Machine learning-based intelligent mining control method and system

Abstract

This invention discloses a machine learning-based intelligent ore mining control method and system, belonging to the field of intelligent control technology in the mining industry. The method includes: constructing a geological model of the mining area; inputting the geological feature map of the ore body into the geological model for feature extraction, stability assessment, and comprehensive analysis to generate a geological analysis report; using a Bayesian update method, probabilistically fusing the geological feature map of the ore body and the geological analysis report to obtain a geomechanical map; combining the geomechanical map with real-time monitoring data of the mining area to analyze ore mining risk factors and dynamically adjust the ore mining strategy to generate optimized ore mining instructions. This invention helps improve the ability to finely characterize ore body stability, fracture-sensitive areas, and groundwater-affected areas, and enhances the spatial resolution, risk identification accuracy, and instruction generation targeting of the entire mining control process.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology in mining, and in particular to an intelligent ore mining control method and system based on machine learning. Background Technology

[0002] With the development of digital mining and intelligent mining technologies, the mining production process has gradually shifted from traditional experience-based decision-making to data-driven decision-making. Existing mine information systems typically acquire information on mine structure, ore body distribution, and environmental parameters through geological exploration, remote sensing mapping, sensor monitoring, and 3D modeling. These are then combined with graph databases, machine learning, geomechanical analysis, and industrial control technologies to identify and assess ore body stability, fracture evolution, groundwater activity, and mining status. Related technologies are continuously evolving towards integrated multi-source data fusion, spatial topology representation, dynamic risk perception, and adaptive control.

[0003] However, existing technologies in mining scenarios still generally suffer from insufficient coupling between geological environmental data, structural relationship information, and real-time monitoring information. Especially under complex mining conditions, the expression forms, spatial scales, and temporal characteristics of data from different sources vary significantly, making it difficult to form a unified characterization of ore body boundaries, fracture distribution, groundwater channels, and stability risks. Consequently, mining control strategies lack the ability to dynamically respond to the geomechanical state. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a machine learning-based intelligent mining control method for ore to solve the problem of the difficulty in unifying and integrating multi-source geological information and real-time monitoring data in mining and supporting dynamic mining decisions.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, this invention provides a machine learning-based intelligent ore mining control method, comprising: collecting geological environment data of the mining area, and performing data cleaning and noise removal to obtain high-quality mining area data; using spatial correlation analysis, transforming the high-quality mining area data into nodes and edges in a graph database, and identifying ore body boundaries, fracture location information, and groundwater flow paths to obtain a geological feature map of the ore body; constructing a geological model of the mining area, inputting the geological feature map of the ore body into the geological model for feature extraction, stability assessment, and comprehensive analysis to generate a geological analysis report; using a Bayesian update method, probabilistically fusing the geological feature map of the ore body and the geological analysis report to obtain a geomechanical map; combining the geomechanical map with real-time monitoring data of the mining area to analyze ore mining risk factors and dynamically adjust ore mining strategies to generate optimized ore mining instructions.

[0007] As a preferred embodiment of the machine learning-based intelligent ore mining control method of the present invention, the specific steps for collecting geological environment data of the mining area and performing data cleaning and noise removal to obtain high-quality mining area data are as follows: The geological environment data of the mining area includes the geological structure of the mining area, the distribution of ore bodies, and the environmental conditions of the mining area. By using data integrity checks, hash algorithms, and Z-scores to identify geological environment data in mining areas, incomplete, duplicate, and abnormal records are removed. Furthermore, K-means is used to identify and remove interfering data, resulting in high-quality mining area data.

[0008] As a preferred embodiment of the machine learning-based intelligent ore mining control method of the present invention, the step of transforming high-quality mining area data into nodes and edges in a graph database through spatial correlation analysis is as follows: Transform high-quality mining area data into three-dimensional spatial coordinates; Geological structure data of the mining area is extracted using spatial analysis tools, and each mineral layer is mapped to a node according to three-dimensional spatial coordinates; Based on ore body distribution data, the spatial distribution of ore bodies is obtained through remote sensing technology, and each ore body is mapped as an independent node. Spatial analysis methods are used to extract environmental condition data from the mining area and map it into independent nodes in a graph database; Spatial data analysis methods are used to analyze the spatial distance, relative position, and potential physical contact between nodes, and these spatial relationships are converted into edges in a graph database.

[0009] As a preferred embodiment of the machine learning-based intelligent ore mining control method of the present invention, the topological indices between nodes in the computational graph database are used to identify ore body boundaries, fracture location information, and groundwater flow paths to obtain a geological feature map of the ore body. The specific steps are as follows: Each node is connected to other nodes through edges to form a graph structure; Represent the connection relationship between each pair of nodes in the graph as a matrix to construct an adjacency matrix; By traversing each node and its adjacent nodes in the graph using the adjacency matrix, the number of connections between a node and other nodes is calculated to obtain the node degree. By analyzing the graph structure and using topology algorithms, the connection state between each node and its neighboring nodes is analyzed to obtain the clustering coefficient; The shortest path is expanded step by step using Dijkstra's algorithm, and the shortest distance from the starting node to all other nodes is calculated. Based on the clustering coefficient and node degree, the DBSCAN algorithm is used to identify the densely related regions between ore bodies and obtain the ore body boundaries. Based on the shortest path, identify abrupt changes in distance between nodes and ore body nodes, and mark the location information of cracks; By using shortest path analysis, potential channels for groundwater flow are revealed, and by combining the aggregation coefficient, the density of water flow at nodes is revealed, the main channels of groundwater flow are identified, and the groundwater flow path is obtained. Based on the ore body boundary, fracture location information, and groundwater flow path, the geometric shape of the ore body, fracture distribution, and the impact of groundwater flow on mining are extracted through spatial analysis and topological relationships, ultimately obtaining a geological feature map of the ore body.

[0010] As a preferred embodiment of the machine learning-based intelligent ore mining control method of the present invention, the specific steps of constructing a geological model of the mining area, inputting the geological feature map of the ore body into the geological model of the mining area for feature extraction, stability assessment and comprehensive analysis processing, and generating a geological analysis report are as follows: The convolutional neural network model is the basic model; Input layer: Historical geological feature map of ore bodies; Convolutional layers use convolution operations to extract spatial features; The pooling layer reduces the dimensionality of spatial features through max pooling operations; Fully connected layers connect the features extracted by convolutional and pooling layers, mapping the spatial features extracted from deep layers to the prediction task; Combining a multilayer perceptron model, the input layer receives the output of the fully connected layer; The MLP layer inputs the output of the fully connected layer into a single-layer MLP for further feature combination and transformation; The output layer outputs the final result; Finally, a geological model of the mining area was obtained; Input the geological feature map of the ore body into the geological model of the mining area to obtain the stability assessment value of the ore body; Using the stability assessment value of the ore body as the basic indicator, and combining the geometry of the ore body, the distribution of fractures and the flow of groundwater, a multi-factor analysis method is used to analyze and obtain the comprehensive stability risk of the ore body. Collect other geological feature data using remote sensing technology; The stability assessment values ​​of the ore body are integrated with other geological feature data and then visualized. A geological analysis report is generated based on the comprehensive stability risk and visualization results of the ore body.

[0011] As a preferred embodiment of the machine learning-based intelligent ore mining control method of the present invention, the step of probabilistically fusing the ore body geological feature map and the geological analysis report using the Bayesian update method to obtain a geomechanical map is as follows: Topological indices were extracted from the geological feature map of the ore body; Based on historical geological analysis reports, the joint probability distribution of various topological indicators is statistically analyzed to obtain prior probability values; Mechanical parameters were extracted from the geological analysis report and normalized. The normalized mechanical parameters in the geomechanical analysis report are converted into observational data, and a likelihood function is constructed. Calculate the posterior probability based on the prior probability and the likelihood function; The posterior probability is mapped onto the three-dimensional spatial coordinates of the ore body geological feature map. Based on the spatial location of each node and the corresponding posterior probability value, a discrete probability density distribution map is generated. Then, the discrete probability values ​​are spatially interpolated using Kriging interpolation technology to construct a continuous geomechanical map.

[0012] As a preferred embodiment of the machine learning-based intelligent ore mining control method of the present invention, the specific steps of combining geomechanical maps with real-time monitoring data of the mining area to analyze ore mining risk factors and dynamically adjust ore mining strategies to generate optimized ore mining instructions are as follows: Real-time monitoring data of the mining area is collected through sensors on the mining equipment; Synchronize and spatially align real-time monitoring data of the mining area; A spatial hash table is constructed to locate real-time monitoring data in the mining area by bucket. Within the target hash bucket and adjacent hash buckets, the nearest neighbor matching algorithm is used to perform spatial proximity search and node association mapping. This accurately projects each monitoring data to the corresponding position in the geomechanical map and extracts the corresponding ore mining risk factors. By using fuzzy logic to quantitatively analyze various ore mining risk factors, the current mining risk value of the mining area can be obtained and the ore mining strategy can be adjusted in real time. Based on the adjusted ore mining strategy, optimized ore mining instructions are generated through a rule engine and then issued via the MQTT protocol.

[0013] Secondly, this invention provides a machine learning-based intelligent ore mining control system, comprising: a data acquisition module, a data analysis module, a report generation module, a data fusion module, and a mining strategy adjustment module; the data acquisition module is used to collect geological environment data of the mining area, and perform data cleaning and noise removal to obtain high-quality mining area data; the data analysis module is used to transform the high-quality mining area data into nodes and edges in a graph database through spatial correlation analysis, and identify ore body boundaries, fracture location information, and groundwater flow paths to obtain a geological feature map of the ore body; the report generation module is used to construct a geological model of the mining area, input the geological feature map of the ore body into the geological model for feature extraction, stability assessment, and comprehensive analysis processing, and generate a geological analysis report; the data fusion module is used to probabilistically fuse the geological feature map of the ore body and the geological analysis report through Bayesian update method to obtain a geomechanical map; the mining strategy adjustment module is used to combine the geomechanical map with real-time monitoring data of the mining area, analyze ore mining risk factors, and dynamically adjust the ore mining strategy to generate optimized ore mining instructions.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the machine learning-based intelligent ore mining control method described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the machine learning-based intelligent ore mining control method as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By using the Bayesian update method to probabilistically fuse the geological feature map of the ore body with the geological analysis report and construct a continuous geomechanical map, the spatial structural characteristics, topological indices, and mechanical parameters of the ore body can be uniformly mapped to the same probabilistic expression framework. This enables the geomechanical state of different locations within the mining area to have quantifiable, correlated, and spatially traceable expressive capabilities. This step further generates continuous probability distribution results covering three-dimensional space, which helps to improve the ability to finely characterize the stability of the ore body, fracture-sensitive areas, and groundwater-affected areas, and enhances the spatial resolution capability, risk identification accuracy, and command generation targeting of the entire mining control process. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a machine learning-based intelligent ore mining control method.

[0019] Figure 2 This is a schematic diagram of a machine learning-based intelligent ore mining control system. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Reference Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides a machine learning-based intelligent ore mining control method, including the following steps: S1. Collect geological environment data of the mining area, and perform data cleaning and noise removal to obtain high-quality mining area data.

[0024] Geological environment data for the mining area includes the geological structure, ore body distribution, and environmental conditions of the mining area.

[0025] By using data integrity checks, hash algorithms, and Z-scores to identify geological environment data in mining areas, incomplete, duplicate, and abnormal records are removed. Furthermore, K-means is used to identify and remove interfering data, resulting in high-quality mining area data.

[0026] It should be noted that, firstly, a data integrity check is performed to identify and remove records lacking necessary information. Then, a hash algorithm is applied to detect duplicates and remove them from the mining area geological environment data. The Z-score method is used to identify and exclude abnormal records whose deviation from the average exceeds a set threshold. Next, the K-means clustering algorithm is used to analyze the preprocessed mining area geological environment data. By identifying data points that do not belong to the mining area geological environment data, interfering data is determined and removed. Finally, a clean, accurate, and representative high-quality mining area dataset is obtained.

[0027] S2. Through spatial correlation analysis, high-quality mining area data is transformed into nodes and edges in a graph database.

[0028] Transform high-quality mining area data into three-dimensional spatial coordinates.

[0029] It should be noted that by using appropriate spatial interpolation methods (such as Kriging interpolation, inverse distance weighting, etc.) based on high-quality mining area data with known sample points, the values ​​of unknown points are predicted, and a continuous geological surface model is constructed. This process takes into account the relative positional relationship between sample points and their attribute values ​​to estimate the attribute value of each location and position it in a three-dimensional spatial coordinate system, thereby realizing the transformation of high-quality mining area data from raw measurement values ​​to three-dimensional spatial coordinates.

[0030] Geological structure data of the mining area is extracted using spatial analysis tools, and each mineral layer is mapped to a node according to three-dimensional spatial coordinates.

[0031] It should be noted that, firstly, geological exploration data and Geographic Information System (GIS) technology are used to identify and extract the spatial extent and attribute information of each mineral layer; then, based on the three-dimensional spatial coordinates of each mineral layer, a specific algorithm is used to accurately map it into independent nodes in the graph database, where each node not only contains the specific coordinate location of the mineral layer, but also is associated with attributes such as mineral layer thickness and mineral type.

[0032] Based on ore body distribution data, the spatial distribution of ore bodies is obtained through remote sensing technology, and each ore body is mapped as an independent node.

[0033] It should be noted that, firstly, high-resolution images of the mining area are acquired using remote sensing technology, and the spatial distribution characteristics of each ore body are identified through image processing and interpretation techniques. Then, based on the three-dimensional spatial coordinates of each ore body obtained from the analysis, it is precisely mapped into independent nodes in the graph database. These nodes not only record the specific location information of the ore body, but also associate key attributes such as the ore body size and mineral composition.

[0034] Spatial analysis methods are used to extract environmental condition data from the mining area and map it into independent nodes in a graph database.

[0035] It should be noted that, firstly, Geographic Information System (GIS) technology is used to parse the data layer containing environmental information and extract key environmental parameters such as topography, vegetation cover, and hydrological features; then, based on the three-dimensional spatial coordinates of the key environmental parameters, a specific algorithm is used to accurately map each environmental feature point into an independent node in the graph database. Each node not only stores specific coordinate information but also associates with relevant environmental attribute data.

[0036] Spatial data analysis methods are used to analyze the spatial distance, relative position, and potential physical contact between nodes, and these spatial relationships are converted into edges in a graph database.

[0037] It should be noted that, firstly, the spatial distance and relative position between each node are calculated to identify whether there is any potential physical contact or proximity between them; then, based on these analysis results, a specific algorithm is used to determine the strength and type of the relationship between the nodes, and these spatial relationships are represented in the form of edges.

[0038] S3. Calculate the topological indicators between nodes in the graph database to identify ore body boundaries, fracture locations, and groundwater flow paths, and obtain a geological feature map of the ore body.

[0039] It should be noted that topological metrics refer to node degree, clustering coefficient, and shortest distance.

[0040] Each node is connected to other nodes through edges to form a graph structure.

[0041] Represent the connection relationship between each pair of nodes in the graph as a matrix to construct the adjacency matrix.

[0042] By traversing each node and its adjacent nodes in the graph using the adjacency matrix, the number of connections between a node and other nodes is calculated to obtain the node degree.

[0043] It should be noted that, firstly, a corresponding row and column are initialized for each node in the graph; then, each node and its adjacent nodes are traversed. If there is a direct connection between two nodes, a value (such as 1) is set in the corresponding position of the adjacency matrix, otherwise it is set to 0; next, the summation operation is performed on each row in the adjacency matrix to calculate the number of connections between each node and other nodes. This total number is the degree of the node.

[0044] By analyzing the graph structure and using topology algorithms, the connection status between each node and its neighboring nodes is analyzed to obtain the clustering coefficient.

[0045] It should be noted that, firstly, the connection status between each node and its neighboring nodes is identified through a topology algorithm. The specific steps are as follows: for a given node, determine the set of all its directly connected neighboring nodes; then calculate the ratio of the actual number of edges between these neighboring nodes to the theoretical maximum possible number of edges. This ratio is the clustering coefficient of the node.

[0046] The shortest path is expanded step by step using Dijkstra's algorithm, calculating the shortest distance from the starting node to all other nodes.

[0047] It should be noted that, firstly, the distance of the starting node is initialized to 0, and the distances of all other nodes are set to infinity, and all nodes are added to the set of unprocessed nodes; then, the unprocessed node with the smallest current distance is selected, and the shortest distances of all its neighboring nodes are updated. If the distance to a neighboring node through the current node is shorter than the known distance, the shortest distance of that neighboring node is updated; this process is repeated to gradually expand the shortest path tree until all nodes have been processed or are unreachable, and finally the shortest distance from the starting node to every node in the graph is obtained.

[0048] Based on the clustering coefficient and node degree, the DBSCAN algorithm is used to identify the densely related regions between ore bodies and obtain the ore body boundaries.

[0049] It should be noted that, firstly, the local density characteristics of each node in the graph are calculated, namely the clustering coefficient and node degree; then, the DBSCAN (density-based spatial clustering application) algorithm is used to identify relatively dense regions between ore bodies based on these density characteristics. Specifically, by setting appropriate distance thresholds and minimum point requirements, the DBSCAN algorithm can discover regions with sufficiently high density and group density-connected nodes into the same cluster, thereby distinguishing different ore body boundaries.

[0050] Based on the shortest path, abrupt changes in distance between the ore body nodes are identified, and crack location information is marked.

[0051] It should be noted that, firstly, the shortest path distance between nodes within the ore body is calculated using Dijkstra's algorithm or other shortest path algorithms; then, these path distance data are analyzed to find abnormally increased distance values, i.e., distance mutation points. Such mutations usually indicate discontinuities or voids in the geological structure; once a significant distance mutation is detected, it is marked as a potential fracture location.

[0052] By using shortest path analysis, potential channels for groundwater flow are revealed, and by combining the aggregation coefficient, the density of water flow at nodes is revealed, the main channels of groundwater flow are identified, and the groundwater flow path is obtained.

[0053] It should be noted that, firstly, Dijkstra's algorithm or other shortest path algorithms are used to calculate the shortest paths between nodes within the ore body to simulate possible groundwater flow channels; then, the density of water flow at each node is evaluated by combining the clustering coefficient. Nodes with higher clustering coefficients indicate that the area has strong connectivity and a high probability of water flow convergence; by analyzing the distribution of these nodes in the network, the main paths or trunk lines of groundwater flow can be identified, and the groundwater flow path can be obtained.

[0054] Based on the ore body boundary, fracture location information, and groundwater flow path, the geometric shape of the ore body, fracture distribution, and the impact of groundwater flow on mining are extracted through spatial analysis and topological relationships, ultimately obtaining a geological feature map of the ore body.

[0055] It should be noted that, firstly, the geometry of the ore body is accurately extracted using spatial analysis techniques, and the stability of the internal structure of the ore body is assessed in conjunction with topological relationships. Next, the impact of fracture location information on the stability of the ore body is analyzed, and key water flow channels that may affect mining operations and their relationship with the ore body structure are identified based on groundwater flow paths. A comprehensive assessment of the geometric characteristics of the ore body, fracture distribution, and groundwater activity is conducted to determine their specific impact on ore mining. Finally, all these analytical results are integrated to form a detailed geological feature map of the ore body.

[0056] S4. Construct a geological model of the mining area, input the geological feature map of the ore body into the geological model of the mining area for feature extraction, stability assessment and comprehensive analysis and processing, and generate a geological analysis report.

[0057] The convolutional neural network model is the basic model; Input layer: Historical geological feature map of ore bodies; Convolutional layers use convolution operations to extract spatial features; The pooling layer reduces the dimensionality of spatial features through max pooling operations; Fully connected layers connect the features extracted by convolutional and pooling layers, mapping the spatial features extracted from deep layers to the prediction task; Combining a multilayer perceptron model, the input layer receives the output of the fully connected layer; The MLP layer inputs the output of the fully connected layer into a single-layer MLP for further feature combination and transformation; The output layer outputs the final result; Finally, a geological model of the mining area was obtained.

[0058] It should be noted that, compared with existing technologies, the geological model of the mining area built based on convolutional neural networks (CNN) and support vector machines (SVM) can handle more complex geological data patterns. By automatically extracting deep spatial features through CNN, the problem of insufficient feature extraction in traditional methods is solved. At the same time, the combination of SVM for accurate classification and regression analysis of high-dimensional data improves the accuracy of prediction. The geological model of the mining area can not only identify the geometry of the ore body and the distribution of fractures, but also assess the impact of groundwater flow on mining, thereby providing more comprehensive and accurate risk assessment and mining strategy recommendations. This significantly improves the depth and breadth of geological analysis of the mining area and provides strong technical support for efficient and safe ore mining.

[0059] By inputting the geological feature map of the ore body into the geological model of the mining area, the stability assessment value of the ore body is obtained.

[0060] It should be noted that the specific expression for obtaining the ore body stability assessment value is as follows: ; in, This is the stability assessment value of the ore body. It is the feature extraction operation of the convolutional neural network model. It is a further transformation of convolutional features by a multilayer perceptron. This is the classification result of the SVM model on the final features. This is a geological feature map of the ore body. These are the weight coefficients for extracting features from the convolutional layer. They are determined by learning the contribution of convolutional features during the training process of the convolutional neural network on historical ore body geological feature maps, and by combining the results of backpropagation and loss function minimization. These are the weighting coefficients for the dimensionality reduction operation in the pooling layer. They are weighted based on the degree of influence of the dimensionality reduction features of the pooling layer on the ore body stability assessment results, combined with the importance of each pooling output feature during model training. These are the combined weight coefficients of each feature in the fully connected layer, determined by iterative optimization using training samples based on the contribution of each combined feature in the fully connected layer to the final classification or regression result. It is a function used in convolutional neural network models to extract spatial features of input data. It is a function in a multilayer perceptron used to further combine and transform the features extracted by the convolutional layers. It is a function used in the support vector machine model to perform the final classification of features processed by the multilayer perceptron.

[0061] Using the stability assessment value of the ore body as the basic indicator, and combining the geometry of the ore body, the distribution of fractures, and the flow of groundwater, a multi-factor analysis method is used to analyze and obtain the comprehensive stability risk of the ore body.

[0062] It should be noted that, firstly, the geometry of the ore body, the distribution of fractures, and the flow of groundwater are determined through geological exploration data. Then, a multi-factor analysis method is used to comprehensively consider the impact on the stability of the ore body. For example, the complexity of the geometry may increase the difficulty of mining, the distribution of fractures indicates potential structural weaknesses, and the flow of groundwater may exacerbate internal erosion of the ore body or trigger landslide risks. Then, statistical analysis or numerical simulation techniques are used to quantify the contribution of each factor to the stability of the ore body, and finally, a comprehensive stability risk that reflects the ore body under the influence of various geological conditions is obtained.

[0063] Collect other geological feature data using remote sensing technology.

[0064] It should be noted that surface images and reflectance data of the mining area are acquired using sensors (such as multispectral scanners, radar, etc.) carried by satellites or aircraft. Then, image processing techniques are used for correction, enhancement, and classification to extract other geological feature data. For example, identifying surface vegetation cover to infer groundwater level or soil moisture, analyzing the spectral characteristics of rocks and minerals to determine their type and distribution, monitoring surface deformation to assess potential geological hazard risks, or observing topographic changes to understand erosion and deposition patterns.

[0065] The stability assessment values ​​of the ore body are integrated with other geological feature data and then visualized.

[0066] It should be noted that, firstly, the collected data (such as the geometry of the ore body, fracture distribution, and groundwater flow paths) are standardized to ensure they can be effectively displayed on the same platform. Next, professional Geographic Information System (GIS) software or advanced visualization tools are used to overlay these data, and the spatial distribution of ore body stability assessment values ​​and other geological features is presented intuitively through color coding, contour lines, and 3D models. Finally, interactive charts and maps are created, allowing users to explore the relationships between different geological parameters and their impact on ore body stability through simple operations.

[0067] A geological analysis report is generated based on the comprehensive stability risk and visualization results of the ore body.

[0068] It should be noted that, firstly, all relevant geological data and analysis results should be compiled, including the geometry of the ore body, fracture distribution, groundwater flow paths, and their impact on the stability of the ore body; then, using this information in conjunction with the comprehensive stability risk value derived from the risk assessment model, a detailed explanation and discussion should be conducted to identify potential risk areas and key influencing factors; next, the data analysis results should be presented using visualization tools such as charts, images, and 3D models to make complex information easy to understand; finally, a geological analysis report should be prepared.

[0069] S5. Using the Bayesian update method, the geological feature map of the ore body and the geological analysis report are probabilistically fused to obtain the geomechanical map.

[0070] Topological indices were extracted from the geological feature map of the ore body.

[0071] Based on historical geological analysis reports, the joint probability distribution of various topological indicators is statistically analyzed to obtain prior probability values.

[0072] It should be noted that the data of various topological indicators (such as node degree, clustering coefficient, etc.) are statistically analyzed, and then the joint probability distribution of various topological indicators is calculated to quantify the possibility of interaction between different indicators. The specific process includes sorting out relevant data from multiple historical reports, determining the value range and frequency of each topological indicator, and then analyzing the correlation or independence between the topological indicators. Through this series of steps, a prior probability value is calculated for each combination of topological indicators.

[0073] Mechanical parameters were extracted from the geological analysis report and normalized.

[0074] It should be noted that, firstly, geomechanical parameters related to the report, such as elastic modulus and Poisson's ratio, are identified and extracted; then, these extracted geomechanical parameters are cleaned and formatted to ensure data consistency and accuracy; finally, normalization techniques are applied to convert mechanical parameters of different scales or magnitudes to a common standard range (e.g., between 0 and 1).

[0075] The normalized mechanical parameters in the geomechanical analysis report are converted into observational data, and a likelihood function is constructed.

[0076] It should be noted that, firstly, specific observational indicators related to the mechanical parameters and geological conditions are identified, such as formation pressure and rock strength, and their values ​​are mapped to actual physical observations or experimental test results. Then, based on these observational indicators and known probability distribution models (such as normal distribution, Poisson distribution, etc.), a likelihood function is constructed. This function quantifies the probability of observing the current geological state under given mechanical parameters.

[0077] The posterior probability is calculated based on the prior probability and the likelihood function.

[0078] It should be noted that, firstly, Bayes' theorem is used to combine prior probability with the likelihood function. The specific process is as follows: For a given mechanical parameter, its prior probability distribution is first determined based on historical data, representing a probability estimate of the mechanical parameter before new observation data is available; then, a likelihood function is constructed using current observation data, reflecting the probability of observing this data under specific parameter conditions; finally, Bayes' formula is applied to calculate the posterior probability.

[0079] The posterior probability is mapped onto the three-dimensional spatial coordinates of the ore body geological feature map. Based on the spatial location of each node and the corresponding posterior probability value, a discrete probability density distribution map is generated. Then, the discrete probability values ​​are spatially interpolated using Kriging interpolation technology to construct a continuous geomechanical map.

[0080] It should be noted that, firstly, a discrete probability density distribution map is generated based on the spatial location of each node and its corresponding posterior probability value. This map shows the probability distribution of mechanical parameters in different regions. Then, Kriging interpolation is used to spatially interpolate these discrete probability values. By considering the positional relationship between nodes and the weight of known data points, the probability value of unsampled points is predicted, thereby constructing a continuous geomechanical map.

[0081] S6. Combine geomechanical maps with real-time monitoring data of the mining area to analyze ore mining risk factors and dynamically adjust ore mining strategies to generate optimized ore mining instructions.

[0082] Real-time monitoring data of the mining area is collected through sensors on the mining equipment.

[0083] It should be noted that the real-time monitoring data of the mining area includes key information from multiple aspects, such as geological conditions (rock hardness, ore body stability), environmental parameters (temperature, humidity, air quality and groundwater level), equipment status (vibration, wear degree, operating efficiency) and safety indicators (ground pressure change, slope stability, gas concentration).

[0084] Real-time monitoring data of the mining area is synchronized in time and aligned spatially.

[0085] It should be noted that the data acquisition time from different sensors is calibrated by a time synchronization algorithm to ensure that all data are aligned on the time axis and to eliminate errors caused by differences in acquisition time. Then, inverse distance weighted (IDW) interpolation is applied for spatial alignment. Based on the known location of the monitoring point and its recorded data value, the data value of the location not directly monitored is estimated. Specifically, points that are closer are given higher weights, making the spatial data distribution more continuous and smooth.

[0086] A spatial hash table is constructed to locate real-time monitoring data in the mining area by bucket. Within the target hash bucket and adjacent hash buckets, the nearest neighbor matching algorithm is used to perform spatial proximity search and node association mapping. This accurately projects each monitoring data point to the corresponding position in the geomechanical map and extracts the corresponding ore mining risk factors.

[0087] It should be noted that the mining area is divided into grids according to the three-dimensional spatial coordinate range in the geomechanical map, and the spatial position of each grid unit is encoded as a corresponding hash index to form a spatial hash table with a one-to-one correspondence between spatial position and hash index. The three-dimensional coordinate information in the real-time monitoring data of the mining area is read, and based on the spatial position encoding result corresponding to the three-dimensional coordinate information, the real-time monitoring data of the mining area is written into the target hash bucket under the corresponding hash index, thereby obtaining the bucket positioning result corresponding to each mining area's real-time monitoring data. The coordinate information of candidate nodes in the geomechanical map is extracted within the target hash bucket and adjacent hash buckets, and the spatial distance between each mining area's real-time monitoring data and the candidate nodes is calculated using the nearest neighbor matching algorithm. Node association mapping is completed according to the principle of minimum spatial distance, accurately projecting each mining area's real-time monitoring data to the corresponding spatial position in the geomechanical map. Based on the projected corresponding spatial position, relevant geological information in the geomechanical map is read, and combined with the real-time monitoring parameters in the mining area's real-time monitoring data, ore mining risk factors such as ground pressure changes, structural instability, and groundwater activity are extracted.

[0088] By using fuzzy logic to quantitatively analyze various ore mining risk factors, the current mining risk value of the mining area can be obtained and the ore mining strategy can be adjusted in real time.

[0089] It should be explained that the ore mining risk factors are mapped to input items in Fuzzy logic. Membership functions in Fuzzy logic are used to fuzzify these risk factors, transforming them into corresponding fuzzy membership values ​​to represent their degree of association with different risk levels. Fuzzy rules in Fuzzy logic are then used to reason about each risk factor, transforming the risk state under the combined effect of these factors into corresponding fuzzy reasoning results. These results are then defuzzified, converting them into comparable quantitative results to obtain the current mining risk value of the mining area. Based on the risk level corresponding to this current risk value, the ore mining strategy is adjusted in real time to obtain the adjusted ore mining strategy.

[0090] It should also be noted that Fuzzy logic is a method for dealing with fuzziness, uncertainty, and problems that are difficult to define precisely. Fuzzy logic differs from the absolute judgment of "yes" or "no" in traditional binary logic. Fuzzy logic uses membership degree to represent the degree to which an object belongs to a certain state. For example, it can represent that a certain risk state has both "lower" and "higher" attributes. Therefore, Fuzzy logic is suitable for describing objects that are continuously changing, have unclear boundaries, or are difficult to classify using fixed thresholds. Fuzzy logic can transform influencing factors that are difficult to quantify directly and precisely into risk results that can be reasoned and compared, thereby enabling a comprehensive judgment of complex objects. In the process of ore mining risk analysis, fuzzy logic can fuzzify, reason by rules, and defuzzify ore mining risk elements, transforming the combined effect of multiple risk factors into the current mining risk value of the mining area, thus providing a basis for real-time adjustment of ore mining strategies.

[0091] Based on the adjusted ore mining strategy, optimized ore mining instructions are generated through a rule engine and then issued via the MQTT protocol.

[0092] It should be noted that, firstly, the key parameters and objectives in the mining strategy (such as reducing risk and improving efficiency) are transformed into specific logical rules. Next, the rule engine automatically derives the optimal set of operating instructions based on current geological conditions, equipment status, and real-time monitoring data. These instructions include, but are not limited to, adjusting equipment operating parameters, planning mining paths, and implementing safety measures. Finally, the rule engine outputs detailed optimized mining instructions and, utilizing the efficient communication capabilities of the MQTT protocol, sends these optimized instructions in real time to various mining equipment and control systems at the mining site.

[0093] This embodiment also provides a machine learning-based intelligent ore mining control system, including: a data acquisition module, a data analysis module, a report generation module, a data fusion module, and a mining strategy adjustment module; The data acquisition module is used to collect geological and environmental data of the mining area, and to perform data cleaning and noise removal to obtain high-quality mining area data. The data analysis module is used to transform high-quality mining area data into nodes and edges in a graph database through spatial correlation analysis methods, and to identify ore body boundaries, fracture location information and groundwater flow paths to obtain geological feature maps of the ore body. The report generation module is used to construct a geological model of the mining area. The geological feature map of the ore body is input into the geological model of the mining area for feature extraction, stability assessment and comprehensive analysis and processing, and to generate a geological analysis report. The data fusion module is used to probabilistically fuse the geological feature map of the ore body and the geological analysis report using the Bayesian update method to obtain the geomechanical map. The mining strategy adjustment module combines geomechanical maps with real-time monitoring data of the mining area to analyze ore mining risk factors and dynamically adjust the ore mining strategy to generate optimized ore mining instructions.

[0094] This embodiment also provides a computer device applicable to the intelligent ore mining control method based on machine learning, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the intelligent ore mining control method based on machine learning as proposed in the above embodiment. The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0095] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the machine learning-based intelligent ore mining control method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0096] In summary, this invention utilizes a Bayesian update method to probabilistically fuse orebody geological feature maps with geological analysis reports, constructing a continuous geomechanical atlas. This unifies the mapping of orebody spatial structural features, topological indices, and mechanical parameters onto a single probabilistic expression framework, enabling quantifiable, correlateable, and spatially traceable representations of the geomechanical states at different locations within the mining area. This step further generates a continuous probability distribution covering three-dimensional space, enhancing the ability to finely characterize orebody stability, fracture-sensitive areas, and groundwater-affected areas. It also strengthens the spatial resolution capabilities, risk identification accuracy, and targeted command generation throughout the entire mining control process.

[0097] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A machine learning-based intelligent ore mining control method, characterized in that, include: Collect geological environment data of the mining area, and perform data cleaning and noise removal to obtain high-quality mining area data; By using spatial correlation analysis, high-quality mining area data is transformed into nodes and edges in a graph database, and the boundaries of the ore body, the location of fractures, and the path of groundwater flow are identified to obtain a geological feature map of the ore body. A geological model of the mining area is constructed. Geological feature maps of the ore bodies are input into the geological model of the mining area for feature extraction, stability assessment and comprehensive analysis and processing, and a geological analysis report is generated. By using the Bayesian update method, the geological feature map of the ore body and the geological analysis report are probabilistically fused to obtain the geomechanical map. By combining geomechanical maps with real-time monitoring data of the mining area, risk factors in ore mining are analyzed, and ore mining strategies are dynamically adjusted to generate optimized ore mining instructions.

2. The intelligent ore mining control method based on machine learning as described in claim 1, characterized in that, The specific steps for collecting geological environmental data of the mining area, and performing data cleaning and noise removal to obtain high-quality mining area data are as follows: The geological environment data of the mining area includes the geological structure of the mining area, the distribution of ore bodies, and the environmental conditions of the mining area. By using data integrity checks, hash algorithms, and Z-scores to identify geological environment data in mining areas, incomplete, duplicate, and abnormal records are removed. Furthermore, K-means is used to identify and remove interfering data, resulting in high-quality mining area data.

3. The intelligent ore mining control method based on machine learning as described in claim 2, characterized in that, The method of spatial correlation analysis transforms high-quality mining area data into nodes and edges in a graph database. The specific steps are as follows: Transform high-quality mining area data into three-dimensional spatial coordinates; Geological structure data of the mining area is extracted using spatial analysis tools, and each mineral layer is mapped to a node according to three-dimensional spatial coordinates; Based on ore body distribution data, the spatial distribution of ore bodies is obtained through remote sensing technology, and each ore body is mapped as an independent node. Spatial analysis methods are used to extract environmental condition data from the mining area and map it into independent nodes in a graph database; Spatial data analysis methods are used to analyze the spatial distance, relative position, and potential physical contact between nodes, and these spatial relationships are converted into edges in a graph database.

4. The intelligent ore mining control method based on machine learning as described in claim 3, characterized in that, The topological indices between nodes in the computational graph database are used to identify ore body boundaries, fracture locations, and groundwater flow paths, thereby obtaining a geological feature map of the ore body. The specific steps are as follows: Each node is connected to other nodes through edges to form a graph structure; Represent the connection relationship between each pair of nodes in the graph as a matrix to construct an adjacency matrix; By traversing each node and its adjacent nodes in the graph using the adjacency matrix, the number of connections between a node and other nodes is calculated to obtain the node degree. By analyzing the graph structure and using topology algorithms, the connection state between each node and its neighboring nodes is analyzed to obtain the clustering coefficient; The shortest path is expanded step by step using Dijkstra's algorithm, and the shortest distance from the starting node to all other nodes is calculated. Based on the clustering coefficient and node degree, the DBSCAN algorithm is used to identify the densely related regions between ore bodies and obtain the ore body boundaries. Based on the shortest path, identify abrupt changes in distance between nodes and ore body nodes, and mark the location information of cracks; By using shortest path analysis, potential channels for groundwater flow are revealed, and by combining the aggregation coefficient, the density of water flow at nodes is revealed, the main channels of groundwater flow are identified, and the groundwater flow path is obtained. Based on the ore body boundary, fracture location information, and groundwater flow path, the geometric shape of the ore body, fracture distribution, and the impact of groundwater flow on mining are extracted through spatial analysis and topological relationships, ultimately obtaining a geological feature map of the ore body.

5. The intelligent ore mining control method based on machine learning as described in claim 4, characterized in that, The construction of the geological model of the mining area involves inputting the geological feature map of the ore body into the geological model for feature extraction, stability assessment, and comprehensive analysis to generate a geological analysis report. The specific steps are as follows: The convolutional neural network model is the basic model; Input layer: Historical geological feature map of ore bodies; Convolutional layers use convolution operations to extract spatial features; The pooling layer reduces the dimensionality of spatial features through max pooling operations; Fully connected layers connect the features extracted by convolutional and pooling layers, mapping the spatial features extracted from deep layers to the prediction task; Combining a multilayer perceptron model, the input layer receives the output of the fully connected layer; The MLP layer inputs the output of the fully connected layer into a single-layer MLP for further feature combination and transformation; The output layer outputs the final result; Finally, a geological model of the mining area was obtained; Input the geological feature map of the ore body into the geological model of the mining area to obtain the stability assessment value of the ore body; Using the stability assessment value of the ore body as the basic indicator, and combining the geometry of the ore body, the distribution of fractures and the flow of groundwater, a multi-factor analysis method is used to analyze and obtain the comprehensive stability risk of the ore body. Collect other geological feature data using remote sensing technology; The stability assessment values ​​of the ore body are integrated with other geological feature data and then visualized. A geological analysis report is generated based on the comprehensive stability risk and visualization results of the ore body.

6. The intelligent ore mining control method based on machine learning as described in claim 5, characterized in that, The process involves probabilistically fusing the geological feature map of the ore body and the geological analysis report using the Bayesian update method to obtain a geomechanical map. The specific steps are as follows: Topological indices were extracted from the geological feature map of the ore body; Based on historical geological analysis reports, the joint probability distribution of various topological indicators is statistically analyzed to obtain prior probability values; Mechanical parameters were extracted from the geological analysis report and normalized. The normalized mechanical parameters in the geomechanical analysis report are converted into observational data, and a likelihood function is constructed. Calculate the posterior probability based on the prior probability and the likelihood function; The posterior probability is mapped onto the three-dimensional spatial coordinates of the ore body geological feature map. Based on the spatial location of each node and the corresponding posterior probability value, a discrete probability density distribution map is generated. Then, the discrete probability values ​​are spatially interpolated using Kriging interpolation technology to construct a continuous geomechanical map.

7. The intelligent ore mining control method based on machine learning as described in claim 6, characterized in that, The process of combining geomechanical maps with real-time monitoring data of the mining area to analyze ore mining risk factors and dynamically adjust ore mining strategies to generate optimized ore mining instructions involves the following steps: Real-time monitoring data of the mining area is collected through sensors on the mining equipment; Synchronize and spatially align real-time monitoring data of the mining area; A spatial hash table is constructed to locate real-time monitoring data in the mining area by bucket. Within the target hash bucket and adjacent hash buckets, the nearest neighbor matching algorithm is used to perform spatial proximity search and node association mapping. This accurately projects each monitoring data to the corresponding position in the geomechanical map and extracts the corresponding ore mining risk factors. By using fuzzy logic to quantitatively analyze various ore mining risk factors, the current mining risk value of the mining area can be obtained and the ore mining strategy can be adjusted in real time. Based on the adjusted ore mining strategy, optimized ore mining instructions are generated through a rule engine and then issued via the MQTT protocol.

8. A machine learning-based intelligent ore mining control system, based on the machine learning-based intelligent ore mining control method according to any one of claims 1 to 7, characterized in that, include: Data acquisition module, data analysis module, report generation module, data fusion module, and mining strategy adjustment module; The data acquisition module is used to collect geological and environmental data of the mining area, and to perform data cleaning and noise removal to obtain high-quality mining area data. The data analysis module is used to transform high-quality mining area data into nodes and edges in a graph database through spatial correlation analysis methods, and to identify ore body boundaries, fracture location information and groundwater flow paths to obtain geological feature maps of the ore body. The report generation module is used to construct a geological model of the mining area. The geological feature map of the ore body is input into the geological model of the mining area for feature extraction, stability assessment and comprehensive analysis and processing, and to generate a geological analysis report. The data fusion module is used to probabilistically fuse the geological feature map of the ore body and the geological analysis report using the Bayesian update method to obtain the geomechanical map. The mining strategy adjustment module combines geomechanical maps with real-time monitoring data of the mining area to analyze ore mining risk factors and dynamically adjust the ore mining strategy to generate optimized ore mining instructions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the machine learning-based intelligent ore mining control method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the machine learning-based intelligent ore mining control method according to any one of claims 1 to 7.

Images