such as density, magnetism, electrical properties, elasticity, and radioactivity, different physical methods and geophysical instruments are used to detect the changes in geological properties at different locations and depths within the mining area.

[0125] Based on the exploration data, physical property parameters of the ore deposit area are extracted, and structural models are generated based on the exploration data of the ore deposit area. Boundary conditions and initial conditions for mineralization calculations are analyzed based on the exploration data of the ore deposit area, so as to generate a mineralization numerical calculation model of the ore deposit area based on the exploration data.

[0126] Calculation model construction module 20: used to extract several index parameters from exploration data and construct a metallogenic numerical calculation model of the deposit based on the several index parameters.

[0127] After resampling the model parameters, the model parameters are forward-modeled using a trained deep neural network surrogate model to obtain forward-modeled data.

[0128] Under the constraints of reasonable values ​​for model parameters, boundary conditions, and initial conditions, the model parameters are resampled at high density and fine intervals. By automatically reading a combination of resampled parameters and combining it with a deep neural network surrogate model, the gravity field, thermal field, flow field, and chemical field of the ore deposit area are solved quickly to obtain the ore-forming condition parameter values ​​corresponding to the input parameters, thus completing a forward modeling calculation of the ore-forming condition parameters.

[0129] Training module 30: Used to output the calculation results of mineralization condition parameters through the mineralization numerical calculation model, and to train the deep neural network model based on the calculation results of mineralization condition parameters.

[0130] The calculation results of each mineralization condition parameter are obtained by the mineralization numerical calculation model. A dataset is generated by combining the exploration data in the mineral deposit area. The machine learning model is trained on the dataset to optimize the accuracy of the machine learning model in predicting mineralization in the mineral deposit area.

[0131] Other data with unknown mineral content will be used as the data to be predicted. A random forest algorithm will be used to split the data for training the machine learning model into training and testing sets, and then the machine learning model will be trained and its hyperparameters optimized to obtain the trained machine learning model.

[0132] Forward modeling module 40: After resampling the model parameters, the forward modeling module performs forward modeling calculations on the model parameters through the trained deep neural network surrogate model to obtain forward modeling data.

[0133] By constructing models and training machine learning models for each mineralization condition parameter within the deposit area, and by combining the calculation results of the mineralization numerical calculation model, different mineralization condition parameters are inverted and extrapolated to obtain a mineralization prediction model with high prediction accuracy.

[0134] Machine learning module 50: Generates a dataset based on the forward modeling data and the exploration data, trains the machine learning model based on the dataset, and obtains the trained machine learning model;

[0135] Judgment module 60: used to output prediction results through the trained machine learning model and detect whether the prediction index of the prediction results has reached the maximum value;

[0136] Output module 70: Used to output prediction results.

[0137] Through particle swarm intelligent optimization iterative inversion calculation, the particle swarm is initialized with discretized resampled simulated parameters. Based on different simulated parameters, i.e. different mineralization condition parameters, the machine learning model is repeatedly trained and constructed. Based on continuous iterative optimization calculation, the next parameter combination value is calculated for forward and inverse calculation until the model with the maximum prediction accuracy index is obtained.

[0138] Furthermore, by combining historical data of the ore deposit area, factors such as the metallogenic background, exploration level, and exploration conditions of the ore deposit area are obtained, the predicted target areas are optimized and ranked, the prospecting prospects and exploration risks are evaluated, and prospecting suggestions are proposed.

[0139] 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, which may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0140] Furthermore, the embodiments of the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for mineral resource prediction based on deep web proxies, characterized by, The prediction method comprises: S11: surveying a deposit through an exploration component, obtaining exploration data of the deposit, extracting a plurality of model parameters from the exploration data, and constructing a metallogenic numerical calculation model of the deposit based on the plurality of model parameters; S12: outputting a metallogenic condition parameter calculation result through the metallogenic numerical calculation model, and training a deep neural network model according to the metallogenic condition parameter calculation result; resampling the model parameters and inputting the resampled model parameters into the trained deep neural network model, and performing forward calculation on the model parameters through the trained deep neural network model to obtain forward data; S13: generating a data set according to the forward data combined with the exploration data, training a machine learning model based on the data set, and obtaining a trained machine learning model; S14: outputting a prediction result through the trained machine learning model, and detecting whether a prediction index quantity of the prediction result reaches a maximum value, if not, returning to step S12; if yes, proceeding to step S15; S15: outputting the prediction result. The outputting a metallogenic condition parameter calculation result through the metallogenic numerical calculation model, and training a deep neural network model according to the metallogenic condition parameter calculation result comprises: performing sensitivity analysis on simulation parameters through a finite element solver to obtain sensitive parameters; resampling the sensitive parameters and inputting the resampled sensitive parameters into the metallogenic numerical calculation model to obtain sensitive simulation parameters; performing high-fidelity numerical simulation operation on the sensitive simulation parameters to output simulation results; generating a training set and a test set according to the simulation results, training a deep neural network model based on the training set and the test set, and obtaining a trained deep neural network model.

2. The deep web agent based mineral resource prediction method of claim 1, wherein, The extracting a plurality of model parameters from the exploration data, and constructing a metallogenic numerical calculation model of the deposit based on the plurality of model parameters comprises: extracting physical property parameters of the deposit from the exploration data; constructing a constitutive model of the deposit according to the exploration data; setting condition parameters of the metallogenic numerical calculation model based on the exploration data; constructing the metallogenic numerical calculation model of the deposit in combination with the physical property parameters, the constitutive model and the condition parameters.

3. The mineral resource prediction method according to claim 2, characterized by, The setting condition parameters of the metallogenic numerical calculation model based on the exploration data comprises: according to the numerical variation of each physical property parameter in the deposit area, obtaining the numerical variation amplitude of each physical property parameter through the variation value of the physical property parameter, and converting the amplitude variation value of each physical property parameter, and setting the stress loading condition and the boundary constraint condition of the metallogenic simulation in combination with the related research data of the metallogenic background and mode.

4. The deep web agent based mineral resource prediction method of claim 1, wherein, The performing sensitivity analysis on simulation parameters through a finite element solver to obtain sensitive parameters comprises: performing sensitivity analysis on simulation parameters through a COMSOL Multiphysics finite element solver to extract sensitive parameters.

5. The deep web agent-based mineral resource prediction method of claim 1, wherein, The performing high-fidelity numerical simulation operation on the sensitive simulation parameters to output simulation results comprises: reserving a plurality of sensitive parameters from a plurality of physical property parameters, and resampling the plurality of sensitive parameters; The sensitive parameter data sample after resampling is operated by the metallogenic numerical calculation model, and corresponding operation results are output based on the sensitive parameters in the metallogenic numerical calculation model; The sensitive simulation parameters are subjected to high-fidelity numerical simulation operation, and simulation results are output.

6. The deep web agent-based mineral resource prediction method of claim 1, wherein, The calculation results are output by the metallogenic numerical calculation model, and the deep neural network model is trained according to the calculation results, and the training set and the test set are generated according to the simulation results, and the deep neural network is trained based on the training set and the test set. The data set is generated according to the forward data and the exploration data, the machine learning model is trained based on the data set, and the trained machine learning model is obtained.

7. The deep web agent based mineral resource prediction method of claim 1, wherein, The random forest algorithm is used to divide the training machine learning data into training set and test set, train the machine learning model and optimize the hyperparameters, and obtain the trained machine learning model. The prediction result is output by the trained machine learning model, and it is detected whether the prediction index quantity of the prediction result reaches the maximum value.

8. The deep web agent based mineral resource prediction method of claim 1, wherein, The metallogenic prediction variable information of unknown ore-bearing data points is fused by the trained machine learning model, the metallogenic posterior probability is calculated, and the preliminary prediction result is output. The parameter operation count result of the machine learning model is obtained, and it is detected whether the prediction index quantity of the preliminary prediction result reaches the maximum value. The prediction device is used to execute the prediction method of any one of claims 1 to 8.

9. A deep web proxy based mineral resource prediction apparatus, characterized by, The exploration module is used to survey the ore deposit by the exploration component, obtain the exploration data of the ore deposit, and extract model parameters based on the exploration data; The calculation model construction module is used to extract a plurality of model parameters from the exploration data, and construct a metallogenic numerical calculation model of the ore deposit based on the plurality of model parameters; The training module is used to output calculation results by the metallogenic numerical calculation model, and train a deep neural network model according to the calculation results; The forward calculation module is used to resample the model parameters, and perform forward calculation on the model parameters by the trained deep neural network model to obtain forward data; The machine learning module is used to generate a data set according to the forward data and the exploration data, train a machine learning model based on the data set, and obtain a trained machine learning model; The judgment module is used to output a prediction result by the trained machine learning model, and detect whether the prediction index quantity of the prediction result reaches the maximum value; The output module is used to output the prediction result. ​

Images