Geologic body occurrence environment visual reconstruction method applied to continuous mining process of underground metal mine

Abstract

According to the geologic body occurrence environment visualization reconstruction method and system applied to the underground metal mine continuous mining process, an underground rock body environment holographic sensing method with multi-scale, multi-physical-quantity and multi-time-resolution information is fused, and a three-dimensional occurrence environment model capable of dynamically evolving in real time is constructed; and accurate guidance and risk prevention and control of the continuous mining process are realized by virtue of a visualization and intelligent feedback mechanism, so that powerful technical support is provided for safe, efficient and intelligent mining of the deep metal mine.

Description

Technical Field

[0001] The present application relates to the field of mining engineering technology, and in particular to a method and system for visually reconstructing the geological environment used in the continuous mining process of underground metal mines. Background Art

[0002] As my country's mineral resource development extends deeper into the earth, continuous mining of underground metal mines has become an important means of resource security. However, deep ore bodies are complex, often accompanied by unfavorable conditions such as high stress, high ground temperature, and high ground pressure. This significantly increases geological structural uncertainty and the risk of induced dynamic disasters (such as rockbursts, roof falls, and structural instability).

[0003] Traditional mining monitoring and environmental awareness methods rely heavily on single-point sensor deployment, empirical fault inference, and manual judgment, failing to meet the practical needs of accurately perceiving and dynamically understanding the deep three-dimensional geological environment. Some mines currently employ microseismic monitoring, geological radar, and borehole logging to detect structural features in mining areas. However, these technologies primarily operate at a local scale, with heterogeneous data types and dispersed sources, making unified modeling and fusion analysis difficult. Existing three-dimensional geological modeling methods often rely on static modeling, failing to reflect the dynamic evolution of geological structure and stress fields under mining disturbances. This makes them ill-suited for the real-time early warning and guidance required under continuous mining conditions. While three-dimensional visualization technology has seen some application in geosciences in recent years, it often focuses on display rather than perception and deduction, lacking dynamic linkage with real-time data streams and ineffectively supporting decision-making optimization during mining operations. In particular, systematic solutions are still lacking for identifying high-risk areas, adjusting mining sequences, and planning risk-avoidance paths. Summary of the Invention

[0004] This application proposes a method and system for visual reconstruction of the geological environment used in the continuous mining process of underground metal mines, which can solve one of the problems existing in the background technology.

[0005] To achieve the above objectives, this application adopts the following technical solutions:

[0006] In a first aspect, a method for visually reconstructing the geological environment of a geological body applied to the continuous mining process of an underground metal mine is provided, the method comprising:

[0007] Collect geological attribute information, experimental mechanical parameter information and engineering environmental response information of underground metal mines;

[0008] Preprocessing the geological attribute information, the experimental mechanical parameter information, and the engineering environment response information to obtain standardized multi-source data;

[0009] Processing the multi-source data using a deep learning algorithm to identify structural features of the geological body, and establishing a static model based on the structural features of the geological body;

[0010] According to the engineering environment response information, dynamically updating the stress field and geological body structure on the basis of the static model to form a dynamic evolution three-dimensional model; and

[0011] Based on the dynamic evolution three-dimensional model, three-dimensional visualization technology is used to reconstruct the geological body occurrence environment.

[0012] Based on the above technical solution, a holographic perception method of underground rock environment is integrated with multi-scale, multi-physical quantity and multi-time resolution information to construct a three-dimensional occurrence environment model that can dynamically evolve in real time. With the help of visualization and intelligent feedback mechanism, precise guidance and risk control of the continuous mining process are achieved, thus providing strong technical support for the safe, efficient and intelligent mining of deep metal mines.

[0013] In a possible design of the first aspect, a deep learning algorithm is used to process the multi-source data to identify geological body structural features, and a static model is established based on the geological body structural features, specifically including:

[0014] For the data presented in the form of grids or tensors in the multi-source data, a convolutional neural network (CNN) is used to process the data to identify the geological body structure; for the data presented in the form of spatial topological relationships in the multi-source data, a graph neural network (GNN) is used to process the data to identify the relationship between the geological body structures; and

[0015] Based on the geological body structures and the relationships between the geological body structures, the static model is constructed using point cloud processing technology and voxel grid algorithm.

[0016] In one possible design approach of the first aspect, the engineering environment response information includes microseismic monitoring data, stress-strain data, and mining disturbance data. Based on the engineering environment response information, the stress field and geological structure are dynamically updated on the basis of the static model to form a dynamically evolving three-dimensional model, specifically including:

[0017] Using the long short-term memory network (LSTM), the time-frequency characteristics of the microseismic monitoring data are learned to predict stress field changes;

[0018] Using an extended Kalman filter (EKF) algorithm to filter and optimize the stress-strain data to correct the stress field; and

[0019] Based on the mining disturbance data, a finite element and discrete element FEM-DEM coupling algorithm is used to simulate the impact of mining disturbance on the geological structure and dynamically adjust the model geometry.

[0020] In a possible design method of the first aspect, according to the engineering environment response information, dynamically updating the stress field and geological body structure on the basis of the static model to form a dynamically evolving three-dimensional model further includes:

[0021] By comparing the residuals of the model-predicted stress field with the stress-strain data, the model parameters are continuously adjusted based on the Bayesian optimization mechanism.

[0022] In a possible design method of the first aspect, based on the dynamic evolution three-dimensional model, three-dimensional visualization technology is used to reconstruct the geological body occurrence environment, specifically including:

[0023] Based on the dynamic evolution 3D model, a 3D modeling platform and a real-time rendering engine are used to present the geological structure, stress field and mining progress in real time; and

[0024] Through slicing technology and dynamic simulation technology, virtual reality VR or augmented reality AR equipment is integrated to interactively display the geological environment.

[0025] In a possible design mode of the first aspect, based on the dynamic evolution three-dimensional model, three-dimensional visualization technology is used to reconstruct the geological body occurrence environment, which also includes:

[0026] Ray casting and isosurface rendering technology are used to transparently overlay and display the stress gradient, microseismic event density and goaf morphology in the dynamic evolution 3D model.

[0027] In a possible design manner of the first aspect, the visual reconstruction method further includes:

[0028] Combine multi-objective optimization with risk assessment algorithms to generate mining strategies and early warning plans.

[0029] In one possible design approach of the first aspect, a mining strategy and early warning plan are generated by combining multi-objective optimization with a risk assessment algorithm, specifically including:

[0030] The non-dominated sorting genetic algorithm NSGA-Ⅱ was used to balance resource recovery rate, safety factor and energy efficiency to optimize the mining sequence.

[0031] Identify early signs of rockbursts, roof falls, and water inrush hazards using support vector machines (SVMs) and random forest (RF) classification models; and

[0032] A digital twin system is constructed to connect the dynamically evolving three-dimensional model with the mine scheduling control system in real time to perform real-time scheduling control.

[0033] In a possible design manner of the first aspect, the preprocessing includes: denoising, data interpolation, spatial alignment, time synchronization and standardization.

[0034] In a second aspect, a visualization reconstruction system for geological body occurrence environment applied to the continuous mining process of underground metal mines is provided, wherein the visualization reconstruction system comprises:

[0035] The data acquisition layer is used to collect geological attribute information, experimental mechanical parameter information and engineering environmental response information of underground metal mines;

[0036] A preprocessing layer, for preprocessing the geological attribute information, the experimental mechanical parameter information and the engineering environment response information to obtain standardized multi-source data;

[0037] A fusion modeling layer is used to process the multi-source data using a deep learning algorithm to identify the structural characteristics of the geological body and establish a static model based on the structural characteristics of the geological body; based on the static model, dynamically update the stress field and geological body structure according to the engineering environment response information to form a dynamically evolving three-dimensional model; and

[0038] The application layer is used to reconstruct the geological body occurrence environment by using three-dimensional visualization technology based on the dynamic evolution three-dimensional model.

[0039] In a third aspect, an electronic device is provided, comprising: a processor, and a memory coupled to the processor, the memory being used to store a computer program; the processor being used to execute the computer program stored in the memory, so that the electronic device performs the visualization reconstruction method as any possible implementation method in the first aspect.

[0040] In a fourth aspect, a computer-readable storage medium is provided, comprising a computer program or instructions, which, when executed on a computer, causes the computer to execute the visualization reconstruction method of any possible implementation of the first aspect.

[0041] In a fifth aspect, a computer program product is provided, comprising: a computer program or instructions, which, when the computer program or instructions are run on a computer, enables the computer to execute the visualization reconstruction method of any possible implementation of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0043] Figure 1 This is the overall architecture diagram provided by the embodiment of the present application;

[0044] Figure 2 This is a cross-scale data fusion flow chart provided in an embodiment of the present application;

[0045] Figure 3 This is a logic diagram of structure recognition and three-dimensional modeling provided by the embodiment of the present application;

[0046] Figure 4 This is a schematic diagram of dynamic modeling and evolution provided by an embodiment of the present application;

[0047] Figure 5 This is the technical roadmap provided by the embodiments of this application. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0049] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and so on, in the specification, claims, and drawings, are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0051] The following is an exemplary description of the method for visually reconstructing the geological body occurrence environment applied to the continuous mining process of underground metal mines in an embodiment of the present application.

[0052] The method for holographic perception and visualization reconstruction of the rock mass environment during continuous mining of an underground metal mine provided in this embodiment includes the following steps:

[0053] S1. Multi-source heterogeneous data collection, collecting multi-source heterogeneous data covering micro, laboratory and engineering scales.

[0054] During the specific implementation, equipment such as scanning electron microscope (SEM) and X-ray diffraction (XRD) are used to obtain information such as the mineral composition, microstructure and crack distribution of the rock. These data provide the basis for subsequent ore body structure analysis.

[0055] Rock mechanical parameters such as uniaxial compressive strength, elastic modulus, and rock stress-strain response are obtained through rock mechanics experiments. These data reflect the physical properties of the rock mass and are important parameters that need to be considered during modeling.

[0056] On-site data collection is carried out using equipment such as microseismic monitoring systems, geological radar, borehole logging, and stress-strain sensors. The engineering data obtained through these devices can provide feedback on the real-time status of the ore body. For example, microseismic data can reflect the fault movement and stress concentration areas of the ore body.

[0057] Among them, the engineering-scale microseismic data comes from the multi-channel microseismic monitoring system deployed in the mining area, with a sampling frequency of not less than 500 Hz and a positioning accuracy of not less than 5 meters.

[0058] S2. Perform preprocessing and standardization operations on the multi-source data collected in S1;

[0059] For microscale data (such as mineral composition, pore structure, and fracture images), image filtering, missing value filling, and grayscale normalization are used to remove noise and extract structural features, and standard deviation normalization is used to improve the comparability between different samples. For laboratory-scale data (such as uniaxial compressive strength, elastic modulus, and stress-strain curves), unit unification and data interpolation are first performed to smooth the measurement curves, and then extreme value normalization is used to map different physical parameters to a unified interval. For engineering-scale data (such as microseismic events, displacement monitoring, and stress-strain field data), coordinate transformation and time synchronization are used to align the spatial and temporal sequences of multi-source signals. At the same time, wavelet denoising and other methods are combined to improve signal quality. Finally, the format is unified and normalization is performed to construct a standardized feature dataset for multiscale modeling.

[0060] During specific implementation, abnormal data and noise values ​​are eliminated to ensure data accuracy, for example, using the Z-Score or IQR (interquartile range) method to remove outliers.

[0061] Because data collected by different devices may use different spatial coordinate systems, spatial coordinate alignment is required. Affine transformation is used to align the coordinates of different data sources to the same reference system. The formula is: T(x',y') = A·T(x,y)+B, where T(x',y') is the transformed coordinate, T(x,y) is the original coordinate, A and B are the rotation matrix and displacement vector, respectively.

[0062] Synchronize data collected at different times. For microseismic data and radar data collected at different times, use interpolation methods (such as linear interpolation or spline interpolation) to synchronize them to the same time point. The interpolation formula is: Where f(t0) and f(t1) are the known data at time points t0 and t1, and f(t) is the interpolated data.

[0063] Since different data have different dimensions, they need to be standardized. The Min-Max normalization method is used to convert the data into a unified standard format. The formula is: Where X is the original data, X' is the standardized data, and X min and X max are the minimum and maximum values ​​of the data, respectively.

[0064] S3. Based on the standardized data processed in S2, perform preliminary geological structure identification and three-dimensional modeling.

[0065] During the specific implementation, deep learning algorithms such as graph neural networks (GNN) are used to extract features and identify structures of the collected multi-source data; GNN can model based on the similarity of different geological features and automatically identify geological structures such as ore bodies, faults, and fissures.

[0066] In specific implementation, deep learning models such as graph neural networks (GNNs) and convolutional neural networks (CNNs) are used for feature extraction and structural modeling for different types of structural recognition data. For data primarily based on spatial topological relationships (such as fault plane distribution, microseismic event coordinate sets, and structural measurement point relationship networks), these data are constructed into a graph structure and then modeled using GNNs to identify potential correlations between fracture connectivity and complex structures. For regular data presented in grid or tensor form (such as geological profiles, three-dimensional stress distribution fields, and displacement heat maps), CNNs are used for modeling, extracting key spatial features such as structural planes and abnormal areas through a local perception mechanism. GNNs are used to construct node feature vectors to represent the spatial adjacency and mechanical property associations of geological bodies. An attention mechanism is used to weightedly fuse the GNN output features to improve the robustness of complex structure recognition.

[0067] Based on the identified structural features, a three-dimensional geological model is constructed using point cloud processing technology and voxel grid algorithms. The point cloud data is processed through a triangulation algorithm (such as Delaunay triangulation) to construct the preliminary geometric shape of the ore body. The model obtained at this time is a static model that represents the spatial distribution and general shape of the ore body.

[0068] S4. Based on the 3D geological model initially established in S3, combined with time-series microseismic monitoring data and mining disturbance data, the model is dynamically updated and evolved.

[0069] Combining real-time microseismic monitoring data with mining disturbance records, a time-series driven algorithm is used to dynamically update the model. The spatiotemporal evolution of microseismic events is modeled through a long short-term memory network (LSTM) to predict the expansion trend of stress concentration areas and potential fracture surfaces. The extended Kalman filter (EKF) is used to fuse the stress-strain field data collected by the sensor to correct the stress distribution and displacement field in the model in real time. Based on the operating parameters of the mining machinery (such as cutting speed and propulsion direction) and the blasting energy distribution, the finite element-discrete element coupling algorithm (FEM-DEM) is used to simulate the impact of mining disturbance on the geological structure and dynamically adjust the model geometry.

[0070] In specific implementation, microseismic monitoring data provides the stress state and local deformation information of the ore body at each time point. By analyzing the time-frequency characteristics of microseismic events, it is determined whether there is fault slip or structural rupture in the ore body. The formula is: Where ΔE(t) represents the change of stress field at time t, S i (t) and S i-1 (t) are the microseismic signal characteristics at time t and t-1 respectively.

[0071] In specific implementation, microseismic monitoring data can provide stress state evolution information inside the ore body at each time point, and reveal the local deformation characteristics of the surrounding rock and the boundary area of ​​the goaf under the action of mining disturbance. The local deformation includes micro-scale shear slip, tensile crack extension and strain anomalies caused by stress redistribution, which are usually manifested as an increase in the number of microseismic events, an increase in magnitude or a sudden change in source parameters.

[0072] The system determines structural anomalies by analyzing the time-frequency characteristics of microseismic events, that is, it comprehensively considers signal characteristics such as the changing trend of event energy and magnitude, the concentration of low-frequency energy in the spectrum, the spatial distribution density, and the inversion results of the source mechanism. If frequent energy release, increasing magnitude, enhanced low-frequency components, and the event location gradually gather to a certain existing structural interface are continuously monitored in a certain spatial area, and at the same time the source tensor reflects an obvious shear-type rupture pattern, it can be determined that there is a fault slip or structural rupture trend in the area, thereby triggering the local update of the model and the disaster warning mechanism.

[0073] The spatiotemporal evolution of microseismic events is modeled based on the long short-term memory network (LSTM) to predict potential stress concentration areas and fracture surface expansion paths. The LSTM network learns the time-frequency characteristics of microseismic signals through time series, and its state update expression is: t =σ(W h x t +W hh h t-1 +b h ), where x t is the input variable at time t (such as microseismic magnitude, location, spectrum energy, etc.), h t is the hidden state vector at the current time t, σ is the activation function, W h , W hh , b h are weight and bias parameters respectively.

[0074] In order to achieve dynamic fusion and model correction of multi-source sensor data, the extended Kalman filter (EKF) algorithm is used to filter and optimize the stress-strain data collected by the sensor. The prediction and update steps of EKF are as follows: renew P k|k =(IK k H k )P k|k-1 ,in is the state variable after fusion (such as stress field distribution), f() is the nonlinear state transfer function, u k is the input of the system at step k (control quantity, such as disturbance parameter), P is the covariance matrix, F k is the Jacobian of the state transfer matrix, Q k is the process noise covariance matrix, K k is the Kalman gain matrix, H k is the Jacobian matrix of the observation model, R k is the observation noise covariance matrix, is the updated state estimate after fusion observation, z k is the observed quantity (sensor strain data), h() is the nonlinear observation function, P k|k is the updated state covariance matrix, and I is the identity matrix.

[0075] To accurately simulate the effects of mining disturbance on underground structures, a coupled finite element and discrete element method (FEM-DEM) approach was used for numerical simulation. The FEM component calculates the stress-strain response of continuous media, with the basic governing equation being: [K]{u} = {F}, where [K] is the stiffness matrix, {u} is the node displacement vector, and {F} is the external load. The DEM, on the other hand, characterizes particle slip and structural fracture within discontinuous media (such as faults and joints), whose mechanical behavior conforms to Newton's second law: where m i is the mass of the i-th particle, is the position vector of the i-th particle, is the acceleration, is the contact force from other particles j, The contact force transmission and strain release mechanism are introduced at the contact interface between FEM and DEM to achieve the coupling simulation between continuous and discontinuous structures.

[0076] The model uses characteristics such as magnitude, event distribution, energy release, and focal mechanism reflected in microseismic data to identify areas of stress concentration and potential locations of structural deformation. It then uses mining disturbance parameters such as blasting energy, propulsion speed, and operating direction as external inputs to drive the simulation module to dynamically calculate the redistribution of the stress field within the ore body. On this basis, the system uses a time-series modeling algorithm to modify the predicted stress state and, based on response simulations under disturbance, determines whether local structural units exhibit abnormal stress increments or discontinuous displacement behavior. If the model's threshold is reached, the stress values ​​and structural geometry of the corresponding area are automatically updated, and the adjusted results are mapped to the three-dimensional model, thereby achieving continuous dynamic evolution of the ore body's structural state.

[0077] Dynamic evolution modeling compares the residuals of the model-predicted stress field with the actual sensor data, and continuously adjusts the model parameters based on the Bayesian optimization mechanism to achieve adaptive improvement of evolution accuracy.

[0078] S5. Based on the dynamically evolving 3D model of S4, the mining environment is reconstructed using 3D visualization technology.

[0079] During specific implementation, the model structure, stress distribution, goaf morphology and microseismic activity are transparently rendered and interactively displayed through a three-dimensional modeling platform and visualization engine, supporting users to understand and analyze the rock environment in real time at different levels and angles. Ray casting and isosurface rendering techniques are used to transparently overlay and display the stress gradient, microseismic event density and goaf morphology in the model. Virtual reality (VR) and augmented reality (AR) devices are integrated to support users to analyze the mining environment in multiple dimensions through gesture interaction, perspective switching and dynamic slicing operations. Dynamic heat maps and particle flow animations are used to display the stress field evolution, crack expansion paths and migration processes of dangerous areas in real time.

[0080] By synchronizing the dynamic changes in the ore body's stress field with the mining progress, on-site personnel can clearly see the real-time status of the mine and adjust the mining plan in a timely manner. The dynamic nature of the model ensures that environmental changes can be reflected immediately and early warnings can be provided during the mining process.

[0081] S6. Conduct intelligent mining guidance and risk assessment based on S5 holographic perception and 3D reconstruction results.

[0082] During specific implementation, by optimizing factors such as mining sequence, ore body stability, and resource recovery rate, mining sequence suggestions and mining path planning are provided, and multi-objective optimization algorithms (such as genetic algorithms or particle swarm optimization algorithms) are used to generate the optimal mining plan based on real-time data.

[0083] By combining real-time monitoring data with modeling results, dangerous areas in mines can be dynamically assessed to predict possible disaster risks. For example, by analyzing stress concentration areas in the ore body, early warnings of dangers such as rock bursts and roof falls can be issued, and mining strategies can be adjusted.

[0084] During implementation, the system, based on real-time modeling results and multi-source monitoring data, comprehensively considers key factors such as mining sequence, orebody stability, and resource recovery rate. It then constructs a multi-objective optimization problem, with resource maximization, optimal safety, and minimal energy consumption as the objective functions. It then uses genetic algorithms or particle swarm optimization to iteratively optimize the mining path and timing plan. Input parameters include the current orebody's geometry, remaining resource distribution, stress state, existing mining progress, and construction capacity boundaries. Evolutionary computation methods are used to search for Pareto optimal solutions within the feasible solution space. The optimal mining strategy is selected based on comprehensive evaluation indicators, and the recommended operation unit execution sequence and mining advancement path are output.

[0085] In terms of risk assessment, the system continuously integrates real-time sensor data with model prediction information to dynamically identify and determine potentially dangerous areas in the mine. By analyzing the spatial gradient changes in the stress field within the ore body, the distribution density of microseismic events, and the magnitude evolution trend, it identifies high-stress areas or structurally weak zones that may be at risk of rockburst, roof fall, or water inrush. If an area shows rapid energy accumulation, frequent microseismic intensification, or sudden changes in the focal mechanism during continuous monitoring, the system identifies it as a potential danger zone and triggers an early warning mechanism. On this basis, the system links risk information with the optimization and scheduling system, dynamically adjusts the mining sequence according to the risk level, and achieves dual optimization of resource efficiency and safety assurance.

[0086] Specifically, the visual reconstruction results are combined with multi-objective optimization and risk assessment algorithms to generate mining strategies and early warning plans. The non-dominated sorting genetic algorithm (NSGA-II) is used to optimize the mining sequence and balance the resource recovery rate, safety factor and energy efficiency. The support vector machine (SVM) and random forest (RF) classification models are used to identify early signs of rock bursts, roof falls and water inrush disasters. A digital twin system is constructed to connect the dynamic model with the mine scheduling control system in real time to achieve a "perception-modeling-decision-execution" closed-loop control.

[0087] Based on a holographic visualization model, the system integrates multi-objective optimization and classification prediction modules for mining strategy generation and disaster early warning, respectively, to achieve scientific guidance of the mining process and intelligent identification of disaster risks. To address the multi-objective requirements of mining, including resource recovery rate, safety factor, and energy efficiency, a non-dominated sorting genetic algorithm (NSGA-II) was introduced to optimize the mining sequence. As a Pareto optimal frontier search algorithm, NSGA-II's core components include population initialization, non-dominated sorting, congestion distance calculation, and tournament selection. The multi-objective fitness function can be defined as: min{f1(x),f2(x),f3(x)}, where f1(x) represents the resource loss rate (desirably minimized); f2(x) represents the structural instability risk indicator corresponding to the mining operation (related to the safety factor); and f3(x) represents the energy consumption per unit of ore mined (kWh / t).

[0088] The individual ranking in NSGA-II adopts the non-dominated level r i , its calculation depends on the relationship between the objective functions; the crowding distance d i Used to maintain the diversity of the solution set, it is calculated as: in are the values ​​of the solutions before and after sorting on the target m, is the maximum and minimum value on the target.

[0089] To identify early signs of disasters such as rockbursts, roof falls, and water inrush, a machine learning-based disaster classification model was constructed. Input features included multidimensional signals such as microseismic frequency, magnitude distribution, stress field gradients, and sensor strain responses. Support vector machine (SVM) and random forest (RF) models were used for training and prediction, respectively.

[0090] The support vector machine distinguishes the two types of disaster patterns by constructing the optimal classification hyperplane. Its basic form is: Where w is the normal vector of the hyperplane, b is the bias term, ξ i is the slack variable and C is the penalty coefficient.

[0091] The Random Forest (RF) model improves the robustness and generalization of classification by constructing multiple decision trees for ensemble voting. The overall prediction result is determined by the voting of each tree: where h t (x) is the predicted output of the t-th tree, and T is the total number of trees.

[0092] The above optimization and classification results can be connected to the scheduling system in real time to form a "modeling-evaluation-decision-making-execution" closed-loop control, realizing safe, efficient and intelligent continuous mining management of mines.

[0093] This embodiment has the following advantages:

[0094] 1. By integrating multi-source heterogeneous data (including microscopic, laboratory, and engineering-scale data), the ability to perceive the continuous mining environment of underground metal mines is comprehensively improved. Unlike existing technologies that rely on a single data source, this embodiment can integrate data from different levels and different physical quantities to achieve a comprehensive understanding of the ore body. This data fusion method not only improves the accuracy of mine environmental information, but also enhances the ability to identify details such as complex geological structures, cracks, faults, etc.

[0095] 2. Collaborative preprocessing and feature extraction of heterogeneous signals to adapt to complex geological conditions: By designing a unified data preprocessing framework and frequency / time domain feature extraction methods, the system can efficiently process active and passive signals with different physical properties, extracting key features with strong compatibility and high robustness. Compared to existing approaches that use static processing flows for a single type of signal, this invention can adapt to complex geological conditions and data variations in a wide range of engineering scenarios.

[0096] 3. Innovations in 3D visualization and reconstruction enable the holographic presentation of complex mining environments. Unlike existing visualization technologies that only support static displays, this embodiment provides a real-time, interactive 3D display platform capable of dynamically displaying the mine's stress field, mining progress, hazardous areas, and more. This not only provides on-site operators with a more intuitive and accurate understanding of the environment, but also provides reliable support for risk assessment and decision-making, effectively improving the safety and efficiency of the mining process.

[0097] 4. Intelligent mining guidance and risk assessment capabilities, through the integration of multi-objective optimization algorithms, automatically optimize mining sequences and avoid hazardous areas based on real-time perception and dynamic modeling results, thereby improving resource recovery and mitigating safety hazards. Compared to traditional mining methods based on manual judgment and empirical decision-making, this intelligent feedback mechanism offers greater flexibility, accuracy, and efficiency, significantly reducing the risk of accidents and improving the overall management of mines.

[0098] The present application also provides a system for visually reconstructing a geological environment for continuous mining of underground metal mines. The system comprises:

[0099] The data acquisition layer is used to collect geological attribute information, experimental mechanical parameter information and engineering environmental response information of underground metal mines;

[0100] A preprocessing layer, for preprocessing the geological attribute information, the experimental mechanical parameter information and the engineering environment response information to obtain standardized multi-source data;

[0101] A fusion modeling layer is used to process the multi-source data using a deep learning algorithm to identify the structural characteristics of the geological body and establish a static model based on the structural characteristics of the geological body; based on the static model, dynamically update the stress field and geological body structure according to the engineering environment response information to form a dynamically evolving three-dimensional model; and

[0102] The application layer is used to reconstruct the geological body occurrence environment by using three-dimensional visualization technology based on the dynamic evolution three-dimensional model.

[0103] An embodiment of the present application also provides an electronic device, comprising: a processor, and a memory coupled to the processor, wherein the memory is used to store a computer program; and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the method described in any one of the above embodiments.

[0104] The electronic device may be a computing device such as a desktop computer, a notebook computer, a PDA, a cloud server, etc. The electronic device may include, but is not limited to, a processor and a memory.

[0105] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, connecting various parts of the entire device using various interfaces and lines.

[0106] The memory may be used to store the computer program, and the processor implements various functions of the electronic device by running or executing the computer program stored in the memory and calling the data stored in the memory.

[0107] The memory may primarily include a program storage area and a data storage area, wherein the program storage area may store an operating system, at least one application required for a function, and the like; and the data storage area may store data created based on the use of the mobile phone, and the like. Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a SmartMedia Card (SMC), a Secure Digital (SD) card, a flash memory card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0108] The embodiment of the present application also provides a storage medium, which is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0109] An embodiment of the present application further provides a computer program product, including: a computer program or instructions, which, when executed on a computer, causes the computer to execute any of the above-mentioned possible implementation methods.

[0110] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications are also considered to be within the scope of protection of the present application.

Claims

1. A method for visual reconstruction of geological body occurrence environment applied to the continuous mining process of underground metal mines, characterized in that: The reconstruction method includes: Collect geological attribute information, experimental mechanical parameter information and engineering environmental response information of underground metal mines; Preprocessing the geological attribute information, the experimental mechanical parameter information, and the engineering environment response information to obtain standardized multi-source data; Processing the multi-source data using a deep learning algorithm to identify structural features of the geological body, and establishing a static model based on the structural features of the geological body; According to the engineering environment response information, dynamically updating the stress field and geological body structure on the basis of the static model to form a dynamic evolution three-dimensional model; and Based on the dynamic evolution three-dimensional model, three-dimensional visualization technology is used to reconstruct the geological body occurrence environment.

2. The visualization reconstruction method according to claim 1, wherein: The multi-source data is processed using a deep learning algorithm to identify the structural features of the geological body, and a static model is established based on the structural features of the geological body, specifically including: For the data presented in the form of grids or tensors in the multi-source data, a convolutional neural network (CNN) is used to process the data to identify the geological body structure; for the data presented in the form of spatial topological relationships in the multi-source data, a graph neural network (GNN) is used to process the data to identify the relationship between the geological body structures; and Based on the geological body structures and the relationships between the geological body structures, the static model is constructed using point cloud processing technology and voxel grid algorithm.

3. The visualization reconstruction method according to claim 1, wherein: The engineering environment response information includes: microseismic monitoring data, stress-strain data, and mining disturbance data. Based on the engineering environment response information, the stress field and geological structure are dynamically updated on the basis of the static model to form a dynamic evolution three-dimensional model, specifically including: Using the long short-term memory network (LSTM), the time-frequency characteristics of the microseismic monitoring data are learned to predict stress field changes; Using an extended Kalman filter (EKF) algorithm to filter and optimize the stress-strain data to correct the stress field; and Based on the mining disturbance data, a finite element and discrete element FEM-DEM coupling algorithm is used to simulate the impact of mining disturbance on the geological structure and dynamically adjust the model geometry.

4. The visualization reconstruction method according to claim 3, wherein: According to the engineering environment response information, the stress field and geological body structure are dynamically updated on the basis of the static model to form a dynamic evolution three-dimensional model, further comprising: By comparing the residuals of the model-predicted stress field with the stress-strain data, the model parameters are continuously adjusted based on the Bayesian optimization mechanism.

5. The visualization reconstruction method according to claim 1, wherein: Based on the dynamic evolution 3D model, 3D visualization technology is used to reconstruct the geological environment, including: Based on the dynamic evolution 3D model, a 3D modeling platform and a real-time rendering engine are used to present the geological structure, stress field and mining progress in real time; and Through slicing technology and dynamic simulation technology, virtual reality VR or augmented reality AR equipment is integrated to interactively display the geological environment.

6. The visualization reconstruction method according to claim 4, wherein: Based on the dynamic evolution 3D model, 3D visualization technology is used to reconstruct the geological body occurrence environment, which also includes: Ray casting and isosurface rendering technology are used to transparently overlay and display the stress gradient, microseismic event density and goaf morphology in the dynamic evolution 3D model.

7. The visualization reconstruction method according to claim 1, wherein: The visualization reconstruction method further includes: Combine multi-objective optimization with risk assessment algorithms to generate mining strategies and early warning plans.

8. The visualization reconstruction method according to claim 5, wherein: Combining multi-objective optimization with risk assessment algorithms, we generate mining strategies and early warning plans, including: The non-dominated sorting genetic algorithm NSGA-Ⅱ was used to balance resource recovery rate, safety factor and energy efficiency to optimize the mining sequence. Identify early signs of rockbursts, roof falls, and water inrush hazards using support vector machines (SVMs) and random forest (RF) classification models; and A digital twin system is constructed to connect the dynamically evolving three-dimensional model with the mine scheduling control system in real time to perform real-time scheduling control.

9. The visualization reconstruction method according to claim 1, wherein: The preprocessing includes: denoising, data interpolation, spatial alignment, time synchronization and standardization.

10. A geological body occurrence environment visualization reconstruction system applied to the continuous mining process of underground metal mines, characterized in that: The visual reconstruction system includes: The data acquisition layer is used to collect geological attribute information, experimental mechanical parameter information and engineering environmental response information of underground metal mines; A preprocessing layer, for preprocessing the geological attribute information, the experimental mechanical parameter information and the engineering environment response information to obtain standardized multi-source data; A fusion modeling layer is used to process the multi-source data using a deep learning algorithm to identify the structural characteristics of the geological body and establish a static model based on the structural characteristics of the geological body; based on the static model, dynamically update the stress field and geological body structure according to the engineering environment response information to form a dynamically evolving three-dimensional model; and The application layer is used to reconstruct the geological body occurrence environment by using three-dimensional visualization technology based on the dynamic evolution three-dimensional model.

Images