Ore grade simulation method and device based on direct sampling method and computer equipment

Abstract

The application discloses a kind of ore grade simulation method, device and computer equipment based on direct sampling method, it is related to geostatistics technical field, it solves the problem that the simulation result precision is lower that the ore grade simulation is carried out based on the multipoint geostatistics method with the algorithm parameter being configured at present.The method comprises the following steps: migrating drill hole data set to simulation grid to obtain drill hole data set with simulation grid structure, then determining training set and test set, and generating training image, using direct sampling method to simulate ore grade, to obtain initial ore grade simulation result, according to initial ore grade simulation result and test set, the key parameters in direct sampling method are optimized using particle swarm optimization algorithm, to obtain optimal key parameters;Based on optimal key parameters, the ore grade simulation is carried out, and the target ore grade simulation result is obtained.

Description

Technical Field

[0001] This application belongs to the field of geostatistics technology, and more specifically, relates to a method, apparatus, computer equipment, and readable storage medium for simulating ore grade based on direct sampling. Background Technology

[0002] Ore grade modeling is a crucial step in the qualitative assessment of mines, and the distribution of ore grades is influenced by a variety of factors, such as location, geological composition, and structural characteristics of the deposit. Geologists and mining engineers have been exploring methods to understand and simulate these geological processes.

[0003] Currently, the Multiple-point Geostatistic (MPS) method is mainly used for ore grade modeling. However, the accuracy of the simulation results obtained by the MPS method is greatly affected by the algorithm parameters. Therefore, selecting appropriate parameter configuration is a crucial step in the process of ore grade simulation using the MPS method. If the selected algorithm parameters have poor sensitivity, the simulation results obtained by the MPS method with those algorithm parameters will be of low accuracy. Summary of the Invention

[0004] In view of this, the present invention provides a method, apparatus, computer equipment and readable storage medium for ore grade simulation based on direct sampling method. The main purpose is to solve the problem that if the selected algorithm parameters have poor sensitivity, the simulation results obtained by multi-point geostatistical method with configured algorithm parameters are of low accuracy.

[0005] By virtue of the first aspect of this application, a method for simulating ore grade based on direct sampling is provided, comprising:

[0006] Obtain the borehole dataset of the ore body to be simulated, and migrate the borehole dataset to the simulation grid to obtain the borehole dataset with the simulation grid structure. The borehole dataset includes multiple borehole data, and each borehole data includes the borehole location and the ore grade corresponding to the borehole location.

[0007] A training set and a test set are determined based on the borehole dataset with the simulated grid structure, and a training image is generated based on the training set, wherein the training set includes multiple first borehole data with a first simulated grid structure, and the test set includes multiple second borehole data with a second simulated grid structure;

[0008] Based on the training images, the training set, and the test set, the ore grade is simulated using the direct sampling method to obtain initial ore grade simulation results. Then, based on the initial ore grade simulation results and the test set, the key parameters in the direct sampling method are optimized using the particle swarm optimization algorithm to obtain the optimal key parameters.

[0009] The optimal key parameters are configured in the direct sampling method to obtain the target direct sampling method. Based on the borehole dataset with simulated grid structure, the training set and the training images, the target direct sampling method is used to simulate the ore grade to obtain the target ore grade simulation result of the ore body to be simulated.

[0010] Optionally, the step of migrating the borehole dataset to a simulated mesh to obtain a borehole dataset with a simulated mesh structure includes:

[0011] Obtain the boundary information of the ore body to be simulated;

[0012] A geological constraint model is established based on the boundary information of the target ore body, and a three-dimensional simulation mesh is determined based on the data characteristics of the borehole dataset.

[0013] The three-dimensional simulation mesh is obtained by trimming the geological body constraint model;

[0014] The borehole dataset includes multiple borehole data points, which are then migrated to the simulated grid to obtain the borehole dataset with a simulated grid structure. The borehole dataset with a simulated grid structure includes multiple grid cells and the grid position and target ore grade corresponding to each grid cell. The target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole position included in a grid cell.

[0015] Optionally, generating training images based on the training set includes:

[0016] The training set includes a first simulated grid, wherein the first simulated grid includes a plurality of first grid cells, and each first grid cell corresponds to a first target ore grade, the first target ore grade being the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell;

[0017] A bounding box is constructed based on the data features of the training set, and the training image is obtained by interpolating the multiple second grid units in the bounding box according to the multiple first ore grades corresponding to the multiple first grid units in the training set using the inverse power distance method. The multiple second grid units are other grid units in the bounding box besides the multiple first grid units.

[0018] Optionally, the step of simulating ore grade using the direct sampling method based on the training images, the training set, and the test set to obtain initial ore grade simulation results includes:

[0019] A second simulation grid is determined based on the test set, wherein the second simulation grid includes a plurality of second grid cells and an actual ore grade value corresponding to each second grid cell, and the actual ore grade value is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a second grid cell;

[0020] The simulation path is determined based on the second simulation grid, and a simulation grid is established based on at least one borehole location included in each second grid cell of the test set, wherein each simulation grid cell includes at least one borehole location.

[0021] Using the training set as a conditional dataset, the direct sampling method is used to traverse each grid cell in the grid to be simulated according to the simulation path. At least one first grid cell that matches each grid cell is queried from the conditional dataset. A first hysteresis vector of each grid cell is calculated based on the at least one first grid cell that matches each grid cell. A first data event value of each grid cell is calculated based on the first hysteresis vector of each grid cell. The conditional dataset includes multiple first grid cells and each first grid cell corresponds to a first target ore grade. The first target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell.

[0022] Based on the direct sampling method, at least one view grid cell that matches each grid cell to be simulated is queried from the training image; a second hysteresis vector of each grid cell to be simulated is calculated based on the at least one view grid cell that matches each grid cell to be simulated; and a second data event value of each grid cell to be simulated is calculated based on the second hysteresis vector of each grid cell to be simulated.

[0023] For each grid cell to be simulated, perform the following operations: calculate the difference between the first data event value and the second data event value based on the direct sampling method; when it is determined that the difference does not reach the target threshold, use the ore grade of the view grid cell as the simulated ore grade value of the grid cell to be simulated.

[0024] The initial ore grade simulation result is obtained based on the ore grade simulation value of each grid cell to be simulated.

[0025] Optionally, the step of optimizing the key parameters in the direct sampling method using a particle swarm optimization algorithm based on the initial ore grade simulation results and the test set to obtain the optimal key parameters includes:

[0026] Based on the particle swarm optimization algorithm, the simulated ore grade value corresponding to each grid cell to be simulated is extracted from the initial ore grade simulation results. A second grid cell matching the grid cell to be simulated is queried from the test set, and the actual ore grade value corresponding to the second grid cell is determined. The mean square error between the simulated ore grade value and the actual ore grade value is calculated. When the mean square error exceeds the global optimum, the key parameters in the direct sampling method are adjusted to obtain updated key parameters. The updated key parameters are configured in the direct sampling method to obtain an intermediate direct sampling method. Based on the training image and the training set... The intermediate direct sampling method is used to simulate ore grade on the test set to obtain intermediate ore grade simulation results. The updated ore grade simulation value corresponding to each grid cell to be simulated is extracted from the intermediate ore grade model simulation results. The updated second grid cell that matches the grid cell to be simulated is queried from the test set. The updated actual ore grade value corresponding to the updated second grid cell is determined. The update mean square error between the updated ore grade simulation value and the updated actual ore grade value is calculated. The above steps are repeated until the update mean square error does not exceed the global optimum value. The updated key parameter is then determined as the optimal key parameter.

[0027] Optionally, after configuring the optimal key parameters in the direct sampling method to obtain the target direct sampling method, the method further includes:

[0028] A validation set is determined based on the borehole dataset with the simulated grid structure. The validation set includes multiple third borehole data with a third simulated grid structure, wherein the multiple third borehole data are other borehole data in the borehole dataset with the simulated grid structure other than the multiple first borehole data and the multiple second borehole data.

[0029] Based on the validation set, the training images, and the training set, the target direct sampling method is used to simulate the ore grade, and the simulation results of the ore grade to be verified are obtained.

[0030] Multiple simulated values ​​of ore grade to be verified are extracted from the simulated results of the ore grade to be verified, wherein each simulated value of ore grade to be verified corresponds to a third grid cell;

[0031] Multiple actual ore grade values ​​are extracted from the verification set, wherein each actual ore grade value corresponds to a third grid cell;

[0032] For each third grid cell, perform the following operation: calculate the difference between the simulated value of the ore grade to be verified and the corresponding actual value of the ore grade, compare the difference with the target threshold, and obtain the comparison result;

[0033] If the difference in the comparison result corresponding to each third grid cell does not exceed the target threshold, then the target direct sampling method is determined to be the optimal direct sampling method;

[0034] If the comparison result of a third grid cell indicates a difference exceeding the target threshold, the particle swarm optimization algorithm is used to further optimize the key parameters in the direct sampling method to obtain updated optimal key parameters. These updated optimal key parameters are then configured in the direct sampling method to obtain an updated target direct sampling method.

[0035] Optionally, a preset ratio is determined, and the borehole dataset with simulated mesh structure is divided into the training set, the test set, and the validation set according to the preset ratio.

[0036] By virtue of a second aspect of this application, an ore grade simulation device based on a direct sampling method is provided, comprising:

[0037] The acquisition module is used to acquire the borehole dataset of the ore body to be simulated, and to migrate the borehole dataset to the simulation grid to obtain a borehole dataset with a simulation grid structure. The borehole dataset includes multiple borehole data, and each borehole data includes the borehole location and the ore grade corresponding to the borehole location.

[0038] A generation module is configured to determine a training set and a test set based on the borehole dataset with a simulated grid structure, and to generate training images based on the training set, wherein the training set includes multiple first borehole data with a first simulated grid structure, and the test set includes multiple second borehole data with a second simulated grid structure.

[0039] An optimization module is used to simulate ore grade using the direct sampling method based on the training image, the training set, and the test set to obtain an initial ore grade simulation result, and to optimize the key parameters in the direct sampling method using a particle swarm optimization algorithm based on the initial ore grade simulation result and the test set to obtain the optimal key parameters.

[0040] The simulation module is used to configure the optimal key parameters in the direct sampling method to obtain the target direct sampling method, and to simulate the ore grade using the target direct sampling method based on the borehole dataset with the simulated grid structure, the training set and the training images to obtain the target ore grade simulation result of the ore body to be simulated.

[0041] By virtue of a third aspect of this application, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described in the first aspect above.

[0042] By virtue of a fourth aspect of this application, a readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any one of the first aspects above.

[0043] By employing the above technical solution, this application provides a method, apparatus, computer equipment, and readable storage medium for ore grade simulation based on direct sampling. First, the borehole data of the ore body to be simulated is preprocessed to obtain a borehole dataset with a simulated grid structure. Then, a training set and a test set are determined based on this borehole dataset. Knowing the training set, it can be used as conditional data, or a training image can be generated based on the training set. Based on the training image and conditional data, the direct sampling method is used to simulate the ore grade on the test set, obtaining an initial ore grade simulation result. Based on the initial ore grade simulation result, a particle swarm optimization algorithm is used to optimize the key parameters in the direct sampling method. By adaptively optimizing the parameters, highly sensitive key parameters can be quickly identified. These key parameters are then configured as the optimal key parameters in the direct sampling method to obtain the target direct sampling method. Finally, for the ore body to be simulated, based on the previously obtained processed borehole dataset, combined with the training set and training images, the ore grade is simulated using the target direct sampling method to obtain the target ore grade simulation result for the ore body to be simulated. Since highly sensitive key parameters can be obtained through the above process, subsequent ore grade simulation using the direct sampling method configured with these key parameters can yield a highly accurate target ore grade simulation result for the ore body to be simulated.

[0044] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0045] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0046] Figure 1 A flowchart of a method for simulating ore grade based on direct sampling, provided in an embodiment of this application, is shown.

[0047] Figure 2 This illustration shows a constraint model, borehole dataset, and a data distribution diagram of a borehole dataset with a simulated grid structure, all provided in an embodiment of this application, for an ore grade simulation method based on direct sampling.

[0048] Figure 3 This illustration shows a schematic diagram of the data distribution of training images for an ore grade simulation method based on direct sampling, provided in an embodiment of this application.

[0049] Figure 4 A flowchart of another ore grade simulation method based on direct sampling provided in an embodiment of this application is shown;

[0050] Figure 5 This paper illustrates a schematic diagram of the geological body constraint model for another ore grade simulation method based on direct sampling, as provided in an embodiment of this application.

[0051] Figure 6 This illustration shows a schematic diagram of the data distribution of the divided borehole dataset in another ore grade simulation method based on direct sampling provided in an embodiment of this application.

[0052] Figure 7 The flowchart illustrates another method for ore grade simulation based on direct sampling, provided in an embodiment of this application, for obtaining optimal key parameters.

[0053] Figure 8 This illustration shows a schematic diagram of the specific iterative process for adaptive optimization of three key parameters in another ore grade simulation method based on direct sampling provided in this application embodiment;

[0054] Figure 9 This paper illustrates another ore grade simulation method based on direct sampling provided in this application, which simulates on a test set and uses the obtained optimal parameters to configure the absolute error cumulative density map of the simulation results.

[0055] Figure 10 This paper illustrates the cumulative density of the absolute error of the simulation results obtained by using the obtained parameter configuration on a validation set in another ore grade simulation method based on direct sampling provided in this application embodiment.

[0056] Figure 11 This illustration shows a comparison of another ore grade simulation method based on direct sampling provided in this application with a target ore body, and the resulting grade distribution diagram.

[0057] Figure 12This illustration shows a variation function diagram of the simulation results obtained by comparing another direct sampling method for ore grade simulation provided in this application with a comparative method for simulating a target ore body.

[0058] Figure 13 A schematic diagram of the structure of an ore grade simulation device based on direct sampling method provided in an embodiment of this application is shown.

[0059] Figure 14 A schematic diagram of the structure of an ore grade simulation device based on direct sampling method provided in an embodiment of this application is shown.

[0060] Figure 15 A schematic diagram of the device structure of a computer device provided in an embodiment of this application is shown. Detailed Implementation

[0061] Various embodiments and features of this application are described herein with reference to the accompanying drawings.

[0062] It should be understood that various modifications can be made to the embodiments described herein. Therefore, the above description should not be considered as limiting, but merely as an example of embodiments. Other modifications within the scope and spirit of this application will be apparent to those skilled in the art.

[0063] The accompanying drawings, which are included in and form part of this specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0064] These and other features of this application will become apparent from the following description of preferred forms of embodiments given as non-limiting examples, with reference to the accompanying drawings.

[0065] It should also be understood that although this application has been described with reference to some specific examples, those skilled in the art can certainly implement many other equivalent forms of this application.

[0066] The above and other aspects, features and advantages of this application will become more apparent when taken in conjunction with the accompanying drawings and in view of the following detailed description.

[0067] Specific embodiments of this application are described thereafter with reference to the accompanying drawings; however, it should be understood that the claimed embodiments are merely examples of this application, which can be implemented in various ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that could obscure the application. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but merely serve as the basis and representative basis for the claims to teach those skilled in the art to use this application in a variety of substantially any suitable detailed structures.

[0068] This specification may use the phrases “in one embodiment,” “in another embodiment,” “in yet another embodiment,” or “in other embodiments,” all of which may refer to one or more of the same or different embodiments according to this application.

[0069] This application provides a method for simulating ore grade based on direct sampling, such as... Figure 1 As shown, the method includes:

[0070] 101. Obtain the borehole dataset of the ore body to be simulated, and migrate the borehole dataset to the simulation grid to obtain a borehole dataset with a simulation grid structure.

[0071] In this embodiment, the executing entity is a system capable of performing an ore grade simulation method. This system needs to acquire a borehole dataset of the ore body to be simulated. This borehole dataset includes multiple borehole data points, each including the borehole location and the corresponding ore grade. The borehole data in the dataset is then preprocessed, specifically migrated to a simulation grid to obtain a borehole dataset with a simulation grid structure. The structure of the processed borehole dataset is as follows: Figure 2 The rightmost simulated mesh structure is shown in Figure 2 From left to right, the data consists of the constraint model of the ore body to be simulated, the borehole dataset, and the borehole dataset after being migrated to a cube (i.e., the borehole dataset with simulated mesh results).

[0072] It is important to note that the ore body to be simulated must first be identified. Before mining the simulated ore body, geological exploration and drilling activities must be conducted to obtain a large amount of reliable borehole data. Borehole data can be understood as various information collected by the drill bit when drilling underground. This information includes core samples, cuttings, borehole coordinates, ore grade, hydrological data, geomagnetic data, geoelectric data, etc. Ore grade here refers to the content of useful components or minerals per unit volume or unit weight of ore. It is generally expressed as a weight percentage (e.g., iron, copper, lead, zinc ores), sometimes in grams per ton (e.g., gold, silver ores), sometimes in grams per cubic meter (e.g., placer gold), and sometimes in grams per liter (e.g., iodine, bromine, chemical raw material minerals). Ore grade is a key indicator for measuring the economic value of a mineral deposit.

[0073] 102. Determine the training and test sets based on the borehole dataset with a simulated grid structure, and generate training images based on the training set.

[0074] In this embodiment, after obtaining a borehole dataset with a simulated mesh structure, a portion of the borehole data can be extracted proportionally as a training set and a portion as a test set. Further, the training set includes multiple first borehole data points with a first simulated mesh structure, and the test set includes multiple second borehole data points with a second simulated mesh structure. Then, training images are generated based on the training set.

[0075] It should be noted that training images can be understood as a priori conceptual model that includes complex geological spatial structures. The specific form of training images is as follows: Figure 3 As shown, in Figure 3 In the middle, from left to right, are the front view, left view, and top view of the training images generated from the training set.

[0076] 103. Based on the training images, training set, and test set, the ore grade is simulated using the direct sampling method to obtain the initial ore grade simulation results. Based on the initial ore grade simulation results and the test set, the key parameters in the direct sampling method are optimized using the particle swarm optimization algorithm to obtain the optimal key parameters.

[0077] In this embodiment, after acquiring the training images, training set, and test set, the training set is determined as the conditional data in the process of simulating ore grade using the direct sampling method. When using the direct sampling method for the first time, the initial values ​​of key parameters need to be pre-configured. In this way, based on the training images, conditional data, and test set, the ore grade simulation can be performed using the direct sampling method to obtain the initial ore grade simulation result. It should be noted that the initial ore grade simulation result is presented in the form of a simulation grid. The simulation grid includes multiple grid cells, and the grid cells have a spatial distribution pattern. That is, each grid cell has coordinate information (corresponding position), and each grid cell corresponds to an initial simulated ore grade value.

[0078] It should be noted that direct sampling (DS) is a type of multi-point geostatistical method. Multi-point geostatistical methods aim to extract spatial patterns directly from training images, rather than describing spatially heterogeneous geometric features through complex calculations. The training images are conceptual models abstracted from real-world phenomena. Direct sampling is a more direct multi-point geostatistical simulation method. This method employs direct sampling, stopping at the first matching data event encountered during the scanning of the training image, and using this as the simulation result. If no match is found, the training image is scanned a predetermined number of times, and the most known matching data event in the training image is selected as the simulation result.

[0079] Particle Swarm Optimization (PSO) is a swarm intelligence optimization algorithm whose goal is to find the optimal solution for all particles within a multi-dimensional hypervolume. First, all particles in the space are assigned initial random positions and velocities. Then, each particle's position is progressively advanced based on its velocity, the known optimal global position in the problem space, and the particle's known optimal position. As computation progresses, particles aggregate or cluster around one or more optimal points by exploring and utilizing known advantageous positions in the search space.

[0080] 104. The optimal key parameters are configured in the direct sampling method to obtain the target direct sampling method. Based on the borehole dataset, training set and training images with simulated grid structure, the target direct sampling method is used to simulate the ore grade and obtain the target ore grade simulation results of the ore body to be simulated.

[0081] In this embodiment of the application, after determining the optimal key parameters, the optimal key parameters can be configured in the direct sampling method, so as to obtain the target direct sampling method. For the ore body to be simulated, the borehole dataset of the ore body to be simulated has been obtained in the early stage. As in the above steps, the training set is used as the conditional data. The ore grade is simulated by combining the training image and the borehole dataset using the target direct sampling method, so as to obtain the target ore grade simulation result of the ore body to be simulated with high accuracy.

[0082] The method provided in this application first preprocesses the borehole data of the ore body to be simulated to obtain a borehole dataset with a simulated grid structure. Then, based on the borehole dataset, a training set and a test set are determined. After knowing the training set, it can be used as conditional data or a training image can be generated based on the training set. Based on the training image and conditional data, the ore grade is simulated on the test set using a direct sampling method to obtain an initial ore grade simulation result. Based on the initial ore grade simulation result, a particle swarm optimization algorithm is used to adaptively optimize the key parameters in the direct sampling method, which can quickly find a key parameter with high sensitivity. This key parameter is configured as the optimal key parameter in the direct sampling method to obtain the target direct sampling method. Finally, for the ore body to be simulated, based on the previously obtained processed borehole dataset, combined with the training set and training image, the target ore grade is simulated using the target direct sampling method to obtain the target ore grade simulation result of the ore body to be simulated. Since a key parameter with high sensitivity can be obtained through the above process, the target ore grade simulation result of the ore body to be simulated can be obtained by using the direct sampling method configured with this key parameter for subsequent ore grade simulation.

[0083] Furthermore, as a refinement and extension of the specific implementation methods of the above embodiments, and in order to fully illustrate the specific implementation process of this embodiment, this application provides a method for simulating ore grade based on direct sampling, such as... Figure 4 As shown, it includes:

[0084] 201. Preprocess the borehole data to obtain a borehole dataset with a simulated mesh structure;

[0085] In this embodiment, the boundary information of the ore body to be simulated can be obtained first. Then, a geological body constraint model can be established based on the boundary information of the target ore body, and a three-dimensional simulation grid can be determined based on the data characteristics of the borehole dataset. Next, the three-dimensional simulation grid is trimmed using the geological body constraint model to obtain a simulation grid. Multiple borehole data included in the borehole dataset are migrated to the simulation grid to obtain a borehole dataset with a simulation grid structure. The borehole dataset with a simulation grid structure includes multiple grid cells and the grid position and target ore grade corresponding to each grid cell. The target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole position included in a grid cell.

[0086] It should be noted that before conducting subsequent ore grade simulations, it is necessary to determine the boundaries of the target ore body through geological surveys and to obtain reliable borehole data through geological exploration and drilling activities. By determining the boundaries of the target ore body, its outline can be determined, thereby establishing a model... Figure 5 The geological constraint model shown in the leftmost figure is used in subsequent simulations. The resulting borehole dataset is as follows: Figure 2 The middle diagram is shown in the image.

[0087] Furthermore, since the direct sampling method used later is essentially a pixel-based sequential simulation method, a simulation grid needs to be constructed in advance. The value of one grid node (also called a grid cell) is simulated point-by-point. The simulation process involves modeling the ore grade in three-dimensional space, requiring the construction of a three-dimensional simulation grid. By observing the data characteristics of the obtained borehole data, a three-dimensional simulation grid can be constructed using multiple 10×10×10m cubic grid cells. In subsequent simulations, each grid cell within this simulation grid is simulated.

[0088] For example, taking copper grade as an example, the simulation mesh is trimmed according to the geological body constraint model, and the simulation mesh obtained after deleting the redundant parts is as follows: Figure 5 As shown, Figure 5 The simulation of the borehole data grid structure is also shown in the image. Figure 5In the image, from left to right, are the front view, left view, and top view of the simulated grid constructed based on the geological body constraint model. Next, the borehole data needs to be migrated to the grid cells. Specifically, this involves iterating through all borehole data in the dataset and assigning each borehole data point to its nearest neighboring grid cell. Then, for each grid cell, the arithmetic mean of the ore grade at all borehole points within that cell is calculated; this mean is the copper grade value for that grid cell. The processed borehole dataset is shown below. Figure 2 As shown in the rightmost image.

[0089] 202. Determine the training and test sets based on the borehole dataset with a simulated grid structure.

[0090] In this embodiment, the borehole data in the borehole dataset with a simulated grid structure can be divided into three parts according to a first preset ratio: a training set, a test set, and a subsequent validation set. Alternatively, the borehole data in the borehole dataset with a simulated grid structure can be divided into two parts according to a second preset ratio: a training set and a test set, depending on actual needs. The first and second preset ratios here have no explicit physical meaning and are only used for distinction; the specific preset ratio can be flexibly chosen according to actual needs.

[0091] For example, a borehole dataset with a simulated grid structure is randomly divided into a training set, a test set, and a validation set in a 6:2:2 ratio. Training images are required in the DS simulation process; the training set is used to generate training images and serves as conditional data during the simulation. The test set is used to obtain the optimal parameter configuration, and the validation set is used to verify the reliability of the obtained parameter configuration. The divided borehole dataset is shown below. Figure 6 As shown, in Figure 6 In the image, from left to right, are the actual data for the training set, test set, and validation set, respectively.

[0092] 203. Generate training images based on the training set.

[0093] In this embodiment of the application, after obtaining the training set, the first simulated grid included in the training set can be determined. The first simulated grid includes multiple first grid cells, and each first grid cell corresponds to a first target ore grade. The first target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell. Then, a bounding box is constructed based on the data features of the training set, and multiple second grid cells in the bounding box are interpolated using the inverse power distance method according to the multiple first ore grades corresponding to the multiple first grid cells in the training set to obtain a training image. Here, the multiple second grid cells are other grid cells in the bounding box besides the multiple first grid cells.

[0094] For example, taking copper grade ore as an example, in the data preprocessing stage, the borehole dataset containing accurate copper grade values ​​has been migrated to a simulated grid. Next, training images need to be generated. The specific process is as follows: a bounding box composed of grid cells is constructed based on the spatial distribution of the borehole dataset. The length, width, and height of the bounding box are the maximum values ​​of the x, y, and z coordinates of the borehole data points in the dataset, respectively. Then, the inverse power distance (IDW) method (with a power of 2 and an effective distance of 5) is used to interpolate the grid cells in the bounding box based on the copper grade values ​​of the training set borehole data. After deleting invalid grid cells, the image is obtained as shown below. Figure 5 The three-dimensional continuous training image shown.

[0095] 204. Based on the training images, training set, and test set, the ore grade is simulated using the direct sampling method to obtain the initial ore grade simulation results.

[0096] In this embodiment, after obtaining the training image, a second simulated grid is first determined based on the test set. The second simulated grid includes multiple second grid cells and the actual ore grade value corresponding to each second grid cell. The actual ore grade value is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a second grid cell. Then, a simulation path is determined based on the second simulated grid, and a grid to be simulated is established based on at least one borehole location included in each second grid cell in the test set. Each grid cell to be simulated includes at least one borehole location. Next, the training set is used as a conditional dataset. Based on the direct sampling method, each grid cell to be simulated is traversed according to the simulation path. At least one first grid cell matching each grid cell to be simulated is queried from the conditional dataset. A first hysteresis vector is calculated for each grid cell to be simulated based on the at least one first grid cell matching each grid cell to be simulated. Finally, the first hysteresis vector of each grid cell to be simulated is calculated based on the first hysteresis vector of each grid cell to be simulated. The first data event value of a grid cell is determined by the condition dataset, which includes multiple first grid cells and a first target ore grade corresponding to each first grid cell. The first target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell. Then, based on the direct sampling method, at least one view grid cell matching each grid cell to be simulated is queried from the training image. A second hysteresis vector of each grid cell to be simulated is calculated based on the at least one view grid cell matching each grid cell to be simulated, and a second data event value of each grid cell to be simulated is calculated based on the second hysteresis vector of each grid cell to be simulated. Next, the following operations are performed for each grid cell to be simulated: the difference between the first data event value and the second data event value is calculated based on the direct sampling method. When it is determined that the difference does not reach the target threshold, the ore grade of the view grid cell is used as the simulated ore grade value of the grid cell to be simulated. Finally, the initial ore grade simulation result is obtained based on the simulated ore grade value of each grid cell to be simulated.

[0097] For example, taking copper grade as an example, after obtaining the training image, the DS method can be used on the grid to be simulated to simulate the copper grade value of each grid cell. During the simulation, borehole data is selected as reliable conditional data to participate in the simulation. Before the simulation starts, the spatial coordinates (x, y, z) of each grid cell in the second simulation grid of the test set need to be mapped into a unique one-dimensional index value index = x + N in row-major order. x ×(y+N y ×z), where Nx and Ny are the maximum dimensions of the second simulation grid x and y, respectively. Then, all index values ​​are randomly sorted and saved into an array, which is called the simulation path.

[0098] Then, DS simulation can be performed on the grid cells in the simulation grid according to the order in the simulation path. The index of the first grid cell to be simulated is retrieved from the simulation path, and the corresponding grid cell is found, denoted as x; the copper grade value of grid cell x is denoted as Z(x). The process of modeling ore grade using the DS method is the process of simulating the Z(x) value. First, it is necessary to find the n conditional data grid cells {x1, x2, ..., x...} that are closest to the index value of x. n}, calculate the lag vector L={h1,…,h n}, where h n =x n -x, the hysteresis vector, indicates the relative position of the conditional data grid cell and the current grid cell to be simulated, x. In the initial stage of the simulation, the copper grade value of the conditional data grid cell (i.e., the first grid cell) has been determined by the borehole data. Therefore, a pattern centered on x can be obtained based on the hysteresis vector and the value of the conditional data, denoted as a data event, i.e.,

[0099] d n (x,L)={Z(x+h1),…,Z(x+h n )}.

[0100] Next, the training images are randomly scanned, and the pattern d centered on grid cell y in the training images is found based on the same hysteresis vector. n (y,L), compare d n (y,L) and d n The distance d{d between (x,L) n (x,L),d n (y,L)}. Since this method simulates the continuous variable of copper grade, the distance is calculated according to the following formula:

[0101]

[0102] in,

[0103] d max =max y∈TI Z(y)-min y∈TI Z(y),

[0104] Among them, h i Let δ be the lag vector and δ be the power series, which is usually 2.

[0105] When d{d n (x,L),d nWhen the distance to (y, L) is less than a given threshold t, the variable Z(y) at y is assigned to x, i.e., Z(x) = Z(y), completing the simulation of the copper grade value of x. If the distance is greater than the threshold, random scanning continues in the training image until the given maximum number of iterations i is reached, at which point the known minimum distance Z(y) is assigned to x. After completing the simulation of the copper grade value of x, x can be removed from the simulation path and treated as newly generated conditional data. Then, the next grid cell in the simulation path is simulated until the simulation path is empty. The above describes the process of simulating ore grade using the DS method.

[0106] 205. Based on the initial ore grade simulation results and the test set, the key parameters in the direct sampling method are optimized using the particle swarm optimization algorithm to obtain the optimal key parameters.

[0107] In this embodiment, after determining the initial ore grade simulation results, the simulated ore grade value corresponding to each grid cell to be simulated can be extracted from the initial ore grade simulation results based on the particle swarm optimization algorithm. A second grid cell matching the grid cell to be simulated is queried from the test set, and the actual ore grade value corresponding to the second grid cell is determined. The mean square error between the simulated ore grade value and the actual ore grade value is calculated. When the mean square error exceeds the global optimum, the key parameters in the direct sampling method are adjusted to obtain updated key parameters. The updated key parameters are then configured in the direct sampling method to obtain intermediate direct sampling. The method involves simulating ore grade using intermediate direct sampling based on training images, training sets, and test sets, obtaining intermediate ore grade simulation results. From these intermediate ore grade simulation results, updated ore grade simulation values ​​are extracted for each grid cell to be simulated. Additionally, updated second grid cells matching the grid cell to be simulated are queried from the test set. The actual updated ore grade value corresponding to the updated second grid cell is determined. The mean square error between the updated ore grade simulation value and the actual updated ore grade value is calculated. This process is repeated until the update mean square error does not exceed the global optimum, at which point the updated key parameters are determined as the optimal key parameters.

[0108] For example, taking copper grade as an example, after analyzing the DS simulation process, the maximum number of neighboring cells *n* directly affects how to obtain data events centered on the current simulated location from the simulation grid; the threshold *t* directly affects whether the data events found from the training image can match the data events in the simulation grid; and the maximum number of iterations *i* directly affects the quality of the suboptimal simulation results. When using the DS method to simulate copper grade, the selection of these three parameters is crucial. It is necessary to find a simple, efficient, and adaptive method to obtain the optimal parameter configuration, rather than subjectively selecting parameter configurations through tedious and repetitive sensitivity analysis. The emergence of the Particle Swarm Optimization (PSO) algorithm can find global or local optima through systematic searching and iteration. Directly selecting key parameters of the DS method based on experience may introduce subjectivity, and the selection process may be cumbersome. Utilizing the advantages of PSO's fast convergence speed and simple implementation, the PSO algorithm can be used to adaptively optimize the parameters in the DS method, making it convenient to determine the parameter configuration.

[0109] It's important to note that the PSO algorithm is based on the concept of particle swarm optimization, where each particle moves independently in a multi-dimensional space. The position of each particle can be considered a candidate solution to the objective function being optimized, and the quality of the solution is evaluated using the fitness value of the objective function. The process of finding the optimal solution to the objective function using the PSO algorithm is essentially the process of particles continuously moving within the search space. During the iteration process, each particle updates its velocity and position based on the individual best value (pbest) and the global best value (gbest) in Equation 3.

[0110]

[0111] In the formula, N represents the number of particles, and i represents the particle index, where i = 1, 2, ..., N. D represents the particle dimension, and d represents the dimension index, where d = 1, 2, ..., D. k represents the number of iterations, ω represents the inertia weight, and c1 and c2 represent the individual learning factor and the group learning factor, respectively. r1 and r2 are random numbers in the range [0, 1], used to introduce randomness in the search. This represents the velocity vector of the i-th particle in the d-th dimension at the k-th iteration. This represents the historical best position of the i-th particle in the d-dimensional space during the k-th iteration. This represents the historical best position of the entire particle swarm in the d-dimensional space at the k-th iteration.

[0112] Furthermore, adaptive optimization is performed on three parameters: the maximum number of neighboring units *n*, the threshold *t*, and the maximum number of iterations *i*. The aim is to find a parameter configuration that balances accuracy and computational efficiency in the DS simulation results. Within the framework of the PSO algorithm, a single particle represents a potential solution, and each dimension corresponds to a different parameter. Currently, three parameters need optimization; therefore, the dimension needs to be set to 3 so that the position of each particle becomes a three-dimensional vector. The fitness value is used to evaluate the quality of the solution obtained by each particle. This method focuses on ore grade modeling, so MSE (mean-square error) is used as the fitness value. MSE is a statistical indicator that quantifies the difference between predicted and actual values. Thus, the optimization process of the PSO algorithm can be viewed as allowing particles to continuously update their position and velocity, searching for solutions with smaller MSE values ​​in the three-dimensional solution space. After simulating the three key parameters of the DS method using the PSO algorithm, the optimal key parameters can be obtained. For ease of understanding, the specific process for obtaining the optimal key parameters can be found in [link to documentation]. Figure 7 .

[0113] 206. The optimal key parameters are configured in the direct sampling method to obtain the target direct sampling method.

[0114] In this embodiment, after obtaining the target direct sampling method, a comparative experiment can be added to determine the reliability and stability of the direct sampling algorithm configured with optimal key parameters. Specifically, firstly, a validation set is determined based on the borehole dataset with the simulated grid structure. The validation set includes multiple third borehole data with a third simulated grid structure, wherein the multiple third borehole data are other borehole data in the borehole dataset with the simulated grid structure, excluding the multiple first borehole data and the multiple second borehole data. Next, based on the validation set, the training image, and the training set, the target direct sampling method is used to simulate ore grade, obtaining the simulated ore grade result to be verified. Then, multiple simulated ore grade values ​​to be verified are extracted from the simulated ore grade result, wherein each simulated ore grade value to be verified is... Each value corresponds to a third grid cell; multiple actual ore grade values ​​are extracted from the verification set, wherein each actual ore grade value corresponds to a third grid cell; then, for each third grid cell, the following operation is performed: calculate the difference between the simulated ore grade value to be verified and the corresponding actual ore grade value, compare the difference with a target threshold, and obtain a comparison result. If the comparison result corresponding to each third grid cell indicates that the difference does not exceed the target threshold, then the target direct sampling method is determined to be the optimal direct sampling method. If the comparison result corresponding to a third grid cell indicates that the difference exceeds the target threshold, then the particle swarm optimization algorithm is used to optimize the key parameters in the direct sampling method to obtain updated optimal key parameters, and the updated optimal key parameters are configured in the direct sampling method to obtain an updated target direct sampling method.

[0115] 207. Based on the borehole dataset, training set and training images with simulated grid structure, the target direct sampling method is used to simulate the ore grade and obtain the target ore grade simulation results of the ore body to be simulated.

[0116] In this implementation, after configuring the optimal key parameters using the direct sampling method, the entire ore body to be simulated can be simulated. This ore grade simulation process is the same as the ore grade simulation process in step 204, except that the test set is replaced with the borehole dataset of the entire ore body to be simulated; this borehole dataset is the processed dataset. Since the ore grade simulation process is the same, it will not be described in detail here. The final simulation result is a high-precision simulation result of the target ore grade of the ore body to be simulated.

[0117] To further illustrate the effectiveness of this application, a DS method using random parameter configurations was added as a comparison method, employing a total of 50 sets of random key parameter configurations. Furthermore, this method focuses on ore grade simulation, and therefore, two traditional spatial interpolation methods were added for comparison: the IDW method with a power of 2 and an effective distance of 5, and the ordinary Kriging method (OK) using a spherical variogram model.

[0118] First, the test set is simulated. The iterative process of the method in this embodiment is as follows: Figure 8 As shown in the figure. After the PSO algorithm converges, the cumulative density plot of the absolute error calculated from the actual values ​​using the simulation results obtained by the method of this application embodiment and the results of the comparison method is shown in the figure. Figure 9 As shown, the absolute error of the simulation results obtained by the method in this embodiment is more concentrated in the low-value region close to 0 than that of other methods, indicating that the method in this embodiment can find parameter configurations that result in higher accuracy of DS simulation results. Then, the parameter configuration is used to simulate the validation set. Figure 10 It can be seen that the obtained parameter configuration allows the DS method to maintain a certain advantage in simulation accuracy compared to the comparative methods. Finally, the entire ore body to be simulated was simulated using this key parameter configuration, and the ore grade distribution of the simulation results is as follows: Figure 11 As shown, the DS simulation results configured using the obtained key parameters are closer to the actual data in terms of ore grade distribution. Figure 12 The variogram shown also confirms this from the perspective of second-order statistics. The results demonstrate that the method described in this application can yield optimal key parameter configurations that result in higher accuracy of the DS method and stronger ability to reconstruct the spatial structure of actual data, thus solving the problem of cumbersome key parameter settings and a lack of unified standards when applying the DS method to ore grade simulation.

[0119] Furthermore, knowing the simulation results of the target ore grade of the ore body to be simulated, the target simulation values ​​of all grid cells in the target ore grade simulation grid can be known. Since the position of each grid cell corresponds to the position of a borehole point, the error of the simulation result can be calculated based on the simulated value and the actual value at the same position. This error is considered an indicator of the accuracy of the method in this embodiment. Assuming there is a grid x in the simulation grid and a corresponding borehole point in the borehole data, the actual value c of x can be known. The method in this embodiment simulates the grid, obtaining a simulated value a for grid x; next, other comparison methods, such as the IDW method, are used to simulate the grid, obtaining an IDW simulated value b for grid x; the errors of a and c, and b and c are calculated respectively, and the smaller error is compared. If the calculated errors of a and c are smaller, it indicates that the method in this embodiment simulates grid x more accurately. In practical applications, the simulated grid and the borehole data overlap in many places. If, after multiple comparisons, the method of this application embodiment is superior to the comparative method in terms of simulation accuracy, it indicates that the method of this application embodiment has high simulation accuracy for copper grade of ore.

[0120] Furthermore, as Figure 1To specifically implement the method, this embodiment of the invention provides an ore grade simulation device based on a direct sampling method, such as... Figure 13 As shown, it includes: acquisition module 301, generation module 302, optimization module 303 and simulation module 304.

[0121] The acquisition module 301 is used to acquire the borehole dataset of the ore body to be simulated, and to migrate the borehole dataset to the simulation grid to obtain a borehole dataset with a simulation grid structure. The borehole dataset includes multiple borehole data, and each borehole data includes the borehole location and the ore grade corresponding to the borehole location.

[0122] The generation module 302 is used to determine a training set and a test set based on the borehole dataset with a simulated grid structure, and to generate a training image based on the training set, wherein the training set includes a plurality of first borehole data with a first simulated grid structure, and the test set includes a plurality of second borehole data with a second simulated grid structure.

[0123] The optimization module 303 is used to simulate ore grade using the direct sampling method based on the training image, the training set, and the test set to obtain an initial ore grade simulation result, and to optimize the key parameters in the direct sampling method using the particle swarm optimization algorithm based on the initial ore grade simulation result and the test set to obtain the optimal key parameters.

[0124] The simulation module 304 is used to configure the optimal key parameters in the direct sampling method to obtain the target direct sampling method, and to simulate the ore grade using the target direct sampling method based on the borehole dataset with the simulated grid structure, the training set and the training image to obtain the target ore grade simulation result of the ore body to be simulated.

[0125] In specific application scenarios, the acquisition module 301 is further used to: acquire the boundary information of the ore body to be simulated; establish a geological body constraint model based on the boundary information of the target ore body, and determine a three-dimensional simulation grid based on the data characteristics of the borehole dataset; trim the three-dimensional simulation grid using the geological body constraint model to obtain the simulation grid; migrate multiple borehole data included in the borehole dataset to the simulation grid to obtain the borehole dataset with a simulation grid structure, wherein the borehole dataset with a simulation grid structure includes multiple grid cells and the grid position and target ore grade corresponding to each grid cell, wherein the target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole position included in a grid cell.

[0126] In a specific application scenario, the generation module 302 is further configured to: determine the first simulated grid included in the training set, wherein the first simulated grid includes a plurality of first grid cells, and each first grid cell corresponds to a first target ore grade, the first target ore grade being the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell; construct a bounding box based on the data features of the training set, and interpolate a plurality of second grid cells in the bounding box using the inverse power distance method according to the plurality of first ore grades corresponding to the plurality of first grid cells in the training set, to obtain the training image, wherein the plurality of second grid cells are other grid cells in the bounding box besides the plurality of first grid cells.

[0127] In specific application scenarios, the optimization module 303 is further configured to: determine a second simulation grid based on the test set, wherein the second simulation grid includes multiple second grid cells and the actual ore grade value corresponding to each second grid cell, the actual ore grade value being the average ore grade of at least one ore grade corresponding to at least one borehole location included in a second grid cell; determine a simulation path based on the second simulation grid, and establish a grid to be simulated based on at least one borehole location included in each second grid cell in the test set, wherein each grid to be simulated cell in the grid to be simulated includes at least one borehole location; use the training set as a conditional dataset, traverse each grid to be simulated cell in the grid to be simulated according to the simulation path based on the direct sampling method, query at least one first grid cell matching each grid to be simulated cell from the conditional dataset, calculate the first hysteresis vector of each grid to be simulated cell based on the at least one first grid cell matching each grid to be simulated cell, and calculate the first hysteresis vector of each grid to be simulated cell based on the first hysteresis vector of each grid to be simulated cell. The conditional dataset includes a plurality of first grid cells and a first target ore grade corresponding to each first grid cell. The first target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell. Based on the direct sampling method, at least one view grid cell matching each grid cell to be simulated is queried from the training image. A second hysteresis vector of each grid cell to be simulated is calculated based on the at least one view grid cell matching each grid cell to be simulated, and a second data event value of each grid cell to be simulated is calculated based on the second hysteresis vector of each grid cell to be simulated. For each grid cell to be simulated, the following operations are performed: the difference between the first data event value and the second data event value is calculated based on the direct sampling method; if the difference does not reach a target threshold, the ore grade of the view grid cell is used as the simulated ore grade value of the grid cell to be simulated; the initial ore grade simulation result is obtained based on the simulated ore grade value of each grid cell to be simulated.

[0128] In specific application scenarios, the optimization module 303 is further configured to: extract the simulated ore grade value corresponding to each grid cell to be simulated from the initial ore grade simulation results based on the particle swarm optimization algorithm; query the second grid cell that matches the grid cell to be simulated from the test set; determine the actual ore grade value corresponding to the second grid cell; calculate the mean square error between the simulated ore grade value and the actual ore grade value; when the mean square error exceeds the global optimum, adjust the key parameters in the direct sampling method to obtain updated key parameters; configure the updated key parameters in the direct sampling method to obtain an intermediate direct sampling method. The training images, the training set, and the test set are used to simulate ore grade using the intermediate direct sampling method to obtain intermediate ore grade simulation results. The updated simulated ore grade value corresponding to each grid cell to be simulated is extracted from the intermediate ore grade model simulation results. An updated second grid cell matching the grid cell to be simulated is queried from the test set. The updated actual ore grade value corresponding to the updated second grid cell is determined. The update mean square error between the updated simulated ore grade value and the updated actual ore grade value is calculated. The above steps are repeated until the update mean square error does not exceed the global optimum value. The updated key parameter is then determined as the optimal key parameter.

[0129] In specific application scenarios, such as Figure 14 As shown, the device also includes a verification module 305.

[0130] The verification module 305 is used to determine a verification set based on the borehole dataset with a simulated grid structure. The verification set includes multiple third borehole data with a third simulated grid structure, wherein the multiple third borehole data are other borehole data in the borehole dataset with a simulated grid structure, excluding the multiple first borehole data and the multiple second borehole data. Based on the verification set, the training image, and the training set, the target direct sampling method is used to simulate ore grade to obtain a simulated ore grade to be verified. Multiple simulated ore grade values ​​to be verified are extracted from the simulated ore grade to be verified, wherein each simulated ore grade value to be verified corresponds to a third grid cell. Multiple actual ore grade values ​​are extracted from the verification set. The actual value is defined as follows: each actual ore grade corresponds to a third grid cell. For each third grid cell, the following operations are performed: the difference between the simulated ore grade to be verified and the corresponding actual ore grade is calculated, and the difference is compared with a target threshold to obtain a comparison result. If the comparison result for each third grid cell indicates that the difference does not exceed the target threshold, then the target direct sampling method is determined to be the optimal direct sampling method. If the comparison result for a third grid cell indicates that the difference exceeds the target threshold, then the particle swarm optimization algorithm is used to further optimize the key parameters in the direct sampling method to obtain updated optimal key parameters, and the updated optimal key parameters are configured in the direct sampling method to obtain an updated target direct sampling method.

[0131] In specific application scenarios, the generation module 302 is also used to: determine a preset ratio and divide the borehole dataset with simulated grid structure into the training set, the test set and the validation set according to the preset ratio.

[0132] The apparatus provided in this application first preprocesses the borehole data of the ore body to be simulated through an acquisition module to obtain a borehole dataset with a simulated grid structure. Then, a generation module determines a training set and a test set based on this borehole dataset. Knowing the training set, it can be used as conditional data or a training image can be generated based on it. An optimization module simulates ore grade on the test set using a direct sampling method based on the training image and conditional data to obtain initial ore grade simulation results. Based on these initial ore grade simulation results, a particle swarm optimization algorithm is used to adaptively optimize the key parameters in the direct sampling method. By quickly identifying highly sensitive key parameters and configuring them as the optimal key parameters in the direct sampling method, the target direct sampling method is obtained. Finally, for the ore body to be simulated, the simulation module uses the previously obtained processed borehole dataset, combined with the training set and training images, to simulate the ore grade through the target direct sampling method, thus obtaining the target ore grade simulation result for the ore body to be simulated. Since highly sensitive key parameters can be obtained through the above process, subsequent ore grade simulation using the direct sampling method configured with these key parameters can yield a highly accurate target ore grade simulation result for the ore body to be simulated.

[0133] It should be noted that other corresponding descriptions of the functional units involved in the ore grade simulation device based on the direct sampling method provided in this application embodiment can be found in the following references. Figure 1 and Figure 4 The corresponding description in [the document] will not be repeated here.

[0134] In an exemplary embodiment, see Figure 15 The invention also provides a computer device including a communication bus, a processor, a memory, and a communication interface. It may also include an input / output interface and a display device, wherein the various functional units can communicate with each other via the bus. The memory stores a computer program, and the processor executes the program stored in the memory to perform the ore grade simulation method based on direct sampling described in the above embodiments.

[0135] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the ore grade simulation method based on direct sampling.

[0136] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented in hardware or by using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0137] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application. Those skilled in the art will understand that the modules in the apparatus of the embodiment can be distributed within the apparatus of the embodiment as described, or can be modified to be located in one or more apparatuses different from this embodiment. The modules of the above-described embodiment can be combined into one module, or further divided into multiple sub-modules.

[0138] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.

Claims

1. A method for simulating ore grade based on direct sampling, characterized in that, include: Obtain the borehole dataset of the ore body to be simulated, and migrate the borehole dataset to the simulation grid to obtain the borehole dataset with the simulation grid structure. The borehole dataset includes multiple borehole data, and each borehole data includes the borehole location and the ore grade corresponding to the borehole location. A training set and a test set are determined based on the borehole dataset with the simulated grid structure, and a training image is generated based on the training set, wherein the training set includes multiple first borehole data with a first simulated grid structure, and the test set includes multiple second borehole data with a second simulated grid structure; Based on the training images, the training set, and the test set, the ore grade is simulated using the direct sampling method to obtain initial ore grade simulation results. Then, based on the initial ore grade simulation results and the test set, the key parameters in the direct sampling method are optimized using the particle swarm optimization algorithm to obtain the optimal key parameters. The optimal key parameters are configured in the direct sampling method to obtain the target direct sampling method. Based on the borehole dataset with simulated grid structure, the training set and the training image, the target direct sampling method is used to simulate the ore grade to obtain the target ore grade simulation result of the ore body to be simulated. The step of simulating ore grade using the direct sampling method based on the training images, the training set, and the test set to obtain initial ore grade simulation results includes: A second simulation grid is determined based on the test set, wherein the second simulation grid includes a plurality of second grid cells and an actual ore grade value corresponding to each second grid cell, and the actual ore grade value is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a second grid cell; The simulation path is determined based on the second simulation grid, and a simulation grid is established based on at least one borehole location included in each second grid cell of the test set, wherein each simulation grid cell includes at least one borehole location. Using the training set as a conditional dataset, the direct sampling method is used to traverse each grid cell in the grid to be simulated according to the simulation path. At least one first grid cell that matches each grid cell is queried from the conditional dataset. A first hysteresis vector of each grid cell is calculated based on the at least one first grid cell that matches each grid cell. A first data event value of each grid cell is calculated based on the first hysteresis vector of each grid cell. The conditional dataset includes multiple first grid cells and each first grid cell corresponds to a first target ore grade. The first target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell. Based on the direct sampling method, at least one view grid cell that matches each grid cell to be simulated is queried from the training image; a second hysteresis vector of each grid cell to be simulated is calculated based on the at least one view grid cell that matches each grid cell to be simulated; and a second data event value of each grid cell to be simulated is calculated based on the second hysteresis vector of each grid cell to be simulated. For each grid cell to be simulated, perform the following operations: calculate the difference between the first data event value and the second data event value based on the direct sampling method; when it is determined that the difference does not reach the target threshold, use the ore grade of the view grid cell as the simulated ore grade value of the grid cell to be simulated. The initial ore grade simulation result is obtained based on the ore grade simulation value of each grid cell to be simulated.

2. The ore grade simulation method based on direct sampling according to claim 1, characterized in that, The step of migrating the borehole dataset to a simulated grid to obtain a borehole dataset with a simulated grid structure includes: Obtain the boundary information of the ore body to be simulated; A geological constraint model is established based on the boundary information of the target ore body, and a three-dimensional simulation mesh is determined based on the data characteristics of the borehole dataset. The three-dimensional simulation mesh is obtained by trimming the geological body constraint model; The borehole dataset includes multiple borehole data points, which are then migrated to the simulated grid to obtain the borehole dataset with a simulated grid structure. The borehole dataset with a simulated grid structure includes multiple grid cells and the grid position and target ore grade corresponding to each grid cell. The target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole position included in a grid cell.

3. The ore grade simulation method based on direct sampling according to claim 1, characterized in that, The step of generating training images based on the training set includes: The training set includes a first simulated grid, wherein the first simulated grid includes a plurality of first grid cells, and each first grid cell corresponds to a first target ore grade, the first target ore grade being the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell; A bounding box is constructed based on the data features of the training set, and the training image is obtained by interpolating the multiple second grid units in the bounding box according to the multiple first ore grades corresponding to the multiple first grid units in the training set using the inverse power distance method. The multiple second grid units are other grid units in the bounding box besides the multiple first grid units.

4. The ore grade simulation method based on direct sampling according to claim 1, characterized in that, The process involves optimizing the key parameters in the direct sampling method using a particle swarm optimization algorithm based on the initial ore grade simulation results and the test set to obtain optimal key parameters, including: Based on the particle swarm optimization algorithm, the simulated ore grade value corresponding to each grid cell to be simulated is extracted from the initial ore grade simulation results. A second grid cell matching the grid cell to be simulated is queried from the test set, and the actual ore grade value corresponding to the second grid cell is determined. The mean square error between the simulated ore grade value and the actual ore grade value is calculated. When the mean square error exceeds the global optimum, the key parameters in the direct sampling method are adjusted to obtain updated key parameters. The updated key parameters are configured in the direct sampling method to obtain an intermediate direct sampling method. Based on the training image and the training set... The intermediate direct sampling method is used to simulate ore grade on the test set to obtain intermediate ore grade simulation results. The updated ore grade simulation value corresponding to each grid cell to be simulated is extracted from the intermediate ore grade model simulation results. The updated second grid cell that matches the grid cell to be simulated is queried from the test set. The updated actual ore grade value corresponding to the updated second grid cell is determined. The update mean square error between the updated ore grade simulation value and the updated actual ore grade value is calculated. The above steps are repeated until the update mean square error does not exceed the global optimum value. The updated key parameter is then determined as the optimal key parameter.

5. The ore grade simulation method based on direct sampling according to claim 1, characterized in that, After configuring the optimal key parameters in the direct sampling method to obtain the target direct sampling method, the method further includes: A validation set is determined based on the borehole dataset with the simulated grid structure. The validation set includes multiple third borehole data with a third simulated grid structure, wherein the multiple third borehole data are other borehole data in the borehole dataset with the simulated grid structure other than the multiple first borehole data and the multiple second borehole data. Based on the validation set, the training images, and the training set, the target direct sampling method is used to simulate the ore grade, and the simulation results of the ore grade to be verified are obtained. Multiple simulated values ​​of ore grade to be verified are extracted from the simulated results of the ore grade to be verified, wherein each simulated value of ore grade to be verified corresponds to a third grid cell; Multiple actual ore grade values ​​are extracted from the verification set, wherein each actual ore grade value corresponds to a third grid cell; For each third grid cell, perform the following operations: calculate the difference between the simulated value of the ore grade to be verified and the corresponding actual value of the ore grade, compare the difference with the target threshold, and obtain the comparison result; If the difference in the comparison result corresponding to each third grid cell does not exceed the target threshold, then the target direct sampling method is determined to be the optimal direct sampling method; If the comparison result of a third grid cell indicates a difference exceeding the target threshold, the particle swarm optimization algorithm is used to further optimize the key parameters in the direct sampling method to obtain updated optimal key parameters. These updated optimal key parameters are then configured in the direct sampling method to obtain an updated target direct sampling method.

6. The ore grade simulation method based on direct sampling according to claim 5, characterized in that, include: A preset ratio is determined, and the borehole dataset with simulated grid structure is divided into the training set, the test set, and the validation set according to the preset ratio.

7. A device for simulating ore grade based on direct sampling, characterized in that, include: The acquisition module is used to acquire the borehole dataset of the ore body to be simulated, and to migrate the borehole dataset to the simulation grid to obtain a borehole dataset with a simulation grid structure. The borehole dataset includes multiple borehole data, and each borehole data includes the borehole location and the ore grade corresponding to the borehole location. A generation module is configured to determine a training set and a test set based on the borehole dataset with a simulated grid structure, and to generate training images based on the training set, wherein the training set includes multiple first borehole data with a first simulated grid structure, and the test set includes multiple second borehole data with a second simulated grid structure. An optimization module is used to simulate ore grade using the direct sampling method based on the training image, the training set, and the test set to obtain an initial ore grade simulation result, and to optimize the key parameters in the direct sampling method using a particle swarm optimization algorithm based on the initial ore grade simulation result and the test set to obtain the optimal key parameters. The simulation module is used to configure the optimal key parameters in the direct sampling method to obtain the target direct sampling method, and to simulate the ore grade using the target direct sampling method based on the borehole dataset with the simulated grid structure, the training set and the training image to obtain the target ore grade simulation result of the ore body to be simulated. The optimization module is further configured to: determine a second simulation grid based on the test set, wherein the second simulation grid includes multiple second grid cells and an actual ore grade value corresponding to each second grid cell, the actual ore grade value being the average ore grade of at least one ore grade corresponding to at least one borehole location included in a second grid cell; determine a simulation path based on the second simulation grid; and establish a grid to be simulated based on at least one borehole location included in each second grid cell in the test set, wherein each grid cell to be simulated includes at least one borehole location; use the training set as a conditional dataset; traverse each grid cell to be simulated according to the simulation path using the direct sampling method; query at least one first grid cell matching each grid cell to be simulated from the conditional dataset; calculate a first hysteresis vector for each grid cell to be simulated based on the at least one first grid cell matching each grid cell to be simulated; and calculate the first hysteresis vector for each grid cell to be simulated based on the first hysteresis vector of each grid cell to be simulated. The conditional dataset includes multiple first grid cells and each first grid cell corresponds to a first target ore grade. The first target ore grade is the average ore grade of at least one ore grade corresponding to at least one borehole location included in a first grid cell. Based on the direct sampling method, at least one view grid cell matching each grid cell to be simulated is queried from the training image. A second hysteresis vector of each grid cell to be simulated is calculated based on the at least one view grid cell matching each grid cell to be simulated. A second data event value of each grid cell to be simulated is calculated based on the second hysteresis vector of each grid cell to be simulated. For each grid cell to be simulated, the following operations are performed: the difference between the first data event value and the second data event value is calculated based on the direct sampling method. When it is determined that the difference does not reach the target threshold, the ore grade of the view grid cell is used as the simulated ore grade value of the grid cell to be simulated. The initial ore grade simulation result is obtained based on the simulated ore grade value of each grid cell to be simulated.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images