Metallogenic prediction method based on multi-modal data fusion and human-machine collaboration

Abstract

The present disclosure relates to the technical field of mineral resource exploration, and particularly provides a mineralization prediction method based on multi-modal data fusion and man-machine collaboration. The method comprises: collecting multi-scale data of a study area and constructing a multi-source spatial database; performing feature extraction from the multi-source spatial database and generating a geological feature layer, a geochemical element anomaly layer and a remote sensing alteration mineral layer; performing spatial registration and rasterization processing on the geological feature layer, the geochemical element anomaly layer and the remote sensing alteration mineral layer to construct a multi-element feature dataset; inputting the multi-element feature dataset into a pre-constructed mineralization prediction model to generate a mineralization favorability prediction map of the study area; performing spatial overlay analysis on the mineralization favorability prediction map and a comprehensive favorable block, and delineating a prospecting target area after performing geological rationality verification on the overlaid region. The present disclosure realizes intelligent fusion of multi-modal data and man-machine collaborative decision-making, and significantly improves the accuracy of mineralization prediction.

Description

Technical Field

[0001] This disclosure relates to the field of mineral resource exploration technology, and in particular to a mineralization prediction method based on multimodal data fusion and human-machine collaboration. Background Technology

[0002] Mineralization prediction is a crucial step in mineral resource exploration, and its accuracy directly impacts prospecting efficiency and costs. With the deepening of geological exploration, a wealth of multi-source data, including geological, geochemical, and remote sensing data, has been accumulated. How to utilize this data to improve the accuracy of mineralization prediction has been a long-standing technical concern in this field.

[0003] In related technologies, mineralization prediction methods mainly rely on the experience and judgment of geological experts. Qualitative analysis of regional geological background, ore-forming models, and ore-controlling elements is conducted, combined with limited exploration data to delineate prospecting target areas. However, because geological, geochemical, and remote sensing data are scattered across maps and reports in different formats and standards, there is a lack of effective fusion methods. A large amount of implicit information remains untapped. Furthermore, different experts may reach significantly different conclusions based on the same data, resulting in highly subjective and poorly repeatable predictions, greatly reducing the accuracy of mineralization predictions. Summary of the Invention

[0004] In view of this, the exemplary embodiments of this disclosure provide a mineralization prediction method based on multimodal data fusion and human-machine collaboration to solve the problems existing in related technologies.

[0005] One aspect of the exemplary embodiments of this disclosure provides a mineralization prediction method based on multimodal data fusion and human-machine collaboration, the method comprising: Multi-scale data of the study area were collected, and a multi-source spatial database containing regional and exploration scales was constructed based on the multi-scale data; the multi-scale data included geological data, geochemical data, and remote sensing data. The geological feature layer is generated by extracting ore-bearing stratigraphic features, ore-controlling fault features, and alteration mineral features from the multi-source spatial database; geochemical element anomaly features are extracted to generate a geochemical element anomaly layer; and remote sensing alteration mineral information is extracted to generate a remote sensing alteration mineral layer. The geological feature layer, geochemical element anomaly layer, and remote sensing alteration mineral layer are spatially registered and rasterized to construct a multi-dimensional feature dataset. The multivariate feature dataset is input into a pre-constructed mineralization prediction model to generate a mineralization favorability prediction map for the study area. The mineralization favorability prediction map is spatially overlaid with the comprehensive favorable blocks, and the geological rationality of the overlaid area is verified before the prospecting target area is delineated.

[0006] In another aspect of exemplary embodiments of this disclosure, a computer device is provided, including a memory, a processor, and a computer program stored in the memory, the processor executing the computer program to implement the methods described in exemplary embodiments of this disclosure.

[0007] In another aspect of exemplary embodiments of this disclosure, a computer-readable storage medium is provided having a computer program / instructions stored thereon that, when executed by a processor, implements the methods described in exemplary embodiments of this disclosure.

[0008] In another aspect of exemplary embodiments of this disclosure, a computer program product is provided, including a computer program / instructions that, when executed by a processor, implement the methods described in exemplary embodiments of this disclosure.

[0009] As will be described in detail below, a metallogenic prediction method based on multimodal data fusion and human-machine collaboration according to an embodiment of this disclosure involves collecting multi-scale data of the study area and constructing a multi-source spatial database containing regional and exploration scales based on the multi-scale data. The multi-scale data includes geological data, geochemical data, and remote sensing data. Ore-bearing stratigraphic features, ore-controlling fault features, and alteration mineral features are extracted from the multi-source spatial database to generate a geological feature layer. Geochemical element anomaly features are extracted to generate a geochemical element anomaly layer. Remote sensing alteration mineral information is extracted to generate a remote sensing alteration mineral layer. The geological feature layer, geochemical element anomaly layer, and remote sensing alteration mineral layer are spatially registered and rasterized to construct a multivariate feature dataset. The multivariate feature dataset is input into a pre-constructed metallogenic prediction model to generate a metallogenic favorability prediction map of the study area. The metallogenic favorability prediction map is spatially overlaid with a comprehensive favorable block, and the geological rationality of the overlaid area is verified to delineate prospecting target areas. Therefore, the mineralization prediction method based on multimodal data fusion and human-machine collaboration provided in this disclosure solves the technical problems of poor prediction accuracy caused by reliance on a single data source or strong subjectivity of expert experience in mineralization prediction. It realizes intelligent fusion of multimodal data and human-machine collaborative decision-making, and significantly improves the objectivity, accuracy and engineering applicability of mineralization prediction. Attached Figure Description

[0010] The above and other objects, features, and advantages of this disclosure will become more apparent from the more detailed description of the embodiments thereof in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this disclosure and form part of the specification. They are used together with the embodiments of this disclosure to explain the disclosure and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0011] Figure 1 This is a flowchart illustrating the principle of the mineralization prediction method provided in the embodiments of this disclosure. Figure 2 This is a schematic diagram of the mineralization prediction method provided in the embodiments of this disclosure; Figure 3 A geochemical engineering progress diagram provided in the embodiments of this disclosure; Figure 4 Remote sensing operational status map provided for embodiments of this disclosure; Figure 5 A schematic diagram illustrating the control effect of regional faults and their secondary faults on the production of lead-zinc deposits, provided in an embodiment of this disclosure. Figure 6 This is a diagram showing the alteration unit and its property structure, and its spatial location relative to the ore deposit point, provided in an embodiment of this disclosure. Figure 7 The 1:250,000 regional scale single-element geochemical map of the study area for the present disclosure is provided in the embodiments of this disclosure. Figure 8 This disclosure provides a 1:50,000 exploration scale geochemical map of the study area for Pb-Zn-Ag elements in an embodiment of the present disclosure. Figure 9 This is a flowchart of the remote sensing extraction technology for mineralization alteration provided in the embodiments of this disclosure; Figure 10 This is a mineralization and alteration distribution map of lead-zinc minerals provided in an embodiment of this disclosure; Figure 11 This is a three-level zoning and high-value point distribution map of geochemical Pb-Zn-Cd-Ag anomalies provided in an embodiment of this disclosure; Figure 12 This is a geochemical mineralization prediction map of carbonate-type lead-zinc mineralization provided in the embodiments of this disclosure; Figure 13 A schematic diagram of a comprehensive mineralization prospect of carbonate-type lead-zinc ore provided in this embodiment of the present disclosure; Figure 14 This is a schematic diagram of Pb-Zn-Ag anomaly characteristics and comprehensive anomalies provided in the embodiments of this disclosure; Figure 15 This is a raster map of intelligent predictive geological-geochemical-remote sensing-sample data for carbonate-type lead-zinc deposits provided in this embodiment of the disclosure; Figure 16 Random forest result diagram for carbonate rock type lead-zinc deposits provided in this embodiment of the disclosure; Figure 17 Support vector machine results for carbonate rock-type lead-zinc deposits provided in this embodiment of the disclosure; Figure 18 A comparison chart of intelligent calculation and expert judgment for carbonate-type lead-zinc ore provided in this embodiment of the disclosure; Figure 19A schematic block diagram of the functional modules of the mineralization prediction device based on multimodal data fusion and human-machine collaboration provided in the embodiments of this disclosure; Figure 20 A structural block diagram of an electronic device provided in an embodiment of this disclosure; Figure 21 A schematic diagram of a computer program product provided in an embodiment of this disclosure. Detailed Implementation

[0012] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0013] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0014] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below. It should be noted that the concepts of "first", "second", etc., used in this disclosure are only used to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0015] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0016] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0017] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0018] Mineralization prediction is a crucial step in mineral resource exploration, and its accuracy directly impacts prospecting efficiency and costs. With the deepening of geological exploration, a wealth of multi-source data, including geological, geochemical, and remote sensing data, has been accumulated. How to utilize this data to improve the accuracy of mineralization prediction has been a long-standing technical concern in this field.

[0019] In related technologies, mineralization prediction methods mainly rely on the experience and judgment of geological experts. Qualitative analysis of regional geological background, ore-forming models, and ore-controlling elements is conducted, combined with limited exploration data to delineate prospecting target areas. However, because geological, geochemical, and remote sensing data are scattered across maps and reports in different formats and standards, there is a lack of effective fusion methods. A large amount of implicit information remains untapped. Furthermore, different experts may reach significantly different conclusions based on the same data, resulting in highly subjective and poorly repeatable predictions, greatly reducing the accuracy of mineralization predictions.

[0020] Therefore, to address the aforementioned issues, this exemplary embodiment provides a mineralization prediction method based on multimodal data fusion and human-machine collaboration. It employs a large language model to perform intelligent text mining on unstructured geological exploration reports, extracting key geological information such as ore deposit locations, strata, and structures to construct a structured dataset. Subsequently, it performs spatial registration, rasterization, and feature fusion on multi-source data from geology, geochemistry, and hyperspectral remote sensing to extract favorable mineralization factors such as ore-bearing strata, regional fault buffer zones, geochemical element anomalies, and alteration mineral distribution. Based on this, it constructs positive and negative sample sets based on known ore deposit locations, and uses random forest and support vector machine algorithms for model training to generate a raster prediction map of mineralization favorableness. Finally, it compares and verifies the intelligent prediction results with expert knowledge and performs spatial overlay analysis to delineate favorable mineral exploration blocks under dual constraints, thereby realizing a fully intelligent mineralization process from data acquisition and feature extraction to target area prediction.

[0021] For example, Figure 1 This is a flowchart illustrating the principle of the mineralization prediction method provided in the embodiments of this disclosure, as follows: Figure 1 As shown, it can specifically include: Step S110: Collect multi-scale data of the study area and construct a multi-source spatial database containing regional and exploration scales based on the multi-scale data. The multi-scale data includes geological data, geochemical data, and remote sensing data.

[0022] In this embodiment, multi-scale data can include geological data, geochemical data, and remote sensing data generated at different exploration stages. Regional-scale data primarily refers to 1:250,000 scale geological survey data, used to construct regional stratigraphic frameworks, identify regional deep faults and the distribution of major magmatic rocks, serving as a basis for metallogenic background analysis. Exploration-scale data primarily refers to 1:50,000 scale geological, geochemical, and mineral data, used to finely characterize stratigraphic contact relationships, secondary fault systems, mineralization alteration zones, and geochemical anomaly concentration centers, serving as a basis for target area delineation. Based on the collected multi-scale data, a multi-source spatial database containing regional and exploration-scale data is constructed, providing a data foundation for subsequent feature extraction.

[0023] Step S120: Extract ore-bearing stratigraphic features, ore-controlling fault features, and alteration mineral features from the multi-source spatial database to generate a geological feature layer; extract geochemical element anomaly features to generate a geochemical element anomaly layer; extract remote sensing alteration mineral information to generate a remote sensing alteration mineral layer.

[0024] In this embodiment, known ore deposit points are spatially overlaid with stratigraphic units to analyze the distribution density of ore deposit points in each stratigraphic unit. Stratigraphic units with ore deposit point densities exceeding a preset threshold are identified as favorable ore-bearing strata. Favorable ore-bearing lithologies are determined based on the lithological specificity of the ore deposits. Spatial distribution areas that simultaneously satisfy favorable ore-bearing strata and lithologies are extracted to generate an ore-bearing stratigraphic feature layer. Next, regional fault structures are extracted from the structural dataset. Spatial distance analysis is performed between known ore deposit points and regional faults to analyze the distance distribution characteristics between ore deposit points and faults. Based on the statistical results, the radius of the fault influence zone is determined, and a buffer zone is established with the regional fault as the center line to generate a fault structure feature layer. Finally, alteration types with mineralization indication significance are selected from the geological dataset. The selected alteration spatial point data are converted into raster format to generate an alteration distribution layer. The aforementioned ore-bearing stratigraphic feature layer, fault structure feature layer, and alteration distribution layer together constitute the geological feature layer.

[0025] Simultaneously, single-element anomaly information of characteristic indicator elements was extracted at both the regional and exploration scales, and valid mineral-induced anomalies were determined based on anomaly discrimination criteria. Anomaly distribution areas meeting the anomaly discrimination criteria were converted into raster format to generate geochemical element anomaly layers, including single-element content distribution layers and element combination anomaly layers.

[0026] Simultaneously, the preprocessed hyperspectral image is spectrally unmixed to extract spectral anomalies of altered minerals related to the target mineral type. The spectral anomalies of altered minerals are then interpreted using remote sensing geology to identify real mineralization alteration information, eliminate non-mineralization false anomalies, and convert the verified altered mineral distribution information into a raster format to generate a remote sensing altered mineral layer.

[0027] Step S130: Spatial registration and rasterization are performed on the geological feature layer, geochemical element anomaly layer, and remote sensing alteration mineral layer to construct a multivariate feature dataset.

[0028] The generated geological feature layer, geochemical element anomaly layer, and remote sensing alteration mineral layer were spatially registered to ensure that all data layers have the same projected coordinate system and spatial extent. Through rasterization and feature layer synthesis techniques, data from different sources and scales were unified to the same raster resolution, constructing a multivariate feature dataset to provide standardized input data for subsequent machine learning modeling.

[0029] Step S140: Input the multivariate feature dataset into the pre-constructed mineralization prediction model to generate a mineralization favorability prediction map of the study area.

[0030] The constructed multivariate feature dataset is input into a pre-built mineralization prediction model to generate a mineralization favorability prediction map for the study area. The pre-built mineralization prediction model is constructed as follows: a training dataset is constructed using known ore deposits as positive samples and areas with no mineralization geological conditions as negative samples; machine learning algorithms are used to train the multivariate feature dataset, and the prediction results of various algorithms are compared and analyzed to select the best-performing model as the pre-built mineralization prediction model.

[0031] For example, machine learning algorithms include at least one of random forest and support vector machine. Random forest integrates multiple decision trees to analyze the intrinsic relationship between ore-controlling factors and mineralization, assesses the importance of each ore-controlling index, and identifies mineralization anomalies through a collective voting mechanism; support vector machine constructs an optimal hyperplane to perform maximum-margin classification in the feature space of ore-controlling data, and uses kernel function technology to map nonlinear relationships to identify the critical boundaries of mineralization anomalies from high-dimensional data.

[0032] Step S150: Perform spatial overlay analysis between the mineralization favorability prediction map and the comprehensive favorable blocks, and delineate the prospecting target area after verifying the geological rationality of the overlay area.

[0033] The generated mineralization favorability prediction map is spatially overlaid with the comprehensive favorable blocks. After the geological rationality of the overlaid area is verified, the prospecting target area is delineated.

[0034] In spatial overlay analysis, the comprehensive favorable block and the mineralization favorable prediction map are spatially overlaid, and the areas where the two overlap are determined as the initial mineral exploration target areas. For areas where the two differ, geological rationality verification is performed, and the initial mineral exploration target areas are revised based on the verification results, ultimately determining the final mineral exploration target areas.

[0035] Based on this, a multi-scale, multi-source spatial database is constructed, and three types of feature layers—geological, geochemical, and remote sensing—are extracted and fused into a multivariate feature dataset. This dataset is then combined with a machine learning model to generate a mineralization favorability prediction map. Finally, spatial overlay analysis and geological plausibility verification are performed with the comprehensive favorable blocks to accurately delineate mineral exploration target areas. This method solves the technical problems in mineralization prediction, such as poor accuracy due to reliance on a single data source or strong subjectivity of expert experience. It achieves intelligent fusion of multimodal data and human-machine collaborative decision-making, significantly improving the objectivity, accuracy, and engineering applicability of mineralization prediction.

[0036] For example, Figure 2 This is a schematic diagram of the mineralization prediction method provided in the embodiments of this disclosure, such as... Figure 2 As shown, it can specifically include: The system collects basic geological, geophysical, geochemical, and remote sensing imagery, as well as historical mineral exploration data within the work area, covering multiple modal formats such as maps, text reports, tabular data, and vector data.

[0037] During data acquisition, multi-scale data generated at different exploration stages are differentiated. This multi-scale data includes regional-scale data, primarily referring to 1:250,000 scale geological survey data (hereinafter referred to as "250,000 data"), used to construct regional stratigraphic frameworks, identify regional deep and large faults and the distribution of major magmatic rocks, serving as a basis for metallogenic background analysis. Multi-scale data also includes exploration-scale data, primarily referring to 1:50,000 scale geological, geochemical, and mineral data (hereinafter referred to as "50,000 data"), used to finely characterize stratigraphic contact relationships, secondary fault systems, mineralization alteration zones, and geochemical anomaly concentration centers, serving as a basis for target area delineation.

[0038] Based on the collected multi-scale data, the regional metallogenic geological background is analyzed, the ore-bearing strata, ore-controlling structures and magmatic rock conditions are determined, and the geological theory is transformed into quantifiable characteristic variables. Specifically, this may include: extracting the spatial distribution of specific ore-bearing strata, identifying regional deep and large faults, constructing a 3-kilometer buffer zone of deep and large faults as ore-controlling structural elements based on the spatial correlation statistics between known ore deposits and faults, extracting favorable structural facies units such as carbonate platform facies, basin facies and rift facies, and forming a set of geological ore-controlling element layers.

[0039] Furthermore, geological ore-controlling elements are spatially registered and integrated with preprocessed geochemical anomaly data and remote sensing alteration information. By rasterizing and synthesizing feature layers from data of different sources and scales, a feature dataset with unified spatial resolution is constructed, which serves as input data for subsequent machine learning modeling.

[0040] Known mineral deposits were used as positive samples, and negative samples were constructed based on the following four criteria: known non-lead-zinc mineral deposits, areas where geological conditions indicate no mineralization, areas where remote sensing mineral information shows no mineralization anomalies, and areas where key geochemical elements have low values.

[0041] Based on this, random forest and support vector machine classification models are constructed using the generated feature dataset. Random forest integrates multiple decision trees to evaluate the importance of various ore-controlling indicators and identifies mineralization anomalies through a collective voting mechanism. Support vector machine constructs an optimal hyperplane and uses kernel function techniques to perform maximum margin classification in a high-dimensional feature space. Through model training and prediction, a raster prediction map of ore-forming favorability covering the entire area is generated.

[0042] The generated intelligent prediction results are spatially superimposed and compared with the prospective mineralized areas based on expert knowledge and manual judgment. The geological rationality of the prediction boundaries is then corrected, and the specific mineralized blocks are finally delineated.

[0043] Based on the above embodiments, in another embodiment provided in this disclosure, the multi-scale data of the study area collected may include: Natural language processing technology is used to extract geological information from geological exploration reports to obtain mineral deposit data; Collect geological data at different scales, and vectorize and structurally define the geological elements of strata, faults, alteration and ore bodies in the geological data to obtain a geological dataset; Geochemical maps are vectorized to extract single-element and combined anomaly information, thus obtaining a geochemical dataset. The hyperspectral remote sensing images were preprocessed with radiometric calibration, atmospheric correction, and geometric correction. Feature alteration minerals were extracted using a spectral library to generate a mineralization alteration distribution layer, thus obtaining a remote sensing dataset.

[0044] In this embodiment, a large language model can be used to perform intelligent text mining on unstructured geological exploration reports to obtain ore deposit data. Specifically, by injecting professional geological corpora, the model is fine-tuned and its retrieval enhanced, enabling the large language model to master professional terminology and contextual logic in fields such as lithostratigraphy, structural geology, and mineral exploration. Structured prompts are designed to clearly define the key information types that the model needs to extract, including: ore deposit name, ore type (including associated minerals), host rock, ore-controlling structures and occurrence, host rock alteration characteristics, ore body morphology and scale, ore mineral composition and assemblage, gangue minerals, and ore structure type.

[0045] An automated processing workflow is built based on Python parsing scripts to capture, clean, validate, and convert the structured data returned by large language models. Finally, various geological elements are seamlessly integrated into standardized tables, and the extracted results are output in standard JSON format, realizing end-to-end intelligent output from unstructured text to structured data tables.

[0046] For mineral deposits not covered by intelligent text mining or requiring verification, supplementary data can be collected manually. Based on the mineral deposit attribute table exported from ArcGIS software, and combined with historical exploration reports of the target area, core geological indicators such as ore-bearing host rocks, ore body size, ore body morphology, occurrence, ore mineral composition, ore structure, and host rock alteration characteristics are extracted. For mineral deposits without names on geological maps, spatial registration and location overlay of multiple geological maps are performed to establish the correlation between the spatial location of the mineral deposits and the geological maps. After supplementing the identification information, the geological information extraction is completed, forming a standardized mineral deposit information dataset.

[0047] Geological data at a scale of 1:250,000 were collected for the study area to construct a regional geological framework. This geological data may include stratigraphic data, structural data, and igneous rock data, specifically including: By using regional geological maps and exploration reports, key information such as stratigraphic age, lithological assemblage, and plate zoning can be extracted to construct a stratigraphic dataset with fine chronological division and clear spatial zoning, providing basic data for metallogenic epoch constraints and mineralized strata location.

[0048] Furthermore, based on the structural elements in geological maps and reports, structural information such as faults, folds, and geological boundaries can be collected to establish a structural database framed by regional deep and large faults. Through standardized processing of structural descriptions from different data sources, the standardized integration of structural types and occurrence data is achieved. After verifying the consistency between vector data and original maps, based on the constructed spatial topological relationships, the spatial correlation characteristics between faults and mineral occurrences are statistically analyzed, and the spatial topological attributes of structural elements such as fault occurrence, mineral occurrence distribution, and dike extension are extracted as inputs for subsequent analysis of favorable mineralization factors.

[0049] Furthermore, based on the collected geological maps and research reports related to igneous rocks in the region, we can extract attribute information such as rock mass morphology, occurrence, lithology, intrusion age, and contact relationship with surrounding rocks. Combined with the regional magmatic evolution sequence, we can establish a classification system for igneous rocks that covers rock mass phases, rock types, and genetic types.

[0050] While collecting geological data at a scale of 1:250,000, geological elements of five types—strata, faults, profiles, alteration, and ore bodies—can be extracted based on data such as mining area exploration reports, geochemical maps, and preliminary survey reports. These elements can then be vectorized and their attributes defined to obtain geological data of the working area at a scale of 1:50,000.

[0051] The stratigraphic elements clearly define the stratigraphic age, lithological assemblage, and spatial distribution characteristics. Fault elements define attribute fields such as fault properties, dip direction, and dip angle, accurately reflecting the tectonic movement mode and the spatial distribution of fault planes. Profile elements set profile attribute and profile number fields; profile attributes record the contact relationship between strata and structures, and profile numbers correspond to the original measured profile numbers, achieving effective correlation between vector data and field logging. Based on alteration elements, an alteration type field is established, uniformly classified into typical categories such as silicification, pyritization, and carbonatization, providing a standardized basis for the study of the relationship between alteration and mineralization. Orebody elements are used to configure orebody attribute and orebody number fields; the orebody attribute field records the combination information of the contained minerals, and the orebody number field corresponds to the original orebody coding system of the mining area, ensuring data traceability and integrated application.

[0052] By standardizing the representation of geological elements in different maps and reports, a geological dataset with consistent data structure and semantic meaning at the exploration scale is constructed.

[0053] In the embodiments, raw data and result maps can be obtained through various geochemical exploration methods such as rock cuttings measurement, stream sediment measurement and soil measurement, and geochemical exploration data of the work area can be collected based on the raw data and result maps.

[0054] First, data screening and vectorization were performed. Existing geological reports, MapGIS data, and images were compared and screened to identify and extract single-element anomaly maps and comprehensive anomaly maps of target elements, including characteristic indicator elements of carbonate-type lead-zinc deposits such as silver (Ag), lead (Pb), zinc (Zn), gold (Au), and cadmium (Cd). The selected map data were then vectorized into a spatially analyzable data format.

[0055] During the vectorization process, key information such as abnormal attributes, abnormal contour lines, abnormal high-value points, and abnormal value ranges are recorded.

[0056] In response to the actual situation of comparing anomalies of multiple elements in the same area, the abnormal data of each single element are separated, created and stored independently, and a differentiated database structure is constructed to facilitate subsequent element combination analysis, matching relationship evaluation and anomaly screening.

[0057] Finally, geochemical data at different scales, including geochemical exploration results at scales of 1:10000, 1:25000, 1:50000, and 1:200000, were integrated to form a multi-scale geochemical dataset covering the entire region, providing data support for regional mineralization prediction and target area delineation.

[0058] For example, Figure 3 A geochemical engineering progress diagram provided for embodiments of this disclosure. (See diagram below.) Figure 3 As shown, geochemical exploration work in the study area covers multiple scales, including 1:10000, 1:25000, 1:50000, 1:200000, and comprehensive anomaly areas. The scope of geochemical exploration work at different scales is distinguished by different map symbols.

[0059] Spatially, 1:200,000 scale geochemical surveys have the widest coverage, distributed in the central-western and eastern parts of the study area, forming the regional geochemical background framework. 1:50,000 scale geochemical surveys are mainly distributed in the central-western, western, and eastern parts of the study area, forming several key exploration blocks. 1:25,000 scale geochemical surveys are scattered throughout the central part of the study area. 1:10,000 scale geochemical surveys are concentrated around known ore deposits in the central and eastern parts of the study area, forming high-precision anomaly verification areas. Comprehensive anomaly areas are mainly distributed in the central and southeastern parts of the study area, covering a relatively wide area.

[0060] In this embodiment, hyperspectral remote sensing image data covering the working area was acquired using the ZY1F satellite (ZY1F) for the extraction and analysis of mineralization and alteration mineral information.

[0061] The hyperspectral imager aboard the ZY1F O2F satellite possesses dual-band imaging capabilities in visible-near-infrared (VNIR) and short-wave infrared (SWIR). The VNIR band ranges from 400 to 1030 nm, with 78 bands and a spectral resolution better than 10 nm. The SWIR band ranges from 1000 to 2500 nm, with 90 bands and a spectral resolution of 20 nm. The overall spatial resolution is uniformly 30 m, and the imaging swath is 60 km. The acquired data exhibits excellent performance in terms of signal-to-noise ratio and radiometric calibration accuracy, making it suitable for the extraction and identification of weak spectral features.

[0062] Furthermore, the acquired hyperspectral image data is sequentially subjected to radiometric calibration, atmospheric correction, and geometric correction to eliminate sensor errors, atmospheric scattering, and absorption effects, restoring the true surface reflectance information. Unsupervised classification-assisted masking technology is then employed to identify and remove interfering information such as snow and water bodies from the images, ensuring the accuracy of subsequent mineral extraction.

[0063] Finally, by combining the USGS standard spectral library and using spectral unmixing techniques such as band ratio calculation and principal component analysis, information on alteration minerals related to lead-zinc mineralization was extracted, including characteristic alteration minerals such as limonite, calcite, siderite, alum, and gypsum, to generate a mineralization alteration distribution layer.

[0064] For example, Figure 4 A remote sensing operational status map provided for embodiments of this disclosure. For example... Figure 4 As shown, the coverage of hyperspectral remote sensing images in the study area is presented in the form of grid-like maps. The coverage of different scenes is distinguished by regular grid boundaries. The image coverage is complete, the grid distribution is uniform, and the overlapping areas are sufficient, providing a full-coverage and high-quality remote sensing data foundation for subsequent mineralization and alteration mineral extraction and mineralization prediction analysis.

[0065] Based on the above embodiments, in another embodiment provided in this disclosure, the above-mentioned extraction of ore-bearing stratigraphic features, ore-controlling fault features, and alteration mineral features from a multi-source spatial database to generate a geological feature layer may include: Based on ore deposit data, the distribution density of ore deposit points in each stratigraphic unit is statistically analyzed. Stratigraphic units with ore deposit point density higher than a preset threshold are identified as favorable ore-bearing strata. Based on the lithological specificity characteristics of the ore deposit, favorable ore-bearing lithology types are determined. Spatial distribution areas that simultaneously satisfy favorable ore-bearing strata and favorable ore-bearing lithology are extracted to generate an ore-bearing stratigraphic feature layer. Regional fault structures are extracted from geological datasets. Spatial distance analysis is performed between ore deposit data and regional faults. The distance distribution characteristics between ore deposits and faults are statistically analyzed. Based on the statistical results, the radius of the fault influence zone is determined. A buffer zone is established with the regional fault as the center line to generate a fault structure feature layer. The geological dataset is used to select alteration types that indicate mineralization. The selected alteration spatial point data is then converted into raster format to generate a geological alteration distribution layer.

[0066] In the embodiments, the geological feature layer may include: an ore-bearing strata feature layer, a fault structure feature layer, and a geological alteration distribution layer.

[0067] For example, in order to meet the needs of different exploration stages, stratigraphic favorable elements are extracted at the regional scale (1:250000) and the exploration scale (1:50000) respectively, generating multi-scale mineralized stratigraphic feature layers.

[0068] For regional-scale feature extraction (1:250000), based on the 1:250000 scale stratigraphic dataset, the spatial distribution areas of favorable ore-bearing stratigraphic units with carbonate lithology are extracted, generating a regional-scale ore-bearing stratigraphic feature layer. This regional-scale ore-bearing stratigraphic feature layer is used for regional metallogenic prospect analysis.

[0069] For exploration-scale feature extraction (1:50000), following the same extraction rules, based on the 1:50000 scale stratigraphic dataset, favorable ore-bearing stratigraphic alignment and favorable lithology screening were performed again to extract the fine spatial distribution areas of favorable ore-bearing strata at the exploration scale, generating an exploration-scale ore-bearing stratigraphic feature layer. This layer has higher resolution and more detailed boundary delineation, and is used for the precise delineation of mineral exploration target areas.

[0070] Finally, the ore-bearing stratigraphic feature layers generated at the regional and exploration scales are spatially registered to ensure they have the same projected coordinate system and spatial extent. In subsequent multi-source data fusion steps, feature layers of appropriate scales can be selected and input into the machine learning model according to the needs of the prediction task.

[0071] For the extraction of favorable stratigraphic elements from 1:250,000 scale geological data, this embodiment takes carbonate-type lead-zinc deposits as an example. Based on the constructed stratigraphic dataset and combined with the analysis of regional metallogenic regularities, favorable ore-bearing strata of carbonate-type lead-zinc deposits are extracted, and an ore-bearing stratigraphic feature layer is generated.

[0072] First, spatial overlay analysis is performed on the known mineral deposit point vector data and stratigraphic unit layers to count the number of mineral deposit points in each stratigraphic unit, and stratigraphic units containing known mineral deposit points are identified as favorable ore-bearing strata.

[0073] Since carbonate-type lead-zinc deposits are mainly hosted in favorable lithological units such as thick-layered limestone-dolomite assemblages of platform facies and alternating sequences of impure carbonate rocks and clastic rocks, carbonate rocks are identified as favorable ore-bearing lithologies based on this specific characteristic of carbonate-type lead-zinc deposits.

[0074] From the constructed stratigraphic dataset, spatial distribution areas of favorable ore-bearing stratigraphic units with carbonate lithology are extracted to generate an ore-bearing stratigraphic distribution vector layer. This vector layer is then converted to raster format, with raster cell values ​​set to 1 to represent favorable ore-bearing stratigraphic distribution areas and 0 to represent unfavorable areas, forming an ore-bearing stratigraphic feature layer, which serves as one of the features input to subsequent machine learning models.

[0075] In this embodiment, based on the constructed tectonic dataset and combined with the analysis of regional mineralization patterns, favorable tectonic elements related to mineralization are extracted from 1:250,000 scale geological data to generate tectonic feature layers.

[0076] First, regional deep and large faults are extracted from the structural dataset as the main structural features controlling the migration of ore-forming fluids. These regional deep and large faults refer to fault structures with an extension length exceeding a preset threshold and a significant cutting depth; in this embodiment, the preset threshold is set to 50 kilometers.

[0077] Spatial distance analysis was performed on the known ore deposit point vector data and the regional fault layer to statistically analyze the distance distribution characteristics between the ore deposit points and the faults. The statistics showed that most ore deposit points in the study area were distributed within a preset distance range from both sides of the regional fault, thus determining the radius of the fault influence zone. In this embodiment, the preset distance range on both sides was set to 3 kilometers.

[0078] Based on the determined radius of the influence domain, a buffer zone is established with the regional fault as the center line, generating a fault influence domain vector layer. Areas falling within the buffer zone in the fault influence domain vector layer are marked as fault influence zones.

[0079] In an alternative approach, secondary fractures with an extension length less than a preset threshold can also be extracted from the constructed dataset to generate a secondary fracture distribution layer. These secondary fractures can serve as indicators of local fluid channels and mineral deposition sites. In this embodiment, the preset threshold is set to 50 kilometers.

[0080] Finally, the generated fracture buffer layer is converted to a raster format, with a raster cell value of 1 indicating it is within the fracture influence region and 0 indicating it is outside the influence region, forming a fracture structural feature layer. If secondary fractures need to be included, the secondary fracture layer can also be converted to a raster format, or it can be overlaid and blended with the buffer layer.

[0081] Furthermore, in the extraction of favorable structural features, tectonic facies information can be incorporated in addition to fault structures. Based on regional metallogenic geology, carbonate platform facies, basin facies, and rift facies are identified as favorable tectonic facies types. The spatial distribution areas of these favorable tectonic facies are extracted from the tectonic facies dataset to generate a tectonic facies feature layer, which is then used as auxiliary features input into the machine learning model.

[0082] For the extraction of favorable stratigraphic features from 1:50,000 scale geological data, a higher resolution mineralized stratigraphic feature layer can be generated based on the constructed 1:50,000 scale geological dataset, following the same extraction rules as at the regional scale, for the fine delineation of mineral exploration target areas.

[0083] Specifically, the known mineral deposit point vector data at a scale of 1:50000 are spatially overlaid with stratigraphic unit layers of the same scale to count the number of mineral deposit points in each stratigraphic unit, and stratigraphic units containing known mineral deposit points are identified as favorable ore-bearing strata.

[0084] Furthermore, based on the lithological specificity characteristics of carbonate-type lead-zinc deposits, carbonate rocks were identified as favorable ore-bearing lithologies.

[0085] Finally, from the 1:50,000 scale stratigraphic dataset, spatial distribution areas of favorable ore-bearing stratigraphic units with carbonate lithology were extracted to generate a vector map layer of ore-bearing stratigraphic distribution. This vector map layer was converted to raster format, with the raster resolution matching the 1:50,000 geochemical data. Raster cell values ​​were set to 1 to represent favorable ore-bearing stratigraphic distribution areas and 0 to represent unfavorable areas, forming a 50,000 scale ore-bearing stratigraphic feature layer.

[0086] For the extraction of structural attributes and favorable mineralization elements from 1:50,000 scale geological data, fault structural elements can be extracted based on the constructed 1:50,000 scale structural dataset to generate structural feature layers.

[0087] Specifically, regional fault structures are first extracted from the constructed dataset. In this embodiment, the identified regional faults are mainly thrust faults and concealed faults.

[0088] Furthermore, based on the mineralization pattern of carbonate-type lead-zinc deposits controlled by fault structures, buffer zones are established along both sides of the regional faults to generate a fault influence domain layer. In this embodiment, the buffer zone width can be set to 500 meters.

[0089] Finally, the fracture buffer layer is converted to a raster format with a raster resolution matching the 1:50000 geochemical data. The raster cell value is set to 1 to indicate that it is within the fracture influence zone and 0 to indicate that it is outside the influence zone, thus forming a 50,000-scale fracture structural feature layer.

[0090] For example, Figure 5 This diagram illustrates the control effect of regional faults and their secondary faults on the production of lead-zinc deposits, as provided in the embodiments of this disclosure.

[0091] For the extraction of alteration attribute structure and mineralization favorable elements from 1:50,000 scale geological data, mineralization-related alteration elements can be extracted based on the constructed alteration dataset to generate a geological alteration distribution layer.

[0092] Specifically, representative alteration types are selected from the collected alteration types based on their mineralization indication significance. In this embodiment, the selected alteration types may include: calcite alteration, epidote alteration, chloritization, sericitization, limonite alteration, and pyrite alteration. Among these, calcite alteration, epidote alteration, chloritization, and sericitization represent a medium-to-low temperature hydrothermal alteration combination, while limonite alteration and pyrite alteration are direct indicators of sulfide mineralization.

[0093] Among them, carbonate-type lead-zinc deposits form alteration mineral assemblages with genetic indicative significance during their mineralization evolution, fully recording the complete evolutionary sequence from the hydrothermal mineralization stage to the supergene oxidation stage. Gypsum and alum, as oxidation products of sulfides, respectively mark the oxidizing environment and the decomposition process of galena; smithsonite, a typical product of sphalerite undergoing carbonatization, directly indicates the existence of deep primary zinc ore bodies; limonite, as a major component of the widely developed iron cap, is formed by the oxidation of iron sulfides such as pyrite, constituting the most intuitive surface prospecting indicator; the abundant occurrence of calcite represents the carbonatization process in hydrothermal or supergene environments. In this alteration mineral assemblage, limonite, smithsonite, and alum together constitute a direct indicator assemblage for finding primary ore bodies, which has important practical significance for determining mineralization types, evaluating the degree of ore erosion, and guiding exploration engineering deployment.

[0094] Then, the selected alteration spatial point data are converted into raster format to generate density distribution layers or existence indicator layers of various alterations.

[0095] Finally, multiple alteration layers are overlaid and merged to generate a geological alteration distribution layer, reflecting the development degree of alteration assemblages.

[0096] For example, Figure 6 The alteration unit and property structure diagram and its spatial location relative to the ore deposit point are provided in the embodiments of this disclosure.

[0097] Based on the above embodiments, in another embodiment provided in this disclosure, the extraction of geochemical element anomaly features and the generation of a geochemical element anomaly layer may include: Single-element anomaly information of feature indicator elements is extracted at both the regional and exploration scales. Effective mineral-induced anomalies are determined based on anomaly discrimination criteria. The anomaly discrimination criteria include: having obvious element concentration centers, good spatial overlap between the main ore-forming elements and associated elements, and exhibiting complete horizontal zoning characteristics. Anomaly distribution areas that meet the anomaly identification criteria are converted into raster format to generate geochemical element anomaly layers.

[0098] In this embodiment, taking the geochemical exploration of carbonate-type lead-zinc deposits as an example, an anomaly identification model with significant deposit specificity was established by integrating geochemical data at a regional scale of 1:250,000 and an exploration scale of 1:50,000. At the regional scale, the model focuses on revealing the spatial coupling characteristics between the regional distribution patterns of elements and the metallogenic geological background. Through systematic analysis of the characteristic elemental combinations of Pb-Zn-Ag-Cd carbonate-type lead-zinc deposits, regional metallogenic anomaly identification criteria were established. At the exploration scale, the model focuses on the precise location of the internal structure of the anomaly and the mineralization center, achieving fine delineation of the prospecting block.

[0099] Specifically, for the extraction of geochemical features at a regional scale of 1:250,000, characteristic indicator element combinations of carbonate-type lead-zinc deposits can be extracted based on 1:250,000 scale geochemical data. In this embodiment, Pb, Zn, Ag, and Cd are selected as core indicator elements, with Pb and Zn being the main ore-forming elements and Ag and Cd being characteristic associated elements.

[0100] For each indicator element, a single-element anomaly map is constructed, with the lower limit of the anomaly set based on the statistical distribution characteristics of the element's content. An effective mineralized anomaly should possess the following characteristics: a significant elemental concentration center, a spatial overlap between the main ore-forming element and associated elements, and a complete horizontal zonation sequence (Zn-Pb-Ag from the inside out). When these anomaly characteristics coincide with a favorable geological tectonic setting, prospective mineralization zones can be reliably delineated.

[0101] Single-element anomaly maps that meet the anomaly identification criteria are converted into raster format to generate a geochemical element anomaly layer. The geochemical element anomaly layer includes layers showing the content distribution of each single element and layers showing element combination anomalies. The element combination anomaly layer can be obtained by combining multiple single-element anomaly layers through band synthesis.

[0102] For example, Figure 7 A 1:250,000 regional scale single-element geochemical map of the study area for the present disclosure. (See attached map.) Figure 7 As shown, the geochemical fields of Pb, Zn, Ag, and Cd exhibit a significant spatial differentiation characteristic of high concentrations in the south and low concentrations in the north. Their concentration centers show a clear coupling relationship with the spatial distribution of known lead-zinc deposits: in areas with known deposits (points), Pb and Zn elements show high-intensity anomalies (deep red color bands); while areas without mineralization show low background characteristics (predominantly blue color bands). The spatial overlap of the elemental anomalies is good, clearly indicating potential mineralization centers.

[0103] In this embodiment, geochemical features can be extracted based on the acquired 1:50,000 scale geochemical data to construct a higher resolution geochemical feature layer. In this embodiment, stream sediment measurement data is used, and Pb, Zn, and Ag are selected as the core indicator element combination for carbonate-type lead-zinc deposits. Pb and Zn, as the main ore-forming elements, directly characterize the presence of mineralization, while Ag, as a characteristic associated element, reflects the intensity of ore-forming fluid activity.

[0104] Among them, effective mineral-induced anomalies should have the following typical characteristics: significant anomaly intensity with obvious element concentration centers; moderate spatial range, neither too dispersed nor too limited; good elemental overlap, with the main ore-forming elements and associated element anomalies highly overlapping in space; and complete horizontal zoning, presenting a zoning sequence of Zn-Pb-Ag from the outside to the inside.

[0105] Single-element anomaly maps that meet the anomaly identification criteria are converted into raster format to generate a geochemical element anomaly layer. The geochemical element anomaly layer includes layers showing the content distribution of each single element and layers showing element combination anomalies. The element combination anomaly layer can be obtained by combining multiple single-element anomaly layers through band synthesis.

[0106] When the aforementioned mineral anomalies are spatially coupled with favorable ore-bearing strata and regional fault structures, the area is identified as a favorable block for mineral exploration.

[0107] For example, Figure 8 A 1:50000 exploration scale geochemical map of the study area for Pb-Zn-Ag elements provided in this embodiment of the disclosure. Figure 8 As shown, in areas with known ore deposits (points), high-intensity anomaly zones are characterized by deep red to orange-red hues in the chromatogram, visually representing the significant enrichment of ore-forming elements, while areas without mineralization generally exhibit blue to light blue hues. This spatial consistency between the geochemical field structure and the geological mineralization conditions further verifies the effectiveness of the selected elemental assemblages in indicating carbonate-type lead-zinc deposits.

[0108] Based on the above embodiments, in another embodiment provided in this disclosure, the extraction of remote sensing alteration mineral information and the generation of a remote sensing alteration mineral layer may include: Spectral unmixing was performed on the preprocessed hyperspectral images to extract spectral anomalies of altered minerals associated with the target mineral type; Remote sensing geological interpretation of spectral anomalies in altered minerals is used to identify genuine mineralization and alteration information and eliminate non-mineralization false anomalies. The verified alteration mineral distribution information is converted into a raster format to generate a remote sensing alteration mineral layer.

[0109] Based on the acquired 1:50,000 scale hyperspectral remote sensing image data, information on alteration minerals related to mineralization was extracted to generate remote sensing feature layers.

[0110] For example, Figure 9 A flowchart of the remote sensing extraction technology for mineralization alteration provided in this disclosure is shown below. Figure 9 As shown, the original hyperspectral images acquired by the ZY-1 satellite were first subjected to radiometric calibration, atmospheric correction, and geometric correction to eliminate sensor errors, atmospheric scattering, and absorption effects, and to restore the true reflectivity information of the Earth's surface.

[0111] Subsequently, unsupervised classification-assisted masking technology was used to identify and remove interfering information such as snow and water bodies in the image, ensuring the accuracy of subsequent mineral extraction.

[0112] Furthermore, by combining the USGS spectral library and employing spectral unmixing techniques such as band ratio calculation and principal component analysis, spectral anomalies of alteration minerals associated with lead-zinc mineralization are extracted. In this embodiment, the key alteration minerals extracted may include limonite, calcite, siderite, alum, and gypsum.

[0113] Finally, remote sensing geological interpretation was performed on the extracted spectral anomalies to conduct anomaly verification, mineral classification, and correlation analysis of key ore-controlling factors, identify genuine mineralization and alteration information, and eliminate non-mineralization false anomalies.

[0114] Through systematic remote sensing geological interpretation, a distribution map of target mineralization and alteration in the study area was finally generated. Non-mineralization false anomalies were eliminated, and the real mineralization and alteration zones were accurately located, providing precise spatial guidance for subsequent comprehensive mineral exploration prediction and field exploration deployment.

[0115] For example, Figure 10 This is a mineralization and alteration distribution map of lead-zinc minerals provided in an embodiment of this disclosure.

[0116] Based on the above embodiments, in another embodiment provided in this disclosure, the above-mentioned mineralization prediction method based on multimodal data fusion and human-machine collaboration may further include: Based on the regional scale, the geological feature layer, the geochemical element anomaly layer, and the remote sensing alteration mineral layer are spatially overlaid and analyzed. The areas where the three overlap in space are identified as prospective mineralization areas. Based on the exploration scale, in the prospective mineralization area, the geological feature layer, the geochemical element anomaly layer, and the remote sensing alteration mineral layer are spatially overlaid again, and the areas where the three overlap in space are identified as comprehensive favorable blocks.

[0117] In this embodiment, the geologically favorable areas are first delineated. Information on favorable ore-bearing rock bodies, ore-bearing strata, volcanic rock distribution, structural features, and alteration zones is integrated and processed. Based on regional metallogenic regularities, areas that simultaneously meet the following conditions are initially delineated as geologically favorable areas: located within the distribution range of favorable ore-bearing strata, located within the influence zone of regional faults, and have developed alteration mineral assemblages related to mineralization.

[0118] Taking carbonate-type lead-zinc deposits as an example, in order to delineate the geological prospective areas of carbonate-type lead-zinc deposits, data such as stratigraphic groups, lithological combinations, structural characteristics and tectonic backgrounds are integrated. In stratigraphic groups, the distribution range of favorable ore-bearing strata is selected. In lithological combinations, limestone is selected. In terms of structural characteristics, regional major faults are selected. In terms of tectonic backgrounds, basins, carbonate platforms and rift valleys are selected, further narrowing down the prediction area of ​​condensed geological mineralization.

[0119] Subsequently, favorable geochemical zones were delineated. Based on the constructed 1:50,000 geochemical element anomaly layer, the anomaly distribution information of key indicator elements Pb, Zn, and Ag was extracted. Favorable geochemical zones were delineated according to the following anomaly discrimination criteria: significant anomaly intensity and obvious element concentration centers, good spatial alignment between the main ore-forming elements and associated elements, and a complete horizontal zonation characteristic from the outside to the inside of Zn-Pb-Ag.

[0120] Taking carbonate-type lead-zinc deposits as an example, in order to delineate the geochemical prospective areas of carbonate-type lead-zinc deposits, the system integrates the anomalous data of four key indicator elements, namely Pb, Zn, Ag, and Cd, and optimizes the target area range by analyzing the spatial overlap relationship of elements, the distribution of concentration centers, and the zoning characteristics.

[0121] For example, Figure 11 This is a three-level zoning and high-value point distribution map of geochemical Pb-Zn-Cd-Ag anomalies provided in an embodiment of this disclosure. (See attached map.) Figure 11 As shown, a three-level anomaly classification scheme is adopted, with the normal distribution mean plus 1.5-2 times the standard deviation (corresponding to approximately 86%-95% confidence interval) as the lower limit of the third-level anomaly. Specifically, Pb (30×10⁻⁶) is set. -6 40×10 -6 60×10 -6 ), Zn (70×10) -6 90×10 -6 130×10 -6 ), Ag (60×10 -6 100×10 -6 180×10 -6 ), Cd (350×10 -6 500×10 -6 800×10 -6 The banding threshold of ).

[0122] Furthermore, by combining remote sensing images and DEM data, the source areas of anomalies were traced along the geochemical high-value points towards the direction of increasing elevation, and the geochemical prospective areas of carbonate rock-type lead-zinc deposits were scientifically delineated.

[0123] For example, Figure 12 A carbonate-type lead-zinc geochemical mineralization prediction map provided for embodiments of this disclosure.

[0124] Finally, the aforementioned geologically favorable areas, geochemically favorable areas, and remotely sensing favorable areas were spatially overlaid to extract the areas where they overlapped spatially as prospective mineralization areas. Based on this, and considering constraints such as verification of known mineral deposits, rationality checks by geological experts, and engineering feasibility considerations, target areas were optimized, ultimately delineating several prospecting target areas as key areas for subsequent exploration projects.

[0125] For example, Figure 13 A schematic diagram of a comprehensive mineralization prospect of carbonate-type lead-zinc ore provided in this embodiment of the disclosure. (See diagram below.) Figure 13 As shown, by integrating multi-source information such as geological structure, multi-element geochemical combination anomalies, topographic features and distribution of known mineral deposits, and through spatial overlay analysis and comprehensive judgment, four potential areas for comprehensive mineralization in the region were delineated.

[0126] Based on the delineation of comprehensive mineralization prospective areas, this embodiment further conducts detailed prediction of favorable mineralization blocks within the prospective areas to accurately locate prospecting target areas.

[0127] First, geologically favorable blocks are screened. Based on the constructed 1:50,000 scale geological dataset, the spatial distribution areas of favorable ore-bearing strata are extracted. At the same time, regional faults and their corresponding secondary fault buffer zones are extracted as structurally favorable areas. Areas that simultaneously meet the conditions of ore-bearing strata and structural favorable conditions are initially delineated as geologically favorable blocks.

[0128] Secondly, favorable geochemical blocks are delineated. Based on the constructed 1:50,000 geochemical data, anomaly distribution information of key indicator elements Pb, Zn, and Ag is extracted, and anomaly lower limits are set according to the statistical characteristics of different map sheets. In this embodiment, differentiated anomaly thresholds are set for different prospective areas, and anomaly areas with high contrast, good elemental alignment, and clear concentration centers are selected as favorable geochemical blocks, reflecting the potential spatial location of surface element enrichment and mineralization.

[0129] For example, the lower limit of anomalies for key indicator elements Pb, Zn, and Ag in a certain mineralization prospective area can be set as Ag > 100 ppm, Pb > 60 ppm, and Zn > 180 ppm; the lower limit of anomalies in another mineralization prospective area can be set as Ag > 60 ppm, Pb > 30.74 ppm, and Zn > 66 ppm.

[0130] For example, Figure 14 This is a schematic diagram of the abnormal features and comprehensive anomalies of Pb-Zn-Ag provided in the embodiments of this disclosure.

[0131] Next, favorable remote sensing blocks were identified. Based on the constructed hyperspectral remote sensing feature layer, information on alteration mineral assemblages related to lead-zinc mineralization was extracted, including characteristic alteration minerals such as limonite, calcite, siderite, galena, and gypsum. Among them, calcite alteration and siderite alteration are indicators of near-ore hydrothermal alteration, reflecting the reaction between ore-forming fluids and carbonate host rocks; limonite alteration is a surface oxidation product, often appearing as an "iron cap," indicating the presence of shallow mineralization; galena is an oxidation product of galena and is a direct indicator mineral of lead. Areas with dense distribution and well-developed assemblages of the above alteration minerals were identified as favorable remote sensing blocks.

[0132] In addition, topographic information such as mountain shadows and valley morphology can be combined to help identify the distribution of structures and lithological contact relationships, infer the source areas of geochemical anomalies and the location of alteration and mineralization, and provide directional basis for mineralization prediction.

[0133] Finally, the aforementioned geologically favorable blocks, geochemically favorable blocks, and remotely sensing favorable blocks are spatially overlaid to extract areas where multi-source information overlaps spatially as comprehensive favorable blocks. Based on this, topographic and geomorphological features are used to help identify tectonic distribution and lithological contact relationships, inferring geochemical anomaly source areas and alteration / mineralization locations. Simultaneously, limiting factors such as mining rights settings, ecological protection red lines, and resource development and utilization conditions are considered to conduct comprehensive block selection and hierarchical evaluation.

[0134] Based on the above embodiments, in another embodiment provided in this disclosure, the above-mentioned spatial registration and rasterization of the geological feature layer, geochemical element anomaly layer, and remote sensing alteration mineral layer to construct a multivariate feature dataset may include: In this embodiment, regarding dataset construction, firstly, geochemical data of key lead-zinc ore prediction elements such as Pb, Zn, and Ag are uniformly synthesized into multi-band TIF format raster images to construct a comprehensive geochemical baseline map. Secondly, geological elements such as favorable ore-bearing strata and regional fault buffer zones are converted into raster format and synthesized with geochemical data using band analysis. Finally, information on alteration minerals such as gypsum, smithsonite, limonite, calcite, and alum extracted from hyperspectral remote sensing is also incorporated into a raster dataset, completing the multi-source data fusion with geochemical and geological information to construct a complete spatial analysis dataset.

[0135] For example, Figure 15 This is a raster map of intelligent predictive geological, geochemical, and remote sensing data for carbonate-type lead-zinc deposits provided in this embodiment of the disclosure.

[0136] In the sample labeling stage, geological criteria were used for sample division. Positive samples were directly selected from known lead-zinc deposits as mineralization samples. Negative samples were determined based on four comprehensive criteria: first, known non-lead-zinc mineral deposits; second, areas determined to be mineral-free based on geological conditions; third, areas showing no mineralization anomalies based on remote sensing mineral information; and fourth, areas with low values ​​of key geochemical elements, as well as areas determined to be mineral-free in regional predictions at a scale of 1:250,000.

[0137] Furthermore, unified spatial registration was implemented for multi-source raster data from geology, geochemistry, and remote sensing to ensure that all data layers have the same projected coordinate system and spatial extent. Through raster resampling technology, each data layer was unified to a spatial dimension of 300×300 pixels, achieving strict consistency in the number of rows and columns, thus establishing a standardized data foundation for subsequent band synthesis and model input. In the sample data processing stage, 30 mineralized and non-mineralized sample points (15 positive and 15 negative samples) were simultaneously converted to raster format, maintaining a spatial dimension of 300×300 pixels to ensure complete spatial matching between sample data and feature data. Based on this, multi-band raster synthesis of geological structures, geochemical elements, and remotely sensed alteration minerals was completed, constructing a unified multi-source feature dataset.

[0138] Based on the above embodiments, in another embodiment provided in this disclosure, the above-mentioned inputting a multivariate feature dataset into a pre-constructed mineralization prediction model to generate a mineralization favorability prediction map of the study area may include: In this embodiment, based on the delineation of prospective mineralization areas and comprehensive favorable blocks, this embodiment further employs machine learning algorithms to predict favorable mineralization blocks, thereby achieving in-depth mining of multi-source data and accurate identification of mineralization anomalies.

[0139] Specifically, the study area selected for machine learning prediction is consistent with the aforementioned block analysis area. Based on the actual coverage of the 1:50,000 scale data, areas with high spatial overlap in geological, geochemical, and remote sensing data are selected as machine learning prediction areas to ensure the coordination and consistency of multi-source data in spatial resolution and geographic registration.

[0140] Based on this, two machine learning algorithms, random forest and support vector machine, were used to train the mineralization prediction model. After the model training was completed, the multivariate feature dataset of the study area was input into the mineralization prediction model to generate 300×300 pixel mineralization favorability prediction maps of random forest and support vector machine, respectively.

[0141] For example, Figure 16 A random forest result diagram for carbonate rock-type lead-zinc deposits provided in an embodiment of this disclosure. Figure 17 The image shows the support vector machine results for carbonate rock-type lead-zinc deposits provided in the embodiments of this disclosure.

[0142] Based on the above embodiments, in another embodiment provided in this disclosure, the above-mentioned spatial overlay analysis of the mineralization favorability prediction map and the comprehensive favorable block, followed by geological rationality verification of the overlay area and delineation of the prospecting target area, may include: By spatially overlaying the comprehensive favorable block with the mineralization favorable prediction map, the overlapping area is determined as the initial mineral exploration target area. For areas where the two differ, a geological rationality verification is performed, and the initial prospecting target area is revised based on the verification results to determine the prospecting target area.

[0143] In this embodiment, a spatial overlay analysis is performed between the comprehensive favorable area and the mineralization favorability prediction map. The comprehensive favorable area is a geologically, geochemically, and remotely sensed favorable area determined through multi-scale spatial overlay analysis based on expert knowledge. The mineralization favorability prediction map is a high-potential area prediction result output by a machine learning model.

[0144] In spatial overlay analysis, the first step is to identify the overlapping areas between the two maps. The comprehensive favorable blocks are overlaid with the high-potential areas in the mineralization favorability prediction map, and the overlapping areas are extracted as initial prospecting target areas. These initial prospecting target areas simultaneously meet the geological favorable conditions determined by expert knowledge and the data-driven characteristics identified by machine learning, thus possessing a high probability of mineralization.

[0145] For areas where the two differ, namely areas that exist in the comprehensive favorable blocks but not in the machine learning high-potential areas, and areas that exist in the machine learning high-potential areas but not in the comprehensive favorable blocks, geological rationality verification is performed separately.

[0146] For areas that are present in the comprehensive favorable blocks but not in the high-potential areas identified by machine learning, geological experts will review them based on regional metallogenic regularities, the distribution characteristics of known ore deposits, and field exploration experience. If the area possesses clear metallogenic geological conditions, such as favorable ore-bearing strata, intersections of ore-controlling faults, or typical alteration assemblages, it will be added to the prospecting target area. If the geological conditions of the area are insufficient to support mineralization, it will be identified as a non-mineralized anomaly area and needs to be excluded.

[0147] For areas that are present in the high-potential areas identified by machine learning but not in the comprehensive favorable blocks, geological experts will conduct geological validation of the machine learning predictions. If the geological conditions of the area are consistent with known metallogenic regularities and there are unidentified geological clues, such as concealed structures or weak alteration information, then these will be added to the prospecting target area as new prospecting clues. If the geological conditions of the area do not conform to the regional metallogenic regularities, or if field verification shows no mineralization, then it will be judged as a false anomaly and needs to be removed.

[0148] Based on the above verification results, the initial prospecting target area is corrected, the omitted areas confirmed by verification are supplemented, the false anomaly areas confirmed by verification are eliminated, and the target area boundary is adjusted to better match the superposition characteristics of geological conditions and prediction results.

[0149] Therefore, the final identified mineral exploration target area is a comprehensive result that has undergone geological rationality verification under the dual constraints of expert knowledge and machine learning, and has high reliability and engineering deployment value.

[0150] For example, Figure 18 A comparison chart of intelligent calculation and expert judgment for carbonate-type lead-zinc ore provided in this embodiment of the disclosure. (See figure) Figure 18 As shown, a spatial comparison analysis was conducted between the mineralization favorability prediction map output by the support vector machine model and the mineralization favorable blocks manually delineated based on expert knowledge. The two maps showed a high degree of consistency in spatial distribution, with significant spatial agreement between the high-potential areas delineated by machine learning and the manually delineated blocks. This consistency verifies the complementary value of data-driven methods and expert knowledge systems: machine learning predictions provide reliable quantitative support for traditional geological assessment, while expert knowledge effectively verifies the geological rationality of the model results. Together, they construct a dual constraint mechanism of intelligent computing and expert assessment, significantly improving the scientific rigor and engineering applicability of mineralization prediction.

[0151] One or more technical solutions provided in the exemplary embodiments of this disclosure construct a multi-scale, multi-source spatial database, extract three types of feature layers—geological, geochemical, and remote sensing—and fuse them into a multi-feature dataset. They then combine this dataset with a machine learning model to generate a mineralization favorability prediction map. Finally, they perform spatial overlay analysis and geological rationality verification on this map with a comprehensive favorable block constructed based on expert knowledge, thereby achieving precise delineation of mineral exploration target areas.

[0152] Therefore, the mineralization prediction method based on multimodal data fusion and human-machine collaboration provided in the exemplary embodiments of this disclosure solves the technical problems in mineralization prediction such as reliance on a single data source, strong subjectivity of expert experience, difficulty in effectively integrating multi-source heterogeneous data, and poor accuracy of prediction results. It realizes intelligent fusion of multimodal data and human-machine collaborative decision-making, and significantly improves the objectivity, accuracy and engineering applicability of mineralization prediction.

[0153] The foregoing primarily describes the solutions provided by exemplary embodiments of this disclosure. It is understood that, in order to achieve the above functions, the electronic device includes corresponding hardware structures and / or software modules for performing each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein, this disclosure can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware 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 implementation should not be considered beyond the scope of this disclosure.

[0154] The exemplary embodiments of this disclosure can divide the electronic device into functional units according to the above method examples. For example, each function can be divided into a separate functional module, or two or more functions can be integrated into a single processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in the exemplary embodiments of this disclosure is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0155] By dividing each functional module according to its corresponding function, an exemplary embodiment of this disclosure provides a mineralization prediction device based on multimodal data fusion and human-machine collaboration. This mineralization prediction device based on multimodal data fusion and human-machine collaboration can be a server or a chip applied to a server. Figure 19 This is a schematic block diagram of the functional modules of the mineralization prediction device based on multimodal data fusion and human-machine collaboration provided in an embodiment of this disclosure. Figure 19 As shown, the mineralization prediction device 1900 based on multimodal data fusion and human-machine collaboration includes: The data acquisition module 1910 is used to collect multi-scale data of the study area and construct a multi-source spatial database containing regional and exploration scales based on the multi-scale data; the multi-scale data includes geological data, geochemical data and remote sensing data. Data processing module 1920 is used to extract ore-bearing stratigraphic features, ore-controlling fault features, and alteration mineral features from the multi-source spatial database to generate a geological feature layer; extract geochemical element anomaly features to generate a geochemical element anomaly layer; and extract remote sensing alteration mineral information to generate a remote sensing alteration mineral layer. The data processing module 1920 is also used to perform spatial registration and rasterization processing on the geological feature layer, geochemical element anomaly layer and remote sensing alteration mineral layer to construct a multi-dimensional feature dataset. The data processing module 1920 is also used to input the multivariate feature dataset into a pre-constructed mineralization prediction model to generate a mineralization favorability prediction map of the study area. The data processing module 1920 is also used to perform spatial overlay analysis of the mineralization favorability prediction map and the comprehensive favorable blocks, and delineate the prospecting target area after verifying the geological rationality of the overlay area.

[0156] In another embodiment provided in this disclosure, the data acquisition module 1910 is further configured to extract geological information from geological exploration reports using natural language processing technology to obtain ore deposit point data; collect geological data at different scales, and perform vectorization processing and attribute structure definition on the geological elements of strata, faults, alteration, and ore bodies in the geological data to obtain a geological dataset; perform vectorization processing on geochemical maps, extract single-element anomaly information and combined anomaly information to obtain a geochemical dataset; perform radiometric calibration, atmospheric correction, and geometric correction preprocessing on hyperspectral remote sensing images, and extract characteristic alteration minerals in combination with a spectral library to generate a mineralization and alteration distribution layer to obtain a remote sensing dataset.

[0157] In another embodiment provided in this disclosure, the data processing module 1920 is further configured to: statistically analyze the distribution density of ore deposit points in various stratigraphic units based on the ore deposit point data; label stratigraphic units with ore deposit point densities higher than a preset threshold as favorable ore-bearing strata; determine favorable ore-bearing lithology types based on the lithological specificity characteristics of the ore deposit; extract spatial distribution areas that simultaneously satisfy the favorable ore-bearing strata and the favorable ore-bearing lithology; generate an ore-bearing stratigraphic feature layer; extract regional fault structures from the geological dataset; perform spatial distance analysis between the ore deposit point data and regional faults; statistically analyze the distance distribution characteristics between ore deposit points and faults; determine the radius of the fault influence zone based on the statistical results; establish a buffer zone with the regional fault as the center line; generate a fault structure feature layer; and select alteration types with mineralization indication significance from the geological dataset, convert the selected alteration spatial point data into a raster format, and generate a geological alteration distribution layer.

[0158] In another embodiment provided in this disclosure, the data processing module 1920 is further configured to extract single-element anomaly information of feature indicator elements at both the regional and exploration scales; determine effective mineralized anomalies based on anomaly discrimination criteria; the anomaly discrimination criteria include: having obvious element concentration centers, good spatial alignment of anomalies between main ore-forming elements and associated elements, and exhibiting complete horizontal zoning characteristics; converting the anomaly distribution areas that meet the anomaly discrimination criteria into a raster format to generate a geochemical element anomaly layer.

[0159] In another embodiment provided in this disclosure, the data processing module 1920 is further configured to perform spectral unmixing on the preprocessed hyperspectral image, extract spectral anomalies of alteration minerals related to the target mineral type; perform remote sensing geological interpretation on the spectral anomalies of alteration minerals, identify real mineralization alteration information, and eliminate non-mineralization false anomalies; and convert the verified alteration mineral distribution information into a raster format to generate a remote sensing alteration mineral layer.

[0160] In another embodiment provided in this disclosure, the data processing module 1920 is further configured to perform spatial overlay analysis on the comprehensive favorable block and the mineralization favorable prediction map, determine the overlapping area as the initial prospecting target area; perform geological rationality verification on the area where the two differ, and correct the initial prospecting target area according to the verification result to determine the prospecting target area.

[0161] In another embodiment provided in this disclosure, the data processing module 1920 is further configured to perform spatial overlay analysis on the geological feature layer, the geochemical element anomaly layer, and the remote sensing alteration mineral layer based on a regional scale, and determine the area where the three overlap in space as a prospective mineralization area; based on an exploration scale, in the prospective mineralization area, perform spatial overlay analysis on the geological feature layer, the geochemical element anomaly layer, and the remote sensing alteration mineral layer again, and determine the area where the three overlap in space as the comprehensive favorable block.

[0162] Exemplary embodiments of this disclosure also provide an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the electronic device to perform a method according to an embodiment of this disclosure.

[0163] Exemplary embodiments of this disclosure also provide a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform a method according to embodiments of this disclosure.

[0164] Figure 20 The structural block diagrams of the electronic devices provided in embodiments of this disclosure are described below. The electronic device 2000, which can serve as a server or client of this disclosure, is an example of a hardware device applicable to various aspects of this disclosure. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the disclosure described and / or claimed herein.

[0165] like Figure 20As shown, the electronic device 2000 includes a computing unit 2001, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 2002 or a computer program loaded from a storage unit 2008 into a random access memory (RAM) 2003. The RAM 2003 may also store various programs and data required for the operation of the electronic device 2000. The computing unit 2001, ROM 2002, and RAM 2003 are interconnected via a bus 2004. An input / output (I / O) interface 2005 is also connected to the bus 2004.

[0166] Multiple components in electronic device 2000 are connected to I / O interface 2005, including: input unit 2006, output unit 2007, storage unit 2008, and communication unit 2009. Input unit 2006 can be any type of device capable of inputting information to electronic device 2000. Input unit 2006 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of electronic device. Output unit 2007 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 2008 may include, but is not limited to, disk and optical disk. Communication unit 2009 allows electronic device 2000 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and / or chipsets, such as Bluetooth™ devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0167] The computing unit 2001 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 2001 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 2001 performs the various methods and processes described above. The various methods described above can all be implemented as computer software programs, which are tangibly contained in a machine-readable medium, such as storage unit 2008. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 2000 via ROM 2002 and / or communication unit 2009.

[0168] Figure 21The diagram illustrates a computer program product provided in an embodiment of this disclosure. An exemplary embodiment of this disclosure also provides a computer program product 2100, including a computer program 2101, wherein the computer program 2101, when executed by a computer's processor, is used to cause the computer to perform a method according to an embodiment of this disclosure.

[0169] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0170] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0171] As used in this disclosure, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0172] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0173] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0174] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0175] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this disclosure are performed, in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a terminal, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center integrating one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video disc (DVD); or it can be a semiconductor medium, such as a solid-state drive (SSD).

[0176] Although this disclosure has been described in conjunction with specific features and embodiments, it will be apparent that various modifications and combinations can be made therein without departing from the spirit and scope of this disclosure. Accordingly, this specification and drawings are merely exemplary illustrations of the disclosure as defined by the appended claims and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this disclosure. It is obvious that those skilled in the art can make various alterations and modifications to this disclosure without departing from its spirit and scope. Thus, this disclosure is also intended to include any such modifications and modifications that fall within the scope of the claims of this disclosure and their equivalents.

Claims

1. A mineralization prediction method based on multimodal data fusion and human-machine collaboration, characterized in that, The method includes: Multi-scale data of the study area were collected, and a multi-source spatial database containing regional and exploration scales was constructed based on the multi-scale data; the multi-scale data included geological data, geochemical data, and remote sensing data. The geological feature layer is generated by extracting ore-bearing stratigraphic features, ore-controlling fault features, and alteration mineral features from the multi-source spatial database; geochemical element anomaly features are extracted to generate a geochemical element anomaly layer; and remote sensing alteration mineral information is extracted to generate a remote sensing alteration mineral layer. The geological feature layer, geochemical element anomaly layer, and remote sensing alteration mineral layer are spatially registered and rasterized to construct a multi-dimensional feature dataset. The multivariate feature dataset is input into a pre-constructed mineralization prediction model to generate a mineralization favorability prediction map for the study area. The mineralization favorability prediction map is spatially overlaid with the comprehensive favorable blocks, and the geological rationality of the overlaid area is verified before the prospecting target area is delineated.

2. The method according to claim 1, characterized in that, The multi-scale data collected for the study area includes: Natural language processing technology is used to extract geological information from geological exploration reports to obtain mineral deposit data; Collect geological data at different scales, and vectorize and structurally define the geological elements of strata, faults, alteration, and ore bodies in the geological data to obtain a geological dataset; Geochemical maps are vectorized to extract single-element and combined anomaly information, thus obtaining a geochemical dataset. The hyperspectral remote sensing images were preprocessed with radiometric calibration, atmospheric correction, and geometric correction. Feature alteration minerals were extracted using a spectral library to generate a mineralization alteration distribution layer, thus obtaining a remote sensing dataset.

3. The method according to claim 2, characterized in that, The step of extracting ore-bearing stratigraphic features, ore-controlling fault features, and alteration mineral features from the multi-source spatial database to generate a geological feature layer includes: Based on the ore deposit data, the distribution density of ore deposit points in each stratigraphic unit is statistically analyzed. Stratigraphic units with ore deposit point density higher than a preset threshold are identified as favorable ore-bearing strata. Based on the lithological specificity characteristics of the ore deposit, the favorable ore-bearing lithology type is determined. The spatial distribution area that simultaneously satisfies the favorable ore-bearing strata and the favorable ore-bearing lithology is extracted to generate an ore-bearing stratigraphic feature layer. Regional fault structures are extracted from the geological dataset. Spatial distance analysis is performed between the ore deposit data and the regional faults. The distance distribution characteristics between the ore deposits and the faults are statistically analyzed. The radius of the fault influence zone is determined based on the statistical results. A buffer zone is established with the regional faults as the center line to generate a fault structure feature layer. The geological dataset is used to select alteration types that have mineralization indicative significance. The selected alteration spatial point data is then converted into raster format to generate a geological alteration distribution layer.

4. The method according to claim 2, characterized in that, The extraction of geochemical element anomaly features and the generation of a geochemical element anomaly layer include: Single-element anomaly information of feature indicator elements is extracted at both the regional and exploration scales. Effective mineral-induced anomalies are determined based on anomaly discrimination criteria. The anomaly discrimination criteria include: having obvious element concentration centers, good spatial overlap between the main ore-forming elements and associated elements, and exhibiting complete horizontal zoning characteristics. The anomalous distribution areas that meet the aforementioned anomaly discrimination criteria are converted into a raster format to generate a geochemical element anomaly layer.

5. The method according to claim 2, characterized in that, The step of extracting remotely sensed alteration mineral information and generating a remotely sensed alteration mineral layer includes: Spectral unmixing was performed on the preprocessed hyperspectral images to extract spectral anomalies of altered minerals associated with the target mineral type; Remote sensing geological interpretation is performed on the spectral anomalies of the altered minerals to identify the true mineralization alteration information and eliminate non-mineralization false anomalies; The verified alteration mineral distribution information is converted into a raster format to generate a remote sensing alteration mineral layer.

6. The method according to claim 1, characterized in that, The process of spatially overlaying the predicted mineralization favorability map with the comprehensive favorable blocks, and then delineating the prospecting target area after geological rationality verification of the overlay area, includes: The comprehensive favorable block and the mineralization favorable prediction map are spatially overlaid and analyzed. The area where the two overlap is determined as the initial mineral exploration target area. For areas where the two differ, a geological rationality verification is performed, and the initial prospecting target area is corrected based on the verification results to determine the prospecting target area.

7. The method according to claim 6, characterized in that, The method further includes: Based on the regional scale, the geological feature layer, the geochemical element anomaly layer, and the remote sensing alteration mineral layer are spatially overlaid and analyzed. The areas where the three overlap in space are identified as prospective mineralization areas. Based on the exploration scale, in the prospective mineralized area, the geological feature layer, the geochemical element anomaly layer, and the remote sensing alteration mineral layer are spatially overlaid again, and the areas where the three overlap in space are identified as the comprehensive favorable blocks.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the method of claim 1.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instruction is executed by the processor, it implements the method of claim 1.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the method of claim 1.

Images