Geological big data metallogenic prediction cloud platform method

Abstract

This invention discloses a geological big data mineralization prediction cloud platform method. This invention relates to the field of geological exploration technology, and enables the unified integration and standardized processing of multi-source geological data, breaking down data silos and significantly improving the utilization efficiency and sharing capability of geological data, providing comprehensive and standardized basic data support for mineralization prediction. Through the deep integration of association rule mining and mineralization element analysis, a high-precision mineralization prediction model with multi-scale and multi-element fusion is constructed, effectively capturing regional mineralization patterns and significantly improving the accuracy and reliability of predicting favorable mineralization areas. Employing federated learning and privacy-preserving computation technologies, while ensuring the privacy and security of original data from various institutions, cross-regional multi-institutional collaborative training is achieved, fully integrating dispersed geological resources and professional knowledge, and promoting collaborative innovation in mineralization prediction technology.

Description

Technical Field

[0001] This invention relates to the field of geological exploration technology, specifically a geological big data mineralization prediction cloud platform method. Background Technology

[0002] Metallogenic prediction is a professional technique that uses the theories and related technical methods of basic geology and mineral deposit geology to analyze the metallogenic conditions and prospecting information in a region (or mining area) and infer the possible mineral deposits and their basic characteristics.

[0003] Existing mineralization prediction methods largely rely on manual processing and analysis of scattered geological, geophysical, and geochemical data from multiple sources. This results in low data integration efficiency and information silos caused by inconsistent formats. Furthermore, traditional prediction models are often built on local data and lack the ability to deeply mine large-scale heterogeneous data, limiting prediction accuracy due to insufficient data scale and computing resources. In addition, existing technologies lack a unified cloud platform, making it difficult to achieve cross-regional data sharing, collaborative model training, and real-time prediction updates. This fails to meet the demands of modern mineralization prediction for efficiency, accuracy, and dynamism, hindering the intelligent and large-scale application of mineralization prediction. Therefore, this invention provides a geological big data mineralization prediction cloud platform method. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a geological big data mineralization prediction cloud platform method to solve the aforementioned problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a geological big data mineralization prediction cloud platform method, comprising:

[0006] The data access and standardization module is used to access multi-source heterogeneous geological data, standardize the data, and clean the data.

[0007] The cloud storage management module is used to uniformly store and efficiently manage standardized and cleaned geological data, build a distributed and scalable cloud storage architecture, and support the secure storage and fast retrieval of massive amounts of geological data.

[0008] The mineralization prediction model module is used to extract multi-dimensional features, mine association rules, and analyze mineralization elements from standardized geological data. It achieves efficient training and parameter optimization of the model based on distributed computing resources, and constructs a multi-scale, multi-element fusion mineralization prediction model by combining regional mineralization regularities and geological theoretical knowledge.

[0009] The collaborative training and update module supports distributed collaborative training across multiple institutions and regions. It uses federated learning technology to achieve model parameter sharing and joint optimization while protecting the data privacy of all participants. At the same time, based on the new geological data and the latest metallogenic theoretical research results continuously accessed by the platform, it automatically triggers incremental updates and iterative upgrades of the model to ensure that the metallogenic prediction model always maintains high accuracy and timeliness. In addition, it provides a model version management function to record model parameters and training logs at different stages, which facilitates users to review and compare the data, and supports the selection of historical models for prediction applications as needed.

[0010] The visualization and interaction module is used to present standardized geological data, intermediate results of mineralization prediction models, and final prediction results in a multi-dimensional and intuitive way.

[0011] The security module is used to ensure data security and stable system operation throughout the entire platform process.

[0012] Preferably, the multi-source heterogeneous geological data includes, but is not limited to, regional geological survey data, mineral exploration reports, geophysical exploration data, geochemical analysis data, remote sensing image data, and borehole logging data.

[0013] Preferably, the data access and standardization module includes:

[0014] The multi-source data access unit is used to access heterogeneous data from multiple channels, achieving full coverage access to various geological data sources and laying the foundation for subsequent standardized processing.

[0015] The data cleaning unit is used to identify and process missing values, outliers, and duplicate values ​​in multi-source heterogeneous geological data.

[0016] Data standardization units are used to standardize the format, encoding, units, and accuracy of multi-source heterogeneous geological data.

[0017] The data quality assessment unit is used to conduct multi-dimensional quality assessments of the cleaned and standardized geological data, and to feed back the assessment results to the data cleaning unit and the data standardization unit in real time, thereby optimizing the process.

[0018] Preferably, the cloud storage management module includes:

[0019] Storage units are used to build a geological data storage pool based on a distributed file system.

[0020] The data indexing and retrieval unit is used to establish spatial, attribute, and temporal indexes for geological data, enabling multi-dimensional and rapid retrieval of stored data and meeting the need for efficient data acquisition in mineralization prediction.

[0021] The data security and backup unit is used to encrypt and store the geological data, and to perform off-site disaster recovery backup and incremental backup on a regular basis.

[0022] The storage resource dynamic scheduling unit is used to monitor the usage of storage resources in real time, realize the tiered storage of hot and cold data, and improve the utilization efficiency of storage resources and the system response speed.

[0023] Preferably, the mineralization prediction model module includes:

[0024] The feature extraction unit is used to extract targeted features from various standardized geological data, enabling accurate extraction and quantification of features from different types of data.

[0025] The association rule mining unit is used to perform association analysis on the multi-dimensional geological features output by the feature extraction unit to mine the potential dependencies and combination rules among ore-forming elements.

[0026] The metallogenic element analysis unit is used to systematically sort out and deeply analyze metallogenic elements based on the multi-dimensional geological features output by the feature extraction unit and the core association rule set filtered by the association rule mining unit, combined with regional metallogenic theory and metallogenic models of typical deposits.

[0027] Preferably, the construction of the mineralization prediction model specifically includes the following steps:

[0028] Step 1: Model Architecture Selection: Based on the element weight system and association rule set output by the metallogenic element analysis unit, and combined with the regional metallogenic regularity and data characteristics, select a suitable model architecture.

[0029] Step 2, Training Dataset Construction: The standardized geological data, the multi-dimensional features output by the feature extraction unit, and the core association rule set selected by the association rule mining unit are divided into training set, validation set and test set in a ratio of 8:1:1.

[0030] Step 3, Distributed Collaborative Training: Utilize the distributed computing resources of the cloud platform to initiate a multi-organization, cross-regional collaborative training process;

[0031] Step 4, Model Validation and Evaluation: Use the validation set to evaluate the performance of the trained model. The core metrics include prediction accuracy, AUC value, confusion matrix, and spatial prediction accuracy.

[0032] Step 5: Model parameter optimization: Based on the evaluation results of the validation set, the model hyperparameters are adjusted using methods such as grid search and Bayesian optimization;

[0033] Step 6, Model Integration and Deployment: Encapsulate the optimized model into a callable service interface and integrate it into the mineralization prediction model module of the cloud platform.

[0034] Preferably, the collaborative training and update module includes:

[0035] The collaborative training unit is used to build a federated learning framework for multiple institutions and regions. In response to the privacy protection requirements of geological data, differential privacy and homomorphic encryption technologies are used to encrypt the transmission process of model parameters, so as to realize efficient collaborative training in heterogeneous devices and data environments, and finally output the jointly optimized global mineralization prediction model parameters.

[0036] The incremental update unit is used to monitor the quality and quantity of newly accessed geological data on the platform in real time, and automatically initiates the incremental update process of the model based on the preset update trigger threshold. It adopts transfer learning and fine-tuning technology to train the new data on the basis of the existing pre-trained model, retaining the original knowledge of the model while quickly adapting to the characteristics of the new data. It dynamically adjusts the feature weights and association rule thresholds of the model in combination with the latest metallogenic theoretical research results, ensuring that the model always keeps up with the latest geological research progress. It accelerates the incremental update process through distributed computing resources and reduces the time cost of model iteration.

[0037] The model version management unit is used to assign a unique version identifier to each metallogenic prediction model generated by training or incremental update, and associates and stores the model's training parameters, datasets used, evaluation metrics, and update log data to achieve redundant storage with multiple copies to ensure data reliability and ensure efficient transmission and access of geological data within the platform.

[0038] Preferably, the visual interaction module includes:

[0039] The basic data visualization unit is used to perform thematic visualization of standardized multi-source geological data;

[0040] The intermediate results visualization unit is used to present key intermediate results in the process of building the mineralization prediction model. Interactive charts help users understand the internal logic of the model and the relationship between mineralization elements.

[0041] The mineralization prediction results visualization unit is used to intuitively present the final mineralization prediction results. It uses a hierarchical color-coded spatial distribution map to show the mineralization probability level of different regions, marks the boundaries and numbers of favorable mineralization target areas, and supports clicking on the target area to view detailed attribute information. It also provides a comparative visualization function of prediction results of multiple model versions, which helps users evaluate the prediction differences and effects of different models by overlaying layers or displaying them side by side.

[0042] The 3D Geology and Prediction Visualization Unit is used to construct regional 3D geological models, integrate multi-source 3D data such as strata, structures, ore bodies, and boreholes, and realize the 3D spatial distribution of strata lithology, 3D reconstruction of ore body burial depth and morphology, and display of the 3D extension characteristics of tectonic zones.

[0043] Preferably, the security module includes: a data encryption unit, an access control unit, and a data protection unit; wherein, the data encryption unit uses end-to-end encryption technology to perform high-strength encryption on the transmitted and stored geologically sensitive data; the access control unit assigns refined operation permissions to users of different institutions and positions based on a role-based access control (RBAC) model; and the data protection unit combines privacy computing technology of federated learning to avoid direct exposure of raw data during model collaborative training.

[0044] Preferably, the steps include:

[0045] S1. Data Access and Standardization: By connecting to multiple data sources such as geological survey databases, geophysical exploration systems, and geochemical analysis platforms through interfaces, and using the unified data format of XML / JSON, the raw data is cleaned, deduplicated, filled with missing values, and converted to a coordinate system to form a standardized geological dataset.

[0046] S2. Distributed cloud storage management: Utilizing a combination of distributed file systems and relational databases, it enables secure storage and efficient retrieval of large-scale geological data, supporting hierarchical access and data sharing for users across regions.

[0047] S3. Construction of intelligent mineralization prediction model: Based on deep learning frameworks (such as TensorFlow and PyTorch), the model integrates convolutional neural networks (CNN) to extract spatial geological features and recurrent neural networks (RNN) to analyze time series evolution patterns, and constructs a multimodal mineralization prediction model to conduct probability assessment of favorable mineralization areas.

[0048] S4. Collaborative Training and Dynamic Updates: Through the federated learning mechanism of the cloud platform, collaborative training of models across multiple institutional nodes is achieved, which avoids leakage of raw data while improving the generalization ability of the model; and scheduled tasks are set to automatically trigger model parameter updates based on newly added geological data to ensure the timeliness of prediction results.

[0049] S5. Visualized and interactive display: Using WebGIS technology, the prediction results are presented intuitively in the form of heat maps, 3D geological models, etc. It supports user-defined query conditions (such as mineralization type, regional range) and provides automatic generation of prediction reports.

[0050] S6. Security Assurance: The platform ensures the security of data and models through measures such as encrypted data transmission, access control, and log auditing.

[0051] Beneficial effects

[0052] Compared with the prior art, the present invention has the following advantages:

[0053] To achieve unified integration and standardized processing of multi-source geological data, break down data silos, significantly improve the utilization efficiency and sharing capability of geological data, and provide comprehensive and standardized basic data support for mineralization prediction;

[0054] By deeply integrating association rule mining with metallogenic element analysis, a high-precision metallogenic prediction model with multi-scale and multi-element fusion is constructed, which effectively captures regional metallogenic regularities and significantly improves the accuracy and reliability of metallogenic favorable areas prediction.

[0055] By employing federated learning and privacy-preserving computation technologies, cross-regional multi-institutional collaborative training is achieved while ensuring the privacy and security of the original data of each institution. This fully integrates scattered geological resources and professional knowledge, and promotes collaborative innovation in mineralization prediction technologies.

[0056] The full-process visualization and interactive function covers basic data, intermediate results and prediction results, intuitively presents geological features and metallogenic laws, helps geologists quickly understand the model logic, screen metallogenic target areas, and significantly improves decision-making efficiency.

[0057] Establishing a model incremental update and version management mechanism can quickly adapt to newly added data and cutting-edge mineralization theories, ensuring that the model continues to evolve and always maintains high timeliness and applicability;

[0058] A comprehensive security system, through technologies such as end-to-end encryption and refined access control, effectively safeguards the privacy, integrity, and compliance of geological data, providing a solid security barrier for collaborative business operations among multiple institutions. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the modules of the present invention;

[0060] Figure 2 This is a flowchart illustrating the construction process of the mineralization prediction model in this invention;

[0061] Figure 3 This is a flowchart of the method of the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Please see Figure 1-3 A geological big data mineralization prediction cloud platform method, comprising:

[0064] The data access and standardization module is used to access multi-source heterogeneous geological data, standardize the data, and clean the data.

[0065] The cloud storage management module is used to uniformly store and efficiently manage standardized and cleaned geological data, build a distributed and scalable cloud storage architecture, and support the secure storage and fast retrieval of massive amounts of geological data.

[0066] The mineralization prediction model module is used to extract multi-dimensional features, mine association rules, and analyze mineralization elements from standardized geological data. It achieves efficient training and parameter optimization of the model based on distributed computing resources, and constructs a multi-scale, multi-element fusion mineralization prediction model by combining regional mineralization regularities and geological theoretical knowledge.

[0067] The collaborative training and update module supports distributed collaborative training across multiple institutions and regions. It uses federated learning technology to achieve model parameter sharing and joint optimization while protecting the data privacy of all participants. At the same time, based on the new geological data and the latest metallogenic theoretical research results continuously accessed by the platform, it automatically triggers incremental updates and iterative upgrades of the model to ensure that the metallogenic prediction model always maintains high accuracy and timeliness. In addition, it provides a model version management function to record model parameters and training logs at different stages, which facilitates users to review and compare the data, and supports the selection of historical models for prediction applications as needed.

[0068] The visualization and interaction module is used to present standardized geological data, intermediate results of mineralization prediction models, and final prediction results in a multi-dimensional and intuitive way.

[0069] The security module is used to ensure data security and stable system operation throughout the entire platform process.

[0070] Specifically, multi-source heterogeneous geological data includes, but is not limited to, regional geological survey data, mineral exploration reports, geophysical exploration data, geochemical analysis data, remote sensing image data, and borehole logging data. The collection of these data provides a comprehensive and multi-dimensional foundational data source for subsequent data standardization processing, covering key dimensions required for mineralization prediction, such as geological background, mineral distribution, geophysical and chemical characteristics, remote sensing image features, and borehole geological information. This ensures that the data access and standardization modules can obtain complete original data materials, providing sufficient data support for the distributed storage and efficient management of the cloud storage management module. It also lays a solid data foundation for subsequent collaborative training and visualization interaction, improving the scientific rigor and accuracy of mineralization prediction on the entire cloud platform.

[0071] Specifically, the data access and standardization module includes:

[0072] The multi-source data access unit is used to access heterogeneous data from multiple channels, achieving full coverage access to various geological data sources and laying the foundation for subsequent standardized processing.

[0073] The data cleaning unit is used to identify and process missing values, outliers, and duplicate values ​​in multi-source heterogeneous geological data.

[0074] Data standardization units are used to standardize the format, encoding, units, and accuracy of multi-source heterogeneous geological data.

[0075] The data quality assessment unit is used to conduct multi-dimensional quality assessments on the cleaned and standardized geological data, and feeds back the assessment results to the data cleaning unit and the data standardization unit in real time to optimize the process and ensure that the output data fully meets the high-quality requirements of mineralization prediction and analysis.

[0076] Through the effective cooperation of the above-mentioned units, the data access and standardization module can efficiently and accurately complete the entire process of multi-source heterogeneous geological data processing and output high-quality standardized data that meets the requirements of mineralization prediction and analysis.

[0077] Specifically, the cloud storage management module includes:

[0078] Storage units are used to build a geological data storage pool based on a distributed file system.

[0079] The data indexing and retrieval unit is used to establish spatial, attribute, and temporal indexes for geological data, enabling multi-dimensional and rapid retrieval of stored data and meeting the need for efficient data acquisition in mineralization prediction.

[0080] The data security and backup unit is used to encrypt and store the geological data, and regularly perform off-site disaster recovery backups and incremental backups to ensure the security and recoverability of the data.

[0081] The storage resource dynamic scheduling unit is used to monitor the usage of storage resources in real time, realize the tiered storage of hot and cold data, and improve the utilization efficiency of storage resources and the system response speed.

[0082] Through the effective coordination of the aforementioned units, the cloud storage management module can achieve elastic storage and intelligent management of massive, multi-source standardized geological data. This ensures the security, integrity, and accessibility of the data, while significantly improving data retrieval efficiency and storage resource utilization through dynamic resource scheduling and multi-dimensional indexing technologies. It provides reliable data infrastructure support for the distributed computing of the mineralization prediction model module, the cross-institutional data sharing of the collaborative training and update module, and the real-time data access of the visualization interaction module, ensuring that the entire cloud platform can operate efficiently and stably when processing large-scale geological data.

[0083] Specifically, the mineralization prediction model module includes:

[0084] The feature extraction unit is used to extract targeted features from various standardized geological data, covering spatial structural features such as stratigraphy and lithology and tectonic zone distribution in regional geological survey data; attribute features such as mineralization point distribution and ore grade in mineral exploration reports; geophysical features such as gravity anomalies and magnetic anomalies in geophysical exploration data; geochemical features such as elemental anomaly combinations and background value to anomaly ratio in geochemical analysis data; remote sensing features such as linear structural density and alteration information extraction in remote sensing image data; and borehole features such as core lithology sequence, ore body thickness and burial depth in borehole logging data. It employs spatial analysis algorithms (such as buffer analysis and overlay analysis), statistical analysis methods (such as outlier detection and cluster analysis), and deep learning models (such as convolutional neural networks and attention mechanism models) to achieve accurate extraction and quantification of features from different types of data.

[0085] The association rule mining unit is used to perform association analysis on the multi-dimensional geological features output by the feature extraction unit, and to uncover the potential dependencies and combination patterns among ore-forming elements. It employs classical association rule algorithms (such as Apriori and FP-Growth) combined with spatial association mining techniques (such as the spatial association rule mining algorithm GSP and spatial co-occurrence pattern mining) to identify frequent itemsets and strong association rules among ore-related elements such as stratigraphic lithology, structural features, geophysical anomalies, geochemical anomalies, and remote sensing alteration information for geological spatial data at different scales. For example, it mines the association relationship between specific stratigraphic lithology and multi-element anomaly combinations, the spatial co-occurrence patterns of tectonic zone distribution and mineralization point distribution, and the synergistic occurrence patterns of magnetic and gravity anomalies. Simultaneously, it verifies and filters the mined association rules by incorporating regional metallogenic theory, eliminating redundant rules without geological significance and retaining the core association rule set that has guiding value for ore-forming prediction. The final output association rule set will serve as an important input to the ore-forming prediction model, providing crucial association basis for the model to identify favorable ore-forming areas.

[0086] The metallogenic element analysis unit is used to systematically sort out and deeply analyze metallogenic elements based on the multi-dimensional geological features output by the feature extraction unit and the core association rule set filtered by the association rule mining unit, combined with regional metallogenic theory and metallogenic models of typical deposits. Specifically, this includes: classifying ore-forming elements (such as ore-controlling structural elements, lithological elements, geochemical anomaly elements, geophysical anomaly elements, remote sensing alteration elements, etc.) to clarify the role mechanism of each element in the ore-forming process; using the Analytic Hierarchy Process (AHP), fuzzy comprehensive evaluation method, and machine learning feature importance assessment models (such as random forests and gradient boosting trees) to quantify the contribution of different ore-forming elements and construct an ore-forming element weight system; analyzing the spatial superposition effect and temporal evolution law of ore-forming elements to identify the element combination characteristics of favorable ore-forming areas (such as tectonic convergence zones and multi-element anomaly superposition zones, alteration zones and magnetic anomaly spatial coupling zones, etc.); and utilizing the distributed computing capabilities of cloud platforms to achieve parallel analysis and dynamic updating of ore-forming elements in multiple regions and at multiple scales, providing accurate element inputs and weight basis for the construction of ore-forming prediction models, and ensuring that the models can accurately capture the key laws of regional ore formation.

[0087] Through the effective cooperation of the above-mentioned units, the metallogenic prediction model module can integrate multi-dimensional geological features and core correlation rules, and combine regional metallogenic theory to construct a high-precision metallogenic prediction model that integrates multiple scales and multiple elements.

[0088] Specifically, the construction of a mineralization prediction model includes the following steps:

[0089] Step 1: Model Architecture Selection: Based on the element weight system and association rule set output by the metallogenic element analysis unit, and combined with the regional metallogenic regularity and data characteristics, select a suitable model architecture. For multi-element fusion scenarios, ensemble learning models (such as random forests and gradient boosting trees) or deep learning models (such as convolutional neural networks and graph neural networks) can be used. For scenarios with strong spatial correlation, prioritize models that support spatial feature modeling (such as spatial convolutional networks and geographic weighted regression models) to ensure that the model can effectively capture the spatial distribution and combination patterns of geological elements.

[0090] Step 2: Training Dataset Construction: The standardized geological data, multi-dimensional features output by the feature extraction unit, and core association rule set selected by the association rule mining unit are divided into training set, validation set, and test set in a ratio of 8:1:1. During the division process, the spatial heterogeneity of the data should be considered. A stratified sampling method should be used to retain the data distribution characteristics of different regions and different mineralization types to avoid sample bias affecting the model's generalization ability.

[0091] Step 3, Distributed Collaborative Training: Utilizing the distributed computing resources of the cloud platform, initiate a multi-institutional, cross-regional collaborative training process. Each participant performs local model training based on local data, and transmits model parameters to the central node using encrypted methods via federated learning. The central node aggregates the parameters and feeds them back to each participant, repeating the iteration until the model converges. During training, logs such as model loss values ​​and accuracy changes are recorded in real time to provide a basis for subsequent optimization.

[0092] Step 4, Model Validation and Evaluation: The trained model is evaluated using a validation set. Key metrics include prediction accuracy, AUC value, confusion matrix, and spatial prediction accuracy (such as Kappa coefficient). At the same time, geological experts are invited to verify the geological rationality of the prediction results, eliminating predictions that do not conform to regional metallogenic theories, and ensuring that the metallogenic favorable areas output by the model have practical geological significance.

[0093] Step 5: Model parameter optimization: Based on the evaluation results of the validation set, adjust the model hyperparameters (such as learning rate, regularization coefficient, and number of network layers) using methods such as grid search and Bayesian optimization; to address the overfitting problem, introduce dropout mechanism or data augmentation techniques (such as spatial rotation and noise addition) to improve the model's generalization ability; to address the underfitting problem, increase the feature dimension or adjust the model complexity until the model performance reaches the preset standard.

[0094] Step 6, Model Integration and Deployment: Encapsulate the optimized model into a callable service interface and integrate it into the mineralization prediction model module of the cloud platform.

[0095] The above steps ensure the construction of a high-precision mineralization prediction model that conforms to regional metallogenic laws and integrates multi-dimensional geological features and core correlation rules.

[0096] Specifically, the collaborative training and update module includes:

[0097] The collaborative training unit is used to build a multi-institutional, cross-regional federated learning framework, supporting two collaborative training modes: horizontal federation (different samples with the same features) and vertical federation (different features with the same samples). To address the privacy protection requirements of geological data, differential privacy and homomorphic encryption technologies are used to encrypt the transmission process of model parameters, ensuring that the original data of each participant is not leaked. This enables efficient collaborative training in heterogeneous equipment and data environments, and finally outputs the jointly optimized global mineralization prediction model parameters.

[0098] The incremental update unit is used to monitor the quality and quantity of newly added geological data on the platform in real time. It automatically initiates the incremental update process of the model based on preset update trigger thresholds (such as when the proportion of newly added data exceeds 10% or when key metallogenic element data is updated). It adopts transfer learning and fine-tuning techniques to train the new data on the existing pre-trained model, retaining the original knowledge of the model while quickly adapting to the characteristics of the new data. It dynamically adjusts the feature weights and association rule thresholds of the model in conjunction with the latest metallogenic theoretical research results (such as the discovery of new metallogenic models and key ore-controlling elements) to ensure that the model always keeps up with the latest geological research progress. It accelerates the incremental update process through distributed computing resources and reduces the time cost of model iteration.

[0099] The model version management unit is used to assign a unique version identifier to each metallogenic prediction model generated by training or incremental update, and associates and stores the model's training parameters, datasets used, evaluation metrics, and update log data to achieve redundant storage with multiple copies to ensure data reliability and ensure efficient transmission and access of geological data within the platform.

[0100] Through the effective coordination of the aforementioned units, the collaborative training and update module can achieve efficient collaborative training across multiple institutions and regions while ensuring data privacy and security, thus promoting the continuous iteration and optimization of the metallogenic prediction model. The incremental update mechanism quickly adapts to newly acquired geological data and cutting-edge metallogenic theoretical achievements, ensuring the model maintains high accuracy and timeliness. Simultaneously, with its comprehensive version management function, the module enables traceability and flexible access throughout the model's entire lifecycle, providing stable, reliable, and continuously evolving model support for the cloud platform's metallogenic prediction business, and assisting geologists in efficiently conducting metallogenic prediction research and practice.

[0101] Specifically, the visual interaction module includes:

[0102] The basic data visualization unit is used to thematically visualize standardized multi-source geological data, covering thematic maps of stratigraphic and lithological spatial distribution of regional geological survey data, gravity / magnetic anomaly contour maps of geophysical exploration data, elemental anomaly heat maps of geochemical analysis data, overlay maps of alteration information of remote sensing image data, and 3D profile maps of borehole trajectories of borehole logging data. It supports layered display and switching according to data type, spatial range, and time dimension, helping users intuitively grasp the regional geological background and data distribution characteristics.

[0103] The intermediate results visualization unit is used to present key intermediate results in the construction process of the mineralization prediction model, including bar charts of the importance of each mineralization element feature output by the feature extraction unit, a core association rule network diagram (nodes represent mineralization elements and edges represent association strength) after screening by the association rule mining unit, and element weight radar charts and spatial superposition effect heat maps generated by the mineralization element analysis unit. Interactive charts (such as scalable network diagrams and switchable weight comparison charts) help users understand the internal logic of the model and the relationship between mineralization elements.

[0104] The mineralization prediction results visualization unit is used to intuitively present the final mineralization prediction results. It uses a hierarchical color-coded spatial distribution map to display the mineralization probability level of different regions (such as high, medium and low levels, corresponding to different color gradients), marks the boundaries and numbers of favorable mineralization target areas, and supports clicking on the target area to view detailed attribute information (including mineralization probability value, combination of main mineralization control elements, association rule basis, etc.). At the same time, it provides a comparative visualization function of prediction results of multiple model versions, which helps users evaluate the prediction differences and effects of different models by overlaying layers or displaying them side by side.

[0105] The 3D Geology and Prediction Visualization Unit is used to construct regional 3D geological models, integrate multi-source 3D data such as strata, structures, ore bodies, and boreholes, and realize the 3D spatial distribution of stratigraphic lithology, 3D reconstruction of ore body burial depth and morphology, and display of 3D extension features of tectonic zones. It supports rotation, scaling, translation, and sectioning operations of 3D scenes (such as vertical sectioning to view geological structures at different underground depths), and overlays mineralization prediction results onto the 3D model to intuitively present the spatial distribution of underground mineralization favorable areas, providing decision support from a 3D perspective for exploration deployment.

[0106] The effective coordination of the above units enables the visualization and interaction module to provide users with full-process visualization support from basic data to prediction results, covering two-dimensional thematic displays and three-dimensional spatial reconstruction. Through interactive operation and multi-dimensional presentation, it helps geologists intuitively understand the characteristics of geological data, the internal logic of the model, and the spatial distribution patterns of mineralization prediction results.

[0107] Specifically, the security module includes: a data encryption unit, an access control unit, and a data protection unit. The data encryption unit uses end-to-end encryption technology to encrypt the transmitted and stored geologically sensitive data with high strength to prevent data leakage or tampering. The access control unit uses a role-based access control (RBAC) model to assign granular operating permissions to users in different institutions and positions, restricting unauthorized access to core data and models. The data protection unit combines privacy computing technology of federated learning to avoid direct exposure of raw data during model collaborative training and strictly adheres to data security regulations.

[0108] Through the effective coordination of the data encryption unit, access control unit, and data protection unit, the security module can comprehensively safeguard the privacy, integrity, and compliance of geological data, building a solid security barrier for data sharing, collaborative training, and mineralization prediction operations across multiple institutions and regions within the cloud platform.

[0109] Working principle:

[0110] Uploaded regional geological surveys, mineral exploration reports, geophysical / chemical exploration data, remote sensing images, borehole logging, and other multi-source data are cleaned, format converted, and standardized by the preprocessing unit to form a unified and standardized dataset, which is then stored in a distributed storage system.

[0111] The mineralization prediction model module sequentially activates the feature extraction unit, the association rule mining unit, and the mineralization element analysis unit to accurately extract multi-dimensional geological features, explore the potential correlations between mineralization elements, construct a mineralization element weight system, and then generate a high-precision mineralization prediction model through steps such as model architecture selection and distributed collaborative training.

[0112] The collaborative training and update module enables privacy-preserving collaborative training across multiple institutions and regions through a federated learning framework. It also performs incremental updates to the model based on new data and cutting-edge mineralization theories, and achieves full lifecycle traceability of the model through version management.

[0113] The visualization and interaction module presents basic data, intermediate results, and prediction results in a straightforward manner in the form of two-dimensional thematic maps and three-dimensional geological models, supporting interactive analysis and target area selection by users.

[0114] The security module safeguards data privacy and model security throughout the process, ensuring compliance of multi-institutional data sharing and collaborative training through end-to-end encryption, RBAC access control, and federated learning privacy computing technology.

[0115] Ultimately, the platform outputs a graded and colored spatial distribution map of mineralization probability and detailed attribute information of the target area, providing geologists with accurate mineralization prediction decision support and promoting the improvement of the efficiency and accuracy of mineral resource exploration.

[0116] Specifically, it includes the following steps:

[0117] S1. Data Access and Standardization: By connecting to multiple data sources such as geological survey databases, geophysical exploration systems, and geochemical analysis platforms through interfaces, and using the unified data format of XML / JSON, the raw data is cleaned, deduplicated, filled with missing values, and converted to a coordinate system to form a standardized geological dataset.

[0118] S2. Distributed cloud storage management: Utilizing a combination of distributed file systems and relational databases, it enables secure storage and efficient retrieval of large-scale geological data, supporting hierarchical access and data sharing for users across regions.

[0119] S3. Construction of intelligent mineralization prediction model: Based on deep learning frameworks (such as TensorFlow and PyTorch), the model integrates convolutional neural networks (CNN) to extract spatial geological features and recurrent neural networks (RNN) to analyze time series evolution patterns, and constructs a multimodal mineralization prediction model to conduct probability assessment of favorable mineralization areas.

[0120] S4. Collaborative Training and Dynamic Updates: Through the federated learning mechanism of the cloud platform, collaborative training of models across multiple institutional nodes is achieved, which avoids leakage of raw data while improving the generalization ability of the model; and scheduled tasks are set to automatically trigger model parameter updates based on newly added geological data to ensure the timeliness of prediction results.

[0121] S5. Visualized and interactive display: Using WebGIS technology, the prediction results are presented intuitively in the form of heat maps, 3D geological models, etc. It supports user-defined query conditions (such as mineralization type, regional range) and provides automatic generation of prediction reports.

[0122] S6. Security Assurance: Through measures such as encrypted data transmission, access control, and log auditing, the security of platform data and models is ensured. This method effectively solves problems such as difficulty in data integration, low prediction accuracy, and poor collaboration in existing technologies, providing an efficient and accurate mineralization prediction solution for the geological exploration field.

[0123] Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0124] 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.

[0125] 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 geological big data mineralization prediction cloud platform method, characterized in that, include: The data access and standardization module is used to access multi-source heterogeneous geological data, standardize the data, and clean the data. The cloud storage management module is used to uniformly store and efficiently manage standardized and cleaned geological data, build a distributed and scalable cloud storage architecture, and support the secure storage and fast retrieval of massive amounts of geological data. The mineralization prediction model module is used to extract multi-dimensional features, mine association rules, and analyze mineralization elements from standardized geological data. It achieves efficient training and parameter optimization of the model based on distributed computing resources, and constructs a multi-scale, multi-element fusion mineralization prediction model by combining regional mineralization regularities and geological theoretical knowledge. The collaborative training and update module supports distributed collaborative training across multiple institutions and regions. It uses federated learning technology to achieve model parameter sharing and joint optimization while protecting the data privacy of all participants. At the same time, based on the new geological data and the latest metallogenic theoretical research results continuously accessed by the platform, it automatically triggers incremental updates and iterative upgrades of the model to ensure that the metallogenic prediction model always maintains high accuracy and timeliness. In addition, it provides a model version management function to record model parameters and training logs at different stages, which facilitates users to review and compare the data, and supports the selection of historical models for prediction applications as needed. The visualization and interaction module is used to present standardized geological data, intermediate results of mineralization prediction models, and final prediction results in a multi-dimensional and intuitive way. The security module is used to ensure data security and stable system operation throughout the entire platform process.

2. The geological big data mineralization prediction cloud platform method according to claim 1, characterized in that, The multi-source heterogeneous geological data includes, but is not limited to, regional geological survey data, mineral exploration reports, geophysical exploration data, geochemical analysis data, remote sensing image data, and borehole logging data.

3. The geological big data mineralization prediction cloud platform method according to claim 1, characterized in that, The data access and standardization module includes: The multi-source data access unit is used to access heterogeneous data from multiple channels, achieving full coverage access to various geological data sources and laying the foundation for subsequent standardized processing. The data cleaning unit is used to identify and process missing values, outliers, and duplicate values ​​in multi-source heterogeneous geological data. Data standardization units are used to standardize the format, encoding, units, and accuracy of multi-source heterogeneous geological data. The data quality assessment unit is used to conduct multi-dimensional quality assessments of the cleaned and standardized geological data, and to feed back the assessment results to the data cleaning unit and the data standardization unit in real time, thereby optimizing the process.

4. The geological big data mineralization prediction cloud platform method according to claim 1, characterized in that, The cloud storage management module includes: Storage units are used to build a geological data storage pool based on a distributed file system. The data indexing and retrieval unit is used to establish spatial, attribute, and temporal indexes for geological data, enabling multi-dimensional and rapid retrieval of stored data and meeting the need for efficient data acquisition in mineralization prediction. The data security and backup unit is used to encrypt and store the geological data, and to perform off-site disaster recovery backup and incremental backup on a regular basis. The storage resource dynamic scheduling unit is used to monitor the usage of storage resources in real time, realize the tiered storage of hot and cold data, and improve the utilization efficiency of storage resources and the system response speed.

5. The geological big data mineralization prediction cloud platform method according to claim 1, characterized in that, The mineralization prediction model module includes: The feature extraction unit is used to extract targeted features from various standardized geological data, enabling accurate extraction and quantification of features from different types of data. The association rule mining unit is used to perform association analysis on the multi-dimensional geological features output by the feature extraction unit to mine the potential dependencies and combination rules among ore-forming elements. The metallogenic element analysis unit is used to systematically sort out and deeply analyze metallogenic elements based on the multi-dimensional geological features output by the feature extraction unit and the core association rule set filtered by the association rule mining unit, combined with regional metallogenic theory and metallogenic models of typical deposits.

6. The geological big data mineralization prediction cloud platform method according to claim 1, characterized in that, The construction of the mineralization prediction model specifically includes the following steps: Step 1: Model Architecture Selection: Based on the element weight system and association rule set output by the metallogenic element analysis unit, and combined with the regional metallogenic regularity and data characteristics, select a suitable model architecture. Step 2, Training Dataset Construction: The standardized geological data, the multi-dimensional features output by the feature extraction unit, and the core association rule set selected by the association rule mining unit are divided into training set, validation set and test set in a ratio of 8:1:

1. Step 3, Distributed Collaborative Training: Utilize the distributed computing resources of the cloud platform to initiate a multi-organization, cross-regional collaborative training process; Step 4, Model Validation and Evaluation: Use the validation set to evaluate the performance of the trained model. The core metrics include prediction accuracy, AUC value, confusion matrix, and spatial prediction accuracy. Step 5: Model parameter optimization: Based on the evaluation results of the validation set, the model hyperparameters are adjusted using methods such as grid search and Bayesian optimization; Step 6, Model Integration and Deployment: Encapsulate the optimized model into a callable service interface and integrate it into the mineralization prediction model module of the cloud platform.

7. The geological big data mineralization prediction cloud platform method according to claim 1, characterized in that, The collaborative training and update module includes: The collaborative training unit is used to build a federated learning framework for multiple institutions and regions. In response to the privacy protection requirements of geological data, differential privacy and homomorphic encryption technologies are used to encrypt the transmission process of model parameters, so as to realize efficient collaborative training in heterogeneous devices and data environments, and finally output the jointly optimized global mineralization prediction model parameters. The incremental update unit is used to monitor the quality and quantity of newly accessed geological data on the platform in real time, and automatically initiates the incremental update process of the model based on a preset update trigger threshold. It adopts transfer learning and fine-tuning techniques to train the new data on the existing pre-trained model, retaining the original knowledge of the model while quickly adapting to the characteristics of the new data. It dynamically adjusts the feature weights and association rule thresholds of the model in combination with the latest metallogenic theoretical research results, ensuring that the model always keeps up with the latest geological research progress. It accelerates the incremental update process through distributed computing resources, reducing the time cost of model iteration. The model version management unit is used to assign a unique version identifier to each metallogenic prediction model generated by training or incremental update, and associates and stores the model's training parameters, datasets used, evaluation metrics, and update log data to achieve redundant storage with multiple copies to ensure data reliability and ensure efficient transmission and access of geological data within the platform.

8. The geological big data mineralization prediction cloud platform method according to claim 1, characterized in that, The visual interaction module includes: The basic data visualization unit is used to perform thematic visualization of standardized multi-source geological data; The intermediate results visualization unit is used to present key intermediate results in the process of building the mineralization prediction model. Interactive charts help users understand the internal logic of the model and the relationship between mineralization elements. The mineralization prediction results visualization unit is used to intuitively present the final mineralization prediction results. It uses a hierarchical color-coded spatial distribution map to show the mineralization probability level of different regions, marks the boundaries and numbers of favorable mineralization target areas, and supports clicking on the target area to view detailed attribute information. It also provides a comparative visualization function for prediction results of multiple model versions, which helps users evaluate the prediction differences and effects of different models by overlaying layers or displaying them side by side. The 3D Geology and Prediction Visualization Unit is used to construct regional 3D geological models, integrate multi-source 3D data such as strata, structures, ore bodies, and boreholes, and realize the 3D spatial distribution of strata lithology, 3D reconstruction of ore body burial depth and morphology, and display of the 3D extension characteristics of tectonic zones.

9. The geological big data mineralization prediction cloud platform method according to claim 1, characterized in that, The security module includes a data encryption unit, an access control unit, and a data protection unit. The data encryption unit uses end-to-end encryption technology to encrypt the transmitted and stored geologically sensitive data with high strength. The access control unit assigns refined operation permissions to users of different institutions and positions based on a role-based access control (RBAC) model. The data protection unit combines privacy computing technology of federated learning to avoid direct exposure of raw data during model collaborative training.

10. A method for a geological big data mineralization prediction cloud platform according to any one of claims 1-9, characterized in that, Includes the following steps: S1. Data Access and Standardization: By connecting to multiple data sources such as geological survey databases, geophysical exploration systems, and geochemical analysis platforms through interfaces, and using the unified data format of XML / JSON, the raw data is cleaned, deduplicated, filled with missing values, and converted to a coordinate system to form a standardized geological dataset. S2. Distributed cloud storage management: Utilizing a combination of distributed file systems and relational databases, it enables secure storage and efficient retrieval of large-scale geological data, supporting hierarchical access and data sharing for users across regions. S3. Construction of intelligent mineralization prediction model: Based on deep learning frameworks (such as TensorFlow and PyTorch), the model integrates convolutional neural networks (CNN) to extract spatial geological features and recurrent neural networks (RNN) to analyze time series evolution patterns, and constructs a multimodal mineralization prediction model to conduct probability assessment of favorable mineralization areas. S4. Collaborative Training and Dynamic Updates: Through the federated learning mechanism of the cloud platform, collaborative training of models across multiple institutional nodes is achieved, which avoids leakage of raw data while improving the generalization ability of the model; and scheduled tasks are set to automatically trigger model parameter updates based on newly added geological data to ensure the timeliness of prediction results. S5. Visualized and interactive display: Using WebGIS technology, the prediction results are presented intuitively in the form of heat maps, 3D geological models, etc. It supports user-defined query conditions and provides an automatic prediction report generation function. S6. Security Assurance: The platform ensures the security of data and models through measures such as encrypted data transmission, access control, and log auditing.

Images