A three-dimensional mineralization prediction method for concealed ore bodies in deep margins

Abstract

The present invention relates to the field of deep exploration technology for mineral resource exploration, and in particular to a three-dimensional mineralization prediction method for deep side hidden ore bodies. The content includes: obtaining basic data of deep side hidden ore bodies, and preprocessing to obtain preprocessed basic data; establishing a three-dimensional geological model based on the preprocessed basic data; based on the three-dimensional geological model, quantitatively and qualitatively analyzing the preprocessed basic data to obtain comprehensive mineralization control characteristics; combining the comprehensive mineralization control characteristics, constructing a three-dimensional mineralization prediction model, and using the output of the three-dimensional mineralization prediction model as the prediction result; according to the prediction result, delineating the prospecting target area, and performing three-dimensional mineralization prediction of deep side hidden ore bodies. The method solves the technical problems of the traditional prediction method in the three-dimensional mineralization prediction of deep side hidden ore bodies, that is, the analysis and processing of deep side ore body data is not accurate enough, and the prediction accuracy is low.

Description

Technical Field

[0001] The invention relates to the technical field of deep exploration of mineral resources, and in particular to a three-dimensional mineralization prediction method for a deep side concealed ore body. Background Art

[0002] With the continuous development of mineral resources, the number of shallow ore bodies that are easy to mine has gradually decreased, and global mineral resource exploration has begun to shift to the detection of deep and hidden ore bodies. Deep and hidden ore bodies are characterized by deep burial, strong concealment, and complex distribution. Traditional two-dimensional mineralization prediction methods and shallow exploration methods are gradually unable to cope with these deep and hidden ore bodies. In order to improve the detection capability of deep and hidden ore bodies, three-dimensional mineralization prediction technology has gradually become an important development direction in the exploration field.

[0003] The 3D mineralization prediction technology integrates multiple sources of data such as geology, geophysics, and geochemistry, and uses mathematical modeling and computer technology to predict underground ore bodies in space. However, the existing 3D mineralization prediction technology still faces many challenges when dealing with deep and hidden ore bodies: first, the acquisition cost of deep and hidden ore body data is high, and the data is sparse and discontinuous. How to effectively pre-process and fuse these multi-source heterogeneous data to establish an accurate 3D geological model has become one of the difficulties in current research.

[0004] In summary, the traditional prediction methods have technical problems in the three-dimensional mineralization prediction of deep and hidden ore bodies, such as inaccurate analysis and processing of deep and hidden ore body data and low prediction accuracy. Summary of the invention

[0005] The present invention provides a three-dimensional mineralization prediction method for deep side concealed ore bodies, so as to solve the technical problems that the traditional prediction method has inaccurate analysis and processing of deep side ore body data and low prediction accuracy in the three-dimensional mineralization prediction of deep side concealed ore bodies.

[0006] A three-dimensional mineralization prediction method for a deep side concealed ore body of the present invention specifically includes the following technical solutions:

[0007] A three-dimensional mineralization prediction method for a deep side concealed ore body comprises the following steps:

[0008] S1. Obtain basic data of the deep and side concealed ore bodies and preprocess them to obtain preprocessed basic data; establish a three-dimensional geological model based on the preprocessed basic data; based on the three-dimensional geological model, conduct quantitative and qualitative analysis on the preprocessed basic data to obtain comprehensive mineralization control characteristics;

[0009] S2. Combined with the comprehensive mineralization control characteristics, a three-dimensional mineralization prediction model is constructed, and the output of the three-dimensional mineralization prediction model is used as the prediction result; based on the prediction result, the prospecting target area is delineated, and three-dimensional mineralization prediction of the deep and hidden ore bodies is carried out.

[0010] Preferably, the S1 specifically includes:

[0011] Based on the preprocessed basic data, a boundary enhancement interpolation algorithm is introduced; in the implementation process of the boundary enhancement interpolation algorithm, interpolation coefficients are introduced to construct an optimized interpolation function.

[0012] Preferably, the S1 specifically includes:

[0013] In the process of optimizing the calculation of the interpolation function, based on the boundary point data set in the preprocessed basic data, the boundary enhancement coefficient is introduced to control the influence range of the boundary points on the interpolation area. The specific formula is:

[0014] ,

[0015] in, It is in Boundary Points The interpolation result of is the border enhancement coefficient that adjusts the border enhancement effect; represents an unknown data point; is a known boundary point dataset.

[0016] Preferably, the S1 specifically includes:

[0017] Through the Lagrangian function, the interpolation error term and the boundary condition constraint term are introduced to optimize the interpolation coefficient and obtain the optimal solution of the interpolation coefficient. Based on the optimal solution of the interpolation coefficient, the optimized interpolation function is further obtained. The specific formula of the Lagrangian function is as follows:

[0018] ,

[0019] in, is the Lagrangian function; It is The actual value of the known data points; is the interpolation coefficient; is the kernel function, which represents the known data points With known data points similarities between; They are and known data points; It is Lagrange multipliers; is the total number of known data points; Indicates the total number of known boundary point data; represents the interpolation error term; is the boundary condition constraint; is the control parameter in the kernel function.

[0020] Preferably, the S2 specifically includes:

[0021] The specific construction process of the three-dimensional mineralization prediction model includes:

[0022] The first step is to collect historical data of hidden ore bodies in the deep side;

[0023] The second step is to map the distribution values ​​of the ore bodies in the historical data of the deep and side hidden ore bodies to obtain the ore body distribution values ​​after interpolation processing;

[0024] The third step is to generate the mineralization probability distribution based on the interpolated ore body distribution value;

[0025] The fourth step is to obtain the rate of change of the mineralization probability distribution over time through spatiotemporal evolution processing based on the mineralization probability distribution, and update the mineralization probability distribution based on the rate of change of the mineralization probability distribution over time;

[0026] The fifth step is to perform recursive convolution evolution processing based on the updated mineralization probability distribution to obtain the prediction result of mineralization probability distribution;

[0027] The sixth step is to estimate the ore body resources based on the predicted results of mineralization probability distribution.

[0028] Preferably, the S2 specifically includes:

[0029] In the process of distribution mapping, the ore body distribution values ​​in the historical data of the deep side hidden ore bodies are spatially interpolated by the multidimensional nonlinear geometric transformation interpolation algorithm to obtain the ore body distribution values ​​after interpolation processing; the multidimensional nonlinear geometric transformation interpolation formula is:

[0030] ,

[0031] in, It is a time-space point The ore body distribution value after interpolation at ; In the historical data of the deep side concealed ore body, it is located at the time and space point The volume of the ore body; is located at a point in space and time The volume of the ore body is and The second derivative of direction; is the geometric weight function; is the integration variable; A time-space point in the historical data representing a concealed ore body at a deep margin.

[0032] Preferably, the S2 specifically includes:

[0033] In the process of spatiotemporal evolution processing, nonlinear stochastic partial differential equations are introduced to simulate the spatiotemporal changes of comprehensive mineralization control characteristics.

[0034] Preferably, the S2 specifically includes:

[0035] In the process of recursive convolution evolution processing, a recursive convolution evolution network is designed to obtain the prediction results of mineralization probability distribution; the formula of the recursive convolution evolution network is:

[0036] ,

[0037] in, is the recursive convolutional evolution network for space-time points The prediction results of the probability distribution of mineralization at the location; is the activation function; is the weight of the convolution kernel, indicating the The convolutional layer and the Time step The weight coefficient between ; It is The convolutional layer has a time step Output when is the convolution operation symbol; is the number of convolutional layers; is the number of time steps.

[0038] Preferably, the S2 specifically includes:

[0039] A multi-dimensional segmented spatial integration algorithm is introduced to estimate the ore body resources by combining the predicted results of mineralization probability distribution and the changes in ore body density.

[0040] The beneficial effects of the technical solution of the present invention are:

[0041] 1. With the help of 3D modeling software and boundary enhancement interpolation algorithm, the 3D geological model of the mining area can be constructed quickly and accurately. By generating spatially continuous geological bodies through layer-by-layer interpolation technology, coupled with the identification and modeling of fault structures and alteration zones, the possible location of the ore body can be accurately determined, especially for the modeling of deep and marginal ore bodies, which solves the limitations of traditional 2D mineralization prediction methods.

[0042] 2. Construct a three-dimensional mineralization prediction model, introduce nonlinear stochastic partial differential equations and time-space evolution formulas, and dynamically simulate the evolution of the mineralization process over time and space. The three-dimensional mineralization prediction model not only takes into account the generation, diffusion and attenuation process of the ore body, but also introduces random noise terms to simulate the unpredictable factors in the geological environment. The three-dimensional mineralization prediction model is more accurate than the traditional static model and can reflect the enrichment and change trend of the ore body in different time and space.

[0043] 3. Through the multi-dimensional segmented spatial integration algorithm, combined with the predicted results of the mineralization probability distribution of the ore body and the change of the ore body density, the ore body resources can be accurately estimated. Different from the traditional resource estimation method, the multi-dimensional segmented spatial integration algorithm takes into account the temporal and spatial changes of the ore body density, making the estimation of the ore body resources more in line with the actual situation, especially suitable for complex deep and marginal mining areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The present invention provides a flow chart of a three-dimensional mineralization prediction method for a deep side concealed ore body. DETAILED DESCRIPTION

[0045] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0047] The specific scheme of the three-dimensional mineralization prediction method for a deep side concealed ore body provided by the present invention is described in detail below with reference to the accompanying drawings.

[0048] See attached Figure 1 , which shows a flow chart of a three-dimensional mineralization prediction method for a deep side concealed ore body provided by an embodiment of the present invention, the method comprising the following steps:

[0049] S1. Obtain basic data of the deep side concealed ore body and perform preprocessing to obtain preprocessed basic data; establish a three-dimensional geological model based on the preprocessed basic data; based on the three-dimensional geological model, perform quantitative and qualitative analysis on the preprocessed basic data to obtain comprehensive mineralization control characteristics;

[0050] Obtain the basic data of the deep and side hidden ore bodies, the basic data include geological data, geophysical data, geochemical data; in order to ensure the accuracy of modeling, preprocess the basic data to obtain preprocessed basic data; the preprocessed basic data include drilling data, surface measurement data, geophysical profile data, boundary point data, etc. The preprocessing process includes data cleaning, data coordinate conversion, data interpolation and smoothing. The methods used in the preprocessing process are all technical means well known to those skilled in the art, and will not be repeated here.

[0051] Based on the preprocessed basic data, a three-dimensional geological model of the deep and side concealed ore body is established using three-dimensional geological modeling software (such as Maptek Vulcan and LeapfrogGeo). The specific implementation includes: importing the preprocessed basic data into the three-dimensional geological modeling software to ensure that the preprocessed basic data can accurately reflect the real spatial position; reconstructing the three-dimensional morphology of the geology based on the preprocessed basic data using the modeling software, especially key geological units such as structural faults and alteration zones; and using the three-dimensional geological modeling software to generate a spatially continuous geological body through layer-by-layer interpolation; then based on the position of the fault structure, identifying and modeling the key structural belts and mineralized belts, further locating the possible three-dimensional spatial position of the ore body, and realizing the establishment of a three-dimensional geological model.

[0052] In order to avoid the boundary fuzziness problem caused by interpolation when processing complex geological units, a boundary enhancement interpolation algorithm is introduced, and the preprocessed basic data (i.e. known data points) are used as the input of the boundary enhancement interpolation algorithm; the boundary enhancement interpolation algorithm avoids the problem of excessive smoothing during the interpolation process by introducing boundary constraints. The specific implementation process is as follows:

[0053] The first step is to construct an optimized interpolation function to interpolate unknown data points; the optimized interpolation function is expressed as:

[0054] ,

[0055] in, is the output of the optimized interpolation function, indicating the unknown data point The interpolated value at ; is the total number of known data points, providing the basic data for interpolation; is the interpolation coefficient, reflecting the value of each known data point For unknown data points The degree of influence of is determined by optimization algorithms such as the Lagrange multiplier method, indicating that The contribution of known data points to the interpolation result; It is a control parameter in the kernel function, which is used to adjust the width of the kernel function and determine the decay rate of the kernel function; This is the kernel function part; in the kernel function, trigonometric functions are introduced ; Make the influence of data points with larger distance on the interpolation result more controllable; Exponential function Used to control the overall attenuation trend of the kernel function, ensuring that the farther the distance, the smaller the impact;

[0056] Furthermore, in order to ensure that the optimized interpolation function can accurately reflect the boundary data at the boundary points, a boundary enhancement condition is introduced; in order to ensure the strictness of this condition, a boundary enhancement coefficient is introduced to further control the influence range of the boundary points on the interpolation area through an exponential function. The specific formula is:

[0057] ,

[0058] in, It is in Boundary Points The interpolation result of It is the boundary enhancement coefficient that adjusts the boundary enhancement effect, and is used to control the influence range of the boundary point on the interpolation area; is a known boundary point data set, which is taken from the preprocessed basic data; the exponential function is used to control the influence of the distance between the unknown data point and the boundary point on the interpolation. The addition of the logarithmic function ensures that when the unknown data point is far away from the boundary point, the influence of the boundary condition gradually decays, preventing the boundary point from causing excessive interference to the interpolation area;

[0059] To solve for the interpolation coefficients , the Lagrange multiplier method is used. By constructing a Lagrange function containing interpolation error terms and boundary condition constraints, the interpolation coefficients are optimized. The form of the Lagrange function is as follows:

[0060] ,

[0061] in, is the Lagrangian function used to minimize the interpolation error and ensure the constraints of boundary conditions; It is The actual value of the known data points; is the interpolation coefficient, indicating the The contribution of known data points to the interpolation result; is the kernel function, which represents the known data points With known data points The similarity between them is controlled by the kernel function, and the influence of distance is controlled. The closer the distance, the greater the influence. They are and known data points; is the Lagrange multiplier, which is used to introduce boundary enhancement conditions to ensure that the optimized interpolation function is at the boundary points The constraints of satisfying known data points are met; Indicates the total number of known boundary point data; represents the interpolation error term, that is, the numerical difference between the actual value of the known data point and the interpolated value of the known data point; the second term It is the boundary condition constraint term, and the Lagrange multiplier is introduced to ensure that the values ​​of the boundary points strictly meet the constraint conditions.

[0062] In order to obtain the final interpolation coefficients , based on the Lagrangian function Respectively and Find the partial derivatives and get a set of nonlinear equations. Solve this set of nonlinear equations to get the interpolation coefficients The optimal solution of is obtained, and the optimized interpolation function is further obtained to realize the boundary enhancement interpolation processing.

[0063] After the three-dimensional geological model is established, the pre-processed basic data is quantitatively and qualitatively analyzed to obtain comprehensive mineralization control characteristics;

[0064] The quantitative analysis is to perform numerical processing on the pre-processed basic data, extract the mineralization control characteristics therein, and obtain the quantitative analysis results; the main process includes:

[0065] Quantification of mineralization factors: Determine the key factors controlling the formation of ore bodies based on expert experience, such as tectonic activity, distribution of alteration zones, extension direction of fault zones, distribution of magmatic intrusions, etc. The mineralization factors correspond to geometric features or geophysical anomaly areas in the three-dimensional geological model; each mineralization factor must be quantitatively assigned in three-dimensional space;

[0066] Fracture density calculation: calculate the density of fractures in the geological area of ​​the deep side concealed ore body through spatial analysis (such as Voronoi diagram), and map its numerical distribution to the three-dimensional geological model;

[0067] Quantification of geochemical anomaly areas: quantify the anomaly areas of mineral content in rocks or soil in three dimensions through multivariate statistical analysis (such as principal component analysis), extract their spatial distribution, and use spatial interpolation algorithms (such as inverse distance weighted method) to estimate the grade in combination with their mineral content;

[0068] Quantification of geophysical anomalies uses geological inversion techniques (such as magnetic inversion and gravity inversion) to numerically analyze geophysical data such as gravity, magnetic and resistivity, and combines convolutional neural networks to extract three-dimensional distribution characteristics.

[0069] The above process adopts existing technology, which will not be described in detail here.

[0070] Furthermore, based on the quantitative analysis results, geostatistical methods (such as correlation analysis, principal component analysis, etc.) are used to study the relationship between various mineralization elements and find the main factors affecting mineralization. Through the Kriging interpolation method, data prediction is performed for unknown areas and a spatial interpolation model is generated to estimate the mineralization potential of deep and marginal areas.

[0071] The qualitative analysis is to interpret the quantitative analysis results in combination with geological experience and metallogenic theory, identify the key areas of metallogenic control, and obtain qualitative analysis results; specifically, it includes:

[0072] Mineralization pattern recognition, based on the results of quantitative analysis, applies existing rule induction technology, and combines the genetic type of the ore deposit (such as hydrothermal mineralization, sedimentary mineralization, etc.) to identify possible mineralization patterns. For example, in the hydrothermal mineralization pattern, the overlapping area of ​​the fault zone and the alteration zone may be a high probability area for mineralization;

[0073] The favorable mineralization area is delineated by qualitatively interpreting the spatial distribution law of the quantitative analysis results, combining the regional tectonic background with the mineralization theory, and using fuzzy logic reasoning to delineate the favorable mineralization area. The area will show a specific spatial form in the three-dimensional geological model, such as a mineralized belt extending along a fault.

[0074] The above quantitative analysis results and qualitative analysis results are combined to obtain comprehensive mineralization control characteristics.

[0075] S2. Combined with the comprehensive mineralization control characteristics, a three-dimensional mineralization prediction model is constructed, and the output of the three-dimensional mineralization prediction model is used as the prediction result; based on the prediction result, the prospecting target area is delineated, and three-dimensional mineralization prediction of the deep and hidden ore bodies is carried out.

[0076] The process of constructing a three-dimensional mineralization prediction model is as follows:

[0077] The first step is data preparation, which involves collecting historical data of hidden ore bodies in deep and marginal areas, including geological exploration data, geophysical data, and geochemical data. The historical data of hidden ore bodies in deep and marginal areas come from previous exploration reports, field measurement data, and existing mining area research results. At the same time, the historical data of hidden ore bodies in deep and marginal areas are converted into a format and uniformly input into the database to form a comprehensive data set, which provides a basis for the construction of a three-dimensional mineralization prediction model.

[0078] The second step is distribution mapping, mapping the ore body distribution values ​​in the historical data of the deep side hidden ore body to the three-dimensional grid unit. The three-dimensional grid unit is based on the three-dimensional geological model, dividing the deep side hidden ore body into equidistant three-dimensional grid units, and using the multi-dimensional nonlinear geometric transformation interpolation algorithm to perform spatial interpolation on the ore body distribution value to obtain the interpolated ore body distribution value, so as to realize mapping the ore body distribution value in the historical data of the deep side hidden ore body to the three-dimensional grid unit. The multi-dimensional nonlinear geometric transformation interpolation formula is:

[0079] ,

[0080] in, It is a time-space point The ore body distribution value after interpolation at ; In the historical data of the deep side concealed ore body, it is located at the time and space point The volume of the ore body; is located at a point in space and time The volume of the ore body is and The second derivative of the direction physically represents the rate of change of the ore body distribution in these two directions; is a geometric weight function that describes the space-time point To the time and space point The influence of time and space distance on the interpolation results of multidimensional nonlinear geometric transformation usually depends on the geological structure and time evolution law; is the integration variable; A time and space point in the historical data representing a concealed ore body at a deep edge;

[0081] The third step is to generate the mineralization probability distribution. The specific implementation process is as follows: the ore body distribution value after interpolation processing After standardization, it is introduced into the probability density function of ore body distribution to obtain the time-space point Ore-forming probability distribution ,The specific form of the probability density function of ore body distribution is defined according to the specific scenario, which will not be elaborated here;

[0082] The fourth step is the spatiotemporal evolution processing. The comprehensive mineralization control characteristics (such as fault zones, alteration zones, etc.) are not static, but evolve in space over time. In order to capture this dynamic behavior, a nonlinear stochastic partial differential equation is introduced to simulate the spatiotemporal changes of the comprehensive mineralization control characteristics. The specific spatiotemporal evolution formula is:

[0083] ,

[0084] in, It is a time-space point The mineralization probability distribution at It is a time-space point The rate of change of the mineralization probability distribution over time indicates the evolution trend of the ore body distribution in the time dimension. If the rate of change is positive, it means that the mineralization probability in the area is increasing and the ore body is enriched; if the rate of change is negative, it means that the ore body is decreasing or being eroded, and the mineralization probability distribution is updated by the rate of change of the mineralization probability distribution over time. After updating the mineralization probability distribution by the rate of change, a probability distribution sequence that changes dynamically over time is obtained, which is used as the input of the subsequent recursive convolutional evolution network; It is the spatial gradient of the mineralization probability distribution, which physically indicates the direction and rate of change of the mineralization probability distribution in space; is a spatial diffusion function used to describe the dynamic changes of the integrated mineralization control characteristics. For example, the expansion of the fault zone will accelerate the diffusion of minerals, while the density of the alteration zone may slow down the diffusion. The spatial diffusion function is obtained according to the expert experience method; is the diffusion term, which is a Laplacian operator, through the dual operation gradient ( ) and divergence ( ) to characterize the spatial diffusion process of mineralization probability; is the accretion term, which indicates the generation rate of the ore body at a certain point in time and space; It is a loss term, describing the natural loss or attenuation of minerals under certain conditions, such as erosion, chemical reaction, etc.; is a random noise term, simulating the fluctuations in mineralization probability due to random changes in the geological environment;

[0085] The fifth step is recursive convolution evolution. In order to further improve the prediction accuracy, a recursive convolution evolution network (RCEN) is designed based on the updated mineralization probability distribution. The recursive convolution evolution network combines the spatial convolution operation of the convolutional neural network with the time series processing capability of the recursive neural network to capture the complex relationship between the ore body in time and space. The core formula of the RCEN network is:

[0086] ,

[0087] in, is the recursive convolutional evolution network for space-time points The predicted results of the mineralization probability distribution are used to determine the distribution location of the potential ore body. The larger the value of the predicted results of the mineralization probability distribution, the higher the possibility of the existence of the ore body in the area. is the activation function; is the weight of the convolution kernel, indicating the The convolutional layer and the Time step The weight coefficient between ; It is The convolutional layer has a time step The output at the time is The input of the layer-by-layer recursive convolutional evolution network reflects the temporal correlation of the recursive structure. In the initial layer, the input of the recursive convolutional evolution network is the spatiotemporal point Updated mineralization probability distribution at , the time and space point Updated mineralization probability distribution at As the initial input of the recursive convolutional evolution network; It is the symbol of convolution operation, which means convolution operation is performed on input data to capture spatial distribution characteristics; is the number of convolutional layers; is the number of time steps; is the time step, which indicates the time interval and is used for time series prediction. The above process can not only predict the static distribution of ore bodies, but also track the dynamic evolution of ore bodies, providing higher accuracy for subsequent resource estimation. The RCEN network can not only perform spatial prediction, but also process multi-dimensional spatiotemporal data, making the prediction results more accurate.

[0088] Finally, after the prediction of the mineralization probability distribution is completed, the ore body resources need to be estimated. Based on the complexity of the mineralization process and the change of ore body density, the multi-dimensional segmented spatial integration algorithm is introduced. The multi-dimensional segmented spatial integration algorithm combines the predicted results of the mineralization probability distribution of the ore body and the change of the ore body density to accurately estimate the ore body resources. Ore body resources The estimation formula is:

[0089] ,

[0090] in, is the estimated ore body resource; It is the density function of the ore body, which indicates the density of the ore body at different spatial points, changing with time; , It is the mineralization time interval, indicating the time range of the mineralization process; It is a three-dimensional spatial integration area, which represents the spatial volume of the mining area to be estimated.

[0091] After the above process, a three-dimensional mineralization prediction model is obtained, and the output of the three-dimensional mineralization prediction model is used as the prediction result. The prediction result includes the location of the potential ore body and its resource volume. The prospecting target area is delineated according to the location of the potential ore body in the prediction result, and the resource distribution of the deep and hidden ore body is generated, thereby realizing the three-dimensional mineralization prediction of the deep and hidden ore body.

[0092] In summary, a three-dimensional mineralization prediction method for deep and concealed ore bodies has been completed.

[0093] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0094] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0095] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A three-dimensional mineralization prediction method for a deep side concealed ore body, characterized in that: The following steps are involved: S1. Obtain basic data of the deep side concealed ore body and preprocess it to obtain preprocessed basic data; establish a three-dimensional geological model based on the preprocessed basic data; the specific implementation includes: Introduce a boundary enhancement interpolation algorithm; in the process of implementing the boundary enhancement interpolation algorithm, introduce an interpolation coefficient and construct an optimized interpolation function; In the process of optimizing the calculation of the interpolation function, based on the boundary point data set in the preprocessed basic data, the boundary enhancement coefficient is introduced to control the influence range of the boundary points on the interpolation area. The specific formula is: ， in, It is in Boundary Points The interpolation result of is the border enhancement coefficient that adjusts the border enhancement effect; represents an unknown data point; is a known boundary point dataset; Through the Lagrangian function, the interpolation error term and the boundary condition constraint term are introduced to optimize the interpolation coefficient and obtain the optimal solution of the interpolation coefficient. Based on the optimal solution of the interpolation coefficient, the optimized interpolation function is further obtained. The specific formula of the Lagrangian function is as follows: ， in, is the Lagrangian function; It is The actual value of the known data points; is the interpolation coefficient; is the kernel function, which represents the known data points With known data points similarities between; They are and known data points; It is Lagrange multipliers; is the total number of known data points; Indicates the total number of known boundary point data; represents the interpolation error term; is the boundary condition constraint; is the control parameter in the kernel function; Based on the three-dimensional geological model, the pre-processed basic data is quantitatively and qualitatively analyzed to obtain comprehensive mineralization control characteristics; S2. Combined with the comprehensive mineralization control characteristics, a three-dimensional mineralization prediction model is constructed. The specific construction process of the three-dimensional mineralization prediction model includes: The first step is to collect historical data of hidden ore bodies in the deep side. The second step is to perform distribution mapping on the ore body distribution values ​​in the historical data of the deep and side hidden ore bodies. In the distribution mapping process, the ore body distribution values ​​in the historical data of the deep and side hidden ore bodies are spatially interpolated by a multidimensional nonlinear geometric transformation interpolation algorithm to obtain the ore body distribution values ​​after interpolation. The multidimensional nonlinear geometric transformation interpolation formula is: ， in, It is a time and space point The ore body distribution value after interpolation at ; In the historical data of the deep side concealed ore body, it is located at the time and space point The volume of the ore body; is located at a point in space and time The volume of the ore body is and The second derivative of direction; is the geometric weight function; is the integration variable; A time and space point in the historical data representing a concealed ore body at a deep edge; The third step is to generate the mineralization probability distribution based on the interpolated ore body distribution value; The fourth step is to obtain the rate of change of the mineralization probability distribution over time through spatiotemporal evolution processing based on the mineralization probability distribution, and update the mineralization probability distribution based on the rate of change of the mineralization probability distribution over time; The fifth step is to perform recursive convolution evolution processing based on the updated mineralization probability distribution. During the recursive convolution evolution processing, a recursive convolution evolution network is designed to obtain the prediction results of the mineralization probability distribution; the formula of the recursive convolution evolution network is: ， in, is the recursive convolutional evolution network for space-time points The prediction results of the probability distribution of mineralization at the location; is the activation function; is the weight of the convolution kernel, indicating the The convolutional layer and the Time step The weight coefficient between ; It is The convolutional layer has a time step Output when is the convolution operation symbol; is the number of convolutional layers; is the number of time steps; Step 6: Estimate the ore body resources based on the predicted results of mineralization probability distribution; The output of the three-dimensional mineralization prediction model is used as the prediction result; based on the prediction result, the prospecting target area is delineated and three-dimensional mineralization prediction of the hidden ore body in the deep side is carried out.

2. A three-dimensional mineralization prediction method for a deep side concealed ore body according to claim 1, characterized in that: The S2 specifically includes: In the process of spatiotemporal evolution processing, nonlinear stochastic partial differential equations are introduced to simulate the spatiotemporal changes of comprehensive mineralization control characteristics.

3. A three-dimensional mineralization prediction method for a deep side concealed ore body according to claim 1, characterized in that: The S2 specifically includes: A multi-dimensional segmented spatial integration algorithm is introduced to estimate the ore body resources by combining the predicted results of mineralization probability distribution and the changes in ore body density.

Images