Modeling method based on digital twin mine

Abstract

The invention relates to the technical field of digital twinning, in particular to a digital twinning-based mine modeling method, which comprises the following steps of: S1, acquiring geological survey data, equipment operation data and material transportation data of a mine; and S2, performing space-time alignment and noise elimination on the geological survey data, the equipment operation data and the material transportation data to form a comprehensive data set. According to the method, space-time alignment, noise elimination and interpolation completion are carried out on mine geological survey, equipment operation and material transportation multi-source data, the problem that traditional modeling data are disordered and inconsistent is solved, and a high-quality data basis is provided for model construction. A causal incidence matrix is generated through a causal inference algorithm, the dynamic causal relationship among geological change, equipment operation and transportation scheduling is accurately described, and the defect that a traditional model cannot accurately reflect the real-time cooperative relationship among the geological change, the equipment operation and the transportation scheduling is overcome.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, specifically to a method for modeling mines based on digital twins. Background Technology

[0002] Digital twin technology has become a core support for the digital transformation of mines. By constructing a real-time interactive channel between virtual space and the physical mine, it enables visualized control and optimization of all elements and processes of the mine. This technology can integrate multi-source data such as mine geology, equipment, and transportation to form a dynamically updated virtual model, providing data support for mining planning and safety management. It is one of the key technologies for the construction of smart and green mines.

[0003] The existing invention patent CN115047833A, "A Digital Twin Factory for Mines and Its Construction Method," proposes the construction of an intelligent sensing system for mines, a digital twin model, and a multi-temporal-scale database, achieving intelligent management and control of mines through virtual-real interaction. This patent focuses on solving the problem of fusion and integration of multi-source heterogeneous data and establishes a relatively complete twin architecture; however, it does not pay attention to the collaborative mapping accuracy of data from different dimensions during the dynamic updating of the model.

[0004] Current technologies require the integration of dynamic data from multiple dimensions, including ore body structure, equipment operation, and material transportation. Model layers corresponding to different data dimensions are often constructed based on their respective data scales. This leads to discrepancies between the real-time changes in the ore body structure and the mapping of equipment operating status and material transportation paths during dynamic model updates. This results in an inaccurate reflection of the real-time collaborative relationship between the three, impacting the accuracy of mining decisions and the rationality of production scheduling. This problem is particularly pronounced in complex mining scenarios, and existing solutions do not offer specific solutions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a digital twin-based mine modeling method that solves the problems of multi-dimensional data collaborative mapping deviation, insufficient adaptability of dynamic model updates, and lack of full-chain traceability in existing technologies.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for modeling mines based on digital twins, comprising: S1. Collect geological survey data, equipment operation data, and material transportation data of the mine; S2. Perform spatiotemporal alignment and noise removal on the geological survey data, equipment operation data, and material transportation data to form a comprehensive dataset; S3. Based on the comprehensive dataset, construct a three-dimensional geological dynamic model of the mine, an intelligent equipment adaptation model, and a transportation dynamic coordination model, and generate a causal correlation matrix representing the dynamic causal relationship between geological changes, equipment operation, and transportation scheduling through a causal inference algorithm. S4. Based on the causal correlation matrix and the current mining scenario, generate model calibration rules and perform preliminary updates to the equipment intelligent adaptation model and the transportation dynamic collaboration model; S5. Perform virtual simulation on the preliminary update scheme to identify potential conflicts and generate conflict resolution schemes; S6. Based on the conflict resolution scheme, perform incremental updates on the three-dimensional geological dynamic model of the mine, the intelligent equipment adaptation model, and the transportation dynamic collaborative model, and establish a full-link traceability archive. S7. The updated model is evaluated for accuracy using real-time feedback data from the physical mine, and a closed-loop evolution process is initiated if the evaluation results do not meet the standards.

[0007] Furthermore, S2 specifically includes: Use timestamp synchronization protocols to perform time-series alignment of multi-source data; A spatial coordinate transformation algorithm is used to unify data from different coordinate systems to the mine's global coordinate system; Apply a sliding window filter or the isolated forest algorithm to identify and remove outlier data points; Based on the geological correlation of adjacent areas, the trend of equipment operation status, and the logic of transportation routes, missing data is interpolated and completed to form the comprehensive dataset.

[0008] Furthermore, in step S3, a causal correlation matrix representing the dynamic causal relationship among geological changes, equipment operation, and transportation scheduling is generated using a causal inference algorithm. Specifically, this includes: Define geological state variables Equipment state variables and transportation state variables Observations at time step t; Granger causality tests or structural equation modeling are used to calculate the causal strength between variables; Constructing a causal relationship matrix , of which elements Let represent the causal influence coefficient of the i-th state on the j-th state, where i,j∈{G,M,T}; Set a causal transmission threshold τ, when When this is confirmed, a valid causal link from state i to state j is confirmed to exist.

[0009] Furthermore, in step S4, model calibration rules are generated based on the causal correlation matrix and the current mining scenario, specifically including: A historical scenario database containing different mining stages, geological conditions, and production loads is pre-built; Extract scene categories from the historical scene database that have a similarity to the current real-time geological conditions, equipment operation data, and transportation load data that meets a preset threshold. The calibration parameter set corresponding to the scenario category is retrieved from the preset calibration rule library. The calibration parameter set includes model parameter calibration threshold, causal transmission efficiency coefficient and cooperative response time threshold. The calibration parameter set is applied to the causal correlation matrix to generate model calibration rules suitable for the current scenario.

[0010] Furthermore, S5 specifically includes: Using the current state of the three-dimensional geological dynamic model of the mine as the initial conditions, load the equipment performance limit parameters and the transportation network topology; The initial update plan is simulated and executed in a time-series manner within a simulation environment. During the monitoring and simulation process, check whether the surrounding rock stress exceeds the limit, the equipment load rate exceeds the standard, or the transportation path congestion index exceeds the preset safety threshold. If any exceedance is detected, a potential conflict is determined, and a multi-objective optimization solver is activated. With mining efficiency, safety threshold, cost control, and equipment life as optimization objectives, at least two feasible conflict resolution solutions are generated.

[0011] Furthermore, the multi-objective optimization solver is activated, with mining efficiency, safety threshold, cost control, and equipment lifespan as optimization objectives, to generate multiple feasible conflict resolution schemes, specifically including: Define a multi-objective optimization function: ; Where x is the vector of decision variables to be optimized. These represent mining efficiency loss, safety risk index, increase in operating costs, and equipment wear rate, respectively. This is the corresponding target weight vector; Solve the Pareto front solution set under the constraint g(x)≤0, where the constraints include equipment physical limits, transportation network capacity, and geological safety boundaries; The solutions that satisfy the causal correlation matrix verification conditions are selected from the Pareto front solution set as the conflict resolution schemes.

[0012] Furthermore, S6 specifically includes: Identify the model nodes involved in the changes in the conflict resolution scheme and the scope of their causal relationships; Incremental updates are completed by adjusting parameters or modifying structures only on the model nodes and the affected associated nodes. Generate a unique identifier for this update operation and record the following information in the traceability archive: update trigger source type, causal relationship matrix version number, matching calibration rule identifier, simulation pre-run input and output data, conflict resolution scheme decision variable values, and the final adopted update instruction sequence.

[0013] Furthermore, in step S7, the updated model is evaluated for accuracy using real-time feedback data from the physical mine, specifically including: Through intelligent sensing systems deployed in physical mines, actual geological deformation data, real-time equipment operating data, and transport vehicle location data are continuously collected. The actual collected data is compared with the output data of the virtual model in the same spatiotemporal coordinates; Calculate the degree of causal relationship matching Data mapping accuracy and decision execution deviation rate Three evaluation indicators; If any evaluation indicator exceeds the preset tolerance range, the accuracy evaluation is deemed to have failed.

[0014] Furthermore, the calculation of the causal relationship matching degree Data mapping accuracy and decision execution deviation rate The three evaluation indicators include: Calculate the degree of causal relationship matching for: ; Where N is the number of sampling points, This represents the actual state change of the physical entity at the k-th sampling point. The state change predicted by the virtual model based on the causal correlation matrix; Calculate data mapping accuracy To effectively match the ratio of the number of data points to the total number of data points; Calculate the decision execution deviation rate This is the normalized Euclidean distance between the actual action and the model's recommended action.

[0015] Furthermore, in S7, when the evaluation result fails to meet the standard, a closed-loop evolutionary process is initiated, specifically including: Based on the causal correlation matrix, the deviation propagation path is traced in reverse to locate the root cause of the deviation. The root cause of the deviation includes at least one of data acquisition, causal modeling, calibration rule matching, or simulation pre-run. Targeted optimizations are performed on the root causes of the deviations: if the problem is data acquisition, the sampling frequency of the sensing device is adjusted or the sensor parameters are calibrated; if the problem is causal modeling, the causal inference model is retrained and the causal correlation matrix is ​​updated; if the problem is calibration rules, new scene samples are added to the historical scene database and the calibration rule library is reconstructed. Using the optimized data, model, or rules as new input, the entire process from building a comprehensive dataset to incrementally updating the model is re-executed, completing a closed-loop evolutionary iteration.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention addresses the problem of inconsistent and disorganized data in traditional modeling by performing spatiotemporal alignment, noise removal, and interpolation on multi-source data from mine geological surveys, equipment operation, and material transportation, providing a high-quality data foundation for model construction. It generates a causal correlation matrix through a causal inference algorithm, accurately depicting the dynamic causal relationships among geological changes, equipment operation, and transportation scheduling, overcoming the limitation of traditional models in accurately reflecting the real-time collaborative relationships among these three factors. By matching historical scenarios to generate calibration rules adapted to the current scenario, and combining virtual simulation pre-visualization to identify potential conflicts and generate resolution solutions through multi-objective optimization, it improves the accuracy of mining decisions and the rationality of production scheduling, while reducing safety risks. Incremental updates reduce the computational and time costs of model updates, and full-link traceability archives enhance the traceability and maintainability of the technical solution. A closed-loop evolutionary process continuously optimizes data, models, and rules, ensuring the model maintains high adaptability and accuracy over the long term. Ultimately, while ensuring mine safety, it simultaneously improves mining efficiency, adapting to the digital transformation needs of different types of mines and possessing broad application value. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart of the simulation pre-running and incremental update sub-process of the present invention; Figure 3 This is a flowchart of the accuracy evaluation and closed-loop evolution process of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 This embodiment uses a large open-pit coal mine as the actual application scenario. The mine has a large mining area and complex geological conditions, including multiple coal seams and interlayer structures. During the mining process, multiple equipment such as excavators, loaders, and transport trucks are involved in the coordinated operation. The material transportation path needs to be dynamically adjusted according to the progress of the mining area. Traditional modeling methods are difficult to accurately match the real-time coordination relationship between geological changes, equipment operation and transportation scheduling, resulting in limited mining efficiency and greater difficulty in safety risk prevention and control.

[0020] Based on this, please refer to Figure 1-3 This invention provides a method for modeling mines based on digital twins, which includes the following steps: The first step is data acquisition, which involves collecting geological survey data, equipment operation data, and material transportation data from the mine. Geological survey data is acquired using equipment such as borehole inclinometers, ground-penetrating radar, and 3D laser scanners deployed in different areas of the mine. This data covers coal seam thickness, depth, dip angle, rock compressive strength, surrounding rock stability parameters, and groundwater distribution. Equipment operation data comes from sensors and industrial control systems on various mining equipment, including excavator digging force, operating time, engine speed, hydraulic system pressure, and fault warning information; loader loading capacity, lifting height, and work cycle time; and transport truck speed, load, fuel consumption, engine temperature, and tire pressure. Material transportation data is collected through vehicle-mounted GPS positioning modules, RFID tag identification systems, and a transportation dispatch platform, including transportation origin, destination, route, transport volume, loading and unloading time, and route congestion. To ensure data integrity and timeliness, geological survey data is collected daily, equipment operation data is collected in real-time at a sampling frequency of 10Hz, and material transportation data is synchronized to the data center every 30 seconds.

[0021] Next, the collected multi-source data is processed to achieve spatiotemporal alignment and noise removal of the geological survey data, equipment operation data, and material transportation data, forming a comprehensive dataset. For temporal alignment, the Network Time Protocol (NTP) is used to synchronize and calibrate the timestamps of all data, ensuring consistency across different devices and data types in the time dimension. For example, excavator operation data is precisely matched with corresponding coal seam geological data and transport truck scheduling data. During spatial alignment, a seven-parameter coordinate transformation algorithm is used to transform the geodetic coordinate system of the geological survey data, the equipment local coordinate system of the equipment operation data, and the GPS coordinate system of the material transportation data to the mine's global coordinate system. This coordinate system uses a fixed control point at the mine's mining boundary as its origin, with the X-axis along the mining direction, the Y-axis perpendicular to the mining direction, and the Z-axis vertical, ensuring the spatial uniformity of all data. In the noise removal stage, a sliding window filter is used to process potential abnormal fluctuations in geological survey data. The window size is set according to the data type; for example, the window size for coal seam thickness data is set to 5 sampling points. Outliers deviating from the mean by more than 3 times the standard deviation are removed by calculating the mean of the data within the window. For isolated outliers in equipment operation data and material transportation data, the isolated forest algorithm is used for identification, with an outlier ratio threshold of 5%, meaning that only extreme outliers accounting for no more than 5% are removed to avoid excessive removal that could lead to the loss of useful information. For missing values ​​that may occur during data acquisition, interpolation is used to complete the data based on the geological correlation between adjacent areas, the trend of equipment operation status, and the flow logic of transportation routes. For example, for missing rock compressive strength data in a certain area, Kriging interpolation is used to complete the data based on the survey results of adjacent boreholes. For transient missing values ​​in equipment operation data, linear interpolation is used to fill in the missing values ​​based on valid data from the preceding and following times. For missing transportation route data, reasonable inferences are made based on the transportation network topology and historical route data to complete the data. Finally, a comprehensive dataset with complete structure, spatiotemporal consistency, and reliable data is formed. This data processing effectively reduces inconsistencies and interference factors in multi-source heterogeneous data, laying a high-quality data foundation for subsequent model construction and solving the problem of insufficient model accuracy caused by messy data in traditional modeling.

[0022] Subsequently, model construction and causal relationship analysis were conducted based on the comprehensive dataset. Specifically, a three-dimensional geological dynamic model of the mine, an intelligent equipment adaptation model, and a dynamic transportation coordination model were constructed based on the comprehensive dataset. A causal correlation matrix representing the dynamic causal relationship between geological changes, equipment operation, and transportation scheduling was generated using a causal inference algorithm. The three-dimensional geological dynamic model of the mine was constructed using 3D modeling software combined with geological survey data. Based on data such as coal seam distribution, rock properties, and geological structures from the comprehensive dataset, a refined three-dimensional geological model was generated using tetrahedral mesh generation technology. The model can reflect the spatial morphology, physical and mechanical properties, and dynamic change characteristics of the geological body. For example, as mining progresses, the model can update parameters such as the remaining coal seam thickness and surrounding rock stress distribution in real time. The intelligent equipment adaptation model is based on equipment operation data, combined with equipment design parameters and performance indicators. The model can output optimal operating parameters and adaptation schemes for the equipment according to changes in geological conditions and production task requirements. For example, based on the rock compressive strength of different areas, the digging force and operating frequency of the excavator can be adjusted to ensure efficient and stable equipment operation. The transportation dynamic collaboration model is built based on material transportation data and transportation network topology. It can simulate the driving status of transportation vehicles and route congestion in real time, and optimize transportation scheduling schemes in combination with production plans to achieve efficient collaboration in material transportation.

[0023] In the process of generating the causal correlation matrix, the geological state variables are first defined. Equipment state variables and transportation state variables The observations at time step t, where This includes parameters such as the change in coal seam thickness at time step t, the increase in surrounding rock stress, and the degree of rock fragmentation. This includes parameters such as equipment load rate, operating efficiency, and failure rate at time step t. This includes parameters such as transportation route efficiency, traffic volume changes, and congestion duration at time step t. Subsequently, the Granger causality test algorithm is used to calculate the causal strength between variables. Specifically, it determines the causal relationship by testing whether the lagged value of one variable can significantly explain the current value of another variable. For example, it tests... Can it significantly explain If the changes in geological conditions indicate a Granger causal relationship between the changes and equipment operation, then the causal strength between other variables can be calculated similarly. Based on the calculated causal strengths, a causal relationship matrix is ​​constructed. , of which elements Let represent the causal influence coefficient of state i on state j, where i,j∈{G,M,T}, and its matrix form is as follows: ; in The coefficient representing the causal influence of the geological state itself. This represents the causal influence coefficient of geological conditions on equipment conditions. This represents the causal influence coefficient of geological conditions on transportation conditions, and the other elements follow the same logic. Next, a causal transmission threshold τ is set. The threshold τ is determined based on statistical analysis of historical data. By calculating the distribution of the absolute values ​​of all causal influence coefficients, the critical value of the 95% confidence interval is taken as τ. For example, after statistical analysis, τ = 0.3. At that time, it was confirmed that there is a valid causal link from state i to state j. This causal relationship matrix can accurately characterize the dynamic causal relationship among geological changes, equipment operation and transportation scheduling, providing a basis for subsequent model calibration and optimization, and effectively solving the problem of inaccurate characterization of the synergistic relationship among the three in traditional models.

[0024] After completing model construction and generating the causal correlation matrix, model calibration rules are generated based on the causal correlation matrix and the current mining scenario, and the equipment intelligent adaptation model and the transportation dynamic collaborative model are initially updated. First, a historical scenario database containing different mining stages, geological conditions, and production loads is pre-constructed. The scenario categories in the database are obtained by classifying and organizing mining data from the past five years, covering different stages such as initial mining, mid-term mining, and late-term mining; different geological conditions such as gentle geology, complex geology, and fractured zone geology; and different production loads such as low load, medium load, and high load. Each scenario category corresponds to complete geological state data, equipment operation data, transportation load data, and corresponding model operation parameters. Then, a cosine similarity algorithm is used to calculate the similarity between the current real-time geological state, equipment operation data, and transportation load data and each scenario category in the historical scenario database. A preset similarity threshold of 0.8 is set, and scenario categories with a similarity greater than 0.8 are extracted as matching scenarios. The calibration parameter set corresponding to the matching scenario category is retrieved from a preset calibration rule library. This calibration parameter set includes model parameter calibration thresholds, causal transmission efficiency coefficients, and collaborative response time thresholds. The model parameter calibration thresholds are determined based on the optimal parameter range for equipment operation in the matching scenario; for example, the calibration threshold for excavator load rate is set to [60%, 85%]. The causal transmission efficiency coefficient is set based on the actual transmission effect of causal relationships in the matching scenario; for example, the causal transmission efficiency coefficient of geological conditions on equipment operation is set to 0.9. The collaborative response time threshold is set based on the response speed requirements of transportation scheduling to changes in equipment operation in the matching scenario; for example, it is set to 30 seconds. The calibration parameter set is applied to the causal correlation matrix. By adjusting the weight coefficients of corresponding elements in the matrix, model calibration rules suitable for the current scenario are generated. For example, if the current scenario is high-load production and complex geological conditions, the weight of the causal influence coefficient of geological conditions on equipment operation and transportation scheduling is appropriately increased to strengthen the constraint effect of geological changes on the model. Based on the generated model calibration rules, the operating parameters of the equipment intelligent adaptation model and the scheduling strategy of the transportation dynamic collaborative model are initially updated, enabling the model to initially adapt to the actual needs of the current mining scenario.

[0025] After the initial update is completed, a virtual simulation is performed on the initial update plan to identify potential conflicts and generate conflict resolution solutions. Using the current state of the mine's 3D geological dynamic model as initial conditions, the equipment performance limit parameters and transportation network topology are loaded into the simulation environment. The simulation environment is built using the Unity engine, enabling high-fidelity visualization simulation of the mine scene. The equipment performance limit parameters are determined based on the equipment's design specifications and actual operational test data. For example, the excavator's maximum load is set to 50 tons, and its maximum operating radius to 15 meters; the loader's maximum load capacity is set to 10 tons, and its maximum lifting height to 6 meters; the transport truck's maximum load capacity is set to 100 tons, and its maximum speed to 60 km / h. The transportation network topology is constructed based on the actual mine transportation road layout, including main roads, branch roads, loading and unloading points, and other key nodes, clearly defining the connection relationships between nodes and road capacity. The simulation environment proceeds sequentially over time, with a time step of 1 second, simulating the execution of the initial update plan and monitoring key parameters such as surrounding rock stress, equipment load rate, and transportation route congestion indicators in real time. The safety threshold for surrounding rock stress is set at 30 MPa, the safety threshold for equipment load rate is 90%, and the safety threshold for transportation route congestion index is an average vehicle waiting time of 10 minutes. If any parameter is detected to exceed the corresponding safety threshold, a potential conflict is determined.

[0026] When a potential conflict is detected, a multi-objective optimization solver is activated, with mining efficiency, safety threshold, cost control, and equipment lifespan as optimization objectives, generating at least two feasible conflict resolution solutions. First, the multi-objective optimization function is defined: ; Where x is the vector of decision variables to be optimized, including equipment operating parameter adjustment values, transportation route change plans, mining progress adjustment amounts, etc. ,in, This represents the loss of mining efficiency, defined as the difference between the updated mining efficiency and the optimal mining efficiency. The safety risk index is calculated by weighting the values ​​of excessive surrounding rock stress, excessive equipment load rate, and excessive transportation route congestion. This indicates the increase in operating costs, including the increase in equipment energy consumption, the increase in transportation costs, etc. It represents the equipment wear rate, calculated based on the degree of deviation between the equipment's operating parameters and its rated parameters; As the corresponding target weight vector, the weight of each target is determined using the analytic hierarchy process (AHP). Based on the mine's current needs of prioritizing safety, followed by efficiency, while balancing cost and equipment lifespan, the weights are set... , , , The Pareto front solution set is solved under the constraint g(x)≤0. Constraints include equipment physical limits (such as maximum equipment load and maximum speed), transportation network capacity (such as the maximum number of vehicles allowed on a road), and geological safety boundaries (such as the ultimate stress value of the surrounding rock). The NSGA-III algorithm is used to solve the Pareto front solution set. Solutions that satisfy the causal correlation matrix verification conditions are selected as conflict resolution schemes. The selection criterion is that the matching degree between the model output result corresponding to the solution and the causal relationship described by the causal correlation matrix is ​​greater than 0.9, ensuring that the conflict resolution scheme does not disrupt the inherent causal logic among the three elements. Through virtual simulation and conflict resolution, potential risks can be identified and avoided before the actual implementation of the scheme, ensuring the safety and stability of the mining process.

[0027] After generating conflict resolution solutions, incremental updates are performed on the mine's 3D geological dynamic model, equipment intelligent adaptation model, and transportation dynamic collaboration model based on these solutions, and a full-link traceability archive is established. First, a causal correlation matrix analysis is conducted to identify the model nodes involved in the changes within the conflict resolution solution and their causal impact range. For example, if the conflict resolution solution involves adjusting the excavator's operating position and the transport truck's travel path, the corresponding model nodes are the excavator operating parameter node in the equipment intelligent adaptation model and the path planning node in the transportation dynamic collaboration model. Affected related nodes include the surrounding rock stress calculation node for the corresponding operating area in the geological dynamic model and the vehicle scheduling node in the transportation dynamic collaboration model. Only these model nodes and affected related nodes are adjusted or structurally modified, eliminating the need for a comprehensive update of the entire model. This incremental update effectively reduces the computational and time costs of model updates and improves their efficiency.

[0028] A unique identifier is generated for this update operation, using the format "Update Date + Update Type + Random 6-Digit Number", such as "20240520-Conflict Resolution Update-123456". Simultaneously, relevant information is recorded in the end-to-end traceability archive. This information includes the update trigger source type (e.g., simulation pre-run finding excessive surrounding rock stress), the causal correlation matrix version number (e.g., V2.3), the matching calibration rule identifier (e.g., high-load complex geology calibration rule-001), the simulation pre-run input and output data (including initial model parameters, key monitoring data during simulation, conflict identification results, etc.), the decision variable values ​​of the conflict resolution scheme (e.g., excavator operating position adjustment coordinates, transport truck route change nodes, etc.), and the final adopted update instruction sequence (e.g., equipment parameter adjustment instructions, transport scheduling instructions, etc.). This end-to-end traceability archive provides a complete traceability basis for subsequent model maintenance, problem investigation, and optimization improvements, enhancing the traceability and maintainability of the technical solution.

[0029] Finally, the updated model's accuracy is evaluated using real-time feedback data from the physical mine, and a closed-loop evolution process is initiated if the evaluation results fail to meet the standards. An intelligent sensing system deployed in the physical mine continuously collects actual geological deformation data, real-time equipment operating condition data, and transport vehicle location data. This system includes deformation monitoring sensors deployed around the mining area, operating condition sensors installed on key equipment components, and vehicle positioning base stations arranged along transport routes. The data collection frequency is consistent with the data acquisition steps to ensure data timeliness and comparability. The actual collected data is compared with the output data of the virtual model in the same spatiotemporal coordinates to calculate the causal relationship matching degree. Data mapping accuracy and decision execution deviation rate Three evaluation indicators.

[0030] Causal relationship matching degree The calculation formula is: ; Where N is the number of sampling points, This represents the actual state change of the physical entity at the k-th sampling point. This refers to the state change predicted by the virtual model based on the causal correlation matrix. The value range is [0,1], and the closer it is to 1, the higher the degree of causal relationship matching.

[0031] Data mapping accuracy The calculation formula is: ; in, To effectively match the number of data points, that is, the number of data points where the deviation between the virtual model's output data and the actual collected data is within the allowable range, Total number of data points The value range is [0,1], and the closer it is to 1, the higher the accuracy of the data mapping.

[0032] Decision execution deviation rate The calculation formula is: ; Where M represents the number of dimensions of the decision-making action. For the actual action value of the decision in dimension m, Recommend action values ​​for the model making decisions in the m-th dimension. The value range is [0,1], and the closer it is to 0, the lower the decision execution deviation rate.

[0033] Set preset tolerance ranges for three evaluation indicators, among which... , , If any evaluation indicator exceeds the preset tolerance range, the accuracy evaluation is deemed to have failed.

[0034] When the accuracy assessment fails to meet the standard, a closed-loop evolutionary process is initiated. First, the deviation propagation path is traced backward based on the causal correlation matrix. A fishbone diagram analysis method is used to locate the root cause of the deviation. The root cause includes at least one of the following: data acquisition, causal modeling, calibration rule matching, or simulation pre-running. If the root cause is a data acquisition problem, the sampling frequency of the sensing equipment is adjusted or the sensor parameters are calibrated. For example, the sampling frequency of the surrounding rock stress sensor in a certain area is adjusted from 10Hz to 20Hz, or the load sensor of the equipment with accuracy drift is recalibrated. If the root cause is a causal modeling problem, the causal inference model is retrained based on the latest actual data, the model's hyperparameters are adjusted, and the causal influence coefficients in the causal correlation matrix are updated. If the root cause is a calibration rule problem, the current new scenario samples and corresponding optimal model parameters are added to the historical scenario database, and the calibration rule library is reconstructed using machine learning algorithms to optimize the matching logic of calibration parameters. If the root cause is a simulation pre-running problem, the physical parameter settings of the simulation environment are adjusted, the algorithm parameters of the multi-objective optimization solver are optimized, and the accuracy of conflict identification and resolution is improved. Using the optimized data, model, or rules as new input, the entire process from building the comprehensive dataset to incremental model updates is re-executed, completing a closed-loop evolutionary iteration. Through this closed-loop evolutionary process, model performance can be continuously optimized, the model's adaptability to physical mines can be constantly improved, and the model can be ensured to maintain high accuracy and reliability over the long term.

[0035] Through the implementation of the above steps, the constructed digital twin mine model in this embodiment can accurately depict the dynamic collaborative relationship between geological changes, equipment operation and transportation scheduling. The dynamic update capability and adaptability of the model are effectively improved, and the problem of multi-dimensional data collaborative mapping deviation in traditional models is effectively solved. Under the premise of ensuring mining safety, mining efficiency and production scheduling rationality are improved simultaneously.

[0036] Example 2 To enable those skilled in the art to fully understand and implement this invention, the specific implementation principle of this invention is further explained below in conjunction with a specific application scenario.

[0037] This embodiment selects an underground metal mine as the application scenario. The mine has a large mining depth, complex geological structure, and adverse geological bodies such as faults and fracture zones. During the mining process, it is necessary to pay close attention to the stability of the surrounding rock and the safe operation of equipment. Material transportation mainly relies on underground belt conveyors and mining trucks. The transportation route is greatly restricted by the layout of the mining area and geological conditions. This invention is used to model a digital twin mine and verify its practical application effect.

[0038] First, data acquisition is conducted. Geological survey data is obtained through underground drilling, geological logging, and acoustic testing, including ore body morphology, ore grade, surrounding rock lithology, fault location and scale, and rock mechanical parameters. Equipment operation data covers the operating parameters of underground rock drills, loaders, belt conveyors, and mining trucks, such as drilling depth, drilling speed, and impact pressure of the rock drills; conveying speed, conveying capacity, and motor temperature of the belt conveyors; and driving speed, load, and braking frequency of the mining trucks. Material transportation data includes the transportation volume of ore and waste rock, transportation routes, loading and unloading point locations, belt conveyor start and stop times, and mining truck dispatch records. All data acquisition equipment is explosion-proof and dustproof, adaptable to the harsh underground environment, and the acquired data is transmitted in real time to the ground data center via industrial Ethernet.

[0039] The collected multi-source data underwent spatiotemporal alignment and noise removal. For temporal alignment, Precise Time Protocol (PTP) was used to ensure that the timestamp error of data from different areas and types of equipment underground was controlled within 1 millisecond. Spatial alignment employed a combination of local and global coordinate system transformations. First, the equipment data from each area underground was transformed to the local coordinate system of that area, and then the local coordinate system was transformed to the mine's global coordinate system using regional control points, ensuring accurate spatial matching. For noise removal, outliers in the geological survey data were processed using a combination of a sliding window filter (with a window size of 8 sampling points) and an isolated forest algorithm. The sliding window filter first removed high-frequency noise, and then the isolated forest algorithm removed extreme outliers. For missing values ​​in equipment operation data and material transportation data, cubic spline interpolation was used to complete the missing values ​​based on the continuity of equipment operation and the logic of the transportation path, ultimately forming a high-quality comprehensive dataset.

[0040] Based on a comprehensive dataset, a three-dimensional geological dynamic model, an intelligent equipment adaptation model, and a transportation dynamic collaboration model for the mine were constructed. The three-dimensional geological dynamic model, built using Surpac software, clearly displays the spatial distribution and dynamic changes of geological bodies such as ore bodies, surrounding rocks, and faults. The intelligent equipment adaptation model optimizes the calculation models for parameters such as equipment load rate and energy consumption, taking into account the characteristics of the underground equipment's working environment. The transportation dynamic collaboration model, considering the complexity of the underground transportation network, uses graph theory algorithms to construct the transportation path topology, achieving optimal allocation of transportation resources. A causal correlation matrix is ​​generated using the Granger causality test algorithm, defining geological state variables. Including parameters such as changes in surrounding rock stress and fault activity intensity, and equipment state variables. Including parameters such as equipment failure rate and operating efficiency, and transportation status variables. Including parameters such as transportation capacity utilization and route congestion index, and setting a causal transmission threshold τ=0.35, the constructed causal correlation matrix effectively reflects the intrinsic relationship between underground geological changes, equipment operation and transportation scheduling.

[0041] Based on the causal correlation matrix and the current underground mining scenario, model calibration rules are generated, and the intelligent equipment adaptation model and the dynamic transportation coordination model are initially updated. The historical scenario database covers different mining stages such as underground development, preparation, and recovery, and different geological conditions such as normal geology, fault-affected areas, and fractured zones, as well as low, medium, and high production load conditions. The cosine similarity algorithm is used to match the current scenario and retrieve the corresponding calibration parameter set, including model parameter calibration thresholds, causal transmission efficiency coefficients, and collaborative response time thresholds. For example, the calibration threshold for equipment load rate is set to [55%, 80%], the causal transmission efficiency coefficient of geological conditions on transportation scheduling is set to 0.85, and the collaborative response time threshold is set to 45 seconds. The initial model update is completed based on the calibration rules.

[0042] A virtual simulation was conducted to preview the initial update plan. The simulation environment was built using the UE4 engine, recreating scenarios such as underground roadways, equipment layout, and transportation networks. Equipment performance limits were loaded, such as setting the maximum drilling depth of the rock drill to 30 meters, the maximum conveyor capacity of the belt conveyor to 1000 tons / hour, and the maximum load capacity of the mining truck to 80 tons. The surrounding rock stress safety threshold was set at 25 MPa, the equipment load rate safety threshold at 85%, and the transportation path congestion index safety threshold at an average vehicle waiting time of 15 minutes. During the simulation, fault activity in a certain area caused the surrounding rock stress to reach 28 MPa, exceeding the safety threshold, indicating a potential conflict. The multi-objective optimization solver is started, and a multi-objective optimization function is defined, where the objective weight vector w=[0.2,0.45,0.2,0.15]. Under the constraints of equipment physical limits, transportation network capacity, and geological safety boundaries, the NSGA-Ⅲ algorithm is used to solve the Pareto front solution set. Two feasible conflict resolution schemes are selected. Scheme 1 is to adjust the mining sequence in this area and mine the stable area on the other side of the fault first. Scheme 2 is to optimize the equipment operating parameters, reduce the mining intensity in this area, and adjust the transportation path to avoid stress concentration areas.

[0043] Based on conflict resolution scheme two, incremental updates are performed on the model. The model nodes involved in the changes are identified, including the surrounding rock stress calculation node in the geological dynamic model, the mining intensity parameter node in the equipment intelligent adaptation model, and the path planning node in the transportation dynamic collaboration model. Only these nodes and the affected related nodes are adjusted in terms of parameters. A unique identifier "20240610-Downhole Conflict Resolution Update-654321" is generated, and the update trigger source type, causal correlation matrix version number, matching calibration rule identifier, simulation pre-run data, conflict resolution scheme decision variable values, and update instruction sequence are recorded in the full-link traceability archive.

[0044] The system collects actual geological deformation data, real-time equipment operating data, and transport vehicle location data through an underground intelligent sensing system. This data is then compared with the output data from a virtual model to calculate three evaluation indicators. The results show the degree of causal relationship matching. The deviation was below the preset tolerance range of 0.85, indicating that the accuracy assessment did not meet the standard. A closed-loop evolutionary process was initiated, tracing the propagation path of the deviation back through the causal correlation matrix. The root cause was found to be a problem with causal modeling; specifically, the causal influence coefficient of geological conditions on equipment operation in the current causal correlation matrix did not fully consider the suddenness of fault activity. To address this issue, the causal inference model was retrained based on the latest collected fault activity data and equipment operation data. The model hyperparameters were adjusted, and the corresponding elements in the causal correlation matrix were updated. The optimized causal correlation matrix was then used as the new input, and the entire process from comprehensive dataset construction to incremental model updates was re-executed, completing the closed-loop evolutionary iteration.

[0045] To further verify the application effect of the present invention, a comparative analysis of the model performance before and after implementation was conducted, and the relevant data are shown in the table below:

[0046] As can be seen from the table, after the implementation of this invention, the digital twin mine model has been significantly improved in terms of the accuracy of collaborative relationship characterization, dynamic update adaptability, decision support effectiveness, and safety risk identification capability. It effectively solves the problem of multi-dimensional data collaborative mapping deviation in complex mining scenarios of underground metal mines, providing strong support for safe and efficient mining. Through practical application verification, this invention has good practicality and adaptability, can meet the needs of digital transformation of different types of mines, and has broad application value.

[0047] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0048] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for modeling mines based on digital twins, characterized in that, include: S1. Collect geological survey data, equipment operation data, and material transportation data of the mine; S2. Perform spatiotemporal alignment and noise removal on the geological survey data, equipment operation data, and material transportation data to form a comprehensive dataset; S3. Based on the comprehensive dataset, construct a three-dimensional geological dynamic model of the mine, an intelligent equipment adaptation model, and a transportation dynamic coordination model, and generate a causal correlation matrix representing the dynamic causal relationship between geological changes, equipment operation, and transportation scheduling through a causal inference algorithm. S4. Based on the causal correlation matrix and the current mining scenario, generate model calibration rules and perform preliminary updates to the equipment intelligent adaptation model and the transportation dynamic collaboration model; S5. Perform virtual simulation on the preliminary update scheme to identify potential conflicts and generate conflict resolution schemes; S6. Based on the conflict resolution scheme, perform incremental updates on the three-dimensional geological dynamic model of the mine, the intelligent equipment adaptation model, and the transportation dynamic collaborative model, and establish a full-link traceability archive. S7. The updated model is evaluated for accuracy using real-time feedback data from the physical mine, and a closed-loop evolution process is initiated if the evaluation results do not meet the standards.

2. The method for modeling mines based on digital twins according to claim 1, characterized in that, S2 specifically includes: Use timestamp synchronization protocols to perform time-series alignment of multi-source data; A spatial coordinate transformation algorithm is used to unify data from different coordinate systems to the mine's global coordinate system; Apply a sliding window filter or the isolated forest algorithm to identify and remove outlier data points; Based on the geological correlation of adjacent areas, the trend of equipment operation status, and the logic of transportation routes, missing data is interpolated and completed to form the comprehensive dataset.

3. The method for modeling mines based on digital twins according to claim 1, characterized in that, In step S3, a causal correlation matrix representing the dynamic causal relationship among geological changes, equipment operation, and transportation scheduling is generated using a causal inference algorithm. Specifically, this includes: Define geological state variables Equipment state variables and transportation state variables Observations at time step t; Granger causality tests or structural equation modeling are used to calculate the causal strength between variables; Constructing a causal relationship matrix , of which elements Let represent the causal influence coefficient of the i-th state on the j-th state, where i,j∈{G,M,T}; Set a causal transmission threshold τ, when When this is confirmed, a valid causal link from state i to state j is confirmed to exist.

4. The method for modeling mines based on digital twins according to claim 1, characterized in that, In step S4, model calibration rules are generated based on the causal correlation matrix and the current mining scenario, specifically including: A historical scenario database containing different mining stages, geological conditions, and production loads is pre-built; Extract scene categories from the historical scene database that have a similarity to the current real-time geological conditions, equipment operation data, and transportation load data that meets a preset threshold. The calibration parameter set corresponding to the scenario category is retrieved from the preset calibration rule library. The calibration parameter set includes model parameter calibration threshold, causal transmission efficiency coefficient and cooperative response time threshold. The calibration parameter set is applied to the causal correlation matrix to generate model calibration rules suitable for the current scenario.

5. The method for modeling mines based on digital twins according to claim 1, characterized in that, S5 specifically includes: Using the current state of the three-dimensional geological dynamic model of the mine as the initial conditions, load the equipment performance limit parameters and the transportation network topology; The initial update plan is simulated and executed in a time-series manner within a simulation environment. During the monitoring and simulation process, check whether the surrounding rock stress exceeds the limit, the equipment load rate exceeds the standard, or the transportation path congestion index exceeds the preset safety threshold. If any exceedance is detected, a potential conflict is determined, and a multi-objective optimization solver is activated. With mining efficiency, safety threshold, cost control, and equipment life as optimization objectives, at least two feasible conflict resolution solutions are generated.

6. The method for modeling mines based on digital twins according to claim 5, characterized in that, The multi-objective optimization solver is activated, with mining efficiency, safety threshold, cost control, and equipment lifespan as optimization objectives, to generate multiple feasible conflict resolution solutions, specifically including: Define a multi-objective optimization function: ； Where x is the vector of decision variables to be optimized. These represent mining efficiency loss, safety risk index, increase in operating costs, and equipment wear rate, respectively. This is the corresponding target weight vector; Solve the Pareto front solution set under the constraint g(x)≤0, where the constraints include equipment physical limits, transportation network capacity, and geological safety boundaries; The solutions that satisfy the causal correlation matrix verification conditions are selected from the Pareto front solution set as the conflict resolution schemes.

7. The method for modeling mines based on digital twins according to claim 1, characterized in that, S6 specifically includes: Identify the model nodes involved in the changes in the conflict resolution scheme and the scope of their causal relationships; Incremental updates are completed by adjusting parameters or modifying structures only on the model nodes and the affected associated nodes. Generate a unique identifier for this update operation and record the following information in the traceability archive: update trigger source type, causal relationship matrix version number, matching calibration rule identifier, simulation pre-run input and output data, conflict resolution scheme decision variable values, and the final adopted update instruction sequence.

8. The method for modeling mines based on digital twins according to claim 1, characterized in that, In step S7, the accuracy of the updated model is evaluated using real-time feedback data from the physical mine, specifically including: Through intelligent sensing systems deployed in physical mines, actual geological deformation data, real-time equipment operating data, and transport vehicle location data are continuously collected. The actual collected data is compared with the output data of the virtual model in the same spatiotemporal coordinates; Calculate the degree of causal relationship matching Data mapping accuracy and decision execution deviation rate Three evaluation indicators; If any evaluation indicator exceeds the preset tolerance range, the accuracy evaluation is deemed to have failed.

9. The method for modeling mines based on digital twins according to claim 8, characterized in that, The calculation of causal relationship matching degree Data mapping accuracy and decision execution deviation rate The three evaluation indicators include: Calculate the degree of causal relationship matching for: ； Where N is the number of sampling points, This represents the actual state change of the physical entity at the k-th sampling point. The state change predicted by the virtual model based on the causal correlation matrix; Calculate data mapping accuracy To effectively match the ratio of the number of data points to the total number of data points; Calculate the decision execution deviation rate This is the normalized Euclidean distance between the actual action performed and the action recommended by the model.

10. The method for modeling mines based on digital twins according to claim 1, characterized in that, In S7, a closed-loop evolution process is initiated when the evaluation result fails to meet the standard, specifically including: Based on the causal correlation matrix, the deviation propagation path is traced in reverse to locate the root cause of the deviation. The root cause of the deviation includes at least one of data acquisition, causal modeling, calibration rule matching, or simulation pre-run. Targeted optimizations are performed on the root causes of the deviations: if the problem is data acquisition, the sampling frequency of the sensing device is adjusted or the sensor parameters are calibrated; if the problem is causal modeling, the causal inference model is retrained and the causal correlation matrix is ​​updated; if the problem is calibration rules, new scene samples are added to the historical scene database and the calibration rule library is reconstructed. Using the optimized data, model, or rules as new input, the entire process from building a comprehensive dataset to incrementally updating the model is re-executed, completing a closed-loop evolutionary iteration.

Images