A method and system for intelligent delineation of gold ore prospecting target areas based on multi-source geological data

Abstract

This application provides a method and system for intelligent delineation of gold prospecting target areas based on multi-source geological data, belonging to the field of mineral exploration technology. This application collects remote sensing images, geophysical exploration data, and surface rock chemical data associated with gold mineralization. Then, based on the mineralization characteristics of gold, it corrects for spatial distribution differences of gold elements and decomposes mineral spectral mixing information to enhance mineralization-related characteristic information. Next, based on the characteristic information, physical exploration data, and geochemical data, it mines the data feature correlation relationships to construct a mineralization correlation network and inputs the characteristic information to obtain mineralization correlation parameters. Combining the spatial distribution characteristics of multi-source geological data, it calculates mineralization probability values ​​point by point, generating a continuous mineralization probability surface. Finally, it filters areas with probability values ​​higher than a preset threshold, and uses a typical gold deposit model to lock in corresponding closed areas, recording coordinate boundary information to generate a gold prospecting target area map, thus achieving intelligent and accurate delineation of gold prospecting target areas.

Description

Technical Field

[0001] This application relates to the field of mineral exploration technology, and in particular to a method and system for intelligent delineation of gold prospecting target areas based on multi-source geological data. Background Technology

[0002] Intelligent delineation of gold prospecting target areas is one of the core technologies in the field of mineral exploration, which can efficiently integrate multi-source geological information to assist in prospecting decisions. With the gradual depletion of shallow gold resources, the demand for prospecting in deep and complex geological conditions is surging, and this type of method shows broad application prospects in improving prospecting efficiency and reducing exploration costs.

[0003] Currently, existing intelligent delineation methods for gold prospecting targets based on multi-source geological data mostly employ machine learning models to integrate remote sensing, geophysical, and geochemical data. They generate mineralization potential maps by analyzing data correlation characteristics to delineate target areas. Among these, the random forest model is widely used in practical exploration scenarios due to its suitability for small to medium-sized datasets and ease of deployment.

[0004] However, such methods often handle geological spatial features implicitly, making it difficult to accurately characterize mineralization-related features and deep correlations between data. This results in insufficient matching between the generated target area map and the actual mineralized structure, limiting the accuracy of mineral exploration. Therefore, existing technologies suffer from the technical problem of unsatisfactory target area delineation for gold exploration. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for intelligent delineation of gold prospecting target areas based on multi-source geological data, so as to solve the problem of poor accuracy in the delineation of gold prospecting target areas in the existing technology.

[0006] To address the aforementioned technical problems, in a first aspect, this application provides a method for intelligent delineation of gold prospecting target areas based on multi-source geological data, comprising:

[0007] Collect multi-source geological data, including remote sensing image data associated with gold mineralization, geophysical exploration data of the earth, and geochemical data of surface rocks;

[0008] Based on the mineralization characteristics of gold deposits and combined with the remote sensing image data, the spatial distribution differences of gold elements are corrected and the mineral spectral mixing information is decomposed to enhance the feature information related to gold mineralization in the multi-source geological data.

[0009] Based on the aforementioned feature information, physical exploration data, and geochemical data, deep-seated correlations between various data features are explored, and a mineralization correlation network is constructed based on these correlations. The feature information is then input into the mineralization correlation network to obtain the mineralization correlation parameters of the gold deposit.

[0010] Based on the mineralization correlation parameters and the spatial distribution characteristics of the multi-source geological data, the mineralization probability value of gold deposits is calculated point by point within the geographic space covered by the multi-source geological data, and a continuously distributed mineralization probability surface is dynamically generated based on the mineralization probability value.

[0011] The regions with mineralization probability values ​​higher than a preset threshold are selected from the mineralization probability surface. Combined with the spatial structure pattern corresponding to a typical gold deposit model, closed regions with high mineralization probability and consistent structural characteristics are locked. The coordinate boundary information of the closed regions is recorded, and a gold prospecting target area map is generated based on the coordinate boundary information.

[0012] Optionally, based on the feature information, physical exploration data, and geochemical data, deep-level correlations between various data features are mined, and a mineralization correlation network is constructed based on these correlations; the feature information is input into the mineralization correlation network to obtain the mineralization correlation parameters of the gold deposit, including:

[0013] The feature information, the anomalous signals in the physical exploration data, and the element combination information in the geochemical data are uniformly resampled to grid cells with the same spatial resolution.

[0014] Each grid cell is assigned a node, and based on the similarity measurement results of the feature vectors between nodes, connection edges are established between similar nodes to initially form an associated network topology.

[0015] By iteratively optimizing the topology of the associated network and dynamically adjusting the weight values ​​of the connecting edges, an associated network reflecting the contribution of different data features to mineralization is generated.

[0016] Using the feature information as input, forward propagation calculations are performed in the association network to generate gold deposit mineralization association parameters at the nodes of each grid cell.

[0017] Optionally, by iteratively optimizing the topology of the associated network and dynamically adjusting the weight values ​​of the connecting edges, an associated network reflecting the contribution of different data features to mineralization is generated, including:

[0018] Based on the feature vector differences of the node pairs connected by the connecting edge, the initial association strength of each connecting edge is calculated. From the network topology, connecting edges with an initial association strength higher than a preset threshold are identified to form a core connecting edge set.

[0019] The weight adjustment factor is calculated based on the spatial proximity of the corresponding node of each connection edge in the core connection edge set in the grid cell, and the changing trend of the feature vector of the node in continuous iteration;

[0020] Based on the weight adjustment factor, the difference between high-weight and low-weight connecting edges is reduced, and the weight values ​​of all connecting edges are adjusted synchronously.

[0021] Repeat the weight adjustment process until the standard deviation of the overall weight distribution of the association network changes less than a preset tolerance in three consecutive iterations, thus obtaining an initial association network with stable weights.

[0022] In the initial association network, the total amount of feature information received by each node through different connection edges is counted as the node activation degree;

[0023] Based on the node activation degree, the initial association network is pruned, removing all connection edges whose node activation degree is lower than that of a preset node, thereby generating an association network that reflects the magnitude of the contribution of different data features to mineralization.

[0024] Optionally, based on the mineralization correlation parameters and the spatial distribution characteristics of the multi-source geological data, the mineralization probability value of gold deposits is calculated point-by-point within the geographic space covered by the multi-source geological data, and a continuously distributed mineralization probability surface is dynamically generated based on the mineralization probability value, including:

[0025] Using the center point of each grid cell as the calculation point, the mineralization correlation parameters of the calculation point and the spatial distribution pattern of the mineralization correlation parameters within a certain range around it are obtained;

[0026] By combining the correlation parameter values ​​of known ore deposit points and non-ore deposit points, a probability calculation model is established, and the correlation parameter values ​​of each calculation point are substituted to calculate the initial mineralization probability value of each point;

[0027] Based on the spatial distribution characteristics of the geological structure extension direction and the rock mass contact zone morphology in the multi-source geological data, the initial mineralization probability value is spatially constrained and corrected to obtain the corrected mineralization probability value.

[0028] Using spatial interpolation, all the corrected mineralization probability values ​​are fitted to form a continuous and smooth mineralization probability surface.

[0029] Optionally, based on the spatial distribution characteristics of the geological structure extension direction and the contact zone morphology of the rock mass in the multi-source geological data, the initial mineralization probability value is spatially constrained and corrected to obtain a corrected mineralization probability value, including:

[0030] Identify the extended trajectories of major geological structural lines within the geographic space and calculate the vertical distance from each calculation point to the nearest major geological structural line;

[0031] Based on the boundary morphology of the rock mass contact zone, calculate the vertical distance from each calculation point to the nearest rock mass contact zone boundary;

[0032] Based on the vertical distance from the calculation point to the geological structure line and the vertical distance to the rock mass contact zone, the geological structure influence factor and the rock mass contact influence factor are calculated respectively.

[0033] Multiply the geological structure influence factor and the rock mass contact influence factor to obtain the spatial constraint weight of each calculation point;

[0034] The initial mineralization probability value of each calculation point is multiplied by the corresponding spatial constraint weight to generate the corrected mineralization probability value.

[0035] Optionally, regions with mineralization probability values ​​higher than a preset threshold are selected from the mineralization probability surface. Combined with the spatial structure pattern corresponding to a typical gold deposit model, closed regions with high mineralization probability and consistent structural characteristics are identified. The coordinate boundary information of these closed regions is recorded. Based on the coordinate boundary information, a gold prospecting target area map is generated, including:

[0036] On the mineralization probability surface, a probability threshold is set, and all regions with mineralization probability values ​​higher than the probability threshold are extracted as candidate regions.

[0037] The spatial morphology, area size, and relative positional relationship between each candidate region and known fault structures and lithological boundaries are compared with the spatial structure pattern of a typical gold deposit model to calculate the degree of matching.

[0038] Based on the calculation results, candidate regions with matching degrees exceeding the set standard are selected as preferred target regions, and the continuous boundary coordinates of each preferred target region are determined.

[0039] On the digital map, polygonal areas are drawn based on the boundary coordinates of the preferred target areas, and grade information is marked to generate a gold prospecting target area map.

[0040] Optionally, based on the mineralization characteristics of gold deposits and in conjunction with the remote sensing image data, the spatial distribution differences of gold elements are corrected, and the mineral spectral mixing information is decomposed to enhance the feature information related to gold mineralization in the multi-source geological data, including:

[0041] Based on the gold content in the geochemical data, the concentration variation pattern of gold in different geographical units was analyzed, and the boundary threshold between the background value and the outlier value of gold was calculated.

[0042] Based on the boundary threshold, the gold content in each geographical unit is calculated by difference to generate a spatial variation map reflecting the degree of gold enrichment.

[0043] For the remote sensing image data, mixed pixels containing multiple mineral spectral features are identified, and single mineral endmember spectral curves related to gold mineralization are extracted.

[0044] Based on the spectral curve of the single mineral endmember, spectral decomposition is performed on each of the mixed pixels to calculate the proportion of each endmember mineral in the mixed pixel and generate a mineral abundance map.

[0045] By integrating the spatial variation map and the mineral abundance map, the intensity of information related to gold mineralization is enhanced through feature overlay, and the enhanced feature information is output.

[0046] Secondly, this application provides an intelligent delineation system for gold prospecting target areas based on multi-source geological data, comprising:

[0047] The acquisition module is used to acquire multi-source geological data, including remote sensing image data associated with gold mineralization, physical exploration data of the earth, and geochemical data of surface rocks.

[0048] The enhancement module is used to enhance the feature information related to gold mineralization in the multi-source geological data by correcting the spatial distribution differences of gold elements and decomposing mineral spectral mixing information based on the mineralization characteristics of gold deposits and in combination with the remote sensing image data.

[0049] A construction module is used to mine the deep-seated correlations between various data features based on the feature information, physical exploration data, and geochemical data, and to construct a mineralization correlation network based on the correlations; the feature information is input into the mineralization correlation network to obtain the mineralization correlation parameters of gold deposits;

[0050] The calculation module is used to calculate the mineralization probability value of gold deposits point by point within the geographic space covered by the multi-source geological data, based on the mineralization correlation parameters and the spatial distribution characteristics of the multi-source geological data, and dynamically generate a continuously distributed mineralization probability surface based on the mineralization probability value.

[0051] The generation module is used to filter out areas with mineralization probability values ​​higher than a preset threshold from the mineralization probability surface, and, in combination with the spatial structure pattern corresponding to a typical gold deposit model, lock in closed areas with high mineralization probability and consistent structural characteristics, record the coordinate boundary information of the closed areas, and generate a gold prospecting target area map based on the coordinate boundary information.

[0052] Thirdly, this application provides an electronic device, comprising:

[0053] Memory, used to store computer programs;

[0054] A processor is configured to execute the computer program to implement the steps of the intelligent delineation method for gold prospecting target areas based on multi-source geological data as described in the first aspect above.

[0055] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the intelligent delineation method for gold prospecting target areas based on multi-source geological data as described in the first aspect above.

[0056] The intelligent delineation method for gold prospecting target areas based on multi-source geological data provided in this application can provide comprehensive and fundamental data source support for subsequent prospecting analysis by collecting remote sensing images, geophysical exploration data, and surface rock geochemical data associated with gold mineralization. By correcting for spatial distribution differences of gold elements and decomposing mineral spectral mixing information based on gold mineralization characteristics, it can strengthen mineralization-related characteristics and reduce interference from invalid data. By mining deep correlations among various data features and constructing a mineralization correlation network, and obtaining mineralization correlation parameters by inputting feature information, it can accurately capture the mineralization correlation patterns between data and provide a basis for mineralization probability calculation. By combining the spatial distribution characteristics of multi-source data to calculate mineralization probability values ​​point by point and generating a continuous mineralization probability surface, it can intuitively present the spatial distribution of regional mineralization potential. By screening high-probability areas and combining them with the spatial structure of typical deposit models, it can lock the target closed area to generate a target area map, which can accurately delineate prospecting target areas and improve the targeting of prospecting.

[0057] Furthermore, feature information, physical exploration anomaly signals, and geochemical elemental assemblage information are resampled to grid cells with the same spatial resolution. Nodes are assigned to each grid cell, and connection edges are established based on the similarity of node feature vectors to form a preliminary network topology. Through iterative optimization and adjustment of connection edge weights, an association network reflecting the mineralization contribution of data features is generated. Input feature information is then used for forward propagation calculations to obtain the mineralization association parameters for each grid cell. This step achieves unified spatial adaptation of multi-source data, constructs an association network that closely matches mineralization realities, accurately quantifies the impact of different data features on mineralization, and makes the generated mineralization association parameters more reliable. This provides accurate data support for subsequent mineralization probability calculations and target area delineation, further improving the accuracy of mineral exploration target area delineation. Attached Figure Description

[0058] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 A flowchart illustrating an intelligent delineation method for gold prospecting target areas based on multi-source geological data, provided in this application embodiment;

[0060] Figure 2 A flowchart illustrating the specific implementation of an intelligent delineation method for gold prospecting target areas based on multi-source geological data, provided in this application embodiment;

[0061] Figure 3 This is a schematic diagram of the structure of an intelligent delineation system for gold prospecting target areas based on multi-source geological data, provided in an embodiment of this application. Detailed Implementation

[0062] In gold prospecting target area delineation, existing intelligent methods based on multi-source geological data often handle geological spatial features in a concealed manner, failing to accurately capture key information related to mineralization and the deep connections between various data. This results in the generated target area map not matching the actual mineralized structure, significantly reducing the accuracy of prospecting and making it difficult to meet the needs of prospecting under deep and complex geological conditions. Therefore, there is an urgent need for a technical solution that can improve the accuracy of target area delineation.

[0063] To address the aforementioned issues, this invention proposes an intelligent delineation method for gold prospecting target areas based on multi-source geological data. The core of this method is to accurately capture mineralization patterns by selectively processing multi-source data and constructing a mineralization correlation network. This method first enhances mineralization-related feature information, then mines deep data correlations to construct a network to obtain mineralization parameters, and subsequently generates a mineralization probability surface. Combined with typical deposit structures, it identifies high-potential areas. By explicitly processing spatial features and deepening data correlation analysis, it overcomes the accuracy shortcomings of existing methods, achieving precise delineation of gold prospecting target areas and providing reliable support for efficient mineral exploration.

[0064] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0065] The core of this application is to provide an intelligent delineation method for gold prospecting target areas based on multi-source geological data. A flowchart of one specific implementation method is shown below. Figure 1 As shown, the method includes:

[0066] S101. Collect multi-source geological data.

[0067] Multi-source geological data refers to a collection of various types of geological information used for gold prospecting analysis, including remote sensing imagery data associated with gold mineralization, geophysical exploration data, and geochemical data of surface rocks. Remote sensing imagery data is used to capture the spectral and morphological characteristics of regional surface and shallow geology; geophysical exploration data is used to reflect the differences in the physical properties of underground geological bodies; and geochemical data of surface rocks is used to reflect the elemental content and combination characteristics of surface rocks, providing a basis for judging mineralization potential.

[0068] In one specific implementation, the above-mentioned multi-source geological data is obtained by combining the collection of specialized equipment with the retrieval of existing geological databases: remote sensing image data is acquired by high-resolution satellite or airborne remote sensing platform, covering the target mineral exploration area; geophysical exploration data is collected through on-site exploration using specialized equipment such as gravimeters and magnetometers, capturing the differences in physical properties of underground geological bodies; and surface rock geochemical data is obtained by collecting rock samples from the target area through field systems, followed by laboratory elemental detection and composition analysis, ensuring that the collected data is accurately correlated with gold mineralization characteristics.

[0069] S102. Based on the mineralization characteristics of gold deposits and combined with the remote sensing image data, the spatial distribution differences of gold elements are corrected, and the mineral spectral mixing information is decomposed to enhance the characteristic information related to gold mineralization in the multi-source geological data.

[0070] Among them, mineral spectral mixing information refers to the comprehensive information formed by the superposition of multiple mineral spectra within a single pixel of a remote sensing image, used to reflect the combination of minerals within the pixel area. The core purpose of this step is to filter out invalid data interference and highlight key features related to gold deposit formation by specifically processing the spatial distribution of gold elements and mineral spectral information.

[0071] Optionally, step S102 may specifically include the following steps:

[0072] S1021. Based on the gold content in the geochemical data, analyze the concentration variation pattern of gold in different geographical units, and calculate the boundary threshold between the background value and the outlier value of gold.

[0073] Among them, the background value of gold refers to the average content of gold in normal rocks in the target area that have not been affected by mineralization; the outlier value refers to the gold content that is significantly higher than the background value and may indicate mineralization activity; the boundary threshold is the critical value that distinguishes the background value from the outlier value and is used to define the abnormal area of ​​gold enrichment.

[0074] S1022. Based on the boundary threshold, perform difference calculation on the gold content in each geographical unit to generate a spatial variation map reflecting the degree of gold enrichment.

[0075] Among them, the geographical unit refers to the target area sub-block divided according to a certain spatial scale, which facilitates the accurate analysis of the spatial distribution of elements; the spatial variation map is used to intuitively present the enrichment or depletion of gold elements in different areas relative to the background value.

[0076] S1023. For the remote sensing image data, identify mixed pixels containing multiple mineral spectral features, and extract the single mineral endmember spectral curves related to gold mineralization.

[0077] Among them, mixed pixels refer to pixels in remote sensing images that simultaneously contain multiple minerals and superimposed spectral features; single mineral endmember spectral curves refer to the characteristic spectral curves of a certain pure mineral, which are the core basis for distinguishing different minerals. Endmember minerals related to gold mineralization include pyrite, quartz, etc.

[0078] S1024. Based on the single mineral endmember spectral curve, perform spectral decomposition operation on each of the mixed pixels to calculate the proportion of each endmember mineral in the mixed pixel and generate a mineral abundance map.

[0079] Among them, the spectral decomposition operation refers to the process of breaking down the comprehensive spectrum of mixed pixels into the end-member spectra of each individual mineral; the mineral abundance map is used to display the content ratio of each mineralization-related mineral in different regions, and intuitively reflects the distribution of mineral combinations that are favorable for mineralization.

[0080] S1025. Integrate the spatial variation map and the mineral abundance map, enhance the information intensity related to gold mineralization through feature overlay, and output the enhanced feature information.

[0081] Feature overlay refers to spatially correlating and fusing gold enrichment information with ore-forming mineral assemblage information, making ore-related features more prominent, reducing interference from irrelevant information, and improving the accuracy of subsequent analysis.

[0082] In this embodiment, the spatial distribution differences of gold elements are first corrected through S1021-S1022, then the mineral spectral mixing information is decomposed through S1023-S1024, and finally integrated and superimposed through S1025 to complete the enhancement of mineralization-related characteristic information and form a complete processing flow.

[0083] As an example, firstly, rock geochemical data of the target mineral exploration area were collected using the S1021 system. A gold exploration area in a mountainous region, with an area of ​​50 km × 50 km, was selected as the target area. Fresh rock samples from the surface (0-20 cm) were collected using a grid sampling method, with one sampling point every 2 km, for a total of 625 sampling points. After laboratory digestion and inductively coupled plasma mass spectrometry (ICP-MS), the gold content data for each sampling point was obtained, and the data was presented as a correspondence between the sampling point coordinates and the gold content in ppm.

[0084] Subsequently, the gold content of all sampling points within each 1km × 1km geographic unit was statistically analyzed. After eliminating extreme outliers using an iterative method, the statistical parameters of the gold content in each unit were calculated to determine the boundary threshold between background and outlier values. The boundary threshold can be calculated using the background value plus standard deviation method commonly used in geochemistry, which will not be elaborated upon in this application. Assuming that the average gold content of each unit in the target area is 0.05ppm, the standard deviation is 0.02ppm, and the confidence coefficient is 2.5, this value is suitable for low-to-medium disturbance geological scenarios, and the confidence coefficient typically ranges from 2 to 3. The specific calculation process is as follows: first, the product of the standard deviation and the confidence coefficient is calculated, i.e., 2.5 × 0.02ppm = 0.05ppm; then, this result is added to the average value, 0.05ppm + 0.05ppm = 0.1ppm. Finally, areas within each geographic unit with gold content higher than 0.1ppm are identified as anomalously enriched areas, and a classification data table of unit numbers and whether anomalies are identified is generated simultaneously.

[0085] The above example is only one example of this application. In actual applications, the confidence coefficient and sampling point density can be adjusted according to the needs. This application does not limit this.

[0086] Secondly, geographic units were divided into 1km × 1km scales using S1022, with unit numbers named using a horizontal-vertical coordinate system, for example, from X01Y01 to X50Y50. Based on the gold content data of each unit obtained from S1021, the difference between the actual gold content of each unit and the background value of 0.05ppm was calculated, generating a difference dataset. The data format is a correspondence between unit number, gold content (ppm), and the difference (ppm) between the gold content and the background value. A positive difference indicates that the gold content of that unit is relatively enriched compared to the background value, and the larger the difference, the higher the enrichment. A negative or zero difference indicates depletion. Based on this difference dataset, a spatial variation map was drawn using geographic information system software. Different color gradients were used to mark the difference ranges: red marked areas with a difference ≥ 0.05ppm, orange marked areas with a difference between 0.01-0.05ppm, and blue marked areas with a difference < 0ppm, clearly identifying the core areas with high gold enrichment and defining the key areas for subsequent mineral spectroscopic analysis.

[0087] Next, a 30m×30m hyperspectral remote sensing image of the target area was acquired via S1023. The image data format was an ENVI standard hdr+dat file, containing spectral information in the 200-2500nm band. The image was preprocessed using the pure pixel index method, with 1000 iterations and a threshold of 0.01. Pure pixels containing only a single mineral and mixed pixels containing multiple minerals were identified, generating a pixel classification map. The data format was a correspondence between pixel coordinates and pixel type, with pixel types categorized into pure and mixed.

[0088] Subsequently, pure pixels were removed, and for the remaining mixed pixels, the constructed mineral spectral library was used. This library contains standard spectral curves of more than 10 minerals, including pyrite, quartz, and calcite, with data in the form of mineral name, band number, and corresponding spectral reflectance correlation data. A spectral matching algorithm was used to accurately extract single-endmember spectral curves of pyrite and quartz, which are closely related to gold mineralization, ensuring the relevance and accuracy of subsequent spectral decomposition.

[0089] Then, based on the extracted pyrite and quartz endmember spectral curves in S1024, a linear spectral decomposition algorithm is used to calculate the mixed pixels. The core of this algorithm is to decompose the comprehensive spectral reflectance of each mixed pixel in each band into a weighted sum of the spectral reflectance of the two endmember minerals and their corresponding abundance coefficients. The spectral decomposition calculation of the mineral proportion can be obtained using a linear spectral decomposition model, which will not be elaborated on in this application.

[0090] After computation, a mineral abundance dataset for each mixed cell is generated. The data format includes cell coordinates and corresponding percentages of pyrite and quartz abundance. The requirement is that the sum of the abundance percentages of the two minerals within a single cell is ≥80%, while the percentages of other minerals are negligible. Based on this dataset, a mineral abundance map is drawn using GIS software, marking areas with pyrite abundance ≥30% and quartz abundance ≥50%, visually presenting the combination and distribution characteristics of mineral-rich minerals. The above example is merely one illustration of this application. In practical applications, nonlinear spectral decomposition algorithms can also be used, while adjusting the abundance percentage thresholds; this application does not limit this approach.

[0091] Finally, in step S1025, the spatial variation map of gold enrichment generated in S1022 and the mineral abundance map generated in S1024 are overlaid in the same GIS coordinate system, and the spatial intersection analysis algorithm is used to filter overlapping areas. The filtering criteria are a gold element difference ≥ 0.03 ppm, and a pyrite content ≥ 30% and a quartz content ≥ 50%. By enhancing the feature signal intensity of this area and suppressing interference from irrelevant areas, an enhanced feature information dataset is generated. The data format is the correspondence between the target area coordinates, gold element enrichment level, and mineral assemblage proportion. Finally, this dataset and the corresponding visualization map are output, completing the entire process of step S102 and providing accurate input for subsequent mineralization correlation analysis.

[0092] In another specific implementation, the median method can be used to calculate the background value and boundary threshold of gold elements, and the probability density function method can be used for spectral decomposition to adapt to data sources with different geological conditions and improve the applicability of the method.

[0093] This application effectively enhances mineralization-related characteristics, reduces interference from invalid information, and improves the reliability of subsequent mineralization analysis and target area delineation by correcting spatial differences in gold elements, decomposing mineral spectral information, and integrating and superimposing it.

[0094] S103. Based on the aforementioned feature information, physical exploration data, and geochemical data, the deep-seated correlations between various data features are mined, and a mineralization correlation network is constructed based on these correlations; the feature information is input into the mineralization correlation network to obtain the mineralization correlation parameters of the gold deposit.

[0095] Among them, the mineralization correlation network is a network model that describes the correlation logic between multi-source geological data features and gold mineralization. Nodes correspond to grid cell data, and connecting edges reflect the correlation strength between features, which is used to quantify the impact of data features on mineralization.

[0096] Optionally, such as Figure 2 As shown, step S103 may specifically include the following steps:

[0097] S1031. The feature information, the anomalous signals in the physical exploration data, and the element combination information in the geochemical data are uniformly resampled to grid cells with the same spatial resolution.

[0098] Resampling is a process that adjusts data of different resolutions to a uniform spatial scale to eliminate spatial dimensional differences between multi-source data and ensure that data can be fused and analyzed within the same grid cell. Physical exploration anomaly signals refer to physical parameter signals that deviate from regional background values ​​and may indicate geological anomalies. Element combination information refers to the elemental combinations and content ratios related to gold mineralization.

[0099] S1032. Assign a node to each of the grid cells, and establish connection edges between similar nodes based on the similarity measurement results of the feature vectors between nodes, so as to initially form an associated network topology.

[0100] In this context, a node is the basic unit of data corresponding to a grid cell in the mineralization association network, used to carry the multi-source data features within the grid; a feature vector is a set of vectors formed by quantizing the multi-type data features within a grid cell; and the association network topology refers to the network framework composed of nodes and connecting edges, which only reflects the connection relationship between nodes and does not assign edge weights.

[0101] S1033. By iteratively optimizing the topology of the associated network, the weight values ​​of the connecting edges are dynamically adjusted to generate an associated network that reflects the contribution of different data features to mineralization.

[0102] Specifically, step S1033 may include the following processes: Based on the feature vector differences of the node pairs connected by the connecting edges, calculate the initial association strength of each connecting edge; identify connecting edges with initial association strength higher than a preset threshold from the association network topology to form a core connecting edge set; calculate a weight adjustment factor based on the spatial proximity of the connecting nodes corresponding to each connecting edge in the core connecting edge set within the grid cell, and the changing trend of the node's feature vector in continuous iterations; reduce the difference between high-weight and low-weight connecting edges based on the weight adjustment factor, and simultaneously adjust the weight values ​​of all connecting edges; repeat the weight adjustment process until the standard deviation of the overall weight distribution of the association network changes less than a preset tolerance in three consecutive iterations, resulting in a weight-stable initial association network; in the initial association network, count the total amount of feature information received by each node through different connecting edges, as the node activation degree; prune the initial association network based on the node activation degree, removing all connecting edges whose node activation degree is lower than a preset threshold, generating an association network reflecting the magnitude of different data features' contribution to mineralization.

[0103] In the above steps, the weight value is used to characterize the contribution of the association relationship corresponding to the connection edge to the mineralization. The higher the weight, the greater the contribution. The initial association strength is the basic association degree of the connection edge calculated based on the differences in node features. The node activation degree refers to the total amount of feature information received by the node through the connection edge, which is used to filter the connection relationships that are valuable to mineralization.

[0104] S1034. Using the feature information as input, perform forward propagation calculations in the association network to generate gold ore mineralization association parameters at the nodes of each grid cell.

[0105] Among them, forward propagation computation refers to the process of passing the input feature information layer by layer along the connection edges of the associated network and outputting the result through the weight superposition operation between nodes; mineralization correlation parameters are the core indicators for quantifying the correlation between grid units and gold mineralization, and directly reflect the mineralization potential within the unit.

[0106] In this embodiment, firstly, in S1031, the enhanced feature information, physical exploration anomaly signals, and geochemical element combination information are resampled to uniformly adapt data from different sources and with different resolutions to grid cells of the same spatial scale, thus achieving spatial alignment and fusion preparation of multi-source data. Then, in S1032, a dedicated node is assigned to each grid cell, the multi-source features of the nodes are quantified to form feature vectors, similar nodes are screened through similarity measurement and connection edges are established to construct an initial association network topology that only reflects the node connection relationship. Subsequently, in S1033, the initial association strength of the connection edges is calculated, the weight values ​​are iteratively adjusted, and pruning is performed based on node activation to complete network optimization and highlight the differences in the contribution of data features to mineralization. Finally, in S1034, the enhanced feature information is input into the optimized network, and through forward propagation weighted operation, parameters for quantifying mineralization association are generated at each node, forming a complete closed-loop process from data integration to parameter output.

[0107] As an example, firstly, multi-source data is processed through S1031. A mountainous gold exploration area consistent with the previous description is selected. The enhanced feature information output from S102, the anomalous signals in the gravity exploration data, and the gold-sulfur-silicon elemental combination information in the geochemical data are uniformly resampled to 1km×1km grid cells, maintaining consistency with the geographic unit scale described earlier to ensure spatial data alignment and form a unified dataset of "grid cells + multi-source data features". The above example is merely one illustration of this application; in practical applications, the grid resolution can be adjusted according to requirements, and this application does not impose any limitations on this.

[0108] Secondly, by assigning a node to each grid cell using S1032, the multi-source data features corresponding to each node are quantized into a 12-dimensional feature vector, covering 12 core indicators such as gold enrichment, pyrite content, and gravity anomaly. Taking two adjacent grid nodes as an example, their feature vectors are as follows: and The similarity between the two nodes is calculated based on the cosine similarity formula. The details of the formula will not be elaborated here. The similarity threshold is set to 0.7, and connection edges are established between nodes with similarity higher than this threshold.

[0109] The specific calculation process is as follows: First, calculate the dot product of the two vectors, the result is 8.92; then calculate the magnitudes of the two vectors respectively. , The product of their moduli is 3.42 × 3.38 = 11.56. The final result calculated using the cosine similarity formula is... The result is higher than the threshold of 0.7, therefore, a connection edge is established between the two nodes. This method is used to calculate the similarity between all nodes, generating a total of 2500 nodes and 3860 connection edges, thus initially forming an associated network topology. The above example is merely one illustration of this application; in practical applications, the feature vector dimension and similarity threshold can also be adjusted, and this application does not impose any limitations on this.

[0110] Next, the network is optimized using S1033. First, the initial association strength of each connection edge is calculated based on the differences in node feature vectors, using the reciprocal of Euclidean distance. After calculating the Euclidean distance using the Euclidean distance formula, the reciprocal is then taken to obtain the initial association strength. The details of the formula are not elaborated further in this application. Taking nodes i and j as an example, the differences in each dimension of the 12-dimensional feature vectors of the two nodes are [0.02, 0.02, 0.02, ..., 0.01]. The sum of the squares of the differences in each dimension is calculated to be 0.001. Then, the square root is taken to obtain the Euclidean distance of 0.0316. The initial association strength is calculated as follows: The initial association strength threshold was set to 0.6, and 2120 core connection edges were selected to form a set.

[0111] Next, the weight adjustment factor is calculated. Combining the spatial proximity of the core connecting edge nodes and the iterative change trend of the feature vectors, adjacent grid cells are assigned a coefficient of 0.8, and non-adjacent cells are assigned a coefficient of 0.5. The adjustment factor is calculated based on a weighted sum formula, with weight proportions of 0.6 for the spatial proximity coefficient and 0.4 for the average feature vector change trend. Assuming a core connecting edge corresponds to adjacent nodes with a spatial proximity coefficient of 0.8, and the absolute values ​​of the feature vector changes for the two nodes are 0.12 and 0.16 respectively, the average change is first calculated as follows: The adjustment factor is then calculated using the weighted sum formula. .

[0112] The weight values ​​are adjusted according to the adjustment factor to reduce the difference between high and low weights. This process is repeated iteratively until the standard deviation of the overall weight distribution of the network changes by less than 0.01 after three consecutive iterations, resulting in a stable initial interconnection network. The activation degree of each node is calculated, and an activation degree threshold of 5.0 is set. 320 connection edges with activation degrees lower than this value are removed to generate the final interconnection network. The above example is only one example of this application. In practical applications, the threshold and coefficients can be adjusted, and this application does not limit this.

[0113] Finally, the enhanced feature information is input into the optimized correlation network through S1034 for forward propagation calculation. Each node receives the feature information of the connected nodes and superimposes it according to the weight to generate the mineralization correlation parameters of each grid cell. The parameter range is 0-10. The higher the value, the stronger the mineralization correlation, providing accurate indicators for subsequent mineralization probability calculation.

[0114] In another specific implementation, the similarity measure can adopt the Pearson correlation coefficient algorithm, and the weight adjustment factor can increase the metallogenic geological background coefficient to adapt to complex geological scenarios and improve the targeting of network optimization.

[0115] This application integrates multi-source data, constructs and optimizes a mineralization correlation network, accurately mines the deep correlation between data features and mineralization, and outputs reliable mineralization correlation parameters, laying the foundation for subsequent mineralization probability analysis.

[0116] S104. Based on the mineralization correlation parameters and the spatial distribution characteristics of the multi-source geological data, calculate the mineralization probability value of gold deposits point by point within the geographic space covered by the multi-source geological data, and dynamically generate a continuously distributed mineralization probability surface based on the mineralization probability value.

[0117] Among them, the mineralization probability value is the core indicator for quantifying the likelihood of gold deposit formation at a given calculation point; a higher value indicates greater mineralization potential. The mineralization probability surface is a continuous and smooth graph formed by fitting the probability values ​​of discrete calculation points through spatial interpolation, used to visually present the spatial variation patterns of regional mineralization potential.

[0118] Optionally, step S104 may specifically include the following steps:

[0119] S1041. Using the center point of each grid cell as the calculation point, obtain the mineralization correlation parameters of the calculation point and the spatial distribution pattern of the mineralization correlation parameters within a certain range around it.

[0120] The calculation point is the center point of the grid cell selected as the reference location for probability calculation, ensuring that the calculation results uniformly cover the target area. The spatial distribution pattern refers to the changing trend and clustering characteristics of the mineralization-related parameters within the area surrounding the calculation point, which is used to help determine the spatial reliability of the parameters and provide spatial context information for subsequent probability calculations.

[0121] S1042. By combining the correlation parameter values ​​of known ore deposit points and non-ore points, a probability calculation model is established, and the correlation parameter values ​​of each calculation point are substituted to calculate the initial mineralization probability value of each point.

[0122] Among them, known ore deposits are locations within the target area where gold deposits have been proven to exist, while non-ore deposits are locations where no gold deposits have been verified to form. Both serve as sample data for model training. The probabilistic calculation model is a predictive model built based on the sample data, used to correlate mineralization parameters with mineralization probability, and its core function is to quantify and transform mineralization potential.

[0123] S1043. Based on the spatial distribution characteristics of the geological structure extension direction and the rock mass contact zone morphology in the multi-source geological data, the initial mineralization probability value is spatially constrained and corrected to obtain the corrected mineralization probability value.

[0124] Specifically, step S1043 may include the following processes: identifying the extension trajectory of the main geological structural lines within the geographic space and calculating the vertical distance from each calculation point to the nearest main geological structural line; calculating the vertical distance from each calculation point to the nearest boundary of the rock mass contact zone based on the boundary morphology of the rock mass contact zone; calculating the geological structural influence factor and the rock mass contact influence factor respectively based on the vertical distance from the calculation point to the geological structural line and the vertical distance to the rock mass contact zone; multiplying the geological structural influence factor and the rock mass contact influence factor to obtain the spatial constraint weight of each calculation point; and multiplying the initial mineralization probability value of each calculation point with the corresponding spatial constraint weight to generate a corrected mineralization probability value.

[0125] In the above steps, the direction of geological structure extension refers to the strike trend of geological structures such as faults and folds, while the contact zone of rock masses refers to the boundary area where different types of rock masses meet. Both are key spatial conditions for gold deposit formation. The spatial constraint weight is a correction coefficient that integrates the influence of geological structures and contact zones of rock masses. It is used to adjust the initial probability value so that the result is more in line with the actual mineralization geological laws. The geological structure influence factor and the rock mass contact influence factor quantify the contribution of geological structures and rock mass contact zones to the mineralization probability, respectively. The factor values ​​are positively correlated with the mineralization favorability.

[0126] S1044. Using spatial interpolation, fit all the corrected mineralization probability values ​​to form a continuous and smooth mineralization probability surface.

[0127] Spatial interpolation is a technique that uses mathematical algorithms to interpolate and estimate the probability values ​​of discrete calculation points, fill in the probability information of non-calculation point locations, and finally generate continuous spatial data. Its core purpose is to realize the transformation of discrete data into continuous surfaces and improve the visualization and practicality of the results.

[0128] In this embodiment, firstly, in S1041, the reference position for probability calculation is determined by using the center point of the grid cell as the calculation point, and the spatial distribution pattern of the mineralization correlation parameters of each calculation point and the surrounding range of parameters is obtained. Then, in S1042, a probability calculation model is constructed using known ore deposit points and non-ore points as samples and a logistic regression algorithm is used. The mineralization correlation parameters of each calculation point are substituted into the model to obtain an initial mineralization probability value in the range of 0-1, realizing the preliminary quantification of mineralization potential. Subsequently, in S1043, the influencing factors and spatial constraint weights are calculated in combination with key mineralization conditions such as geological structure and rock mass contact zone to correct the initial mineralization probability value, making the result more consistent with the actual geological mineralization law. Finally, in S1044, the Kriging interpolation method is used to fit and complete the corrected probability value of the discrete calculation points, generating a continuous and smooth mineralization probability surface, forming a complete closed-loop process from data preprocessing, probability quantification, geological correction to visualization output.

[0129] As an example, firstly, the calculation point data is processed using S1041, continuing the 1km×1km grid cell method described earlier. With the center point of each cell as the calculation point, a total of 2500 calculation points are obtained. The mineralization-related parameters for each calculation point are recorded, with a parameter range of 0-10. Simultaneously, a 3km×3km perimeter is delineated around each calculation point. The mean, extreme values, and distribution trends of the mineralization-related parameters within this perimeter are statistically analyzed to determine the spatial distribution pattern of each calculation point. Some calculation points show a clustered distribution of parameters around their perimeter, while others show a dispersed distribution. The above example is merely one illustration of this application; in practical applications, the scale of the perimeter can be adjusted, and this application does not impose any limitations on this.

[0130] Secondly, a probabilistic calculation model was established using S1042. Fifteen known mineral deposits and 30 non-mineralized points within the target area were selected as samples. Mineralization correlation parameters were extracted from the sample points, and a logistic regression algorithm was used to construct the model. The model training process was as follows: the sample parameters were used as input features, mineral deposits were marked as 1, and non-mineralized points were marked as 0 as output labels. The model parameters were optimized using a gradient descent algorithm, and iterative training was conducted until the model's prediction accuracy stabilized, resulting in the final probabilistic calculation model.

[0131] The initial mineralization probability value can be calculated using a logistic regression formula, which will not be elaborated upon in this application. Specifically, the mineralization correlation parameters of the calculation points are substituted into the model to output an initial mineralization probability value ranging from 0 to 1. Taking a calculation point as an example, its mineralization correlation parameter is 7.2, and the calculated initial mineralization probability value after substituting it into the model is 0.65; another calculation point with a correlation parameter of 2.8 has an initial mineralization probability value of 0.18, which conforms to the rule that the higher the correlation parameter, the greater the mineralization probability. The above example is only one example of this application. In practical applications, the sample size and model type can be adjusted, and this application does not limit this.

[0132] Next, by correcting the initial probability value using S1043, the extension trajectories of three main geological structural lines within the target area are identified based on the structural and rock mass information from multi-source geological data. A distance calculation algorithm is then used to obtain the vertical distance from each calculation point to the nearest structural line. Taking a certain calculation point as an example, its vertical distance to the nearest structural line is 1.2 km. Based on the negative correlation between distance and mineralization advantage, a geological structural influence factor is calculated. This influence factor can be obtained using an exponential decay formula, which will not be elaborated upon in this application.

[0133] The threshold for the attenuation of tectonic influence is set at 2km. Substituting the vertical distance of 1.2km into the calculation, the ratio of distance to threshold is first calculated to be 1.2÷2=0.6. Then, the geological tectonic influence factor is calculated to be approximately 0.549 according to the exponential attenuation logic.

[0134] Based on the boundary morphology of the rock mass contact zone, the vertical distance from the calculation point to the nearest rock mass contact zone is calculated to be 0.8 km. The rock mass contact influence factor can also be calculated using the exponential decay formula, which will not be elaborated on in this application. Setting the rock mass influence decay threshold to 1.5 km, and substituting it into the vertical distance of 0.8 km, the ratio of distance to threshold is 0.8 ÷ 1.5 ≈ 0.533. Using the same exponential decay logic, the rock mass contact influence factor is calculated to be approximately 0.587.

[0135] Multiplying the two influencing factors yields the spatial constraint weight, calculated as 0.549 × 0.587 ≈ 0.322. Then, multiplying the initial mineralization probability value of this calculation point (0.65) by the weight (0.322) gives the corrected mineralization probability value: 0.65 × 0.322 ≈ 0.209, completing the probability correction for this calculation point. This method is used to correct all calculation points, ensuring the results conform to geological mineralization patterns.

[0136] Finally, a mineralization probability surface is generated using S1044. The corrected probability values ​​of 2500 calculation points are then fitted using the Kriging interpolation method. The interpolation calculation can be obtained using the Kriging interpolation formula, which will not be elaborated upon in this application. During the interpolation process, the variogram parameter is set based on the spatial correlation between calculation points. The probability value of each non-calculation point is estimated using the interpolation algorithm, ultimately generating a mineralization probability surface with a resolution of 30m×30m covering the entire target area. The surface clearly presents the clustering range of high-probability areas and the distribution characteristics of low-probability areas. The above example is only one example of this application. In practical applications, other methods such as inverse distance weighted interpolation can also be used, and this application does not limit this approach.

[0137] In another specific implementation, the probability calculation model can be constructed using the random forest algorithm to improve the prediction accuracy in complex geological scenarios; spatial interpolation can use the natural neighborhood interpolation method to adapt to irregularly distributed calculation points and further improve the rationality of surface fitting.

[0138] This application generates a continuous mineralization probability surface that conforms to actual geological laws by calculating mineralization probability, correcting geological feature constraints, and fitting spatial interpolation, thus accurately presenting the distribution of regional mineralization potential.

[0139] S105. Select regions with mineralization probability values ​​higher than a preset threshold from the mineralization probability surface, and combine them with the spatial structure pattern corresponding to the typical gold deposit model to lock closed regions with high mineralization probability and consistent structural characteristics. Record the coordinate boundary information of the closed regions, and generate a gold prospecting target area map based on the coordinate boundary information.

[0140] Among them, the typical gold deposit model is a standardized model constructed based on the spatial morphology, structural relationships, and lithological combinations of proven gold deposits, used to provide structural references for favorable mineralization areas. The prospecting target area map is a digital map that marks the boundaries and grade information of preferred target areas, used to visually present key prospecting areas.

[0141] Optionally, step S105 may specifically include the following steps:

[0142] S1051. On the mineralization probability surface, a probability threshold is set, and all regions with mineralization probability values ​​higher than the probability threshold are extracted as candidate regions.

[0143] The probability threshold is a critical value used to distinguish between high and low mineralization potential. It is set in conjunction with the regional mineralization background and exploration needs to screen out areas with a certain probability of mineralization. The candidate area is a preliminary screening area after the probability value meets the standard. It is determined solely based on probability indicators and has not yet considered the adaptability of geological structure.

[0144] S1052. The spatial morphology, area size, and relative positional relationship between each candidate area and known fault structures and lithological boundaries are matched with the spatial structure pattern of a typical gold deposit model.

[0145] Among them, the matching degree is an indicator that quantifies the degree of fit between the candidate area and the structural characteristics of a typical deposit model. The higher the value, the more consistent the structure and the stronger the reliability of mineralization. The spatial structure model, which encompasses the morphological characteristics, scale range, and locational relationship with surrounding geological structures of a typical gold deposit, is the core reference for the matching degree calculation.

[0146] S1053. Based on the calculation results, select candidate regions with matching degrees exceeding the set standard as preferred target regions, and determine the continuous boundary coordinates of each preferred target region.

[0147] The preferred target area is the region that simultaneously meets both the probability threshold and the structural matching standard, and is the final key area for mineral exploration. Continuous boundary coordinates are a precise set of spatial coordinates that delineate the scope of the preferred target area, ensuring that the target area can be accurately located on the digital map.

[0148] S1054. On the digital map, draw a polygonal area based on the boundary coordinates of the preferred target area and mark the grade information to generate a gold prospecting target area map.

[0149] Among them, the grade information is a priority identifier determined by the mineralization probability value and matching degree of the preferred target area. It is used to distinguish the exploration value of different target areas and provide a basis for ranking in field exploration.

[0150] In this embodiment, firstly, by setting a probability threshold in combination with the regional mineralization background in S1051, areas meeting the probability criteria are extracted from the mineralization probability surface as candidate areas, completing the initial screening of high-potential areas; then, by S1052, a suitable typical gold deposit model is retrieved, and features such as the spatial morphology, area, and positional relationship with geological structures of the candidate areas are extracted. A multi-feature weighted algorithm is used to calculate the matching degree between the candidate areas and the model, realizing the geological structure adaptability verification; subsequently, by setting the matching degree standard in S1053, the preferred target areas that simultaneously meet the probability and structural requirements are selected, and the continuous boundary of the target area is delineated and the coordinates are recorded using a spatial boundary extraction algorithm; finally, by S1054, the boundary coordinates are imported into a digital geological map, the target area range is drawn and the priority level is marked, generating a complete gold prospecting target area map.

[0151] As an example, firstly, the mineralization probability surface is processed using S1051. Following the probability surface of the previously discussed mountainous gold exploration area, and considering the regional mineralization background and exploration accuracy requirements, a probability threshold of 0.4 is set. Based on this threshold, all regions with probability values ​​higher than 0.4 are extracted from the surface, resulting in 12 candidate regions. These regions are distributed around major geological structural lines and rock mass contact zones, with each candidate region having an area between 2.5 square kilometers. Up to 8 The above example is merely one example of this application. In practical applications, the probability threshold can be adjusted according to exploration needs, and this application does not limit this.

[0152] Secondly, by calculating the matching degree using S1052, a typical deposit model suitable for the gold deposit type in this area is retrieved. The model clearly shows that the spatial morphology of a typical gold deposit is an irregular ellipse with an area of ​​5-10 square meters. Furthermore, they are mostly distributed within 1 km on both sides of the fault structure and adjacent to the boundary of intermediate-acidic rock mass. The matching degree is calculated using a multi-feature weighted matching algorithm, with the following weights for each feature: spatial morphology 0.3, area size 0.2, and positional relationship with the structural and lithological boundaries 0.5.

[0153] The matching degree can be calculated using a weighted summation formula, which will not be elaborated upon in this application. The core calculation logic is the sum of the products of the feature fit and their corresponding weights. Taking a candidate region as an example, its spatial shape has a fit of 0.8 with the typical model, its area size has a fit of 0.9, and the area of ​​this candidate region is 6.2. It is within the typical model setting of 5-10. Within this range, the positional fit with the fault structure and rock mass boundary is 0.85, the distance to the fault structure is only 0.6 km, and it is adjacent to the boundary of the intermediate-acidic rock mass.

[0154] The specific calculation process is as follows: 0.8×0.3+0.9×0.2+0.85×0.5=0.24+0.18+0.425=0.845, resulting in a matching degree of 0.845 for the candidate region. The matching degree calculation for all 12 candidate regions is performed using this method, with results ranging from 0.32 to 0.86. The above example is merely one illustration of this application; in practical applications, the feature weights and fit criteria can be adjusted, and this application does not impose any limitations on this.

[0155] Next, the preferred target areas were selected using S1053. A matching degree standard of 0.6 was set, and candidate areas with matching degrees higher than this value were selected, resulting in six preferred target areas. A spatial boundary extraction algorithm was then used to delineate continuous boundaries along the probability value abrupt change lines of each preferred target area, recording the coordinates of each inflection point. For example, the boundary coordinates of a certain preferred target area were (118.23°E, 36.52°N), (118.28°E, 36.52°N), (118.28°E, 36.48°N), and (118.23°E, 36.48°N), forming a complete set of boundary coordinates.

[0156] Finally, a mineral exploration target area map is generated using S1054. The boundary coordinates are imported into a digital geological map based on the WGS84 coordinate system, and polygonal areas are drawn according to the coordinate inflection points to accurately delineate the range of six preferred target areas. Combining the mineralization probability value and matching degree of each target area, three levels are defined: the top two target areas in both probability and matching degree are labeled as Level 1, the middle two as Level 2, and the remaining two as Level 3. The core coordinates and key geological background of each target area are also labeled. The final generated mineral exploration target area map clearly presents the location, range, and priority of each target area, providing clear guidance for on-site exploration deployment. The above example is only one example of this application; in practical applications, the matching degree standard and level classification rules can be adjusted, and this application does not limit this.

[0157] In another specific implementation, the matching degree calculation can adopt a neural network algorithm, which automatically learns the matching rules between typical deposit characteristics and candidate areas by training the model, thereby improving the matching accuracy in complex geological scenarios; the classification can add a geological risk coefficient index to further optimize the rationality of target area priority determination.

[0158] This application generates a reliable and practical mineral exploration target map through probability screening, structural matching, and precise annotation, accurately identifying high-value exploration areas, providing scientific guidance for gold mine field exploration, and improving exploration efficiency and targeting.

[0159] Figure 3 This application provides a schematic diagram of a specific implementation of an intelligent delineation system for gold prospecting target areas based on multi-source geological data, as illustrated in the embodiments of this application. Figure 3 The system may include:

[0160] The acquisition module 31 is used to acquire multi-source geological data, including remote sensing image data associated with gold mineralization, physical exploration data of the earth, and geochemical data of surface rocks.

[0161] The enhancement module 32 is used to enhance the feature information related to gold mineralization in the multi-source geological data by correcting the spatial distribution differences of gold elements and decomposing the mineral spectral mixing information based on the mineralization characteristics of gold deposits and in combination with the remote sensing image data.

[0162] The construction module 33 is used to mine the deep-level correlation between various data features based on the feature information, physical exploration data and geochemical data, and construct a mineralization correlation network based on the correlation; the feature information is input into the mineralization correlation network to obtain the mineralization correlation parameters of gold deposits;

[0163] The calculation module 34 is used to calculate the mineralization probability value of gold deposits point by point within the geographic space covered by the multi-source geological data, based on the mineralization correlation parameters and the spatial distribution characteristics of the multi-source geological data, and dynamically generate a continuously distributed mineralization probability surface based on the mineralization probability value.

[0164] The generation module 35 is used to filter out areas with mineralization probability values ​​higher than a preset threshold from the mineralization probability surface, and, in combination with the spatial structure pattern corresponding to a typical gold deposit model, lock in closed areas with high mineralization probability and consistent structural characteristics, record the coordinate boundary information of the closed areas, and generate a gold prospecting target area map based on the coordinate boundary information.

[0165] The intelligent delineation system for gold prospecting target areas based on multi-source geological data in this application embodiment is used to implement the aforementioned intelligent delineation method for gold prospecting target areas based on multi-source geological data. Therefore, the specific implementation of the intelligent delineation system for gold prospecting target areas based on multi-source geological data can be found in the embodiment section of the intelligent delineation method for gold prospecting target areas based on multi-source geological data above. The specific implementation can be referred to the description of the corresponding embodiments, which will not be repeated here.

[0166] This application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the above-described intelligent delineation method for gold prospecting target areas based on multi-source geological data.

[0167] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the intelligent delineation method for gold prospecting target areas based on multi-source geological data described above.

[0168] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0169] The embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the embodiments of the intelligent delineation method for gold prospecting target areas based on multi-source geological data.

[0170] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0171] The above provides a detailed description of the intelligent delineation method and system for gold prospecting target areas based on multi-source geological data provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for intelligent delineation of gold prospecting target areas based on multi-source geological data, characterized in that, include: Collect multi-source geological data, including remote sensing image data associated with gold mineralization, geophysical exploration data of the earth, and geochemical data of surface rocks; Based on the mineralization characteristics of gold deposits and combined with the remote sensing image data, the spatial distribution differences of gold elements are corrected and the mineral spectral mixing information is decomposed to enhance the feature information related to gold mineralization in the multi-source geological data. Based on the aforementioned feature information, physical exploration data, and geochemical data, deep-seated correlations between various data features are explored, and a mineralization correlation network is constructed based on these correlations. The feature information is then input into the mineralization correlation network to obtain the mineralization correlation parameters of the gold deposit. Based on the mineralization correlation parameters and the spatial distribution characteristics of the multi-source geological data, the mineralization probability value of gold deposits is calculated point by point within the geographic space covered by the multi-source geological data, and a continuously distributed mineralization probability surface is dynamically generated based on the mineralization probability value. The regions with mineralization probability values ​​higher than a preset threshold are selected from the mineralization probability surface. Combined with the spatial structure pattern corresponding to a typical gold deposit model, closed regions with high mineralization probability and consistent structural characteristics are locked. The coordinate boundary information of the closed regions is recorded, and a gold prospecting target area map is generated based on the coordinate boundary information.

2. The method according to claim 1, characterized in that, Based on the aforementioned feature information, physical exploration data, and geochemical data, deep-seated correlations between various data features are mined, and a mineralization correlation network is constructed based on these correlations. The feature information is then input into the mineralization correlation network to obtain mineralization correlation parameters for gold deposits, including: The feature information, the anomalous signals in the physical exploration data, and the element combination information in the geochemical data are uniformly resampled to grid cells with the same spatial resolution. Each grid cell is assigned a node, and based on the similarity measurement results of the feature vectors between nodes, connection edges are established between similar nodes to initially form an associated network topology. By iteratively optimizing the topology of the associated network and dynamically adjusting the weight values ​​of the connecting edges, an associated network reflecting the contribution of different data features to mineralization is generated. Using the feature information as input, forward propagation calculations are performed in the association network to generate gold deposit mineralization association parameters at the nodes of each grid cell.

3. The method according to claim 2, characterized in that, By iteratively optimizing the topology of the associated network and dynamically adjusting the weight values ​​of the connecting edges, an associated network reflecting the contribution of different data features to mineralization is generated, including: Based on the feature vector differences of the node pairs connected by the connecting edge, the initial association strength of each connecting edge is calculated. From the network topology, connecting edges with an initial association strength higher than a preset threshold are identified to form a core connecting edge set. The weight adjustment factor is calculated based on the spatial proximity of the corresponding node of each connection edge in the core connection edge set in the grid cell, and the changing trend of the feature vector of the node in continuous iteration; Based on the weight adjustment factor, the difference between high-weight and low-weight connecting edges is reduced, and the weight values ​​of all connecting edges are adjusted synchronously. Repeat the weight adjustment process until the standard deviation of the overall weight distribution of the association network changes less than a preset tolerance in three consecutive iterations, thus obtaining an initial association network with stable weights. In the initial association network, the total amount of feature information received by each node through different connection edges is counted as the node activation degree; Based on the node activation degree, the initial association network is pruned, removing all connection edges whose node activation degree is lower than that of a preset node, thereby generating an association network that reflects the magnitude of the contribution of different data features to mineralization.

4. The method according to claim 1, characterized in that, Based on the aforementioned mineralization correlation parameters and the spatial distribution characteristics of the multi-source geological data, the mineralization probability value of gold deposits is calculated point-by-point within the geographic space covered by the multi-source geological data. Based on the mineralization probability values, a continuously distributed mineralization probability surface is dynamically generated, including: Using the center point of each grid cell as the calculation point, the mineralization correlation parameters of the calculation point and the spatial distribution pattern of the mineralization correlation parameters within a certain range around it are obtained; By combining the correlation parameter values ​​of known ore deposit points and non-ore deposit points, a probability calculation model is established, and the correlation parameter values ​​of each calculation point are substituted to calculate the initial mineralization probability value of each point; Based on the spatial distribution characteristics of the geological structure extension direction and the rock mass contact zone morphology in the multi-source geological data, the initial mineralization probability value is spatially constrained and corrected to obtain the corrected mineralization probability value. Using spatial interpolation, all the corrected mineralization probability values ​​are fitted to form a continuous and smooth mineralization probability surface.

5. The method according to claim 4, characterized in that, Based on the spatial distribution characteristics of the geological structure extension direction and the contact zone morphology of the rock mass in the multi-source geological data, the initial mineralization probability value is spatially constrained and corrected to obtain the corrected mineralization probability value, including: Identify the extended trajectories of major geological structural lines within the geographic space and calculate the vertical distance from each calculation point to the nearest major geological structural line; Based on the boundary morphology of the rock mass contact zone, calculate the vertical distance from each calculation point to the nearest rock mass contact zone boundary; Based on the vertical distance from the calculation point to the geological structure line and the vertical distance to the rock mass contact zone, the geological structure influence factor and the rock mass contact influence factor are calculated respectively. Multiply the geological structure influence factor and the rock mass contact influence factor to obtain the spatial constraint weight of each calculation point; The initial mineralization probability value of each calculation point is multiplied by the corresponding spatial constraint weight to generate the corrected mineralization probability value.

6. The method according to claim 1, characterized in that, From the mineralization probability surface, regions with mineralization probability values ​​higher than a preset threshold are selected. Combined with the spatial structure pattern corresponding to a typical gold deposit model, closed regions with high mineralization probability and consistent structural characteristics are identified. The coordinate boundary information of these closed regions is recorded. Based on this coordinate boundary information, a gold prospecting target area map is generated, including: On the mineralization probability surface, a probability threshold is set, and all regions with mineralization probability values ​​higher than the probability threshold are extracted as candidate regions. The spatial morphology, area size, and relative positional relationship between each candidate region and known fault structures and lithological boundaries are compared with the spatial structure pattern of a typical gold deposit model to calculate the degree of matching. Based on the calculation results, candidate regions with matching degrees exceeding the set standard are selected as preferred target regions, and the continuous boundary coordinates of each preferred target region are determined. On the digital map, polygonal areas are drawn based on the boundary coordinates of the preferred target areas, and grade information is marked to generate a gold prospecting target area map.

7. The method according to claim 1, characterized in that, Based on the mineralization characteristics of gold deposits and combined with the remote sensing image data, the spatial distribution differences of gold elements are corrected, and the mineral spectral mixing information is decomposed to enhance the feature information related to gold mineralization in the multi-source geological data, including: Based on the gold content in the geochemical data, the concentration variation pattern of gold in different geographical units was analyzed, and the boundary threshold between the background value and the outlier value of gold was calculated. Based on the boundary threshold, the gold content in each geographical unit is calculated by difference to generate a spatial variation map reflecting the degree of gold enrichment. For the remote sensing image data, mixed pixels containing multiple mineral spectral features are identified, and single mineral endmember spectral curves related to gold mineralization are extracted. Based on the spectral curve of the single mineral endmember, spectral decomposition is performed on each of the mixed pixels to calculate the proportion of each endmember mineral in the mixed pixel and generate a mineral abundance map. By integrating the spatial variation map and the mineral abundance map, the intensity of information related to gold mineralization is enhanced through feature overlay, and the enhanced feature information is output.

8. A smart delineation system for gold prospecting target areas based on multi-source geological data, characterized in that, include: The acquisition module is used to acquire multi-source geological data, including remote sensing image data associated with gold mineralization, physical exploration data of the earth, and geochemical data of surface rocks. The enhancement module is used to enhance the feature information related to gold mineralization in the multi-source geological data by correcting the spatial distribution differences of gold elements and decomposing mineral spectral mixing information based on the mineralization characteristics of gold deposits and in combination with the remote sensing image data. A construction module is used to mine the deep-seated correlations between various data features based on the feature information, physical exploration data, and geochemical data, and to construct a mineralization correlation network based on the correlations; the feature information is input into the mineralization correlation network to obtain the mineralization correlation parameters of gold deposits; The calculation module is used to calculate the mineralization probability value of gold deposits point by point within the geographic space covered by the multi-source geological data, based on the mineralization correlation parameters and the spatial distribution characteristics of the multi-source geological data, and dynamically generate a continuously distributed mineralization probability surface based on the mineralization probability value. The generation module is used to filter out areas with mineralization probability values ​​higher than a preset threshold from the mineralization probability surface, and, in combination with the spatial structure pattern corresponding to a typical gold deposit model, lock in closed areas with high mineralization probability and consistent structural characteristics, record the coordinate boundary information of the closed areas, and generate a gold prospecting target area map based on the coordinate boundary information.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of the intelligent delineation method for gold prospecting target areas based on multi-source geological data as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the intelligent delineation method for gold prospecting target areas based on multi-source geological data as described in any one of claims 1 to 7.

Images