Prospecting method for tin polymetallic ore based on multivariate geology information superimposed identification

Abstract

This invention provides a method for tin polymetallic ore prospecting based on the overlay and identification of multivariate geoscientific information, belonging to the technical field of tin polymetallic ore prospecting methods. The method includes calculating the Euclidean distance between each grid and known ore deposits, analyzing the Euclidean distance between each grid and known ore deposits, and calculating the adaptive weight coefficient for each grid; calculating the local variation coefficient for each grid in the neighborhood region, and calculating the comprehensive mineralization favorability of each grid; performing discrete wavelet decomposition on the comprehensive mineralization favorability matrix to reconstruct the local anomaly field matrix of each layer, and calculating the comprehensive anomaly index for each grid; and classifying connected target areas to generate a prospecting target area distribution map. By introducing a nonlinear enhancement factor for the adaptive weight coefficient and the local variation coefficient, the weights can be dynamically adjusted according to the similarity between the grid and known ore deposits; a three-layer discrete wavelet decomposition is used to decompose the comprehensive mineralization favorability into local anomaly fields of different scales.

Description

Technical Field

[0001] This invention relates to the technical field of methods for determining the allowable surplus of phosphate fertilizer application, specifically a method for tin polymetallic mineral exploration based on the superposition and identification of multi-source geoscience information. Background Technology

[0002] Tin polymetallic deposit prospecting prediction is typically applied in mineral exploration scenarios with complex metallogenic geological conditions and diverse types of anomalies. In actual prospecting work, it is necessary to comprehensively analyze various geoscientific information such as geochemical element anomalies, geophysical response anomalies, and remote sensing alteration anomalies in order to identify potential metallogenic areas and delineate prospecting target areas. Especially for mineralization systems with strong co-existence or association relationships of polymetallic elements such as tin, tungsten, copper, lead, and zinc, the metallogenic information often simultaneously manifests as multiple responses such as element enrichment, wall rock alteration, and changes in physical properties. Therefore, conducting comprehensive prospecting prediction based on multi-dimensional geoscientific information has become a common technical approach in the exploration of such deposits. In the prior art, patent document CN120447030A discloses a method for initially selecting potential tin polymetallic deposit areas by comprehensively considering multi-source information such as geology, geophysics, geochemistry, and remote sensing, and collecting borehole data and profile data within these areas to construct a comprehensive prediction model. However, although multi-source geoscience information can participate in mineral exploration analysis, the lack of unified feature expression and adaptive fusion mechanism can easily lead to mutual interference between different information sources, thus affecting the stability and accuracy of the comprehensive evaluation results. On the other hand, this method has insufficient ability to enhance the identification of local anomalies, making it difficult to take into account both the regional background field and the local anomaly field, and is prone to problems such as insensitive anomaly identification, high misjudgment rate, and unclear target area boundaries. Therefore, there is an urgent need for a tin polymetallic deposit exploration method based on the superposition and identification of multi-source geoscience information.

[0003] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] The purpose of this invention is to provide a method for tin polymetallic mineral exploration based on the overlay and identification of multi-source geoscientific information, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The method for tin polymetallic deposit exploration based on the overlay identification of multivariate geoscientific information includes the following steps: S1: Determine the scope of the study area, divide the study area into multiple grids, collect the chemical element content, apparent resistivity value at the center of the grid, and reflectivity data of different bands for each grid, and calculate the hydroxyl alteration anomaly index and iron staining alteration anomaly index of each grid based on the reflectivity data. S2: Perform a logarithmic transformation on the chemical element content of each grid, and standardize the logarithmically transformed chemical element content, apparent resistivity value, hydroxyl alteration anomaly index and iron staining alteration anomaly index of each grid to construct a standardized feature vector for each grid. S3: Obtain the standardized feature vectors of known typical mineral deposits, calculate the similarity between each grid and the known typical mineral deposits, determine the adaptive weight coefficient of each grid based on the similarity, delineate the neighborhood region with each grid as the center, determine the local variation coefficient of each grid for the neighborhood region, and calculate the comprehensive mineralization favorability of each grid based on the adaptive weight coefficient, local variation coefficient and feature variables. S4: Construct a comprehensive mineralization favorability matrix based on the comprehensive mineralization favorability of each grid, perform discrete wavelet decomposition on the comprehensive mineralization favorability matrix, extract high-frequency detail coefficients of each layer and reconstruct the local anomaly field matrix of each layer, and calculate the comprehensive anomaly index of each grid based on the value of each grid in the local anomaly field matrix of each layer. S5: Calculate the mineralization prediction value based on the comprehensive anomaly index and comprehensive mineralization favorability of each grid, compare the mineralization prediction value with the comprehensive anomaly threshold, screen out candidate anomaly grids, use the 8 connectivity criteria to perform connectivity analysis on the candidate anomaly grids, obtain the connected target areas, classify the connected target areas, and obtain the mineral exploration target area distribution map.

[0006] Furthermore, the hydroxyl alteration anomaly index and the iron staining alteration anomaly index were calculated, including... Collect the apparent resistivity value of each grid center, as well as the reflectance data of the blue band, red band, near-infrared band, and short-wave infrared band; By comparing the reflectance changes in the shortwave infrared band and the near-infrared band, the alteration response associated with hydroxyl-containing minerals was extracted and defined as the hydroxyl alteration anomaly index. By comparing the reflectance changes in the red and blue bands, the alteration response associated with iron oxide staining was extracted and defined as the iron staining alteration anomaly index.

[0007] Furthermore, chemical elements including tin, tungsten, copper, lead, zinc, silver, fluorine, and boron are used to construct a standardized feature vector for each grid, specifically: Obtain the elemental content values ​​of tin, tungsten, copper, lead, zinc, silver, fluorine, and boron in each grid. The content value of each element in each grid is summed with the detection limit reference value of the corresponding element to obtain the corrected content value of each element. Each of the corrected content values ​​is subjected to logarithmic transformation to obtain the corresponding logarithmic characteristic value of each element; The logarithmic eigenvalues, apparent resistivity values, hydroxyl alteration anomaly index, and iron staining alteration anomaly index corresponding to each element are used as mesh feature variables. Each grid feature variable is standardized to obtain the standardized result for each grid feature variable. The standardized results are combined in a preset order to obtain the standardized feature vector of each grid.

[0008] Further, the adaptive weight coefficients for each grid are calculated, specifically through the following steps: Obtain the standardized eigenvectors of known typical mineral deposits, and calculate the Euclidean distance between the standardized eigenvectors of each grid and the standardized eigenvectors of known typical mineral deposits. Based on the Euclidean distance between each grid and known typical deposits, the theoretical similarity of each grid is calculated using Euclidean distance according to the distance decay method; The theoretical similarity of each grid is input into the S-shaped mapping function for transformation to obtain the adaptive weight coefficients of each grid. The S-shaped mapping function includes a first adjustment parameter and a second adjustment parameter, which are used to adjust the response magnitude and offset of the adaptive weight coefficient to changes in theoretical similarity.

[0009] Further, the local coefficient of variation for each grid is calculated, specifically through the following steps: Centered on each grid, surround it The grid region is defined as the surrounding neighborhood region of the grid. The mean and standard deviation of each feature in all surrounding neighborhood regions are calculated. The standard deviation is divided by the mean to calculate the local coefficient of variation of each feature in each grid.

[0010] Further, the overall mineralization favorability of each grid is calculated, and the specific steps are as follows: Based on adaptive weighting coefficients, local variation coefficients, and standardized feature variables for each grid, the overall mineralization favorability of each grid is calculated: in, Index representing the number of standardized feature variables; Indicates the first The first grid One standardized feature variable; Indicates the first The first grid The local variation coefficients of the standardized feature variables; Indicates the first The overall mineralization advantage of each grid; Indicates the first The adaptive weight coefficients for each grid.

[0011] Furthermore, the three-layer high-frequency detail coefficients are obtained, and the specific steps are as follows: Arrange the comprehensive mineralization favorability of each grid according to the row and column positions of the grid to construct an M-row N-column comprehensive mineralization favorability matrix; Daubechies-4 wavelet is selected as the wavelet basis function. The comprehensive mineralization favorability matrix is ​​first subjected to low-pass filtering and high-pass filtering by row, and then downsampled to obtain the row low-frequency matrix and row high-frequency matrix. The row low-frequency matrix and the row high-frequency matrix are then subjected to low-pass filtering and high-pass filtering by column, and downsampling is performed respectively to obtain the first layer low-frequency approximation coefficient matrix, as well as the first layer horizontal high-frequency detail coefficient matrix, vertical high-frequency detail coefficient matrix and diagonal high-frequency detail coefficient matrix. Using the first-layer low-frequency approximation coefficient matrix as input, repeat the same decomposition operation to obtain the second-layer low-frequency approximation coefficient matrix, as well as the second-layer horizontal high-frequency detail coefficient matrix, vertical high-frequency detail coefficient matrix, and diagonal high-frequency detail coefficient matrix. The matrix includes the third layer of high-frequency detail coefficient matrices in the horizontal, vertical, and diagonal directions; This yields nine high-frequency detail coefficient matrices corresponding to the three-level discrete wavelet decomposition, which characterize the high-frequency detail coefficients.

[0012] Further, the comprehensive anomaly index for each grid is calculated, specifically through the following steps: The high-frequency detail coefficient matrices in the horizontal direction, vertical direction, and diagonal direction obtained from the first, second, and third layers of discrete wavelet decomposition are extracted respectively. A zero matrix with the same size as the high-frequency detail coefficient matrix of the corresponding layer is constructed to replace the low-frequency approximation coefficient matrix of that layer. Inverse wavelet reconstruction is performed on the zero matrix of each layer and the corresponding high-frequency detail coefficient matrix in the horizontal direction, vertical direction, and diagonal direction of the layer to obtain the first layer local anomaly field matrix, the second layer local anomaly field matrix, and the third layer local anomaly field matrix with the same size as the original comprehensive mineralization favorable matrix. Calculate the mean and standard deviation of the local anomaly field matrix for each layer, and determine the anomaly threshold for that layer based on the mean and standard deviation of the corresponding layer. The ratio of the outlier value of each grid in the local anomaly field matrix of each layer to the anomaly threshold of the corresponding layer is processed to obtain the relative anomaly intensity of each grid in each layer. The maximum value of the relative anomaly intensity of each grid in the first, second, and third layers is selected as the comprehensive anomaly index of that grid.

[0013] Further, the comprehensive anomaly threshold is calculated, and the specific steps are as follows: The calculation steps for the comprehensive anomaly threshold include: The mineralization prediction value for each grid is obtained by multiplying the comprehensive anomaly index and comprehensive mineralization favorability of each grid. Statistical analysis was performed on the mineralization prediction values ​​of all grids in the study area to obtain the arithmetic mean and standard deviation of the mineralization prediction values; the comprehensive anomaly threshold was determined by weighted summation of the arithmetic mean and standard deviation of the mineralization prediction values.

[0014] Furthermore, a mineral exploration target area distribution map is generated, specifically as follows: The predicted mineralization values ​​are compared with the comprehensive anomaly threshold of the set prospecting target area. Grids with predicted mineralization values ​​greater than the comprehensive anomaly threshold are initially screened as candidate anomaly grids. The connectivity analysis of the candidate anomaly grids is performed using the 8-connectivity criterion to obtain the connected target area, specifically: All mineralization prediction values ​​are greater than the comprehensive anomaly threshold The grids are marked as candidate anomalous grids, and a blank label matrix with the same size as the original grid is created to record the connected target region number to which each grid belongs. Then, all candidate anomalous grids are traversed in row-major order. When an unmarked candidate anomalous grid is encountered, it is used as the seed point of a new target region, assigned a new target region number, and a breadth-first search is used to recursively check the adjacent grids in the eight directions around the grid, namely up, down, left, right, upper left, upper right, lower left, and lower right. If an adjacent grid is also a candidate anomalous grid and has not yet been marked, it is added to the current target region and assigned the same target region number. At the same time, the search continues to expand outward from the adjacent grid as the center until no more adjacent candidate anomalous grids that meet the conditions are found. The above process is repeated until all candidate anomalous grids are marked. At this time, the set of grids with the same target region number constitutes a connected target region. For each connected target region, calculate the average of all its grid-containing mineralization predictions. The connected target areas are classified and a mineral exploration target area distribution map is generated, specifically: if there are Mark the connected target region as a primary target region; if there are Mark the connected target region as a secondary target region; if there are The connected target region is marked as a level 3 target region, where, This represents the mean of the predicted mineralization values; This represents the standard deviation of the predicted mineralization value.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This scheme collects data on chemical element content, apparent resistivity, and reflectivity, and further calculates the hydroxyl alteration anomaly index and the iron staining alteration anomaly index. It integrates geochemical, geophysical, and remote sensing alteration information into a unified analytical framework. By performing logarithmic transformation and standardization on the chemical element content, standardized feature vectors are constructed for each grid. This reduces the impact of differences in dimensions and distribution among different data sources on the comprehensive analysis results, improves the consistency and comparability of the joint expression of multi-source geoscience information, and solves the problem of the difficulty in uniformly representing multi-source information in existing technologies. By calculating the similarity between each grid and known typical mineral deposits, and combining this with the local coefficient of variation to determine the comprehensive mineralization favorability, and then extracting local anomaly information through discrete wavelet decomposition and calculating the comprehensive anomaly index, this scheme can simultaneously reflect the matching degree between the grid and typical mineralization characteristics, as well as local spatial anomaly characteristics. This enhances the separation ability between regional background and local anomalies, and improves the accuracy of anomaly identification under complex geological conditions. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a graph showing the relationship between the chemical element content values ​​and the logarithmic transformation values ​​of the corresponding regions. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0018] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0019] Example: Please see Figures 1-2 The present invention provides a technical solution: The method for tin polymetallic deposit exploration based on the overlay identification of multivariate geoscientific information includes the following steps: S1: Determine the scope of the study area, divide the study area into multiple grids, collect the chemical element content, apparent resistivity value at the center of the grid, and reflectivity data of different bands for each grid, and calculate the hydroxyl alteration anomaly index and iron staining alteration anomaly index of each grid based on the reflectivity data. Calculate the hydroxyl alteration anomaly index and the iron staining alteration anomaly index, including Collect the apparent resistivity value of each grid center, as well as the reflectance data of the blue band, red band, near-infrared band, and short-wave infrared band; By comparing the reflectance changes in the shortwave infrared band and the near-infrared band, the alteration response associated with hydroxyl-containing minerals was extracted and defined as the hydroxyl alteration anomaly index. By comparing the reflectance changes in the red and blue bands, the alteration response associated with iron oxide staining was extracted and defined as the iron staining alteration anomaly index.

[0020] The formula used in the above process is as follows: Calculate the hydroxyl alteration anomaly index for each grid based on reflectance data: in, Indicates the first The reflectance values ​​of the shortwave infrared band 2 of each grid; Indicates the first The reflectance values ​​of the near-infrared band for each grid; Indicates the first Hydroxyl alteration anomaly index of each grid; Calculate the iron staining and alteration anomaly index: in, Indicates the first The reflectance values ​​of the red band in each grid; Indicates the first The reflectance values ​​of the blue band in each grid; Indicates the first Iron staining and alteration anomaly index of each grid; Among them, the hydroxyl alteration anomaly index is adopted. The processing is based on the fact that hydroxyl-containing minerals exhibit strong spectral absorption in the short-wave infrared band 2, while displaying high reflectivity in the near-infrared band. When hydroxyl-containing minerals such as kaolinite, sericite, and chlorite are present, the reflectivity of Band 7 decreases significantly, while the reflectivity of Band 5 remains at a high level, leading to molecular... If the value is negative and the absolute value is large, the denominator Keep it positive, so that It exhibits a clear negative anomaly; this ratio form can effectively eliminate the influence of terrain shadows and changes in light intensity on the absolute value of reflectance, highlighting the contrast between hydroxyl alteration information and background features; through ratio calculation, it effectively eliminates the interference of terrain shadows, changes in light intensity, and atmospheric scattering on the absolute value of reflectance, so that the weak spectral absorption signal caused by hydroxyl-containing minerals is significantly enhanced and converted into a quantifiable negative anomaly index, thereby achieving high-precision remote sensing identification of hydrothermal alteration zones such as sericitization, kaolinization, and chloritization; Iron staining and alteration anomaly index adopted The processing is based on the fact that iron-bearing minerals exhibit moderate absorption in the red band and relatively high reflectivity in the blue band. When iron-bearing minerals such as limonite, hematite, and jaundice are present, the reflectivity of Band 4 decreases due to absorption, while the reflectivity of Band 2 changes less, leading to molecular... If the value is negative, the denominator Keep it positive, so that It exhibits a negative anomaly; this ratio processing can enhance the spectral differences between iron staining alteration information and background features such as vegetation and water bodies, while suppressing the interference of lighting conditions and topographic relief on reflectance values; by calculating the band ratio, the spectral responses of vegetation, water bodies and unmineralized soil and rock bodies are suppressed, while amplifying the contrast between the absorption characteristics of iron oxides in the red band and the high reflectance characteristics in the blue band, transforming iron staining alteration information closely related to tin polymetallic mineralization, such as limonite and hematite mineralization, into obvious negative anomaly indices, effectively eliminating the interference of false anomalies.

[0021] S2: Perform a logarithmic transformation on the chemical element content of each grid, and standardize the logarithmically transformed chemical element content, apparent resistivity value, hydroxyl alteration anomaly index and iron staining alteration anomaly index of each grid to construct a standardized feature vector for each grid. The chemical elements include tin, tungsten, copper, lead, zinc, silver, fluorine, and boron. A standardized feature vector is constructed for each grid cell, specifically as follows: Obtain the elemental content values ​​of tin, tungsten, copper, lead, zinc, silver, fluorine, and boron in each grid. The content value of each element in each grid is summed with the detection limit reference value of the corresponding element to obtain the corrected content value of each element. Each of the corrected content values ​​is subjected to logarithmic transformation to obtain the corresponding logarithmic characteristic value of each element; The formula upon which the above process is based is: in, Indicates the first The first grid The value of the content of each chemical element after logarithmic transformation; Indicates the first The first grid The content value of each chemical element ,and Take integers, which represent tin, tungsten, copper, lead, zinc, silver, fluorine, and boron, respectively. Indicates the first Reference values ​​for the detection limits of each chemical element; The logarithmic eigenvalues, apparent resistivity values, hydroxyl alteration anomaly index, and iron staining alteration anomaly index corresponding to each element are used as mesh feature variables. Each grid feature variable is standardized to obtain the standardized result for each grid feature variable. The standardized results are combined in a preset order to obtain the standardized feature vector of each grid.

[0022] In the above process, the above process adopts The reason for this processing is that the distribution of geochemical elements in nature typically follows a log-normal distribution rather than a normal distribution. By using the log-transformed chemical element content, apparent resistivity, hydroxyl alteration anomaly index, and iron staining alteration anomaly index as grid feature variables, the original skewed data can be converted into an approximately normal distribution, satisfying the basic assumptions of subsequent standardization regarding the data distribution pattern. Simultaneously, summing the element content values ​​with the corresponding detection limit reference values ​​effectively avoids the problem of negative infinity or incomputability in logarithmic transformation when element content is close to or below the detection limit, ensuring that all grid data can participate in the calculation. The technical effects of this processing are as follows: Logarithmic transformation compresses the data range of high-content values ​​and amplifies the differences between low-content values, making strong and weak mineralization anomalies comparable in the feature space. This avoids the drawback of high-content elements dominating the feature vector and masking the effective information of low-content elements. Furthermore, this transformation stabilizes the variance of each element, laying the foundation for subsequent Z-score standardization and significantly improving the quality and reliability of multivariate geoscientific information fusion analysis. The preset order is as follows: logarithmic characteristic value of tin, logarithmic characteristic value of tungsten, logarithmic characteristic value of copper, logarithmic characteristic value of lead, logarithmic characteristic value of zinc, logarithmic characteristic value of silver, logarithmic characteristic value of fluorine, logarithmic characteristic value of boron, apparent resistivity value, hydroxyl alteration anomaly index, and iron staining alteration anomaly index.

[0023] In the above embodiments, 20 sets of data on the chemical element content of tin and the logarithmic transformation values ​​of the corresponding regions are given to reflect how the logarithmic transformation values ​​change with the chemical element content of tin, as shown in Table 1: Table 1: Relationship between chemical element content values ​​and corresponding logarithmic transformation values ​​for the region Table 1 above provides... When the tin content is far below the detection limit (0.05 to 0.45 μg / g), the logarithmic transformation values ​​are all negative (-0.26 to -0.02). As the content increases, the negative values ​​gradually approach zero, effectively avoiding the problem of negative infinity when taking the logarithm in the low-content region, while preserving the relative differences between samples below the detection limit. When the tin content is equal to the detection limit (0.50 μg / g), the logarithmic transformation value is 0.00, serving as a clear boundary between background and abnormality. When the tin content increases from 1.00 μg / g to 10.00 μg / g, the logarithmic transformation value increases from 0.18 to 1.02, an increment of 0.84. However, when the tin content increases from 10.00 μg / g to 50.00 μg / g... At μg / g, the logarithmic transformation value only increased from 1.02 to 1.70, an increment of 0.68. The increment in the low content range was significantly greater than that in the high content range, which reflects the compression effect of the logarithmic transformation on the high content region. This reduces the gap between strong mineralization anomalies and medium mineralization anomalies on the transformed numerical scale, and avoids the high content elements dominating the subsequent eigenvectors and masking the mineralization information of other elements.

[0024] S3: Obtain the standardized feature vectors of known typical mineral deposits, calculate the similarity between each grid and the known typical mineral deposits, determine the adaptive weight coefficient of each grid based on the similarity, delineate the neighborhood region with each grid as the center, determine the local variation coefficient of each grid for the neighborhood region, and calculate the comprehensive mineralization favorability of each grid based on the adaptive weight coefficient, local variation coefficient and feature variables. The specific steps for calculating the adaptive weight coefficients for each grid are as follows: Obtain the standardized eigenvectors of known typical mineral deposits, and calculate the Euclidean distance between the standardized eigenvectors of each grid and the standardized eigenvectors of known typical mineral deposits. Based on the Euclidean distance between each grid and known typical deposits, the theoretical similarity of each grid is calculated using Euclidean distance according to the distance decay method; The theoretical similarity of each grid is input into the S-shaped mapping function for transformation to obtain the adaptive weight coefficients of each grid. The S-shaped mapping function includes a first adjustment parameter and a second adjustment parameter, which are used to adjust the response magnitude and offset of the adaptive weight coefficient to changes in theoretical similarity.

[0025] The formula upon which the above process is based is: Calculate the theoretical similarity for each grid cell: in, Indicates bandwidth parameter; Indicates the first Theoretical similarity of the grids; Indicates the first The theoretical Euclidean distance of each grid; In the above process, the formula In the middle, the dependent variable Used to characterize the The theoretical similarity between a grid and known typical mineral deposits is expressed as follows: a larger value indicates that the geoscientific information characteristics of the grid are closer to the known metallogenic environment, and the higher the metallogenic potential; a smaller value indicates a greater difference and a lower probability of metallogenesis. (Independent variable) The distance represents the Euclidean distance between the standardized eigenvector of the i-th grid and the standardized eigenvector of a known typical ore deposit. This distance comprehensively reflects the overall difference between the two grids in eleven characteristic dimensions, including the content of tin, tungsten, copper, lead, zinc, silver, fluorine, and boron, apparent resistivity, hydroxyl alteration anomaly index, and iron staining alteration anomaly index. The smaller the distance, the closer the values ​​of the two grids are in each mineralization indicator element. This conforms to the geological similarity principle of "like attracts like," that is, the combination of geological information characteristics of the known ore deposit location represents a favorable mineralization environment, and the grids that are more similar to this environment are more likely to develop similar ore deposits. This formula uses a Gaussian kernel function and implements an exponential decay mechanism to achieve a monotonically decreasing similarity as distance increases: when... hour (Completely the same), when hour ,when hour ,when hour This smooth distance decay pattern aligns with the geological exploration principle that "similarity decreases with increasing distance," and the bandwidth parameter... It can adaptively adjust based on the average distance between known mineral deposits, making the similarity calculation adaptable to the spatial scale of different mineral clusters. The technical effect is that it transforms multidimensional feature differences into intuitive comparable similarity indicators, providing a quantitative basis for the subsequent determination of adaptive weight coefficients.

[0026] Based on the theoretical similarity of each grid, calculate the adaptive weight coefficient for each grid: in, Indicates the first Adaptive weighting coefficients for each grid; Indicates the first adjustment parameter; This indicates the second adjustment parameter.

[0027] In the above process, the dependent variable Used to characterize the The adaptive weighting coefficients of each grid cell reflect the proportion of geoscientific information from that grid cell that contributes to the subsequent comprehensive mineralization favorability calculation. The larger the value (closer to 1), the more similar the grid is to the known mineralization environment, and its characteristic variables should be given a higher degree of confidence; The smaller the value (closer to 0), the more significant the difference between the grid and the known mineralization environment, and the contributions of its various characteristic variables should be suppressed; Independent variable Indicates the first The theoretical similarity between a grid and known typical mineral deposits is considered. A higher similarity indicates that the combination of geological, geochemical, geophysical, and remote sensing features of the grid closely resembles known mineralization patterns, thus indicating higher reliability of its geoscientific information. Therefore, it should be assigned a greater weighting coefficient, and the two are positively correlated. (Parameters) The first adjustment parameter controls the steepness of the S-curve. The larger the value, the better. right The more sensitive the change, the greater the weight (a slight increase in similarity can lead to a significant increase in weight). The smaller the value, the flatter the curve; parameter As the second adjustment parameter, it controls the horizontal offset position of the S-curve, determining when... hour The baseline value; the formula uses a sigmoid function for the following reasons: First, its output value is naturally limited to between 0 and 1, which meets the domain requirements of the weighting coefficient; second, the sigmoid curve is in the middle region ( The similarity coefficient (approximately 0.3 to 0.7) exhibits an almost linear variation, while showing saturation at both ends (extremely high or low similarity regions). This aligns with geological understanding—for grids highly similar to known deposits, further increasing the similarity has limited effect on improving the weight; for grids significantly different from known deposits, the reduction in weight also tends to level off. The technical advantage of this formula lies in: through the first adjustment parameter... Second adjustment parameter With the cooperation of these factors, the response characteristics of the adaptive weight coefficients to theoretical similarity can be flexibly adjusted, enabling the model to make adaptive adjustments according to the geological complexity of different study areas.

[0028] The specific steps for calculating the local coefficient of variation for each grid are as follows: Centered on each grid, surround it The grid region is defined as the surrounding neighborhood region of the grid. The mean and standard deviation of each feature in all surrounding neighborhood regions are calculated. The standard deviation is divided by the mean to calculate the local coefficient of variation of each feature in each grid.

[0029] The specific steps for calculating the overall mineralization favorability of each grid are as follows: Based on adaptive weighting coefficients, local variation coefficients, and standardized feature variables for each grid, the overall mineralization favorability of each grid is calculated: in, Index representing the number of standardized feature variables; Indicates the first The first grid One standardized feature variable; Indicates the first The first grid The local variation coefficients of the standardized feature variables; Indicates the first The overall mineralization advantage of each grid; Indicates the first The adaptive weight coefficients for each grid.

[0030] In the above process, a 3×3 neighborhood region is used to calculate the local coefficient of variation for each feature. The rationale behind this approach is that tin polymetallic mineralization often exhibits localized aggregation characteristics in space. Mineralized bodies and their surrounding alteration zones lead to drastic changes in geochemical element content, geophysical fields, and remote sensing alteration indices within a relatively small spatial range, manifested as a significant increase in local coefficients of variation. In contrast, the geological information of the background area is relatively evenly distributed spatially, with smaller local coefficients of variation. By defining a 3×3 neighborhood region centered on each grid (corresponding to a 300m×300m area on the actual ground, matching the width of a typical alteration zone in tin polymetallic deposits), the mean and standard deviation of each feature within the neighborhood are calculated. Dividing the standard deviation by the mean yields the local coefficient of variation, which quantitatively characterizes the degree of variation in geological information of each grid relative to its surrounding background. A larger coefficient of variation indicates a higher probability that the grid is located at a geological anomaly boundary or a mineralization alteration center. The above comprehensive mineralization favorableness calculation formula The processing logic is as follows: The adaptive weight coefficients... Standardized feature variables Nonlinear enhancement factor of local variation coefficient The three factors are multiplied together and then summed. The adaptive weight coefficients are included. The overall credibility of all features of the grid is dynamically adjusted based on the similarity between the grid and known mineral deposits; feature variables are standardized. Reflecting the grid in the first Anomaly intensity relative to the overall background (positive values ​​indicate high anomaly, negative values ​​indicate low anomaly); nonlinear enhancement factor Features with large local variability are amplified exponentially, when The time is increased by approximately 1.82 times, when The time-to-time increase is approximately 2.61 times; The technical effect of this formula is that, through the synergistic effect of three factors—the weighting coefficient controls reliability, the standardized variable reflects the intensity of anomalies, and the nonlinear enhancement factor highlights local variations—it achieves the comprehensive extraction and enhancement of favorable mineralization information, and significantly improves the identifiability of weak mineralization signals in complex geological backgrounds.

[0031] S4: Construct a comprehensive mineralization favorability matrix based on the comprehensive mineralization favorability of each grid, perform discrete wavelet decomposition on the comprehensive mineralization favorability matrix, extract high-frequency detail coefficients of each layer and reconstruct the local anomaly field matrix of each layer, and calculate the comprehensive anomaly index of each grid based on the value of each grid in the local anomaly field matrix of each layer. The specific steps to obtain the three-layer high-frequency detail coefficients are as follows: Arrange the comprehensive mineralization favorability of each grid according to the row and column positions of the grid to construct an M-row N-column comprehensive mineralization favorability matrix; Daubechies-4 wavelet is selected as the wavelet basis function. The comprehensive mineralization favorability matrix is ​​first subjected to low-pass filtering and high-pass filtering by row, and then downsampled to obtain the row low-frequency matrix and row high-frequency matrix. The row low-frequency matrix and the row high-frequency matrix are then subjected to low-pass filtering and high-pass filtering by column, and downsampling is performed respectively to obtain the first layer low-frequency approximation coefficient matrix, as well as the first layer horizontal high-frequency detail coefficient matrix, vertical high-frequency detail coefficient matrix and diagonal high-frequency detail coefficient matrix. Using the first-layer low-frequency approximation coefficient matrix as input, repeat the same decomposition operation to obtain the second-layer low-frequency approximation coefficient matrix, as well as the second-layer horizontal high-frequency detail coefficient matrix, vertical high-frequency detail coefficient matrix, and diagonal high-frequency detail coefficient matrix. The matrix includes the third layer of high-frequency detail coefficient matrices in the horizontal, vertical, and diagonal directions; This yields nine high-frequency detail coefficient matrices corresponding to the three-level discrete wavelet decomposition, which characterize the high-frequency detail coefficients.

[0032] The formula upon which the above process is based is: in, Represents the comprehensive mineralization favorableness matrix; Indicates the first Line 1 The overall mineralization favorableness of the series; Indicates the grid row index; Indicates a grid column index; In the above process, the Daubechies-4 wavelet is selected as the wavelet basis function, and the integrated mineralization favorability matrix is ​​subjected to discrete wavelet decomposition to obtain a three-layer high-frequency detail coefficient matrix, specifically: The Daubechies-4 wavelet is chosen as the wavelet basis function because it contains both low-pass and high-pass filter coefficients, which can separate the information in the matrix into low-frequency components reflecting the regional background and high-frequency components reflecting local anomalies. A row-column decomposition is performed on the comprehensive mineralization favorability matrix: first, each row of the comprehensive mineralization favorability matrix is ​​subjected to low-pass and high-pass filtering and downsampling. This downsampling involves retaining a value for every other element in the comprehensive mineralization favorability matrix, resulting in a row low-frequency matrix and a row high-frequency matrix. Then, each column of the row low-frequency matrix and the row high-frequency matrix is ​​subjected to the same filtering and downsampling. After one layer of decomposition, the original matrix is ​​decomposed into four sub-band coefficients: a low-frequency approximation coefficient matrix and three high-frequency detail coefficient matrices, representing local variation information in the horizontal, vertical, and diagonal directions, respectively. The first level of decomposition starts with the original... OK Taking the comprehensive mineralization favorability matrix as input, after the above row-column decomposition, we obtain a matrix of size [value missing]. OK, The first layer decomposes the matrix into a low-frequency approximation coefficient matrix and three high-frequency detail coefficient matrices of the same size. Then, a second layer decomposition is performed, using the low-frequency approximation coefficient matrix obtained in the first layer as the new input matrix, repeating the exact same row-column decomposition operation to obtain a matrix of size [missing information]. OK, The second layer of the column consists of a low-frequency approximation coefficient matrix and three high-frequency detail coefficient matrices; The third-level decomposition uses the low-frequency approximation coefficient matrix obtained in the second level as input, and repeats the row-column decomposition again to obtain a matrix of size [missing information]. OK, The third-level low-frequency approximation coefficient matrix and three high-frequency detail coefficient matrices are obtained. At this point, the three-level decomposition is completed, and the system obtains a total of nine high-frequency detail coefficient matrices, with three high-frequency detail coefficient matrices in each level.

[0033] Among them, the Daubechies-4 wavelet is used to synthesize the mineralization favorability matrix. Three-level discrete wavelet decomposition is performed based on the fact that tin polymetallic mineralization information has multi-scale spatial characteristics. Shallow vein-like mineralization (depth 0 to 200m), mid-to-deep layered ore bodies (depth 200 to 500m), and mineralization at the top of deep concealed rock masses (depth below 500m) correspond to anomalous patterns at different spatial scales. This is achieved by analyzing the matrix... First, perform low-pass filtering and high-pass filtering and downsampling by row, and then perform the same operation by column. One-level decomposition decomposes the original matrix into a low-frequency approximation coefficient matrix (reflecting the regional background) and three high-frequency detail coefficient matrices (representing local change information in the horizontal, vertical and diagonal directions, respectively). The Daubechies-4 wavelet was chosen because of its moderate filter length, good tight support and regularity, which allows it to preserve the spatial morphology of geological anomalies while extracting local anomaly information. The technical effect of this processing is that, through a three-layer hierarchical decomposition, mineralization information at different scales is separated into different decomposition layers. The first layer, with high-frequency detail coefficients, corresponds to the finest-scale shallow anomalies; the second layer corresponds to the intermediate-scale middle-scale anomalies; and the third layer corresponds to the largest-scale deep anomalies. This achieves the ability to extract mineralization anomalies of different depths from superimposed mixed signals, providing crucial technical support for subsequent identification of concealed ore bodies.

[0034] The specific steps for calculating the comprehensive anomaly index for each grid are as follows: The high-frequency detail coefficient matrices in the horizontal direction, vertical direction, and diagonal direction obtained from the first, second, and third layers of discrete wavelet decomposition are extracted respectively. A zero matrix with the same size as the high-frequency detail coefficient matrix of the corresponding layer is constructed to replace the low-frequency approximation coefficient matrix of that layer. Inverse wavelet reconstruction is performed on the zero matrix of each layer and the corresponding high-frequency detail coefficient matrix in the horizontal direction, vertical direction, and diagonal direction of the layer to obtain the first layer local anomaly field matrix, the second layer local anomaly field matrix, and the third layer local anomaly field matrix with the same size as the original comprehensive mineralization favorable matrix. Calculate the mean and standard deviation of the local anomaly field matrix for each layer, and determine the anomaly threshold for that layer based on the mean and standard deviation of the corresponding layer. The ratio of the outlier value of each grid in the local anomaly field matrix of each layer to the anomaly threshold of the corresponding layer is processed to obtain the relative anomaly intensity of each grid in each layer. The maximum value of the relative anomaly intensity of each grid in the first, second, and third layers is selected as the comprehensive anomaly index of that grid.

[0035] The above process is based on the following formula: For the first-layer local anomaly field matrix, calculate its mean and standard deviation, and construct the anomaly threshold of the first-layer local anomaly field matrix: in, , These represent the mean and standard deviation of the local anomaly field matrix in the first layer, respectively. This represents the anomaly threshold of the first-layer local anomaly field matrix; Using the same method, the anomaly thresholds of the second-layer and third-layer local anomaly field matrices are calculated. Based on the value of each grid cell in the local anomaly field matrix of each layer, the relative anomaly intensity of each grid cell in each layer is calculated. in, Indicates the first The relative intensity of anomalies in the first layer of each grid; Indicates the first The relative intensity of anomalies in the second layer of each grid; Indicates the first The relative intensity of anomalies in the third layer of each grid; Represents the coordinates in the local anomaly field matrix of the first layer. Outliers at; Represents the coordinates in the local anomaly field matrix of the second layer. Outliers at; Represents the coordinates in the local anomaly field matrix of the third layer. outliers at; where These represent the x and y indices of the grid, respectively. , And all of them take integer values; , These represent the mean and standard deviation of the local anomaly field matrix in the first layer, respectively. This represents the anomaly threshold of the first-layer local anomaly field matrix.

[0036] In the above process, discrete wavelet decomposition separates the information in the original comprehensive mineralization favorability matrix into coefficients of different scales. The high-frequency detail coefficients carry local anomaly information, while the low-frequency approximation coefficients carry regional background information. By constructing a zero matrix with the same size as the corresponding layer's high-frequency detail coefficient matrix to replace the low-frequency approximation coefficient matrix, and then performing inverse wavelet reconstruction on the zero matrix and the high-frequency detail coefficient matrices in the three directions, the local anomaly information contained in the high-frequency detail coefficients can be extracted separately, generating an anomaly field matrix that only reflects local changes at that scale, thus eliminating the interference of regional background on anomaly identification. The technical effect of this processing is that the first-layer local anomaly field matrix... Reflecting fine anomalies caused by shallow mineralization, the second layer Reflecting medium-scale anomalies in deep ore bodies, the third layer It reflects large-scale anomalies in deep, concealed rock masses and enables the stratified extraction of mineralization information at different depths. use The anomaly thresholds for each layer are calculated in the following manner, along with the relative intensity of the anomalies. This calculation is based on the fact that the numerical distribution of the local anomaly field matrix in each layer follows statistical laws, where the mean and standard deviation reflect the background level and dispersion of the local anomalies in that layer. Setting the anomaly threshold to the mean plus twice the standard deviation allows for the screening of significant anomaly areas with a confidence level of approximately 95% or higher under the assumption of a normal distribution, effectively distinguishing genuine mineralization anomalies from background noise. The dimensionless relative intensity of the anomaly is obtained by dividing the anomaly value of each grid by the threshold for that layer. A value greater than 1 indicates a significant anomaly in that grid at that layer, while a value less than 1 indicates an insignificant anomaly. Finally, the maximum value of the relative intensity of the anomalies in each grid across the three layers is selected as the comprehensive anomaly index. The technical advantage of this approach is that as long as a grid exhibits a significant mineralization anomaly at any depth level, it can be identified as a potential favorable mineralization unit, avoiding the potential for overlooking deep-seated hidden mineralization information during single-scale anomaly screening.

[0037] S5: Calculate the mineralization prediction value based on the comprehensive anomaly index and comprehensive mineralization favorability of each grid, compare the mineralization prediction value with the comprehensive anomaly threshold, screen out candidate anomaly grids, use the 8 connectivity criteria to perform connectivity analysis on the candidate anomaly grids, obtain the connected target areas, classify the connected target areas, and obtain the mineral exploration target area distribution map.

[0038] The specific steps for calculating the comprehensive anomaly threshold are as follows: The calculation steps for the comprehensive anomaly threshold include: The mineralization prediction value for each grid is obtained by multiplying the comprehensive anomaly index and comprehensive mineralization favorability of each grid. Statistical analysis was performed on the mineralization prediction values ​​of all grids in the study area to obtain the arithmetic mean and standard deviation of the mineralization prediction values; the comprehensive anomaly threshold was determined by weighted summation of the arithmetic mean and standard deviation of the mineralization prediction values.

[0039] Generate a mineral exploration target area distribution map, specifically: The formula upon which the above process is based is: in, Indicates the first The comprehensive anomaly index of each grid; Indicates the first Mineralization prediction values ​​for each grid; Calculate the mineralization prediction values ​​for all grids in the study area, and statistically analyze the mean and standard deviation. Set the comprehensive anomaly threshold as follows: in, Indicates the overall anomaly threshold; This represents the mean of the predicted mineralization values; This represents the standard deviation of the predicted mineralization values; In the above process, the following methods are adopted The calculation of the mineralization prediction values ​​for each grid is based on the comprehensive mineralization favorability. This reflects the nonlinear weighting intensity of the grid based on the overlay of multivariate geoscientific information, while the comprehensive anomaly index... This reflects the degree of development of local anomalies at different depth levels within the grid; both characteristics depict the grid's mineralization potential from different perspectives. The focus is on the similarity between the combination of geological information and known mineral deposits, as well as the degree of local variation. The focus is on the relative intensity of multi-scale anomalies. Multiplying the two creates a complementary constraint: only when the grid simultaneously possesses a high intensity of geoscientific information overlay ( Larger) and significant multiscale local anomalies ( When the value is relatively large, its mineralization prediction value Only then can a higher level be achieved; if any one factor is low, the product will be lowered, thus effectively suppressing false anomalies caused by a single factor being high while another factor is low. The technical effect of this processing is that it integrates mineralization indicator information from two different dimensions, improving the reliability and anti-interference ability of mineralization prediction; The above process adopts The comprehensive anomaly threshold is set based on the mineralization prediction values ​​of all grids within the study area. This constitutes a statistical distribution, where Represents the overall background level of the region. This represents the degree of dispersion of the predicted values; the comprehensive anomaly threshold can be set as the mean plus... This threshold, representing 1 / 3 standard deviation, is based on statistical analysis of multiple known tin polymetallic ore clusters in South China. At this threshold, the true positive rate—the proportion of known deposits correctly identified as anomalies within their respective grids—is approximately [percentage missing]. The false positive rate, i.e., the proportion of non-mineral grid cells that are mistakenly identified as abnormal, is approximately The threshold value is the optimal threshold for classification performance; this threshold is higher than the mean but not excessively high, effectively filtering out high-potential grids that significantly deviate from the background as candidate anomaly grids, while eliminating a large number of low-value background grids, reducing the computational load of subsequent connectivity analysis. The technical advantage of this process is that it achieves objectivity and automation in anomaly screening using statistical thresholds, avoiding subjective bias from manually setting thresholds and ensuring the comparability of prediction results between different study areas.

[0040] The predicted mineralization values ​​are compared with the comprehensive anomaly threshold of the set prospecting target area. Grids with predicted mineralization values ​​greater than the comprehensive anomaly threshold are initially screened as candidate anomaly grids. The connectivity analysis of the candidate anomaly grids is performed using the 8-connectivity criterion to obtain the connected target area, specifically: All mineralization prediction values ​​are greater than the comprehensive anomaly threshold The grids are marked as candidate anomalous grids, and a blank label matrix with the same size as the original grid is created to record the connected target region number to which each grid belongs. Then, all candidate anomalous grids are traversed in row-major order. When an unmarked candidate anomalous grid is encountered, it is used as the seed point of a new target region, assigned a new target region number, and a breadth-first search is used to recursively check the adjacent grids in the eight directions around the grid, namely up, down, left, right, upper left, upper right, lower left, and lower right. If an adjacent grid is also a candidate anomalous grid and has not yet been marked, it is added to the current target region and assigned the same target region number. At the same time, the search continues to expand outward from the adjacent grid as the center until no more adjacent candidate anomalous grids that meet the conditions are found. The above process is repeated until all candidate anomalous grids are marked. At this time, the set of grids with the same target region number constitutes a connected target region. For each connected target region, calculate the average of all its grid-containing mineralization predictions. The connected target areas are classified and a mineral exploration target area distribution map is generated, specifically: if there are Mark the connected target region as a primary target region; if there are Mark the connected target region as a secondary target region; if there are The connected target region is marked as a level 3 target region, where, This represents the mean of the predicted mineralization values; This represents the standard deviation of the predicted mineralization value.

[0041] In the above process, the connectivity criterion was used to analyze the connectivity of candidate anomaly grids. This was based on the fact that tin polymetallic mineralization typically exhibits a clustered distribution in space, with a single ore body or mineralization zone occupying multiple consecutive adjacent grid cells in a horizontal projection, rather than an isolated single grid point. By examining the adjacent grids in eight directions (up, down, left, right, upper left, upper right, lower left, lower right) around each candidate anomaly grid, spatially adjacent candidate anomaly grids can be merged into the same connected target area, forming a mineralization cluster region with actual geological significance. A breadth-first search method was used to expand layer by layer from the seed point, ensuring that all grids meeting the connectivity conditions were completely covered. The technical effect of this process is that it transforms discrete candidate anomaly grids into continuous, compact target area polygons, while deleting isolated target areas with fewer than four grids. This effectively eliminates false anomalies caused by single-grid noise and random disturbances, improving the geological reliability of the target area delineation. use , and Three grading standards are used to classify connected target areas into Level 1, Level 2, and Level 3. The basis for this classification is that different levels of target areas reflect the degree of deviation of the average mineralization prediction value within the target area from the overall background level. The average prediction value of the Level 1 target area... Reach the mean Plus Double standard deviation This area represents a highly significant anomalous cluster, with the highest probability of mineralization, and should be prioritized for engineering verification; the average predicted value of the secondary target area falls within the comprehensive anomaly threshold. and The area between these two points is considered a moderately significant anomaly and is therefore a secondary priority for validation; the average predicted value of the level 3 target area is lower than [the specified value]. However, since all of its constituent meshes satisfy The candidate conditions still have certain mineral exploration potential and can be used as backup verification targets. The technical effect of this treatment is that it realizes the quantitative classification and ranking of mineral exploration target areas, provides a clear priority basis for exploration deployment, and enables limited drilling and engineering verification resources to be invested in the target areas with the highest probability of mineralization, thereby improving the economic benefits and success rate of exploration work.

[0042] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0043] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0044] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for tin polymetallic mineral exploration based on the overlay and identification of multivariate geoscientific information, characterized by the following steps: include: S1: Determine the scope of the study area, divide the study area into multiple grids, collect the chemical element content, apparent resistivity value at the center of the grid, and reflectivity data of different bands for each grid, and calculate the hydroxyl alteration anomaly index and iron staining alteration anomaly index of each grid based on the reflectivity data. S2: Perform a logarithmic transformation on the chemical element content of each grid, and standardize the logarithmically transformed chemical element content, apparent resistivity value, hydroxyl alteration anomaly index and iron staining alteration anomaly index of each grid to construct a standardized feature vector for each grid. S3: Obtain the standardized feature vectors of known typical mineral deposits, calculate the similarity between each grid and the known typical mineral deposits, determine the adaptive weight coefficient of each grid based on the similarity, delineate the neighborhood region with each grid as the center, determine the local variation coefficient of each grid for the neighborhood region, and calculate the comprehensive mineralization favorability of each grid based on the adaptive weight coefficient, local variation coefficient and feature variables. S4: Construct a comprehensive mineralization favorability matrix based on the comprehensive mineralization favorability of each grid, perform discrete wavelet decomposition on the comprehensive mineralization favorability matrix, extract high-frequency detail coefficients of each layer and reconstruct the local anomaly field matrix of each layer, and calculate the comprehensive anomaly index of each grid based on the value of each grid in the local anomaly field matrix of each layer. S5: Calculate the mineralization prediction value based on the comprehensive anomaly index and comprehensive mineralization favorability of each grid, compare the mineralization prediction value with the comprehensive anomaly threshold, screen out candidate anomaly grids, use the 8 connectivity criteria to perform connectivity analysis on the candidate anomaly grids, obtain the connected target areas, classify the connected target areas, and obtain the mineral exploration target area distribution map.

2. The method for tin polymetallic ore prospecting based on multi-source geoscientific information overlay identification according to claim 1, characterized in that, Calculate the hydroxyl alteration anomaly index and the iron staining alteration anomaly index, including Collect the apparent resistivity value of each grid center, as well as the reflectance data of the blue band, red band, near-infrared band, and short-wave infrared band; By comparing the reflectance changes in the shortwave infrared band and the near-infrared band, the alteration response associated with hydroxyl-containing minerals was extracted and defined as the hydroxyl alteration anomaly index. By comparing the reflectance changes in the red and blue bands, the alteration response associated with iron oxide staining was extracted and defined as the iron staining alteration anomaly index.

3. The method for tin polymetallic ore prospecting based on multivariate geoscientific information overlay identification according to claim 2, characterized in that, The chemical elements include tin, tungsten, copper, lead, zinc, silver, fluorine, and boron. A standardized feature vector is constructed for each grid cell, specifically as follows: Obtain the elemental content values ​​of tin, tungsten, copper, lead, zinc, silver, fluorine, and boron in each grid. The content value of each element in each grid is summed with the detection limit reference value of the corresponding element to obtain the corrected content value of each element. Each of the corrected content values ​​is subjected to logarithmic transformation to obtain the corresponding logarithmic characteristic value of each element; The logarithmic eigenvalues, apparent resistivity values, hydroxyl alteration anomaly index, and iron staining alteration anomaly index corresponding to each element are used as mesh feature variables. Each grid feature variable is standardized to obtain the standardized result for each grid feature variable. The standardized results are combined in a preset order to obtain the standardized feature vector of each grid.

4. The method for tin polymetallic ore prospecting based on multi-source geoscientific information overlay identification according to claim 1, characterized in that, The specific steps for calculating the adaptive weight coefficients for each grid are as follows: Obtain the standardized eigenvectors of known typical mineral deposits, and calculate the Euclidean distance between the standardized eigenvectors of each grid and the standardized eigenvectors of known typical mineral deposits. Based on the Euclidean distance between each grid and known typical deposits, the theoretical similarity of each grid is calculated using Euclidean distance according to the distance decay method; The theoretical similarity of each grid is input into the S-shaped mapping function for transformation to obtain the adaptive weight coefficients of each grid. The S-shaped mapping function includes a first adjustment parameter and a second adjustment parameter, which are used to adjust the response magnitude and offset of the adaptive weight coefficient to changes in theoretical similarity.

5. The method for tin polymetallic mineral exploration based on multi-source geoscientific information overlay and identification according to claim 1, characterized in that, The specific steps for calculating the local coefficient of variation for each grid are as follows: Centered on each grid, surround it The grid region is defined as the surrounding neighborhood region of the grid. The mean and standard deviation of each feature in all surrounding neighborhood regions are calculated. The standard deviation is divided by the mean to calculate the local coefficient of variation of each feature of each grid.

6. The method for tin polymetallic ore prospecting based on multi-source geoscientific information overlay identification according to claim 5, characterized in that, The specific steps for calculating the overall mineralization favorability of each grid are as follows: Based on adaptive weighting coefficients, local variation coefficients, and standardized feature variables for each grid, the overall mineralization favorability of each grid is calculated: in, Index representing the number of standardized feature variables; Indicates the first The first grid One standardized feature variable; Indicates the first The first grid The local variation coefficients of the standardized feature variables; Indicates the first The overall mineralization advantage of each grid; Indicates the first The adaptive weight coefficients for each grid.

7. The method for tin polymetallic ore prospecting based on multi-source geoscientific information overlay identification according to claim 6, characterized in that, The specific steps to obtain the three-layer high-frequency detail coefficients are as follows: Arrange the comprehensive mineralization favorability of each grid according to the row and column positions of the grid to construct an M-row N-column comprehensive mineralization favorability matrix; Daubechies-4 wavelet is selected as the wavelet basis function. The comprehensive mineralization favorability matrix is ​​first subjected to low-pass filtering and high-pass filtering by row, and then downsampled to obtain the row low-frequency matrix and row high-frequency matrix. The row low-frequency matrix and the row high-frequency matrix are then subjected to low-pass filtering and high-pass filtering by column, and downsampling is performed respectively to obtain the first layer low-frequency approximation coefficient matrix, as well as the first layer horizontal high-frequency detail coefficient matrix, vertical high-frequency detail coefficient matrix and diagonal high-frequency detail coefficient matrix. Using the first-layer low-frequency approximation coefficient matrix as input, repeat the same decomposition operation to obtain the second-layer low-frequency approximation coefficient matrix, as well as the second-layer horizontal high-frequency detail coefficient matrix, vertical high-frequency detail coefficient matrix, and diagonal high-frequency detail coefficient matrix. The matrix includes the third layer of high-frequency detail coefficient matrices in the horizontal, vertical, and diagonal directions; This yields nine high-frequency detail coefficient matrices corresponding to the three-level discrete wavelet decomposition, which characterize the high-frequency detail coefficients.

8. The method for tin polymetallic ore prospecting based on multi-source geoscientific information overlay identification according to claim 1, characterized in that, The specific steps for calculating the comprehensive anomaly index for each grid are as follows: The high-frequency detail coefficient matrices in the horizontal direction, vertical direction, and diagonal direction obtained from the first, second, and third layers of discrete wavelet decomposition are extracted respectively. A zero matrix with the same size as the high-frequency detail coefficient matrix of the corresponding layer is constructed to replace the low-frequency approximation coefficient matrix of that layer. Inverse wavelet reconstruction is performed on the zero matrix of each layer and the corresponding high-frequency detail coefficient matrix in the horizontal direction, vertical direction, and diagonal direction of the layer to obtain the first layer local anomaly field matrix, the second layer local anomaly field matrix, and the third layer local anomaly field matrix with the same size as the original comprehensive mineralization favorable matrix. Calculate the mean and standard deviation of the local anomaly field matrix for each layer, and determine the anomaly threshold for that layer based on the mean and standard deviation of the corresponding layer. The ratio of the outlier value of each grid in the local anomaly field matrix of each layer to the anomaly threshold of the corresponding layer is processed to obtain the relative anomaly intensity of each grid in each layer. The maximum value of the relative anomaly intensity of each grid in the first, second, and third layers is selected as the comprehensive anomaly index of that grid.

9. The method for tin polymetallic ore prospecting based on multi-source geoscientific information overlay identification according to claim 8, characterized in that, The specific steps for calculating the comprehensive anomaly threshold are as follows: The calculation steps for the comprehensive anomaly threshold include: The mineralization prediction value for each grid is obtained by multiplying the comprehensive anomaly index and comprehensive mineralization favorability of each grid. Statistical analysis was performed on the mineralization prediction values ​​of all grids in the study area to obtain the arithmetic mean and standard deviation of the mineralization prediction values; the comprehensive anomaly threshold was determined by weighted summation of the arithmetic mean and standard deviation of the mineralization prediction values.

10. The method for tin polymetallic ore prospecting based on multi-source geoscientific information overlay identification according to claim 9, characterized in that, Generate a mineral exploration target area distribution map, specifically: The predicted mineralization values ​​are compared with the comprehensive anomaly threshold of the set prospecting target area. Grids with predicted mineralization values ​​greater than the comprehensive anomaly threshold are initially screened as candidate anomaly grids. The connectivity analysis of the candidate anomaly grids is performed using the 8-connectivity criterion to obtain the connected target area, specifically: All mineralization prediction values ​​are greater than the comprehensive anomaly threshold The grids are marked as candidate anomalous grids, and a blank label matrix with the same size as the original grid is created to record the connected target region number to which each grid belongs. Then, all candidate anomalous grids are traversed in row-major order. When an unmarked candidate anomalous grid is encountered, it is used as the seed point of a new target region, assigned a new target region number, and a breadth-first search is used to recursively check the adjacent grids in the eight directions around the grid, namely up, down, left, right, upper left, upper right, lower left, and lower right. If an adjacent grid is also a candidate anomalous grid and has not yet been marked, it is added to the current target region and assigned the same target region number. At the same time, the search continues to expand outward from the adjacent grid as the center until no more adjacent candidate anomalous grids that meet the conditions are found. The above process is repeated until all candidate anomalous grids are marked. At this time, the set of grids with the same target region number constitutes a connected target region. For each connected target region, calculate the average of all its grid-containing mineralization predictions. The connected target areas are classified and a mineral exploration target area distribution map is generated, specifically: if there are Mark the connected target region as a primary target region; if there are Mark the connected target region as a secondary target region; if there are The connected target region is marked as a level 3 target region, where, This represents the mean of the predicted mineralization values; This represents the standard deviation of the predicted mineralization value.

Images