Deep concealed deposit detection method based on multi-source geological data fusion

Abstract

The invention discloses a deep concealed ore deposit detection method based on multi-source geological data fusion, and the method comprises the following steps: 1, carrying out a dynamic multi-source data fusion algorithm: introducing a fuzzy logic algorithm, dynamically adjusting the data fusion weight according to the feature significance of different geological units, and developing a cross-scale feature pyramid network, hierarchical feature extraction and fusion are carried out on different detection depth data; 2, optimizing an intelligent model based on prior knowledge; step 3, constructing a real-time interactive three-dimensional modeling system; 4, establishing an efficient multi-means verification system; and step 5, intelligent integration of the exploration process. According to the method, multi-source data are fused through fuzzy logic dynamic weight adjustment, the mineralization mode and the reinforcement learning optimization model are combined, three-dimensional modeling and multi-means combined verification are updated in real time, and the effects of accurate fusion of the multi-source data, dynamic model adjustment and optimization, efficient and accurate modeling, short verification period and low cost are achieved; and the detection efficiency and scale of the deep concealed ore deposit are improved.

Description

Technical Field

[0001] This invention relates to the field of mineral resource exploration, and in particular to a method for detecting deep concealed mineral deposits based on the fusion of multi-source geological data. Background Technology

[0002] Detection of deep, concealed mineral deposits is a key challenge in the field of mineral resource exploration. Currently, the application of technologies in this field has significant limitations. At the level of single-technology applications, geological methods are limited by the detection depth and cannot effectively reach mineral deposit information deep underground; geophysical methods have insufficient ability to identify deep anomalies, which can easily lead to the omission of deep mineralization information; geochemical methods are easily affected by surface environmental factors, which greatly reduces the accuracy of the data; and traditional remote sensing technology, due to its limited resolution, cannot provide detailed surface and shallow feature references for deep detection.

[0003] Regarding multi-source data fusion technology, although attempts have been made to integrate various types of data such as geological, geophysical, geochemical, and remote sensing data, there are still significant shortcomings. Existing data fusion methods are relatively fixed, mostly employing pre-defined principal component analysis (PCA) and independent component analysis (ICA). This fixed processing approach cannot flexibly adapt to the significant differences in data characteristics under different geological scenarios. Machine learning models exhibit mechanical parameter adjustments. The parameter optimization of models such as support vector machines and convolutional neural networks relies solely on fixed cross-validation methods, without any dynamic tuning based on regional geological characteristics, resulting in inconsistent application effects of the models in different geological regions. In the 3D modeling process, the lack of dynamism is a prominent problem. After the model is built, it requires manual adjustment based on comparison with drilling data, and cannot achieve automated real-time updates, which greatly affects modeling efficiency and accuracy. In terms of the verification process, it mainly relies on drilling and geophysical logging, which have problems such as long verification cycles and high costs, severely restricting the efficiency and scale of deep concealed mineral deposit detection. Therefore, a method for deep concealed mineral deposit detection based on multi-source geological data fusion is proposed. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for detecting deep concealed mineral deposits based on the fusion of multi-source geological data, thereby solving the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for detecting deep concealed mineral deposits based on multi-source geological data fusion, comprising the following steps: Step 1: Implementation of the dynamic multi-source data fusion algorithm: Fuzzy logic algorithm is introduced to dynamically adjust data fusion weights based on the saliency of features of different geological units. At the same time, a cross-scale feature pyramid network is developed to perform hierarchical feature extraction and fusion of data at different exploration depths. Step 2: Optimization of the intelligent model based on prior knowledge: The mineralization pattern is transformed into a constraint and embedded into the machine learning model training process. Furthermore, during the model validation phase, a reinforcement learning algorithm is introduced to automatically adjust the model configuration. Step 3: Construction of a Real-Time Interactive 3D Modeling System Develop a streaming data processing interface to input real-time data into 3D modeling software to achieve incremental model updates, and build an interactive interface for 3D geological models to support visualization analysis. Step 4: Establishment of an efficient multi-method verification system: The suspicious area was scanned using a drone equipped with a hyperspectral sensor, and multi-parameter joint verification was carried out by combining geoelectrochemical measurements and microseismic monitoring technology. Step 5: Intelligent Integration of the Exploration Process Develop a fully automated platform that integrates data acquisition, fusion, modeling, and verification, and integrate an economic evaluation module into the system to generate exploration decision recommendations; In the implementation of the dynamic multi-source data fusion algorithm, for igneous rock distribution areas, the weight of geophysical data is enhanced by analyzing the saliency of features in the region, such as seismic wave velocity and magnetic anomaly; in sedimentary rock areas, the weight of geochemical data is emphasized based on saliency features such as element content and ratio. The cross-scale feature pyramid network constructs a multi-level feature extraction module to process data from different detection depths, such as shallow remote sensing spectra and deep electromagnetic responses, and performs hierarchical feature extraction and fusion. By employing a dynamic multi-source data fusion algorithm, weights are dynamically adjusted based on the characteristics of different geological units. A hierarchical feature extraction method using a cross-scale feature pyramid network effectively improves the accuracy and adaptability of data fusion, solving the problem of insufficient adaptability in traditional fixed algorithms. Intelligent model optimization based on prior knowledge transforms mineralization patterns into constraints embedded in model training, enhancing the geological rationality of prediction results. Furthermore, automatic adjustment of model configuration through reinforcement learning algorithms improves the generalization ability of cross-regional exploration. The construction of a real-time interactive 3D modeling system enables real-time updates and visualization analysis of exploration data, improving modeling efficiency and accuracy. An efficient multi-method verification system utilizes UAV hyperspectral scanning, geoelectrochemical measurements, and microseismic monitoring technologies to cross-verify model prediction results from multiple dimensions, improving the accuracy and reliability of verification. Intelligent integration of the exploration process, through a fully automated platform and economic evaluation module, achieves standardization and efficiency improvement in the exploration process.

[0006] Preferably, in the implementation of the dynamic multi-source data fusion algorithm, the weight of geophysical data is increased in igneous rock distribution areas, while geochemical data is emphasized in sedimentary rock areas, in order to solve the problem of insufficient adaptability of traditional fixed algorithms; To specifically implement the data weight adjustment for igneous and sedimentary rock areas, a geological unit classifier can be constructed to automatically identify the area type based on geological maps or existing drilling data. When identified as an igneous rock area, the system automatically increases the fusion weight of geophysical data (such as gravity and magnetic methods); when identified as a sedimentary rock area, the weight of geochemical data (such as elemental concentration and isotope ratios) is emphasized. The flexibility and adaptability of data fusion are achieved through a dynamic weight adjustment algorithm. Dynamically adjusting data fusion weights effectively solves the problem of insufficient adaptability of traditional fixed algorithms. In igneous rock areas, increasing the weight of geophysical data can more accurately capture structural information deep underground; while in sedimentary rock areas, focusing on geochemical data helps to reveal more subtle geological changes and mineralization characteristics. This data fusion method based on geological unit characteristics not only improves the efficiency and accuracy of data processing, but also significantly enhances the algorithm's adaptability to complex geological environments, providing more reliable technical support for the detection of deep concealed mineral deposits.

[0007] Preferably, the cross-scale feature pyramid network is used to perform cross-scale feature fusion on data from different detection depths, such as shallow remote sensing spectra and deep electromagnetic responses, to improve the comprehensive characterization capability of mineralization information at different depths. In the cross-scale feature pyramid network, in order to achieve cross-scale feature fusion of data from different detection depths, such as shallow remote sensing spectra and deep electromagnetic responses, a multi-level feature extraction module can be constructed. For data from different detection depths, shallow remote sensing feature extraction layers and deep electromagnetic feature extraction layers are designed respectively. By skipping connections and feature fusion layers, feature information from different scales can be organically integrated to improve the comprehensive characterization ability of mineralization information at different depths. Cross-scale feature pyramid networks play a crucial role in the detection of deep concealed mineral deposits. They effectively integrate data from different detection depths, such as shallow remote sensing spectroscopy and deep electromagnetic responses. Through cross-scale feature fusion technology, they enhance the comprehensive characterization of mineralization information at different depths. This network design enables detection methods to capture mineralization information more comprehensively, reducing blind spots caused by data limitations. Furthermore, cross-scale feature fusion helps improve the accuracy and reliability of detection, providing geological exploration personnel with richer mineralization information for reference, thus supporting more scientific exploration decisions. In addition, the application of this network promotes the intelligent and automated development of deep concealed mineral deposit detection technology, improves exploration efficiency, reduces exploration costs, and is of great significance for the sustainable development and utilization of mineral resources.

[0008] Preferably, in the intelligent model optimization based on prior knowledge, the mineralization mode includes the ore-controlling law of hydrothermal mineralization zone. By transforming it into constraint conditions and embedding it into the machine learning model training process of random forest and deep learning network, the geological rationality of the prediction results is enhanced. In the optimization of intelligent models based on prior knowledge, for the ore-controlling laws of hydrothermal metallogenic belts, the geological data of hydrothermal metallogenic belts are first collected and organized, and the ore-controlling elements such as structure, lithology, and alteration are analyzed. Then, these ore-controlling elements are transformed into mathematical models or logical rules, which are embedded as constraints into the training process of machine learning models such as random forests and deep learning networks. During model training, the model output results are forced to conform to these geological constraints, thereby enhancing the geological rationality of the prediction results. In the optimization of intelligent models based on prior knowledge, mineralization patterns such as the ore-controlling laws of hydrothermal mineralization zones are transformed into constraints and embedded into the training process of machine learning models such as random forests and deep learning networks. This approach significantly improves the geological rationality of the prediction results. It not only makes full use of the experience and knowledge of geological experts, but also achieves in-depth mining and precise analysis of complex geological data through the automatic learning capabilities of machine learning algorithms. This method, which combines prior knowledge with machine learning, effectively avoids the geological inconsistencies that may exist in traditional models and improves the accuracy and reliability of deep concealed mineral deposit detection.

[0009] Preferably, during the model verification stage, the reinforcement learning algorithm automatically adjusts the model configuration of the number of neural network layers and kernel function parameters based on the drilling verification results, thereby improving the generalization ability of cross-regional exploration. The specific implementation of the reinforcement learning algorithm in the model validation stage is as follows: a feedback mechanism based on drilling validation results is constructed. When there is a deviation between the new drilling data and the model prediction results, the algorithm automatically analyzes the deviation and adjusts the number of neural network layers and kernel function parameters to optimize the model configuration. This process, through iterative training, enables the model to gradually adapt to the geological features of different regions, thereby improving the generalization ability of cross-regional exploration. The automatic adjustment mechanism of reinforcement learning algorithms in the model validation stage significantly improves the intelligence level of deep concealed mineral deposit detection. By dynamically adjusting the number of neural network layers and kernel function parameters based on drilling validation results, the algorithm can continuously optimize the model configuration to better adapt to complex and changing geological conditions. This adaptive adjustment not only enhances the model's generalization ability to geological features of different regions but also reduces the need for manual intervention, improving detection efficiency and accuracy. At the same time, this mechanism continuously optimizes model performance through iterative learning, providing more reliable and efficient technical support for geological exploration.

[0010] Preferably, in the construction of the real-time interactive 3D modeling system, the core mineral data obtained from drilling and the real-time data of well logging resistivity are input into the 3D modeling software through the spatial kriging interpolation method, and the mineral deposit distribution prediction is dynamically updated to form a closed-loop automation of "detection-verification-correction". In the construction of a real-time interactive 3D modeling system, in order to realize the dynamic updating of ore deposit distribution prediction, it is necessary to develop a spatial kriging interpolation algorithm module and integrate it into the 3D modeling software. This module automatically receives the core mineral data and well logging resistivity data obtained from drilling, uses the spatial kriging interpolation method to calculate the values ​​of unknown areas, updates the 3D geological model, and feeds back the prediction results to the visualization interface in real time, forming a closed-loop automated process of "detection-verification-correction". The construction of a real-time interactive 3D modeling system effectively integrates real-time drilling and logging data through spatial kriging interpolation, significantly improving the timeliness and accuracy of ore deposit distribution prediction. This system not only enables dynamic model updates but also supports real-time analysis and decision-making by exploration personnel through a visual interface, accelerating the exploration process. The resulting "detection-verification-correction" closed-loop automation reduces manual intervention, improves work efficiency, and enhances the credibility of prediction results, providing strong technical support for geological exploration.

[0011] Preferably, the interactive interface of the three-dimensional geological model allows exploration personnel to directly adjust the parameters of stratigraphic contact relationship and geochemical anomaly threshold through graphical operation of sliders and parameter panels, and view the changes in model prediction results in real time, thereby improving modeling efficiency and accuracy. The interactive interface of the 3D geological model is implemented through an integrated graphical user interface framework. The slider control allows explorers to intuitively adjust parameters of stratigraphic contact relationships, such as angles and distances. The parameter panel provides input and modification functions for geochemical anomaly thresholds. The interface backend communicates with the 3D modeling software in real time. Once the parameter adjustment is completed, the model update calculation is triggered, and the results are displayed on the interface in real time, realizing the synchronous display of parameter adjustment and model prediction results. The design of the interactive interface for 3D geological models greatly improves the efficiency and accuracy of modeling in the geological exploration process. Exploration personnel can directly and conveniently adjust key parameters such as stratigraphic contact relationships and geochemical anomaly thresholds through the graphical interface, without the need for in-depth programming or complex data processing. This intuitive operation not only lowers the technical threshold, allowing non-professionals to quickly get started, but also significantly accelerates the speed of model construction and iteration. At the same time, the function of viewing changes in model prediction results in real time allows exploration personnel to instantly assess the impact of different parameter settings on the model, thereby making more scientific and reasonable decisions. This interactive method not only improves work efficiency, but also effectively improves the accuracy of modeling through multiple iterations to optimize model parameters.

[0012] Preferably, in the establishment of the efficient multi-method verification system, high-potential target areas are screened using spectral anomaly identification technology of UAV hyperspectral scanning, and the metal active state measurement and microseismic monitoring technology are combined with geoelectrochemical measurement to cross-validate the model prediction results from multiple dimensions. In establishing an efficient multi-method verification system, for the spectral anomaly identification technology using UAV hyperspectral scanning, a specific mineral spectral feature library can be set up and the scanning data compared to quickly locate the anomaly area; geoelectrochemical measurement uses an electrode grid to monitor changes in the metal activity state in real time; and microseismic monitoring uses a seismograph to capture minute vibration signals and analyze their relationship with ore body activity. The combination of these three methods cross-validates the model prediction from different dimensions to ensure the accuracy of target area selection. The establishment of an efficient multi-method verification system, through the integrated application of UAV hyperspectral scanning, geoelectrochemical measurement, and microseismic monitoring technologies, has enabled multi-dimensional cross-verification of predictions for deep concealed mineral deposits. Hyperspectral scanning can quickly identify spectral anomalies and screen high-potential target areas; geoelectrochemical measurement provides direct evidence of metal activity; and microseismic monitoring captures subtle signals of ore body activity. This multi-method joint verification approach not only improves the reliability of prediction results but also reduces exploration risks, providing a scientific basis for subsequent exploration work.

[0013] Preferably, in the intelligent integration of the exploration process, the fully automated platform achieves seamless connection between data acquisition, fusion, modeling and verification through modular design, reducing human operation errors and improving the standardization and efficiency of the exploration process; In the intelligent integration of the exploration process, the fully automated platform achieves seamless connection between data acquisition, fusion, modeling, and verification through modular design. Specifically, the data acquisition module can integrate various geological exploration equipment to realize automatic data acquisition and transmission; the fusion module adopts a dynamic multi-source data fusion algorithm to extract and fuse hierarchical features of different geological data; the modeling module uses a real-time interactive 3D modeling system to dynamically update the mineral deposit distribution prediction; and the verification module combines UAV hyperspectral scanning, geoelectrochemical measurement, and microseismic monitoring technologies to perform multi-dimensional cross-verification of the model prediction results. The fully automated platform in the intelligent integration of the exploration process achieves seamless connection between data acquisition, fusion, modeling, and verification through modular design, greatly reducing human error. This platform not only improves the standardization of the exploration process and ensures the accuracy and consistency of data processing, but also significantly improves exploration efficiency. Exploration personnel can use this platform to obtain mineral deposit information more quickly and accurately, providing strong support for geological exploration decisions. At the same time, the platform's automation and intelligence features also reduce exploration costs and improve resource utilization efficiency.

[0014] Preferably, the economic evaluation module combines parameters of the predicted scale of the ore deposit and the difficulty of mining to automatically generate exploration priority ranking and resource allocation suggestions, providing a scientific basis for geological exploration decision-making; When implementing the economic evaluation module, it first estimates the predicted size and mining difficulty of the deposit through an algorithm model. Specifically, the predicted size may be based on the volume and grade estimation of the three-dimensional geological model, while the mining difficulty may take into account factors such as the burial depth of the ore body, the complexity of the geological structure, and the hardness of the ore. Subsequently, in combination with these parameters, the module will use preset evaluation indicators and weighting systems to automatically calculate the priority of each exploration area and generate resource allocation suggestions accordingly, providing a scientific basis for decision-makers. The economic evaluation module, by combining key parameters such as the predicted size of the deposit and the difficulty of mining, has achieved the automated generation of exploration priority ranking and resource allocation recommendations. This approach not only greatly improves decision-making efficiency and reduces the subjectivity and error of manual analysis, but also ensures the scientific and rational nature of the decisions. Through this module, exploration companies can allocate resources more effectively, prioritize the development of high-potential, low-risk deposits, thereby optimizing the exploration process, reducing exploration costs, and improving overall economic benefits.

[0015] In summary, compared with existing technologies, this invention provides a method for detecting deep concealed mineral deposits based on multi-source geological data fusion, which has the following beneficial effects: This invention introduces a fuzzy logic algorithm to dynamically adjust the data fusion weights and develops a cross-scale feature pyramid network to perform hierarchical feature extraction and fusion of data from different exploration depths. This achieves a more accurate and flexible fusion effect for multi-source geological data. It can adaptively adjust the data fusion ratio based on the saliency of the features of different geological units, fully explore the effective information of various types of data at different scales, and overcome the shortcomings of existing fixed data fusion methods that cannot adapt to the differences in data features of different geological scenarios. This provides a more reliable data foundation for subsequent exploration and effectively improves the accuracy and reliability of deep concealed mineral deposit exploration. By transforming mineralization patterns into constraints and embedding them into the machine learning model training process, and introducing reinforcement learning algorithms to automatically adjust the model configuration during the model validation stage, dynamic optimization of machine learning model parameters is achieved. This allows for targeted optimization of the model based on regional geological characteristics, ensuring good application performance across different geological regions. This solves the problem of inconsistent application performance caused by the mechanical adjustment of traditional model parameters and the lack of integration with geological characteristics. Simultaneously, a streaming data processing interface was developed to enable real-time data input into 3D modeling software for incremental model updates, and an interactive interface for 3D geological models was constructed to support visual analysis. This achieved dynamic real-time updates of 3D modeling, improving modeling efficiency and accuracy. By utilizing multiple technologies, including UAVs equipped with hyperspectral sensors, multi-parameter joint verification was conducted, establishing an efficient multi-method verification system. This system offers advantages such as shortening the verification cycle and reducing verification costs, overcoming the shortcomings of traditional verification methods, which are characterized by long cycles and high costs. Overall, it significantly improved the efficiency and scale of deep concealed mineral deposit exploration. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of the invention's method for detecting deep, concealed mineral deposits based on the fusion of multi-source geological data. Detailed Implementation

[0017] This invention provides a technical solution: a method for detecting deep, concealed mineral deposits based on multi-source geological data fusion. Please refer to [link to relevant documentation]. Figure 1 It includes the following steps: Step 1: Implementation of the dynamic multi-source data fusion algorithm: Fuzzy logic algorithm is introduced to dynamically adjust data fusion weights based on the saliency of features of different geological units. At the same time, a cross-scale feature pyramid network is developed to perform hierarchical feature extraction and fusion of data at different exploration depths. Step 2: Optimization of the intelligent model based on prior knowledge: The mineralization pattern is transformed into a constraint and embedded into the machine learning model training process. Furthermore, during the model validation phase, a reinforcement learning algorithm is introduced to automatically adjust the model configuration. Step 3: Construction of a Real-Time Interactive 3D Modeling System Develop a streaming data processing interface to input real-time data into 3D modeling software to achieve incremental model updates, and build an interactive interface for 3D geological models to support visualization analysis. Step 4: Establishment of an efficient multi-method verification system: The suspicious area was scanned using a drone equipped with a hyperspectral sensor, and multi-parameter joint verification was carried out by combining geoelectrochemical measurements and microseismic monitoring technology. Step 5: Intelligent Integration of the Exploration Process Develop a fully automated platform that integrates data acquisition, fusion, modeling, and verification, and integrate an economic evaluation module into the system to generate exploration decision recommendations; In the implementation of the dynamic multi-source data fusion algorithm, for igneous rock distribution areas, the weight of geophysical data is enhanced by analyzing the saliency of features in the region, such as seismic wave velocity and magnetic anomaly; in sedimentary rock areas, the weight of geochemical data is emphasized based on saliency features such as element content and ratio. The cross-scale feature pyramid network constructs a multi-level feature extraction module to process data from different detection depths, such as shallow remote sensing spectra and deep electromagnetic responses, and performs hierarchical feature extraction and fusion. By employing a dynamic multi-source data fusion algorithm, weights are dynamically adjusted based on the characteristics of different geological units. A hierarchical feature extraction method using a cross-scale feature pyramid network effectively improves the accuracy and adaptability of data fusion, solving the problem of insufficient adaptability in traditional fixed algorithms. Intelligent model optimization based on prior knowledge transforms mineralization patterns into constraints embedded in model training, enhancing the geological rationality of prediction results. Furthermore, automatic adjustment of model configuration through reinforcement learning algorithms improves the generalization ability of cross-regional exploration. The construction of a real-time interactive 3D modeling system enables real-time updates and visualization analysis of exploration data, improving modeling efficiency and accuracy. An efficient multi-method verification system utilizes UAV hyperspectral scanning, geoelectrochemical measurements, and microseismic monitoring technologies to cross-verify model prediction results from multiple dimensions, improving the accuracy and reliability of verification. Intelligent integration of the exploration process, through a fully automated platform and economic evaluation module, achieves standardization and efficiency improvement in the exploration process.

[0018] Please see Figure 1 In the implementation of the dynamic multi-source data fusion algorithm, the weight of geophysical data is increased in the igneous rock distribution area, and geochemical data is emphasized in the sedimentary rock area, so as to solve the problem of insufficient adaptability of the traditional fixed algorithm. To specifically implement the data weight adjustment for igneous and sedimentary rock areas, a geological unit classifier can be constructed to automatically identify the area type based on geological maps or existing drilling data. When identified as an igneous rock area, the system automatically increases the fusion weight of geophysical data (such as gravity and magnetic methods); when identified as a sedimentary rock area, the weight of geochemical data (such as elemental concentration and isotope ratios) is emphasized. The flexibility and adaptability of data fusion are achieved through a dynamic weight adjustment algorithm. Dynamically adjusting data fusion weights effectively solves the problem of insufficient adaptability of traditional fixed algorithms. In igneous rock areas, increasing the weight of geophysical data can more accurately capture structural information deep underground; while in sedimentary rock areas, focusing on geochemical data helps to reveal more subtle geological changes and mineralization characteristics. This data fusion method based on geological unit characteristics not only improves the efficiency and accuracy of data processing, but also significantly enhances the algorithm's adaptability to complex geological environments, providing more reliable technical support for the detection of deep concealed mineral deposits.

[0019] Please see Figure 1A cross-scale feature pyramid network is used to perform cross-scale feature fusion on shallow remote sensing spectral and deep electromagnetic response data at different detection depths, thereby improving the comprehensive characterization of mineralization information at different depths. In the cross-scale feature pyramid network, in order to achieve cross-scale feature fusion of data from different detection depths, such as shallow remote sensing spectra and deep electromagnetic responses, a multi-level feature extraction module can be constructed. For data from different detection depths, shallow remote sensing feature extraction layers and deep electromagnetic feature extraction layers are designed respectively. By skipping connections and feature fusion layers, feature information from different scales can be organically integrated to improve the comprehensive characterization ability of mineralization information at different depths. Cross-scale feature pyramid networks play a crucial role in the detection of deep concealed mineral deposits. They effectively integrate data from different detection depths, such as shallow remote sensing spectroscopy and deep electromagnetic responses. Through cross-scale feature fusion technology, they enhance the comprehensive characterization of mineralization information at different depths. This network design enables detection methods to capture mineralization information more comprehensively, reducing blind spots caused by data limitations. Furthermore, cross-scale feature fusion helps improve the accuracy and reliability of detection, providing geological exploration personnel with richer mineralization information for reference, thus supporting more scientific exploration decisions. In addition, the application of this network promotes the intelligent and automated development of deep concealed mineral deposit detection technology, improves exploration efficiency, reduces exploration costs, and is of great significance for the sustainable development and utilization of mineral resources.

[0020] Please see Figure 1 In the intelligent model optimization based on prior knowledge, the mineralization mode includes the ore-controlling law of hydrothermal mineralization zone. By transforming it into constraint conditions and embedding it into the machine learning model training process of random forest and deep learning network, the geological rationality of the prediction results is enhanced. In the optimization of intelligent models based on prior knowledge, for the ore-controlling laws of hydrothermal metallogenic belts, the geological data of hydrothermal metallogenic belts are first collected and organized, and the ore-controlling elements such as structure, lithology, and alteration are analyzed. Then, these ore-controlling elements are transformed into mathematical models or logical rules, which are embedded as constraints into the training process of machine learning models such as random forests and deep learning networks. During model training, the model output results are forced to conform to these geological constraints, thereby enhancing the geological rationality of the prediction results. In the optimization of intelligent models based on prior knowledge, mineralization patterns such as the ore-controlling laws of hydrothermal mineralization zones are transformed into constraints and embedded into the training process of machine learning models such as random forests and deep learning networks. This approach significantly improves the geological rationality of the prediction results. It not only makes full use of the experience and knowledge of geological experts, but also achieves in-depth mining and precise analysis of complex geological data through the automatic learning capabilities of machine learning algorithms. This method, which combines prior knowledge with machine learning, effectively avoids the geological inconsistencies that may exist in traditional models and improves the accuracy and reliability of deep concealed mineral deposit detection.

[0021] Please see Figure 1 In the model validation phase, reinforcement learning algorithms automatically adjust the model configuration of the number of neural network layers and kernel function parameters based on the drilling validation results, thereby improving the generalization ability of cross-regional exploration. The specific implementation of the reinforcement learning algorithm in the model validation stage is as follows: a feedback mechanism based on drilling validation results is constructed. When there is a deviation between the new drilling data and the model prediction results, the algorithm automatically analyzes the deviation and adjusts the number of neural network layers and kernel function parameters to optimize the model configuration. This process, through iterative training, enables the model to gradually adapt to the geological features of different regions, thereby improving the generalization ability of cross-regional exploration. The automatic adjustment mechanism of reinforcement learning algorithms in the model validation stage significantly improves the intelligence level of deep concealed mineral deposit detection. By dynamically adjusting the number of neural network layers and kernel function parameters based on drilling validation results, the algorithm can continuously optimize the model configuration to better adapt to complex and changing geological conditions. This adaptive adjustment not only enhances the model's generalization ability to geological features of different regions but also reduces the need for manual intervention, improving detection efficiency and accuracy. At the same time, this mechanism continuously optimizes model performance through iterative learning, providing more reliable and efficient technical support for geological exploration.

[0022] Please see Figure 1 In the construction of the real-time interactive 3D modeling system, the core mineral data obtained from drilling and the real-time data of well logging resistivity are input into the 3D modeling software through the spatial kriging interpolation method, and the mineral deposit distribution prediction is dynamically updated to form a closed-loop automation of "detection-verification-correction". In the construction of a real-time interactive 3D modeling system, in order to realize the dynamic updating of ore deposit distribution prediction, it is necessary to develop a spatial kriging interpolation algorithm module and integrate it into the 3D modeling software. This module automatically receives the core mineral data and well logging resistivity data obtained from drilling, uses the spatial kriging interpolation method to calculate the values ​​of unknown areas, updates the 3D geological model, and feeds back the prediction results to the visualization interface in real time, forming a closed-loop automated process of "detection-verification-correction". The construction of a real-time interactive 3D modeling system effectively integrates real-time drilling and logging data through spatial kriging interpolation, significantly improving the timeliness and accuracy of ore deposit distribution prediction. This system not only enables dynamic model updates but also supports real-time analysis and decision-making by exploration personnel through a visual interface, accelerating the exploration process. The resulting "detection-verification-correction" closed-loop automation reduces manual intervention, improves work efficiency, and enhances the credibility of prediction results, providing strong technical support for geological exploration.

[0023] Please see Figure 1 The interactive interface of the 3D geological model allows exploration personnel to directly adjust the parameters of stratigraphic contact relationship and geochemical anomaly threshold through graphical operations of sliders and parameter panels, and view the changes in model prediction results in real time, thereby improving modeling efficiency and accuracy. The interactive interface of the 3D geological model is implemented through an integrated graphical user interface framework. The slider control allows explorers to intuitively adjust parameters of stratigraphic contact relationships, such as angles and distances. The parameter panel provides input and modification functions for geochemical anomaly thresholds. The interface backend communicates with the 3D modeling software in real time. Once the parameter adjustment is completed, the model update calculation is triggered, and the results are displayed on the interface in real time, realizing the synchronous display of parameter adjustment and model prediction results. The design of the interactive interface for 3D geological models greatly improves the efficiency and accuracy of modeling in the geological exploration process. Exploration personnel can directly and conveniently adjust key parameters such as stratigraphic contact relationships and geochemical anomaly thresholds through the graphical interface, without the need for in-depth programming or complex data processing. This intuitive operation not only lowers the technical threshold, allowing non-professionals to quickly get started, but also significantly accelerates the speed of model construction and iteration. At the same time, the function of viewing changes in model prediction results in real time allows exploration personnel to instantly assess the impact of different parameter settings on the model, thereby making more scientific and reasonable decisions. This interactive method not only improves work efficiency, but also effectively improves the accuracy of modeling through multiple iterations to optimize model parameters.

[0024] Please see Figure 1 In the process of establishing an efficient and multi-method verification system, high-potential target areas are screened by using the spectral anomaly identification technology of UAV hyperspectral scanning, and the metal active state measurement and microseismic monitoring technology are combined with geoelectrochemical measurement to cross-validate the model prediction results from multiple dimensions. In establishing an efficient multi-method verification system, for the spectral anomaly identification technology using UAV hyperspectral scanning, a specific mineral spectral feature library can be set up and the scanning data compared to quickly locate the anomaly area; geoelectrochemical measurement uses an electrode grid to monitor changes in the metal activity state in real time; and microseismic monitoring uses a seismograph to capture minute vibration signals and analyze their relationship with ore body activity. The combination of these three methods cross-validates the model prediction from different dimensions to ensure the accuracy of target area selection. The establishment of an efficient multi-method verification system, through the integrated application of UAV hyperspectral scanning, geoelectrochemical measurement, and microseismic monitoring technologies, has enabled multi-dimensional cross-verification of predictions for deep concealed mineral deposits. Hyperspectral scanning can quickly identify spectral anomalies and screen high-potential target areas; geoelectrochemical measurement provides direct evidence of metal activity; and microseismic monitoring captures subtle signals of ore body activity. This multi-method joint verification approach not only improves the reliability of prediction results but also reduces exploration risks, providing a scientific basis for subsequent exploration work.

[0025] Please see Figure 1 In the intelligent integration of the exploration process, the fully automated platform achieves seamless connection between data acquisition, fusion, modeling and verification through modular design, reducing human operation errors and improving the standardization and efficiency of the exploration process. In the intelligent integration of the exploration process, the fully automated platform achieves seamless connection between data acquisition, fusion, modeling, and verification through modular design. Specifically, the data acquisition module can integrate various geological exploration equipment to realize automatic data acquisition and transmission; the fusion module adopts a dynamic multi-source data fusion algorithm to extract and fuse hierarchical features of different geological data; the modeling module uses a real-time interactive 3D modeling system to dynamically update the mineral deposit distribution prediction; and the verification module combines UAV hyperspectral scanning, geoelectrochemical measurement, and microseismic monitoring technologies to perform multi-dimensional cross-verification of the model prediction results. The fully automated platform in the intelligent integration of the exploration process achieves seamless connection between data acquisition, fusion, modeling, and verification through modular design, greatly reducing human error. This platform not only improves the standardization of the exploration process and ensures the accuracy and consistency of data processing, but also significantly improves exploration efficiency. Exploration personnel can use this platform to obtain mineral deposit information more quickly and accurately, providing strong support for geological exploration decisions. At the same time, the platform's automation and intelligence features also reduce exploration costs and improve resource utilization efficiency.

[0026] Please see Figure 1 The economic evaluation module combines parameters such as the predicted scale of the ore deposit and the difficulty of mining to automatically generate exploration priority ranking and resource allocation suggestions, providing a scientific basis for geological exploration decision-making. When implementing the economic evaluation module, it first estimates the predicted size and mining difficulty of the deposit through an algorithm model. Specifically, the predicted size may be based on the volume and grade estimation of the three-dimensional geological model, while the mining difficulty may take into account factors such as the burial depth of the ore body, the complexity of the geological structure, and the hardness of the ore. Subsequently, in combination with these parameters, the module will use preset evaluation indicators and weighting systems to automatically calculate the priority of each exploration area and generate resource allocation suggestions accordingly, providing a scientific basis for decision-makers. The economic evaluation module, by combining key parameters such as the predicted size of the deposit and the difficulty of mining, has achieved the automated generation of exploration priority ranking and resource allocation recommendations. This approach not only greatly improves decision-making efficiency and reduces the subjectivity and error of manual analysis, but also ensures the scientific and rational nature of the decisions. Through this module, exploration companies can allocate resources more effectively, prioritize the development of high-potential, low-risk deposits, thereby optimizing the exploration process, reducing exploration costs, and improving overall economic benefits.

[0027] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0028] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for detecting deep concealed mineral deposits based on multi-source geological data fusion, characterized in that, Includes the following steps: Step 1: Implementation of the dynamic multi-source data fusion algorithm: Fuzzy logic algorithm is introduced to dynamically adjust data fusion weights based on the saliency of features of different geological units. At the same time, a cross-scale feature pyramid network is developed to perform hierarchical feature extraction and fusion of data at different exploration depths. Step 2: Optimization of the intelligent model based on prior knowledge: The mineralization pattern is transformed into a constraint and embedded into the machine learning model training process. Furthermore, during the model validation phase, a reinforcement learning algorithm is introduced to automatically adjust the model configuration. Step 3: Construction of a Real-Time Interactive 3D Modeling System Develop a streaming data processing interface to input real-time data into 3D modeling software to achieve incremental model updates, and build an interactive interface for 3D geological models to support visualization analysis. Step 4: Establishment of an efficient multi-method verification system: The suspicious area was scanned using a drone equipped with a hyperspectral sensor, and multi-parameter joint verification was carried out by combining geoelectrochemical measurements and microseismic monitoring technology. Step 5: Intelligent Integration of the Exploration Process Develop a fully automated platform that integrates data acquisition, fusion, modeling, and verification, and integrate an economic evaluation module into the system to generate exploration decision recommendations.

2. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: In the implementation of the dynamic multi-source data fusion algorithm, the weight of geophysical data is increased in igneous rock distribution areas, while geochemical data is emphasized in sedimentary rock areas.

3. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: The cross-scale feature pyramid network is used to perform cross-scale feature fusion of shallow remote sensing spectral and deep electromagnetic response data at different detection depths.

4. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: In the intelligent model optimization based on prior knowledge, the mineralization mode includes the ore-controlling law of hydrothermal mineralization zone, which is transformed into constraint conditions and embedded in the training process of random forest and deep learning network machine learning models.

5. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: During the model validation phase, the reinforcement learning algorithm automatically adjusts the model configuration of the number of neural network layers and kernel function parameters based on the drilling validation results.

6. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: In the construction of the real-time interactive 3D modeling system, the core mineral data obtained from drilling and the real-time data of well logging resistivity are input into the 3D modeling software through the spatial kriging interpolation method to dynamically update the ore deposit distribution prediction.

7. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: The interactive interface of the three-dimensional geological model allows exploration personnel to directly adjust the parameters of stratigraphic contact relationship and geochemical anomaly threshold through graphical operations of sliders and parameter panels, and view the changes in model prediction results in real time.

8. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: In establishing the efficient multi-method verification system, high-potential target areas are screened using spectral anomaly identification technology via UAV hyperspectral scanning, combined with geoelectrochemical measurements of metal active states and microseismic monitoring technology.

9. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: In the intelligent integration of the exploration process, the fully automated platform achieves seamless connection between data acquisition, fusion, modeling and verification through modular design.

10. The method for detecting deep concealed mineral deposits based on multi-source geological data fusion according to claim 1, characterized in that: The economic evaluation module automatically generates exploration priority ranking and resource allocation suggestions by combining parameters such as the predicted scale of the deposit and the mining difficulty.

Images