Mineral resource reserve intelligent management method and system based on big data

Abstract

The invention discloses an intelligent mineral resource reserve management method and system based on big data, and the method comprises the steps: obtaining a mineral resource data set, mining the potential correlation between different data modes in the mineral resource data set through a self-attention mechanism, and visualizing key features through an attention weight; based on the cross-modal mineral feature data, introducing a GNN graph neural network to be combined with an InSAR technology, constructing a dynamic three-dimensional geological map, and modeling an ore body, a fault and a rock stratum into graph nodes and edges to represent a spatial topological relation; and generating a reserve report and a mining plan according to the ore body space structure map, encrypting the reserve report and the mining plan by using a Hash algorithm, and uploading the encrypted reserve report and mining plan to a block chain network for storage. The actual situation of mineral resources can be reflected more accurately and comprehensively, and the resource mining efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of mineral resource management, and in particular to a method and system for intelligent management of mineral resource reserves based on big data. Background Art

[0002] Traditional mineral resource data management methods rely heavily on manual analysis and simple statistical models, making it difficult to process massive amounts of data, resulting in inaccurate and unreliable resource reserve calculations. Furthermore, multi-source information, including geological exploration data, mining and transportation data, and environmental and ecological data, is isolated from one another, lacking effective integration and correlation analysis. Furthermore, in terms of resource management, the storage and sharing of reserve reports and mining plans face data security risks, resulting in low traceability and trustworthiness. Summary of the Invention

[0003] The purpose of the present invention is to solve the above problems and to design a method and system for intelligent management of mineral resource reserves based on big data.

[0004] To achieve the above-mentioned purpose, the technical solution of the present invention is that, further, in the above-mentioned intelligent management method of mineral resource reserves based on big data, the intelligent management method of mineral resource reserves includes the following steps: Acquire multi-source heterogeneous data, clean and standardize them in sequence, and obtain a mineral resource dataset; The Transformer’s self-attention mechanism is used to mine potential correlations between different data modalities in the mineral resource dataset. The spatiotemporal Transformer branch is used to process time series data. Key features are visualized through attention weights to obtain cross-modal mineral feature data. Based on the cross-modal mineral feature data, the GNN graph neural network is introduced and combined with InSAR technology to construct a dynamic three-dimensional geological map. The ore bodies, faults and rock layers are modeled as graph nodes, and the edges represent the spatial topological relationships. The spatial structure map of the ore body is output; A reserve report and mining plan are generated based on the spatial structure map of the ore body, and the reserve report and mining plan are encrypted using a hash algorithm and uploaded to the blockchain network for storage.

[0005] Furthermore, in the above-mentioned intelligent management method for mineral resource reserves based on big data, the multi-source heterogeneous data is obtained, and the multi-source heterogeneous data is cleaned and standardized in sequence to obtain a mineral resource dataset, including: Acquire geological exploration data, remote sensing data, mining and transportation data, and environmental and ecological data to obtain multi-source heterogeneous data; The missing core sample parameters in the geological exploration data in the multi-source heterogeneous data are filled in according to the sample data of adjacent depths in the same borehole using the linear interpolation method; The spatiotemporal interpolation algorithm is used in combination with time series data to repair the missing pixel data in remote sensing data caused by cloud cover; The missing data in some time periods of mining and transportation data caused by equipment failure are filled by using the average value of adjacent time periods to obtain cleaned multi-source heterogeneous data.

[0006] Furthermore, in the above-mentioned intelligent management method for mineral resource reserves based on big data, the step of acquiring multi-source heterogeneous data, sequentially cleaning and standardizing the multi-source heterogeneous data to obtain a mineral resource dataset further includes: The wavelet denoising algorithm is used to perform multi-scale decomposition on the signal data in the cleaned multi-source heterogeneous data, remove the high-frequency noise components, and obtain the denoised multi-source heterogeneous data; Establish a data unit conversion dictionary to convert units in data from different sources into a unified standard unit; The outliers in the data were detected by combining the Z-score method and the IQR method, and the outliers were deleted to obtain the mineral resource data set.

[0007] Furthermore, in the above-mentioned intelligent management method for mineral resource reserves based on big data, the potential correlation between different data modalities in the mineral resource dataset is mined through the self-attention mechanism of the Transformer, including: A multimodal Transformer model is constructed, taking the mineral resource dataset as input. The data of each modality is converted into a vector representation of fixed dimension through an embedding layer. A dot product operation is performed on the Query vector of each modality, the Key vector and the Value vector of other modalities to obtain the attention score. After Softmax normalization, the attention weight is obtained, and then the fused feature vector is obtained by weighted summation.

[0008] Furthermore, in the above-mentioned intelligent management method for mineral resource reserves based on big data, the potential correlation between different data modalities in the mineral resource dataset is mined through the self-attention mechanism of the Transformer, including: The time series data is divided into time windows, and the data in each time window is input into the Transformer as a sequence. The temporal self-attention mechanism is used to capture the dependencies between different time points. Through attention weight visualization, the key features in the time and space dimensions are displayed in the form of heat maps to obtain cross-modal mineral feature data.

[0009] Furthermore, in the above-mentioned intelligent management method for mineral resource reserves based on big data, the cross-modal mineral feature data is based on the introduction of GNN graph neural network combined with InSAR technology, including: Ore bodies, faults, and rock formations are modeled as graph nodes, each of which contains rich attribute features, including at least the grade, reserves, and geometric parameters of the ore body, the strike, dip, and throw of the fault, and the lithology, thickness, and porosity of the rock formation. Edges represent the spatial topological relationships between nodes, including the adjacent relationship between ore bodies, the cutting relationship between ore bodies and faults, and the sequence relationship between rock strata. The surface deformation data obtained by InSAR is used as the dynamic attribute of the graph node, and the deformation information of the node is updated in real time. The InSAR data is combined with the GNN graph neural network, and the time dimension is introduced into the graph model to output the spatial structure map of the ore body.

[0010] Furthermore, in the above-mentioned intelligent management method for mineral resource reserves based on big data, generating a reserve report and a mining plan according to the ore body spatial structure map, encrypting the reserve report and the mining plan using a hash algorithm, and uploading them to the blockchain network for storage, includes: The SHA-256 secure hash algorithm is used to perform hash operations on the generated reserve report and mining plan to generate a unique hash value. The encrypted reserve report and mining plan are uploaded to the blockchain network in the form of transactions, and the blockchain consensus mechanism is used to ensure the legitimacy and non-tamperability of the transaction.

[0011] Furthermore, in the intelligent management system for mineral resource reserves based on big data, the intelligent management system for mineral resource reserves includes the following modules: The data acquisition and processing module is used to acquire multi-source heterogeneous data, clean and standardize the multi-source heterogeneous data in sequence, and obtain the mineral resource data set; A cross-modal feature extraction module is used to mine potential correlations between different data modalities in the mineral resource dataset through the self-attention mechanism of the Transformer, process time series data using the spatiotemporal Transformer branch, and visualize key features through attention weights to obtain cross-modal mineral feature data; A feature fusion processing module is used to construct a dynamic three-dimensional geological map based on the cross-modal mineral feature data by introducing a GNN graph neural network and combining it with InSAR technology. The ore bodies, faults, and rock layers are modeled as graph nodes, with edges representing spatial topological relationships, and the spatial structure map of the ore bodies is output. The mineral resource management module is used to generate a reserve report and mining plan based on the spatial structure map of the ore body, encrypt the reserve report and mining plan using a hash algorithm, and upload them to the blockchain network for storage.

[0012] Furthermore, in the intelligent management system for mineral resource reserves based on big data, the cross-modal feature extraction module includes the following submodules: Construct a submodule for building a multimodal Transformer model, which takes the mineral resource dataset as input and converts the data of each modality into a fixed-dimensional vector representation through an embedding layer; A sub-module is obtained, which is used to perform dot product operations on the Query vector of each modality, the Key vector and the Value vector of other modalities to obtain the attention score. After Softmax normalization, the attention weight is obtained, and then the fused feature vector is obtained by weighted summation.

[0013] Furthermore, in the intelligent management system for mineral resource reserves based on big data, the cross-modal feature extraction module includes the following submodules: The input submodule is used to divide the time series data into time windows. The data in each time window is input into the Transformer as a sequence, and the temporal self-attention mechanism is used to capture the dependencies between different time points. The display submodule is used to display the key features in the time and space dimensions in the form of heat maps through attention weight visualization to obtain cross-modal mineral feature data.

[0014] Its beneficial effects are: 1. It can more accurately and comprehensively reflect the actual situation of mineral resources, providing rich and reliable data support for resource reserve assessment and mining decisions. 2. The reserve reports and mining plans generated based on the ore body spatial structure map and cross-modal mineral characteristic data are more scientific and accurate. The reserve calculation takes into account multiple factors, and the resource quality analysis and development and utilization condition assessment are more comprehensive. The mining plan can be reasonably planned according to the actual situation of the ore body and market demand, thereby improving resource mining efficiency. 3. Ensure that the data of the reserve report and mining plan is secure, tamper-proof and traceable. While protecting corporate commercial confidentiality, it realizes the trusted sharing of data between different departments and regulatory agencies, and improves the transparency and collaborative efficiency of mineral resource management. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Various other advantages and benefits will become apparent to those skilled in the art by reading the following detailed description of the preferred embodiment.The accompanying drawings are only for the purpose of illustrating the preferred embodiment and are not to be considered as limiting the present invention.

[0016] Figure 1 Schematic diagram of a first embodiment of a method for intelligent management of mineral resource reserves based on big data in an embodiment of the present invention; Figure 2 Schematic diagram of a second embodiment of a method for intelligent management of mineral resource reserves based on big data in an embodiment of the present invention; Figure 3This is a schematic diagram of the first embodiment of the intelligent management system for mineral resource reserves based on big data in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", "" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0019] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 As shown, the intelligent management method of mineral resource reserves based on big data includes the following steps: Step 101: Acquire multi-source heterogeneous data, clean and standardize the multi-source heterogeneous data in sequence, and obtain a mineral resource dataset; Specifically, in this embodiment, geological exploration data, remote sensing data, mining and transportation data, and environmental and ecological data are acquired to obtain multi-source heterogeneous data; The missing core sample parameters in the geological exploration data in the multi-source heterogeneous data are filled in according to the sample data of adjacent depths in the same borehole using the linear interpolation method; The spatiotemporal interpolation algorithm is used in combination with time series data to repair the missing pixel data in remote sensing data caused by cloud cover; The missing data in some time periods of mining and transportation data caused by equipment failure are filled by using the average value of adjacent time periods to obtain cleaned multi-source heterogeneous data.

[0020] Acquire geological exploration data, remote sensing data, mining and transportation data, and environmental and ecological data to obtain multi-source heterogeneous data; The missing core sample parameters in the geological exploration data in the multi-source heterogeneous data are filled in according to the sample data of adjacent depths in the same borehole using the linear interpolation method; The spatiotemporal interpolation algorithm is used in combination with time series data to repair the missing pixel data in remote sensing data caused by cloud cover; The missing data in some time periods of mining and transportation data caused by equipment failure are filled by using the average value of adjacent time periods to obtain cleaned multi-source heterogeneous data.

[0021] Specifically, (1) Geological exploration data Geological exploration data is obtained through a variety of professional methods, as follows: Drilling data: Different types of drilling rigs (rotary drills, percussion drills) are used to carry out drilling operations in the planned exploration area. The drilling depth is determined according to the exploration objectives and geological conditions.

[0022] Geophysical data Gravity exploration: Use high-precision gravimeters to measure the gravity field strength at various points on the surface. By analyzing gravity anomalies, we can infer the density differences of underground geological structures and identify possible geological bodies such as ore bodies and faults.

[0023] Magnetic prospecting: Use a magnetic meter to measure the intensity and distribution characteristics of the surface magnetic field, and based on the magnetic differences of different rocks and minerals, delineate the scope of the ore body containing magnetic minerals and the geological structure morphology.

[0024] Electrical exploration: including resistivity method, induced polarization method, etc.

[0025] Geochemical data: Soil, rock fragments, stream sediments and other samples are collected on the surface in grids or profiles. The content of chemical elements such as copper, iron, gold and silver in the samples is determined using analytical techniques such as atomic absorption spectroscopy and inductively coupled plasma mass spectrometry.

[0026] (2) Remote sensing data Optical remote sensing data: Spectral data of different wavelengths (visible light, near-infrared, short-wave infrared, etc.) reflected by the Earth's surface are obtained using optical sensors carried by multispectral and hyperspectral satellites (Landsat, Sentinel-2, Hyperion, etc.) and drones.

[0027] Synthetic Aperture Radar (SAR) remote sensing data: SAR satellites (Sentinel-1, Radarsat, etc.) acquire microwave backscatter signals from the Earth's surface. These signals are unrestricted by weather and day / night conditions, can penetrate clouds and some vegetation, and obtain information on surface topography, landforms, and surface deformation. Interferometric SAR (InSAR) technology can monitor subtle surface deformations in mining areas, providing data support for geological disaster warnings and mining subsidence monitoring.

[0028] (3) Mining and transportation data Mining data: Various Internet of Things (IoT) devices, such as sensors, cameras, and RFID tags, are deployed at mining sites. Pressure sensors, speed sensors, and temperature sensors are installed on mining equipment such as rock drills, crushers, and loaders.

[0029] Transportation Data: GPS positioning devices and weighing sensors are installed on ore transport vehicles (trucks, trains) to obtain real-time data such as the vehicle's location, route, speed, and load. At mining ports or railway freight stations, weighing equipment such as floor scales and belt scales are used to record the amount of ore loaded, unloaded, and transported.

[0030] (IV) Environmental and ecological data Atmospheric environmental data: Air quality monitoring stations are deployed in and around the mining area to monitor the concentrations of pollutants such as particulate matter (PM2.5, PM10), sulfur dioxide (SO2), nitrogen oxides (NOx), carbon monoxide (CO), and ozone (O3) in the atmosphere.

[0031] Water environment data: Water quality monitoring sensors are installed in rivers, lakes, groundwater wells and other water bodies within the mining area to monitor water quality indicators such as water temperature, pH value, dissolved oxygen (DO), conductivity, chemical oxygen demand (COD), and ammonia nitrogen (NH3-N) in real time.

[0032] Soil environmental data: Soil samples are collected in and around the mining area according to a certain grid, and the soil pH, organic matter content, heavy metal (lead, cadmium, mercury, arsenic, etc.) content, soil fertility and other indicators are analyzed.

[0033] Ecological data: Through a combination of drone aerial photography and ground surveys, ecological data such as vegetation coverage, vegetation type, and biodiversity (plant species, animal populations) in and around the mining area are obtained.

[0034] 2. Data Processing (1) Data cleaning Data missing processing: For missing core sample parameters (porosity, permeability) in geological exploration data, linear interpolation or cubic spline interpolation is used to fill in the missing data based on sample data at adjacent depths of the same borehole; for missing pixel data in remote sensing data caused by cloud cover, spatiotemporal interpolation algorithms (based on mean interpolation of adjacent pixels, Kriging interpolation) are used in combination with time series data to repair the missing data; for missing data in some time periods caused by equipment failure in mining and transportation data, the average or median of the adjacent time periods is used to fill in the missing data.

[0035] Noise data processing: For the noise in geological exploration data (electromagnetic interference noise in geophysical data), the wavelet noise reduction algorithm is used to perform multi-scale decomposition of the signal, remove high-frequency noise components, and retain the effective signal; for salt and pepper noise and Gaussian noise in remote sensing image data, spatial filtering methods such as median filtering and Gaussian filtering are used for smoothing; for random noise in data collected by mining equipment sensors, sliding window mean filtering or Kalman filtering algorithm is used for noise reduction.

[0036] Outlier detection and correction: A combination of the Z-score and IQR methods is used to detect outliers in various data types. For abnormal grade values ​​in geological exploration data, the Z-score method is first used to initially screen out data points that deviate from the mean by more than three standard deviations. This is then further confirmed using the IQR method (interquartile range). Data identified as outliers is corrected or eliminated based on geostatistical principles and adjacent sample data. For abnormal load data in mining and transportation data (values ​​that significantly exceed the vehicle's rated load), the data is compared with historical data and equipment parameters of the transport vehicle to determine whether it is a sensor failure or human error. If a failure is identified, the data is corrected; if an error is identified, the data is eliminated.

[0037] (2) Data standardization Data format unification: Develop unified data storage format standards. For example, for numerical data, uniformly adopt floating-point or integer formats, and specify the number of digits after the decimal point. For text data, uniformly adopt UTF-8 encoding, standardize the spelling and abbreviations of geological terms, mineral names, equipment models, etc.; for time series data, uniformly adopt the ISO8601 time format ("YYYY-MM-DDHH:MM:SS").

[0038] Data unit conversion: Build a data unit conversion dictionary to convert various units in data from different sources into a unified standard unit. For example, convert pressure units (atm, bar, kPa) in geological exploration data to Pa, convert temperature units (°C, °F) in remote sensing data to K, and convert weight units (tons, kilograms, pounds) in mining and transportation data to kg.

[0039] Data normalization and standardization: Different normalization methods are used for data of different modalities to keep the data in the same numerical range, facilitating subsequent data analysis and model training. For continuous parameters in geological exploration data (grade, porosity), the minimum-maximum normalization method is used to map the data to the [0, 1] interval. For categorical data (lithology, mineral species), one-hot encoding or label encoding methods are used for conversion. For time series data, the Z-score normalization method is used to ensure that the data has zero mean and unit variance.

[0040] Step 102: Use the Transformer’s self-attention mechanism to mine potential correlations between different data modalities in the mineral resource dataset, use the spatiotemporal Transformer branch to process time series data, and visualize key features through attention weights to obtain cross-modal mineral feature data. Specifically, in this embodiment, a multimodal Transformer model is constructed, which takes the mineral resource dataset as input, and converts the data of each modality into a vector representation of fixed dimension through an embedding layer; A dot product operation is performed on the Query vector of each modality, the Key vector and the Value vector of other modalities to obtain the attention score. After Softmax normalization, the attention weight is obtained, and then the fused feature vector is obtained by weighted summation.

[0041] The time series data is divided into time windows, and the data in each time window is input into the Transformer as a sequence. The temporal self-attention mechanism is used to capture the dependencies between different time points. Through attention weight visualization, the key features in the time and space dimensions are displayed in the form of heat maps to obtain cross-modal mineral feature data.

[0042] Specifically, 1. Transformer self-attention mechanism to mine cross-modal associations A multimodal Transformer model is constructed, taking preprocessed geological exploration data, remote sensing data, mining and transportation data, and environmental and ecological data as input. The data for each modality is first converted into a fixed-dimensional vector representation through an embedding layer. In the Transformer encoder, a self-attention mechanism is used to calculate attention weights between data of different modalities, thereby exploring potential connections between different data modalities. Specifically, for each modality's query vector (Query), a dot product operation is performed with the key vector (Key) and value vector (Value) of the other modalities to obtain an attention score. This is then normalized using Softmax to obtain the attention weight, and then a weighted sum is performed to obtain the fused feature vector. By stacking multiple layers of Transformer encoders, the complex nonlinear connections between data of different modalities are continuously captured.

[0043] (2) Spatiotemporal Transformer branch processes time series data A spatiotemporal Transformer branch is designed for time series data such as mining and transportation data, and environmental and ecological data. In the temporal dimension, time series data is divided into time windows, and the data within each time window is input into the Transformer as a sequence. A temporal self-attention mechanism is used to capture dependencies between different time points, such as trends in mining output over time and seasonal fluctuations in environmental indicators. In the spatial dimension, spatial location encoding is incorporated into the Transformer model, combining the geographic coordinates of the mining area. This allows the model to simultaneously consider both temporal and spatial characteristics of the data. For example, for surface deformation monitoring data, the spatiotemporal Transformer can capture deformation characteristics at different locations at different time points and their mutual influence. Using attention weight visualization technology, key features in the temporal and spatial dimensions are displayed in the form of heat maps or matrices, allowing analysts to intuitively identify key time points and key areas that have a significant impact on mineral resource reserve management.

[0044] Step 103: Based on the cross-modal mineral feature data, a GNN graph neural network is introduced and combined with InSAR technology to construct a dynamic three-dimensional geological map. The ore bodies, faults, and rock layers are modeled as graph nodes, and the edges represent the spatial topological relationships. The spatial structure map of the ore body is output; Specifically, in this embodiment, ore bodies, faults, and rock formations are modeled as graph nodes, each of which contains rich attribute features, including at least the grade, reserves, and geometric parameters of the ore body, the strike, dip, and throw of the fault, and the lithology, thickness, and porosity of the rock formation. Edges represent the spatial topological relationships between nodes, including the adjacent relationship between ore bodies, the cutting relationship between ore bodies and faults, and the sequence relationship between rock strata. The surface deformation data obtained by InSAR is used as the dynamic attribute of the graph node, and the deformation information of the node is updated in real time. The InSAR data is combined with the GNN graph neural network, and the time dimension is introduced into the graph model to output the spatial structure map of the ore body.

[0045] Specifically, (1) Combining GNN graph neural network and InSAR technology Graph node definition: Ore bodies, faults, and rock formations are modeled as graph nodes. Each node contains a rich set of attribute features, such as the grade, reserves, and geometric parameters (length, width, and height) of the ore body; the strike, dip, and throw of the fault; and the lithology, thickness, and porosity of the rock formation. These attribute features are derived through the previous data processing and feature extraction steps.

[0046] Edge construction: Edges represent the spatial topological relationships between nodes, including the adjacency between ore bodies, the cutting relationship between ore bodies and faults, and the stratigraphic relationship between rock strata. Edge connectivity is determined by calculating the spatial distance and geometric positional relationships (intersection, inclusion, parallelism, etc.) between nodes. Edges are then assigned corresponding weights, which can represent the inverse of spatial distance or the closeness of spatial relationships.

[0047] InSAR technology applications: Surface deformation data acquired by InSAR is used as dynamic attributes of graph nodes, updating their deformation information in real time. By combining InSAR data with GNN graph neural networks and introducing the time dimension into the graph model, a dynamic graph model is constructed that can reflect the spatial morphological changes and interactions of ore bodies, faults, and rock strata over time.

[0048] (2) Construction of dynamic three-dimensional geological maps Based on the graph structure and node attributes constructed by the GNN (Geo-Native Neural Network), combined with 3D geological modeling techniques (3D mesh modeling and solid modeling) of the mining area, the spatial positions and attribute characteristics of the graph nodes are mapped into 3D space, generating a dynamic 3D geological map. In this 3D geological map, ore bodies, faults, and rock strata are visualized using different colors and shapes, and the edge connections clearly represent the spatial topology. By updating node attribute data in real time (e.g., changes in ore body reserves during mining and shifts in fault position caused by surface deformation), the 3D geological map is dynamically updated, intuitively demonstrating the dynamic evolution of the mining area's geological structure. The final output is a spatial structural map of the ore body, clearly showing its spatial distribution, morphological characteristics, and relationship to surrounding geological bodies.

[0049] Step 104: Generate a reserve report and mining plan based on the ore body spatial structure map, encrypt the reserve report and mining plan using a hash algorithm, and upload them to the blockchain network for storage.

[0050] Specifically, in this embodiment, the SHA-256 secure hash algorithm is used to perform a hash operation on the generated reserve report and mining plan to generate a unique hash value. The encrypted reserve report and mining plan are uploaded to the blockchain network in the form of a transaction, and the legitimacy and non-tamperability of the transaction are ensured through the consensus mechanism of the blockchain.

[0051] Specifically, (1) Reserves report generation Generate a detailed reserve report based on the ore body spatial structure map and the cross-modal mineral characteristic data obtained through mining. The reserve report includes: Overview of mineral resources: geographical location, scope, geological structure background, mineral types and distribution characteristics of the mining area.

[0052] Reserve calculation: Use appropriate reserve calculation methods (geological block method, cross-section method, 3D modeling method) to calculate the reserves of different types of mineral resources (proven reserves, controlled reserves, inferred reserves) based on parameters such as the spatial form, grade distribution and thickness of the ore body, and list the basis, formula and results of the reserve calculation.

[0053] Resource quality analysis: Evaluate the quality of mineral resources, including the grade of major useful elements, the content of impurity elements, the physical properties of the ore (hardness, selectivity), etc., and analyze the impact of resource quality on mining and processing.

[0054] Evaluation of development and utilization conditions: Evaluate the mining technical conditions (burial depth, inclination, hydrogeological conditions), engineering geological conditions (rock stability, fault development), environmental conditions (air, water, soil environmental quality) and economic conditions (mining costs, market demand, economic benefit forecast) of the mining area.

[0055] (2) Formulation of mining plan Based on the reserve report and the actual situation of the mining area, a scientific and reasonable mining plan is formulated, including: Mining sequence: Determine the mining sequence of the ore bodies, such as mining shallow ore bodies first and then deep ore bodies, mining rich ore bodies first and then poor ore bodies, to ensure the rational use of resources and mining safety.

[0056] Mining process: Select the appropriate mining process (open-pit mining, underground mining), and determine the process parameters and equipment selection for each link such as rock drilling, blasting, loading, and transportation based on the geological conditions of the ore body and mining technical requirements.

[0057] Production scale: Determine reasonable annual mining volume and production service life based on mineral resource reserves, market demand and the company's production capacity to ensure the feasibility and economy of the mining plan.

[0058] Safety and Environmental Protection Measures: Develop detailed safety and environmental protection measures, including mine safety management regulations, dust and noise control plans, wastewater and waste gas treatment measures, land reclamation and ecological restoration plans, etc., to ensure that the mining process complies with national and local safety and environmental protection standards.

[0059] (3) Hash algorithm encryption and blockchain storage Hash algorithm encryption: Utilizing secure hash algorithms such as SHA-256, we hash the generated reserve reports and mining plans to generate a unique hash value. This hash value is then bound to the original file to ensure file integrity and immutability. Furthermore, sensitive data (such as mineral resource grade, specific reserve values, and mining costs) is symmetric encrypted using the AES algorithm, while the encryption key is transmitted and stored using asymmetric encryption using the RSA algorithm.

[0060] Blockchain network storage: Select a blockchain platform (consortium chain) suitable for the mining industry and upload encrypted reserve reports and mining plans to the blockchain network as transactions. Each transaction includes the transaction time, data summary (hash value), uploader information, and other information. The blockchain's consensus mechanism (PBFT, PoS) ensures the legitimacy and immutability of transactions. Once stored on the blockchain, data can only be accessed by authorized parties, ensuring secure data sharing and traceability, providing reliable technical support for the management and regulation of mineral resources.

[0061] Its beneficial effects are: 1. It can more accurately and comprehensively reflect the actual situation of mineral resources, providing rich and reliable data support for resource reserve assessment and mining decisions. 2. The reserve reports and mining plans generated based on the ore body spatial structure map and cross-modal mineral characteristic data are more scientific and accurate. The reserve calculation takes into account multiple factors, and the resource quality analysis and development and utilization condition assessment are more comprehensive. The mining plan can be reasonably planned according to the actual situation of the ore body and market demand, thereby improving resource mining efficiency. 3. Ensure that the data of the reserve report and mining plan is secure, tamper-proof and traceable. While protecting corporate commercial confidentiality, it realizes the trusted sharing of data between different departments and regulatory agencies, and improves the transparency and collaborative efficiency of mineral resource management.

[0062] See also Figure 2 In the intelligent management method of mineral resource reserves based on big data, obtaining multi-source heterogeneous data, cleaning and standardizing the multi-source heterogeneous data in sequence, and obtaining a mineral resource dataset include the following steps: Step 201: Acquire geological exploration data, remote sensing data, mining and transportation data, and environmental and ecological data to obtain multi-source heterogeneous data; Step 202: Fill missing core sample parameters in the geological exploration data in the multi-source heterogeneous data with sample data at adjacent depths of the same borehole using a linear interpolation method; Step 203: Using a spatiotemporal interpolation algorithm in combination with time series data to repair missing pixel data in the remote sensing data due to cloud cover; Step 204: Fill in the missing data in some time periods of the mining and transportation data due to equipment failure using the average value of adjacent time periods to obtain cleaned multi-source heterogeneous data.

[0063] The above is an introduction to the embodiment of the intelligent management method of mineral resources reserves based on big data of the present invention. Figure 3 In the intelligent management system of mineral resources reserves based on big data, the intelligent management system of mineral resources reserves includes the following modules: The data acquisition and processing module is used to acquire multi-source heterogeneous data, clean and standardize the multi-source heterogeneous data in sequence, and obtain the mineral resource data set; The cross-modal feature extraction module is used to mine potential correlations between different data modalities in the mineral resource dataset through the self-attention mechanism of the Transformer. It uses the spatiotemporal Transformer branch to process time series data and visualizes key features through attention weights to obtain cross-modal mineral feature data. The feature fusion processing module is used to construct a dynamic three-dimensional geological map based on cross-modal mineral feature data by introducing the GNN graph neural network and combining it with InSAR technology. The ore bodies, faults, and rock layers are modeled as graph nodes, with edges representing spatial topological relationships, and the spatial structure map of the ore bodies is output. The mineral resource management module is used to generate reserve reports and mining plans based on the spatial structure map of the ore body, encrypt the reserve reports and mining plans using a hash algorithm, and then upload them to the blockchain network for storage.

[0064] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent management method for mineral resource reserves based on big data, characterized in that: The method for intelligent management of mineral resource reserves comprises the following steps: Acquire multi-source heterogeneous data, clean and standardize them in sequence, and obtain a mineral resource dataset; The Transformer's self-attention mechanism is used to mine the potential correlations between different data modalities in the mineral resource dataset. The spatiotemporal Transformer branch is used to process time series data. The key features are visualized through attention weights to obtain cross-modal mineral feature data. Based on the cross-modal mineral feature data, the GNN graph neural network is introduced and combined with InSAR technology to construct a dynamic three-dimensional geological map. The ore bodies, faults and rock layers are modeled as graph nodes, and the edges represent the spatial topological relationship. The spatial structure map of the ore body is output; A reserve report and mining plan are generated based on the spatial structure map of the ore body, and the reserve report and mining plan are encrypted using a hash algorithm and uploaded to the blockchain network for storage.

2. The method for intelligent management of mineral resource reserves based on big data according to claim 1, characterized in that: The method of obtaining multi-source heterogeneous data and sequentially cleaning and standardizing the multi-source heterogeneous data to obtain a mineral resource dataset includes: Acquire geological exploration data, remote sensing data, mining and transportation data, and environmental and ecological data to obtain multi-source heterogeneous data; The missing core sample parameters in the geological exploration data in the multi-source heterogeneous data are filled in according to the sample data of adjacent depths in the same borehole using the linear interpolation method; The spatiotemporal interpolation algorithm is used in combination with time series data to repair the missing pixel data in remote sensing data caused by cloud cover; The missing data in some time periods of mining and transportation data caused by equipment failure are filled by using the average value of adjacent time periods to obtain cleaned multi-source heterogeneous data.

3. The method for intelligent management of mineral resource reserves based on big data according to claim 2, characterized in that: The method of acquiring multi-source heterogeneous data, cleaning and standardizing the multi-source heterogeneous data in sequence to obtain a mineral resource dataset further includes: The wavelet denoising algorithm is used to perform multi-scale decomposition on the signal data in the cleaned multi-source heterogeneous data, remove the high-frequency noise components, and obtain the denoised multi-source heterogeneous data; Establish a data unit conversion dictionary to convert units in data from different sources into a unified standard unit; The outliers in the data were detected by combining the Z-score method and the IQR method, and the outliers were deleted to obtain the mineral resource data set.

4. The method for intelligent management of mineral resource reserves based on big data according to claim 1, characterized in that: The Transformer self-attention mechanism is used to mine the potential correlations between different data modalities in the mineral resource dataset, including: A multimodal Transformer model is constructed, taking the mineral resource dataset as input. The data of each modality is converted into a vector representation of fixed dimension through an embedding layer. A dot product operation is performed on the Query vector of each modality, the Key vector and the Value vector of other modalities to obtain the attention score. After Softmax normalization, the attention weight is obtained, and then the fused feature vector is obtained by weighted summation.

5. The method for intelligent management of mineral resource reserves based on big data according to claim 1, characterized in that: The Transformer self-attention mechanism is used to mine the potential correlations between different data modalities in the mineral resource dataset, including: The time series data is divided into time windows, and the data in each time window is input into the Transformer as a sequence. The temporal self-attention mechanism is used to capture the dependencies between different time points. Through attention weight visualization, the key features in the time and space dimensions are displayed in the form of heat maps to obtain cross-modal mineral feature data.

6. The method for intelligent management of mineral resource reserves based on big data according to claim 1, characterized in that: Based on the cross-modal mineral feature data, the GNN graph neural network is introduced and combined with InSAR technology, including: Ore bodies, faults, and rock formations are modeled as graph nodes, each of which contains rich attribute features, including at least the grade, reserves, and geometric parameters of the ore body, the strike, dip, and throw of the fault, and the lithology, thickness, and porosity of the rock formation. Edges represent the spatial topological relationships between nodes, including the adjacent relationship between ore bodies, the cutting relationship between ore bodies and faults, and the sequence relationship between rock strata. The surface deformation data obtained by InSAR is used as the dynamic attribute of the graph node, and the deformation information of the node is updated in real time. The InSAR data is combined with the GNN graph neural network, and the time dimension is introduced into the graph model to output the spatial structure map of the ore body.

7. The method for intelligent management of mineral resource reserves based on big data according to claim 1, characterized in that: The method of generating a reserve report and a mining plan based on the ore body spatial structure map, encrypting the reserve report and the mining plan using a hash algorithm, and uploading them to the blockchain network for storage includes: The SHA-256 secure hash algorithm is used to perform hash operations on the generated reserve report and mining plan to generate a unique hash value. The encrypted reserve report and mining plan are uploaded to the blockchain network in the form of transactions, and the blockchain consensus mechanism is used to ensure the legitimacy and non-tamperability of the transaction.

8. The intelligent management system for mineral resource reserves based on big data is characterized by: The mineral resource reserve intelligent management system includes the following modules: The data acquisition and processing module is used to acquire multi-source heterogeneous data, clean and standardize the multi-source heterogeneous data in sequence, and obtain the mineral resource data set; A cross-modal feature extraction module is used to mine potential correlations between different data modalities in the mineral resource dataset through the self-attention mechanism of the Transformer, process time series data using the spatiotemporal Transformer branch, and visualize key features through attention weights to obtain cross-modal mineral feature data; A feature fusion processing module is used to construct a dynamic three-dimensional geological map based on the cross-modal mineral feature data by introducing a GNN graph neural network and combining it with InSAR technology. The ore bodies, faults, and rock layers are modeled as graph nodes, with edges representing spatial topological relationships, and the spatial structure map of the ore bodies is output. The mineral resource management module is used to generate a reserve report and mining plan based on the spatial structure map of the ore body, encrypt the reserve report and mining plan using a hash algorithm, and upload them to the blockchain network for storage.

9. The intelligent management system for mineral resource reserves based on big data according to claim 8, characterized in that: The cross-modal feature extraction module Includes the following submodules: Construct a submodule for building a multimodal Transformer model, which takes the mineral resource dataset as input and converts the data of each modality into a fixed-dimensional vector representation through an embedding layer; A sub-module is obtained, which is used to perform dot product operations on the Query vector of each modality, the Key vector and the Value vector of other modalities to obtain the attention score. After Softmax normalization, the attention weight is obtained, and then the fused feature vector is obtained by weighted summation.

10. The intelligent management system for mineral resource reserves based on big data according to claim 8, characterized in that: The cross-modal feature extraction module includes the following sub-modules: The input submodule is used to divide the time series data into time windows. The data in each time window is input into the Transformer as a sequence, and the temporal self-attention mechanism is used to capture the dependencies between different time points. The display submodule is used to display the key features in the time and space dimensions in the form of heat maps through attention weight visualization to obtain cross-modal mineral feature data.

Images