Geological restoration method for mining area

Abstract

The invention provides a mining area geology restoration method. The mining area geology restoration method comprises the steps that environmental parameters and reaction state data of a to-be-restored mining area are acquired through a sensing network; based on the environmental parameters and the reaction state data, digital twinborn bodies are constructed and run, and the digital twinborn bodies are used for simulating the migration of carbon dioxide in the to-be-restored mining area and the reaction process of the carbon dioxide and minerals; generating an injection strategy for controlling a carbon sequestration reaction process based on a simulation result of the digital twin; according to the injection strategy, injection equipment is controlled to inject a reaction medium into the to-be-repaired mining area; and performing monitoring and fault prediction on the operation state of the injection equipment, and generating a maintenance instruction. According to the invention, intelligence, precision, high efficiency and multi-target collaboration of the geological restoration process of the mining area are realized.

Description

Technical Field

[0001] This application relates to the field of mine ecological geological restoration, and in particular to a method for geological restoration of mining areas. Background Technology

[0002] In recent years, mineral carbon sequestration (MCS) technology has attracted widespread attention as an emerging geochemical remediation method. Its basic principle is to utilize abundant silicate minerals (such as olivine and serpentine) in mining areas to react with industrially captured carbon dioxide, generating stable carbonate minerals (such as magnesite and calcite), thereby achieving permanent carbon dioxide sequestration while simultaneously neutralizing acidic wastewater and passivating heavy metal ions. This method is considered a highly promising "waste-to-resource" remediation strategy that can address both climate change and the mining environment.

[0003] However, existing mineral carbon sequestration technologies suffer from extensive process control, lacking detailed insights and forward-looking regulation of processes such as carbon dioxide transport, reaction heat effects, and fluid-mineral interactions in complex underground porous media. This results in slow reaction rates, low sequestration efficiency, and high consumption of reagents and energy, leading to poor economic viability. Secondly, the underground environment of mining areas is highly heterogeneous and uncertain. Traditional methods struggle to effectively predict and prevent engineering and environmental risks such as injection well blockage, microseismic events induced by excessive pore pressure, and excessive acidification of the reaction zone, making it difficult to guarantee the safety and reliability of system operation. Furthermore, most existing technologies focus solely on carbon sequestration itself, failing to organically integrate the remediation process with the recycling of waste resources in the mining area.

[0004] In summary, existing mineral carbon sequestration technologies generally lack sufficient understanding and control over the complex system of a mining area. They can only perform "blind injection" or passive responses based on limited point data, resulting in core defects such as low sequestration efficiency and high consumption of reagents and energy. There is an urgent need for a new geological remediation method for mining areas to solve these problems. Summary of the Invention

[0005] In view of the aforementioned problems, this application is made in order to provide a geological remediation method for mining areas that overcomes or at least partially solves the aforementioned problems.

[0006] This application discloses a method for geological remediation of a mining area, including the following steps: The environmental parameters and reaction status data of the mining area to be restored are obtained through a sensor network. The sensor network includes pH, Eh, humidity, temperature and pressure sensors deployed at the mining site, as well as periodic exploration data obtained through UAV remote sensing and geophysical exploration. Based on the environmental parameters and reaction status data, a digital twin is constructed and run. The digital twin is used to simulate the migration of carbon dioxide in the mining area to be restored and its reaction with minerals. Based on the simulation results of the digital twin, an injection strategy for controlling the carbon sequestration reaction process is generated; According to the injection strategy, the injection equipment is controlled to inject the reaction medium into the mining area to be repaired; The operating status of the injection equipment is monitored and faults are predicted, and maintenance instructions are generated.

[0007] Furthermore, the digital twin is a multiphysics coupling model based on geology, geochemistry, and fluid dynamics. The step of constructing and running the digital twin based on the environmental parameters and reaction state data specifically includes the following steps: By integrating real-time data and periodic exploration data from the sensor network, a high-precision three-dimensional geological attribute model is constructed using a geostatistical interpolation algorithm. The three-dimensional geological attribute model includes a porosity distribution field, a permeability distribution field, a mineral composition distribution field, and an initial hydrochemical field. A multiphysics coupled mathematical model is established, and the feedback mechanism of bidirectional coupling between fluid dynamics equations, convection-diffusion-reaction equations, geochemical equations and energy conservation equations, as well as the corresponding initial and boundary conditions, are defined. The finite element numerical analysis method is used to discretize and iteratively solve the coupled nonlinear partial differential equations. The simulation is advanced in a time-step manner to dynamically simulate the entire process of plume migration, chemical reaction, temperature change and metal ion activation after carbon dioxide injection. The real-time data of the sensor network is compared with the simulation predictions at the corresponding time step and location. If the error exceeds the preset threshold, the ensemble Kalman filter is activated to back-optimize and adjust the key uncertainty parameters in the model. The calculation results of each time step are extracted to generate a dynamically evolving four-dimensional visualization field, and macroscopic performance parameters are calculated. The four-dimensional visualization field includes four-dimensional images of carbon dioxide saturation field, pH value field, secondary mineral precipitation field and metal enrichment field.

[0008] Furthermore, the step of generating an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin specifically includes the following steps: The simulation results of the digital twin are analyzed and features are extracted to calculate the carbon sequestration rate, reaction efficiency and metal enrichment data, and a comprehensive characterization vector is generated. The reinforcement learning agent receives the comprehensive representation vector, quantifies and evaluates the long-term benefits of different control actions through a reward function, and outputs the optimal injection strategy. The control action is a set of control instructions. The optimal injection strategy is simulated in the digital twin, and the simulation results are compared with the preset security operation boundaries. When the simulation results exceed the safety boundary, the optimal injection strategy is regenerated and the simulation is performed again until an optimal injection strategy that can pass the security check is generated.

[0009] Furthermore, the safe operating boundaries include the maximum allowable wellhead pressure, the allowable pH fluctuation range, the maximum instantaneous carbon dioxide injection rate, and the maximum formation pore pressure threshold. The step of regenerating the optimal injection strategy and performing another simulation when the simulation results exceed the safe boundaries, until an optimal injection strategy that passes the safety verification is generated, specifically includes the following steps: When the simulation results exceed the security boundary, a security verification failure signal is issued along with information indicating that the limit has been exceeded. The current injection strategy is then marked as an invalid strategy and discarded. After receiving a security check failure signal, the reinforcement learning agent applies a corresponding penalty to its decision-making logic based on the information of exceeding the limit, thereby generating a more conservative new strategy. The new strategy was simulated again in the digital twin, and the simulation results were verified for security. When a pre-run result shows that all indicators remain within the safety boundary throughout the pre-run period, a safety verification pass signal is issued, and the strategy is marked as the optimal injection strategy output.

[0010] Furthermore, after the step of issuing a security verification pass signal and marking the strategy as the optimal injection strategy output when a certain pre-simulation result shows that all indicators remain within the safety boundary throughout the pre-simulation period, the method further includes the step of: When the number of iterations of the injection strategy exceeds a preset threshold, the security verification loop is terminated and it is determined that there is no feasible solution under the current constraints. Package all rejected strategies' detailed parameters and their exceeding information to generate a diagnostic report; The diagnostic report is pushed to the engineer through the enterprise communication platform, and the engineer makes a decision and develops a specific human intervention plan.

[0011] Furthermore, the step of controlling the injection device to inject the reaction medium into the mining area to be repaired according to the injection strategy specifically includes the following steps: The injection strategy is compiled into standardized industrial control instructions, and an independent task instruction package containing device ID, target parameters and execution time is generated for each smart injection well. The task instruction packet is sent to the programmable logic controller of the target injection well via industrial Ethernet, and the programmable logic controller returns an acknowledgment signal after completing the instruction verification. The programmable logic controller drives the connected actuators to perform precise actions according to the instructions. During the process, the sensors integrated into the injection well body continuously collect actual operating data; The closed-loop control algorithm compares the deviation between the actual operating data and the target value, and automatically issues fine-tuning instructions to accurately match the target.

[0012] Furthermore, the step of monitoring the operating status and predicting faults of the injection device, and generating maintenance instructions, specifically includes the following steps: Continuously collect real-time operating data such as vibration, pressure, and flow of the injection equipment, and correlate them with environmental data such as media properties. Through feature extraction, key indicators characterizing the health status of the equipment are obtained. Using the device's digital twin, the current operating characteristics are compared with the health baseline in real time. When the characteristic value is found to deviate continuously from the preset threshold, a primary anomaly alarm is triggered. Based on the Long Short-Term Memory Network Fault Prediction Model, time series data is analyzed to predict the probability of a specific fault occurring in the future and its remaining useful life. The prediction results are input into the decision rule engine, which automatically matches the optimal maintenance strategy based on the fault type, probability, and RUL, and generates structured instructions that include maintenance objects, types, resources, and operation guidelines. Maintenance instructions are automatically distributed to the responsible team or autonomous robot in the form of work orders, and feedback data is collected after maintenance is completed to update the predictive model and digital twin, thereby achieving closed-loop learning and continuous optimization.

[0013] This application also discloses a geological restoration system for mining areas, including: The data collection module is used to acquire environmental parameters and reaction status data of the mining area to be repaired through a sensor network. The sensor network includes pH, Eh, humidity, temperature and pressure sensors deployed at the mining site, as well as periodic exploration data acquired through UAV remote sensing and geophysical exploration. A digital twin construction module is used to construct and run a digital twin based on the environmental parameters and reaction state data. The digital twin is used to simulate the migration of carbon dioxide in the mining area to be restored and its reaction with minerals. An injection strategy generation module is used to generate an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin. An execution module is used to control the injection device to inject a reaction medium into the mining area to be repaired, according to the injection strategy. The injection equipment monitoring module is used to monitor the operating status of the injection equipment and predict faults, and generate maintenance instructions.

[0014] This application also discloses a computer device, including a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, it implements the steps of a geological restoration method for a mining area as described above.

[0015] This application also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a geological restoration method for a mining area as described above.

[0016] This application has the following advantages: In the embodiments of this application, addressing the core shortcomings of existing technologies such as low carbon sequestration efficiency and high consumption of reagents and energy, this application provides a method for geological remediation of mining areas, specifically comprising: acquiring environmental parameters and reaction state data of the mining area to be remediated through a sensor network, wherein the sensor network includes pH, Eh, humidity, temperature, and pressure sensors deployed at the mining site, as well as periodic exploration data acquired through UAV remote sensing and geophysical exploration; constructing and operating a digital twin based on the environmental parameters and reaction state data, wherein the digital twin is used to simulate the migration of carbon dioxide in the mining area to be remediated and its reaction process with minerals; generating an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin; controlling the injection equipment to inject reaction media into the mining area to be remediated according to the injection strategy; monitoring the operating status of the injection equipment and predicting faults, and generating maintenance instructions. By acquiring environmental parameters and reaction state data of the mining area to be restored through sensor networks, the shortcomings of existing technologies in "insufficient understanding of complex porous underground media" are addressed, achieving the effect of "comprehensive, multi-dimensional, real-time, and accurate perception of the restoration area." By "constructing and operating a digital twin based on the environmental parameters and reaction state data," the shortcomings of existing technologies in "lacking detailed insights into processes such as carbon dioxide transport, reaction heat effects, and fluid-mineral interactions" are addressed, achieving the effect of "high-fidelity reproduction and prediction of the reaction process of physical entities in virtual space, realizing in-depth analysis of multi-physics coupling processes under complex geological conditions." By "generating an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin," the shortcomings of existing technologies in "process control" are addressed. Overcoming the shortcomings of "extensive and inefficient control" in existing technologies, this system achieves the effect of "intelligent generation and dynamic optimization of injection strategies through artificial intelligence optimization algorithms, significantly improving reaction rate and sealing efficiency, and reducing reagent and energy consumption." By "controlling the injection equipment to inject reaction media into the mine to be repaired according to the injection strategy," it solves the shortcomings of "experience-dependent operation and poor execution accuracy" in existing technologies, achieving the effect of "precise and automated execution of the repair process, ensuring accurate implementation of the optimized strategy." Furthermore, by "monitoring the operating status of the injection equipment and predicting faults, and generating maintenance instructions," it solves the shortcomings of "difficulty in predicting and preventing engineering risks" in existing technologies, achieving the effect of "predictive maintenance, ensuring long-term stable and reliable system operation, and significantly improving safety and reliability." Attached Figure Description

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

[0018] Figure 1 This is a flowchart illustrating the steps of a geological restoration method for a mining area according to an embodiment of this application; Figure 2 This is a structural block diagram of a geological restoration system for mining areas provided in one embodiment of this application; Figure 3 This is a schematic diagram of the structural block of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, features, and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] The inventors discovered through analysis of existing technology that: 1. Existing mineral carbon sequestration technologies have extensive process control and lack detailed insights and forward-looking regulation of processes such as carbon dioxide transport, reaction heat effects, and fluid-mineral interactions in complex underground porous media. This results in slow reaction rates, low sequestration efficiency, and high consumption of reagents and energy, making them uneconomical.

[0021] 2. The underground environment of the mining area is highly heterogeneous and uncertain. Traditional methods are difficult to effectively predict and prevent engineering and environmental risks such as injection well blockage, micro-earthquakes induced by excessive pore pressure, and excessive acidification in the reaction zone. The safety and reliability of the system operation are difficult to guarantee.

[0022] 3. Most existing technologies focus only on carbon sequestration itself and fail to organically combine the remediation process with the recycling of waste resources in mining areas.

[0023] Reference Figure 1 The diagram shows a flowchart of the steps of a geological restoration method for a mining area provided in an embodiment of this application. This application provides a method for geological restoration of a mining area, the method comprising the following steps: S110: Obtain environmental parameters and reaction status data of the mining area to be repaired through a sensor network. The sensor network includes pH, Eh, humidity, temperature, and pressure sensors deployed at the mining site, as well as periodic exploration data obtained through UAV remote sensing and geophysical exploration. S120: Based on the environmental parameters and reaction status data, construct and run a digital twin, which is used to simulate the migration of carbon dioxide in the mining area to be restored and its reaction with minerals; S130: Based on the simulation results of the digital twin, generate an injection strategy for controlling the carbon sequestration reaction process; S140: According to the injection strategy, control the injection device to inject the reaction medium into the mining area to be repaired; S150: Monitor the operating status of the injection equipment and predict faults, and generate maintenance instructions.

[0024] In the embodiments of this application, addressing the core shortcomings of existing technologies such as low carbon sequestration efficiency and high consumption of reagents and energy, this application provides a method for geological remediation of mining areas, specifically comprising: acquiring environmental parameters and reaction state data of the mining area to be remediated through a sensor network, wherein the sensor network includes pH, Eh, humidity, temperature, and pressure sensors deployed at the mining site, as well as periodic exploration data acquired through UAV remote sensing and geophysical exploration; constructing and operating a digital twin based on the environmental parameters and reaction state data, wherein the digital twin is used to simulate the migration of carbon dioxide in the mining area to be remediated and its reaction process with minerals; generating an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin; controlling the injection equipment to inject reaction media into the mining area to be remediated according to the injection strategy; monitoring the operating status of the injection equipment and predicting faults, and generating maintenance instructions. By acquiring environmental parameters and reaction state data of the mining area to be restored through sensor networks, the shortcomings of existing technologies in "insufficient understanding of complex porous underground media" are addressed, achieving the effect of "comprehensive, multi-dimensional, real-time, and accurate perception of the restoration area." By "constructing and operating a digital twin based on the environmental parameters and reaction state data," the shortcomings of existing technologies in "lacking detailed insights into processes such as carbon dioxide transport, reaction heat effects, and fluid-mineral interactions" are addressed, achieving the effect of "high-fidelity reproduction and prediction of the reaction process of physical entities in virtual space, realizing in-depth analysis of multi-physics coupling processes under complex geological conditions." By "generating an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin," the shortcomings of existing technologies in "process control" are addressed. Overcoming the shortcomings of "extensive and inefficient control" in existing technologies, this system achieves the effect of "intelligent generation and dynamic optimization of injection strategies through artificial intelligence optimization algorithms, significantly improving reaction rate and sealing efficiency, and reducing reagent and energy consumption." By "controlling the injection equipment to inject reaction media into the mine to be repaired according to the injection strategy," it solves the shortcomings of "experience-dependent operation and poor execution accuracy" in existing technologies, achieving the effect of "precise and automated execution of the repair process, ensuring accurate implementation of the optimized strategy." Furthermore, by "monitoring the operating status of the injection equipment and predicting faults, and generating maintenance instructions," it solves the shortcomings of "difficulty in predicting and preventing engineering risks" in existing technologies, achieving the effect of "predictive maintenance, ensuring long-term stable and reliable system operation, and significantly improving safety and reliability."

[0025] The following will further describe a geological restoration method for a mining area in this exemplary embodiment.

[0026] As described in step S120, a digital twin is constructed and run based on the environmental parameters and reaction state data. The digital twin is used to simulate the migration of carbon dioxide in the mining area to be restored and its reaction with minerals.

[0027] In one embodiment of the present invention, the specific process of "constructing and running a digital twin based on the environmental parameters and reaction state data" in step S120 can be further described in conjunction with the following description.

[0028] As described in the following steps By integrating real-time data and periodic exploration data from the sensor network, a high-precision three-dimensional geological attribute model is constructed using a geostatistical interpolation algorithm. The three-dimensional geological attribute model includes a porosity distribution field, a permeability distribution field, a mineral composition distribution field, and an initial hydrochemical field. A multiphysics coupled mathematical model is established, and the feedback mechanism of bidirectional coupling between fluid dynamics equations, convection-diffusion-reaction equations, geochemical equations and energy conservation equations, as well as the corresponding initial and boundary conditions, are defined. The finite element numerical analysis method is used to discretize and iteratively solve the coupled nonlinear partial differential equations. The simulation is advanced in a time-step manner to dynamically simulate the entire process of plume migration, chemical reaction, temperature change and metal ion activation after carbon dioxide injection. The real-time data of the sensor network is compared with the simulation predictions at the corresponding time step and location. If the error exceeds the preset threshold, the ensemble Kalman filter is activated to back-optimize and adjust the key uncertainty parameters in the model. The calculation results of each time step are extracted to generate a dynamically evolving four-dimensional visualization field, and macroscopic performance parameters are calculated. The four-dimensional visualization field includes four-dimensional images of carbon dioxide saturation field, pH value field, secondary mineral precipitation field and metal enrichment field.

[0029] It should be noted that the digital twin is a multiphysics coupled model based on geology, geochemistry, and fluid dynamics. The "ensemble Kalman filter" is a sequential data assimilation algorithm that maintains a set of model states to characterize the prediction error covariance. When observation data (i.e., real-time sensor data) arrives, the algorithm updates the entire set of states by calculating the Kalman gain, ensuring that the updated set of states conforms to the model dynamics and approximates the observation data. This process effectively integrates observation information into the model, thereby reducing simulation errors caused by uncertainties in model parameters (such as permeability and reaction rate constant), and significantly improving the prediction accuracy and reliability of the digital twin.

[0030] As an example, the "multiphysics coupling" can be implemented in different ways, such as: combining all governing equations into a large system of equations for simultaneous solution through direct coupling (full coupling); solving each physical field sequentially through sequential coupling (iterative coupling), and iteratively passing coupling variables at each time step until convergence; or decomposing the complex multiphysics problem into a series of single-physics problems for approximate solution through operator splitting. The generation and display of the "four-dimensional visualization field" can be achieved through various technical approaches, such as: post-processing using open-source scientific visualization software like ParaView and VisIt; remote real-time monitoring through an interactive web visualization platform based on WebGL technology; or immersive data visualization and analysis using virtual reality (VR) devices, thereby providing engineers with intuitive and powerful tools for understanding complex underground processes.

[0031] In one specific implementation, the "finite element numerical analysis method" can be implemented using an unstructured mesh: First, a mesh is generated based on the generated three-dimensional geological model, and local mesh refinement technology is used for key areas such as complex geological bodies (e.g., faults, fractures) and the area around injection wells; then, the partial differential equation system is transformed into a large sparse linear algebraic equation system using the Galerkin weighted residual method; finally, combined with a high-performance computing cluster, iterative algorithms such as the preprocessed conjugate gradient method are used for efficient solution, thereby achieving high-resolution simulation of multi-field coupling processes under large-scale complex geological conditions.

[0032] As described in step S130, an injection strategy for controlling the carbon sequestration reaction process is generated based on the simulation results of the digital twin.

[0033] In one embodiment of the present invention, the specific process of "generating an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin" in step S130 can be further described in conjunction with the following description.

[0034] As described in the following steps The simulation results of the digital twin are analyzed and features are extracted to calculate the carbon sequestration rate, reaction efficiency and metal enrichment data, and a comprehensive characterization vector is generated. The reinforcement learning agent receives the comprehensive representation vector, quantifies and evaluates the long-term benefits of different control actions through a reward function, and outputs the optimal injection strategy. The control action is a set of control instructions. The optimal injection strategy is simulated in the digital twin, and the simulation results are compared with the preset security operation boundaries. When the simulation results exceed the safety boundary, the optimal injection strategy is regenerated and the simulation is performed again until an optimal injection strategy that can pass the security check is generated.

[0035] It should be noted that the "comprehensive characterization vector" is a process of condensing the multi-dimensional, multi-scale simulation data output by the digital twin into a machine-understandable and information-rich low-dimensional vector. This vector includes not only macroscopic performance indicators (such as cumulative carbon sequestration) but also key statistics that reflect the spatial distribution characteristics of underground reactions (such as the spatial variance of the pH field and the spatial autocorrelation length of carbon dioxide saturation), thus providing a comprehensive and compact state description for subsequent intelligent decision-making. The "safe operating boundary" is a set of hard constraints set to prevent engineering failure and environmental risks. These boundaries include not only equipment-level hard limits (such as the maximum rated pressure of the injection pump and the maximum opening of the valve) but also geological environment-level soft constraints (such as the upper limit of formation pore pressure set to prevent the induction of microseismic events and the pH fluctuation range set to protect the stability of the surrounding rock). The pre-simulation comparison is to verify the time-series curves of key parameters simulated by the digital twin under strategy-driven conditions against these boundary values ​​one by one.

[0036] As an example, the "reinforcement learning agent" can be implemented through different algorithmic architectures, such as: using algorithms suitable for continuous action spaces like Deep Deterministic Policy Gradient (DDPG) or Soft Actor-Critic (SAC); using Proximal Policy Optimization (PPO) to ensure the stability of the training process; or based on a multi-agent reinforcement learning framework, assigning the task of controlling different injection wells to a cooperative subset of agents to achieve distributed collaborative optimization. The "regeneration of the optimal injection policy" can be implemented through various optimization strategies, such as: using constrained policy optimization methods, directly embedding the safety boundary as a constraint in the policy search space; using a penalty function method, greatly increasing the negative reward for violating the safety boundary in the reward function to guide the agent to actively avoid dangerous areas; or using a projection method, projecting the unsafe policy onto the nearest valid point in the set of safe policies.

[0037] In one specific implementation, the "reward function" can be implemented through a multi-objective weighted summation. The design formula for the reward function R is: R = W_1*ΔCarbonation + W_2*ΔMetalEnrichment - W_3*EnergyCost - W_4*ChemicalCost - W_5*pHViolationPenalty; where ΔCarbonation represents the amount of carbonate minerals generated in this time step, ΔMetalEnrichment represents the increase in the target metal concentration, EnergyCost and ChemicalCost represent energy consumption and reagent cost, respectively, pHViolationPenalty is a penalty term for pH values ​​exceeding the optimal reaction range, and W_1 to W_5 are weighting coefficients pre-set according to the priority of the repair objectives. The goal of the agent is to maximize the expected value of future cumulative rewards through learning, thereby automatically exploring a control strategy that optimally balances storage efficiency, resource recovery, and economic costs.

[0038] As described in the following steps, when the simulation results exceed the security boundary, the optimal injection strategy is regenerated and the simulation is performed again until an optimal injection strategy that can pass the security check is generated.

[0039] In one embodiment of the present invention, the specific process of the above step "when the pre-simulation result exceeds the security boundary, regenerate the optimal injection strategy and perform pre-simulation again until an optimal injection strategy that can pass the security check is generated" can be further described in conjunction with the following description: S1341: When the pre-simulation result exceeds the security boundary, a security verification failure signal is issued along with the over-limit information, and the current injection strategy is marked as an invalid strategy and discarded. S1342: After receiving a security check failure signal, the reinforcement learning agent applies a corresponding penalty to its decision-making logic based on the exceeding information, and generates a more conservative new strategy. S1343: Perform a pre-simulation of the new strategy in the digital twin again, and perform a security verification on the pre-simulation results; S1344: When a pre-run result shows that all indicators remain within the safety boundary throughout the pre-run period, a safety verification pass signal is issued, and the strategy is marked as the optimal injection strategy output.

[0040] It should be noted that the safety operation boundaries include the maximum allowable wellhead pressure, the allowable pH fluctuation range, the maximum instantaneous carbon dioxide injection rate, and the maximum formation pore pressure threshold. The "exceedance information" is a structured data packet that not only indicates which safety indicators (such as wellhead pressure and pH) have exceeded the limits, but also records in detail the specific values ​​of the exceedances, the time of occurrence, the duration, and their location in three-dimensional space. This detailed information provides precise feedback guidance for the agent's subsequent strategy adjustments, rather than a simple binary pass / fail signal. Steps S1343 and S1344 constitute an internal optimization loop, which is conducted entirely in the digital space. A digital twin replaces the real environment in bearing the risks that may arise from poor strategies, realizing a zero-risk learning paradigm of "trial and error in the virtual world, execution in reality." The loop terminates when a strategy is found that is verified as absolutely safe by the digital twin, or when a preset maximum number of iterations is reached, triggering an upper-level alarm mechanism.

[0041] As an example, the "imposing corresponding penalties in its decision-making logic" can be implemented through various mechanisms, such as: dynamically adding a large negative reward term proportional to the severity of exceeding the limit to the reward function; adding a constraint layer to the output layer of the policy network to directly limit the policy's exploration in dangerous directions; modifying the environmental state representation to include the history of exceeding the limit as part of the state input, enabling the agent to "remember" and avoid decision sequences that lead to failure; or using a model-based safety layer to project unsafe actions onto a set of verified safe actions. The determination in step S1344 that "all indicators remain within the safety boundary throughout the entire pre-test period" can be accomplished through rigorous mathematical calculations. For example, for the time series data of each safety indicator, the algorithm checks whether its maximum value (for upper limit indicators such as pressure) or minimum value (for lower limit indicators such as pH) on the entire pre-test time axis still falls within the corresponding safe range. This requires not only that the instantaneous value does not exceed the limit, but also that its changing trend will not cause the risk of exceeding the limit in the near future, thereby ensuring the long-term safety of the policy.

[0042] In one specific implementation, the "generation of a more conservative new policy" can be implemented through constrained policy optimization: after receiving a security check failure signal, the agent's policy optimization objective changes from simply maximizing the cumulative reward to maximizing the reward under the condition of satisfying security constraints. Specifically, the algorithm estimates the constraint cost function (i.e. the expected degree of insecurity) under the current policy and ensures that the expected constraint cost of the new policy is lower than a threshold at each step of the policy update. This drives the policy parameters to be updated in the direction of maintaining high performance while satisfying all security constraints, thereby systematically generating a more conservative but safe new policy.

[0043] As described in the following steps, when a pre-run result shows that all indicators remain within the safety boundary throughout the pre-run period, a safety verification pass signal is issued, and the strategy is marked as the optimal injection strategy output.

[0044] In one embodiment of the present invention, the following description can be used to further illustrate the above-mentioned step "when a certain pre-run result shows that all indicators remain within the safety boundary throughout the entire pre-run period, a safety verification pass signal is issued, and the strategy is marked as the optimal injection strategy output", which further includes: S1345: When the number of iterations of the injection strategy exceeds the preset threshold, terminate the security check loop and determine that there is no feasible solution under the current constraints; S1346: Package all rejected strategies' detailed parameters and their exceeding information to generate a diagnostic report; S1347: The diagnostic report is pushed to the engineer through the enterprise communication platform, and the engineer makes a decision and formulates a specific manual intervention plan.

[0045] It should be noted that the "preset threshold" mentioned in step S1345 is a configurable safety parameter, typically set based on computing resources and the urgency of the operation (e.g., 5-10 times). This mechanism serves as an important safety redundancy for the algorithm, preventing the system from falling into an infinite loop when no feasible solution can be found, ensuring the system remains responsive under all circumstances. This reflects the careful balance between autonomy and reliability in the method of this invention. The "human intervention plan" refers to decisions made by human experts that are outside the system's original set of automated operations. This usually means temporarily stepping out of the current optimization framework and fundamentally changing the "rules of the game," such as maintaining physical facilities (unblocking, replacing), adjusting the threshold of the safety boundary (requiring high-level authorization), or approving the execution of a "barely acceptable" strategy with a reassessed risk-benefit ratio and requiring enhanced human monitoring. This is a key manifestation of the "human" ultimately exercising decision-making power in human-machine collaborative decision-making.

[0046] As an example, the content and format of the "diagnostic report" can be diverse. For example, it can be a structured JSON or XML data file that is easy to parse by other systems; it can be an automatically generated PDF document with pictures and text, including trend charts and 3D visualization screenshots; or it can be a fault ticket that is directly pre-populated in the work order management system, including fields such as device number, problem category, and priority.

[0047] In one specific implementation, the "generating a diagnostic report" can be performed using a rule-based expert system: the system has a built-in knowledge base containing "IF" statements. <condition>THEN <conclusion>rules in the form of IF-THEN statements. For example: IF majority of the rejected strategies failed due to "exceeding wellhead pressure" AND historical data shows that the well flow rate is continuously declining, THEN conclude that "injector well IW-01 is suspected to be severely plugged", AND recommend action as "perform high-pressure water jetting operation". The system automatically executes these rules to reason over the data collected during the iteration, and incorporates the final conclusion and supporting evidence into the diagnostic report, thus providing engineers with a report that is preliminarily analyzed and has actionable insights, rather than a simple list of raw data.

[0048] According to the injection strategy, the injection equipment is controlled to inject the reaction medium into the mining area to be repaired, as described in step S140.

[0049] In an embodiment of the present application, the specific process of the above-mentioned step "controlling the injection equipment to inject the reaction medium into the mining area to be repaired according to the injection strategy" can be further described as follows: The injection strategy is compiled into standardized industrial control instructions, and an independent task instruction package containing the equipment ID, target parameters and execution time is generated for each intelligent injection well; The task instruction package is issued to the programmable logic controller of the target injection well through industrial Ethernet, and the programmable logic controller returns an acknowledgement signal after completing instruction verification; The programmable logic controller drives the connected actuators to perform precise actions according to the instruction content; During the execution process, the sensors integrated in the injection well body continuously collect actual operation data; The deviation between the actual operation data and the target value is compared through a closed-loop control algorithm, and a fine-tuning instruction is automatically issued to accurately match the target.

[0050] It should be noted that the "standardized industrial control instructions" refer to standardized command frames that comply with international general industrial communication protocols (such as OPC UA, Modbus TCP). The compilation process not only completes protocol conversion, but also needs to perform unit conversion (such as converting L / min to pulse frequency recognized by the pump controller), instruction serialization and sorting, and adding time stamp and check code, to ensure the accuracy, timeliness and reliability of the instructions during transmission and execution. The "integrated sensors" constitute the "perception" system of the injection equipment. These sensors are designed in an integrated manner and embedded in key components of the injection well, such as integrating a miniature pressure sensor in the valve body to directly measure the pressure before and after the valve core, and integrating a corrosion-resistant pH sensor in the flow channel near the injection head to monitor the pH change of the injection medium in situ. This design reduces measurement lag and improves the accuracy and real-time performance of the perception data.

[0051] As an example, the "industrial Ethernet" can be implemented based on various network topologies and protocols, such as a PROFINET network with a star topology, which has high real-time performance and determinacy; an EtherNet / IP network with a ring topology, which has link redundancy; or a 5G industrial wireless private network in a remote mining area, which uses the uRLLC (ultra-reliable low-latency communication) feature to realize wireless reliable transmission of instructions. The "closed-loop control algorithm" can take various forms according to different control objectives and object characteristics, such as a classic PID control algorithm for flow control; a model predictive control (MPC) algorithm for complex processes that require feedforward compensation (such as injection agent concentration control); or a fuzzy adaptive PID control algorithm for pH control with strong nonlinearity and large hysteresis. The system has an algorithm module library that can automatically call the most suitable control algorithm instance according to the control objective in the instruction package.

[0052] In a specific implementation, the "driving the connected actuators to perform precise actions" can be implemented through an intelligent injection well. The instruction received by the programmable logic controller (PLC) is "increase the carbon dioxide injection flow to 10.0 L / min". The PLC first solves the instruction, and its built-in PID control algorithm outputs a control signal to the frequency converter to adjust the speed of the diaphragm metering pump motor. At the same time, the PLC synchronously controls the electric regulating valve to increase its opening to a predicted position to reduce the pipeline back pressure. The data of the pump pressure sensor and flowmeter are fed back to the PLC in real time, and the PLC finally accurately stabilizes the flow at 10.0 L / min by fine-tuning the pump speed and valve position. This process is the result of the coordinated action of multiple actuators.

[0053] As described in step S150, the operating state of the injection equipment is monitored and fault prediction is performed, and maintenance instructions are generated.

[0054] In an embodiment of the present application, the specific process of "monitoring the operating state of the injection equipment and predicting faults, and generating maintenance instructions" described in step S150 can be further described as follows: Real-time operating data such as vibration, pressure, and flow of the injection equipment are continuously collected and associated with environmental data such as medium properties, and key indicators representing the health state of the equipment are obtained through feature extraction; Using the digital twin of the equipment, the current operating characteristics are compared with the health baseline in real time. When it is found that the characteristic value continuously deviates from the preset threshold, a primary abnormality alarm is triggered; Based on a long short-term memory network fault prediction model, time series data are analyzed to predict the probability of occurrence of a specific fault in the future and its remaining useful life; The prediction results are input into a decision rule engine, which automatically matches the optimal maintenance strategy according to the fault type, probability, and RUL, and generates structured instructions containing the maintenance object, type, resources, and operation guidelines. The maintenance instructions are automatically distributed to the responsible team or autonomous robots in the form of work orders, and feedback data is collected after the maintenance is completed for updating the prediction model and digital twin, achieving closed-loop learning and continuous optimization.

[0055] It should be noted that the "feature extraction" is a process of converting raw sensor data into characteristic quantities with physical meaning and sensitive to reflect the degradation state of the equipment, for example, performing fast Fourier transform (FFT) on vibration signals to extract the amplitude of a specific frequency band as a feature; calculating the moving average and standard deviation of flow data to represent flow stability; or extracting time-frequency features of non-stationary signals through wavelet transform, which are closely related to potential failure modes such as mechanical wear, fouling, and corrosion of the equipment. The "decision rule engine" is an expert system encapsulating maintenance domain knowledge. Its rules usually use "IF-THEN" logic, for example: IF fault type == "impeller corrosion" AND occurrence probability > 0.8 AND RUL < 7 days THEN maintenance type = "planned replacement" AND resources = "spare parts P-101A, tool kit TK-2" AND priority = "high", the engine will reason comprehensively according to all matching rules, and finally generate an executable, resource-specific maintenance instruction, realizing the automatic conversion from prediction information to maintenance action.

[0056] As an example, the "digital twin of the equipment" can be constructed by different modeling methods, such as: white-box model based on physical laws (e.g. establishing the fluid dynamics equation of the pump and the dynamics equation of the rotor); black-box model trained based on a large amount of historical data (e.g. deep neural network); or gray-box model combining physical principles and data-driven, which can simulate the expected normal behavior of the equipment under given working conditions, providing dynamic and personalized health baseline for anomaly detection. The "closed-loop learning and continuous optimization" can be achieved through various mechanisms, such as: using the performance recovery data of the equipment after maintenance and the real failure modes found through disassembly inspection as labeled training samples to fine-tune the LSTM prediction model regularly; summarizing new knowledge found during maintenance (such as an unexpected wear mode) as new rules and adding them to the knowledge base of the decision rule engine; or calibrating the equipment performance degradation model parameters in the digital twin according to the actual equipment degradation rate, so that it can make more accurate predictions in the future.

[0057] In a specific implementation, the "long short-term memory network-based fault prediction model" can be implemented with multivariate input: the input of the LSTM network is not a single signal, but a multivariate time series matrix, the rows of which are time steps and the columns of which are different dimensional features (such as vibration features, pressure, flow, medium pH). The architecture of the network is designed to learn the long-term dependencies and interaction patterns between these multivariate variables. Through learning from historical failure data, this model can not only predict the failure probability, but also output a residual useful life (RUL) estimate in the form of a probability distribution, thereby quantifying the uncertainty of the prediction.

[0058] For the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts are referred to the part of the method embodiment.

[0059] Referring to Figure 2 , a structural block diagram of a mine area geological repair system provided by an embodiment of the present application is shown; The system specifically comprises: A data collection module 210 is configured to acquire environment parameter and reaction state data of a mine area to be repaired through a sensing network, wherein the sensing network comprises pH, Eh, humidity, temperature, and pressure sensors arranged on a mine site, and periodic exploration data acquired through unmanned aerial vehicle remote sensing and geophysical exploration; A digital twin construction module 220 is configured to construct and run a digital twin based on the environment parameter and reaction state data, wherein the digital twin is configured to simulate migration of carbon dioxide in the mine area to be repaired and a reaction process of the carbon dioxide with minerals; An injection strategy generation module 230 is configured to generate an injection strategy for controlling a carbon sequestration reaction process based on a simulation result of the digital twin; An execution module 240 is configured to control an injection device to inject a reaction medium into the mine area to be repaired according to the injection strategy; An injection device monitoring module 250 is configured to monitor an operating state of the injection device and perform fault prediction, and generate a maintenance instruction.

[0060] In an embodiment of the present application, the digital twin construction module 220 comprises: A first digital twin construction submodule is configured to fuse real-time data and periodic exploration data from the sensing network, and construct a high-precision three-dimensional geological attribute model by using a geostatistical interpolation algorithm, wherein the three-dimensional geological attribute model comprises a porosity distribution field, a permeability distribution field, a mineral composition distribution field, and an initial fluid chemical field; The second digital twin construction submodule is configured to establish a multi-physical field coupling mathematical model, set a feedback mechanism for bidirectional coupling between fluid dynamics equations, convection-diffusion-reaction equations, geochemical equations and energy conservation equations and corresponding initial and boundary conditions; The third digital twin construction submodule is configured to use a finite element numerical analysis method to discretize and iteratively solve the coupled nonlinear partial differential equation set, and advance the simulation in a time-stepping manner to dynamically simulate the whole process of carbon dioxide plume migration, chemical reaction, temperature change and metal ion activation after carbon dioxide injection. The fourth digital twin construction submodule is configured to compare real-time data of the sensor network with simulation prediction values corresponding to a time step and a corresponding position, and if an error exceeds a preset threshold, start a set Kalman filter to reversely optimize and adjust key uncertainty parameters in the model. The fifth digital twin construction submodule is configured to extract calculation results of each time step, generate a dynamic evolution four-dimensional visual field, and calculate macroscopic performance parameters, wherein the four-dimensional visual field includes four-dimensional images of carbon dioxide saturation field, pH value field, secondary mineral precipitation field and metal enrichment field.

[0061] In an embodiment of the present application, the injection strategy generation module 230 includes: The first injection strategy generation submodule is configured to analyze and extract features of simulation results of the digital twin, calculate carbon sequestration rate, reaction efficiency and metal enrichment degree data, and generate a comprehensive feature vector. The second injection strategy generation submodule is configured to enable the reinforcement learning agent to receive the comprehensive feature vector, quantitatively evaluate long-term benefits of different control actions through a reward function, and output an optimal injection strategy, wherein the control action is a set of control instructions. The third injection strategy generation submodule is configured to pre-simulate the optimal injection strategy in the digital twin, and compare pre-simulation results with a preset safe operation boundary. The fourth injection strategy generation submodule is configured to regenerate the optimal injection strategy and pre-simulate again when the pre-simulation results exceed the safe boundary, until an optimal injection strategy that can pass the safety check is generated.

[0062] In an embodiment of the present application, the fourth injection strategy generation submodule includes: The invalid strategy determination unit is configured to issue a safety check failure signal with over-standard information when the pre-simulation results exceed the safe boundary, mark the current injection strategy as an invalid strategy and discard it. The new strategy generation unit is configured to enable the reinforcement learning agent to receive the safety check failure signal, apply a corresponding penalty in its decision logic according to the over-standard information, and generate a more conservative new strategy. an iterative simulation and verification unit configured to perform a simulation of the new strategy in the digital twin again and perform a safety verification on the simulation result; an optimal injection strategy generation unit configured to send a safety verification pass signal when the simulation result of a certain simulation indicates that all indexes remain within the safety boundary during the entire simulation period, and mark the strategy as an optimal injection strategy.

[0063] In an embodiment of the present application, the fourth injection strategy generation sub-module further comprises: a safety verification termination unit configured to terminate the safety verification cycle and determine that there is no feasible solution under the current constraint condition when the number of iterations of the injection strategy exceeds a preset threshold value; a diagnostic report generation unit configured to package detailed parameters of all rejected strategies and their over-standard information, and generate a diagnostic report; a manual intervention unit configured to push the diagnostic report to an engineer through an enterprise communication platform, and make a decision and formulate a specific manual intervention scheme by the engineer.

[0064] In an embodiment of the present application, the execution module 240 comprises: a first execution sub-module configured to compile the injection strategy into a standardized industrial control instruction, and generate an independent task instruction package containing a device ID, a target parameter and an execution time for each intelligent injection well; a second execution sub-module configured to issue the task instruction package to a programmable logic controller of the target injection well through an industrial Ethernet, and return an acknowledgement signal by the programmable logic controller after completing instruction verification; a third execution sub-module configured to drive each actuator connected according to the instruction content by the programmable logic controller to perform accurate actions; a fourth execution sub-module configured to continuously collect actual operation data by the sensors integrated in the injection well body during the execution process; a fifth execution sub-module configured to compare the deviation of the actual operation data from the target value through a closed-loop control algorithm, and automatically issue a fine-tuning instruction to accurately match the target.

[0065] In an embodiment of the present application, the injection device monitoring module 250 comprises, a first injection device monitoring sub-module configured to continuously collect real-time operation data such as vibration, pressure and flow of the injection device, and associate the data with environmental data such as medium properties, and obtain key indexes representing the health state of the device through feature extraction; a second injection device monitoring sub-module configured to perform real-time comparison between the current operation characteristics and the health baseline by using the digital twin of the device, and trigger a primary abnormality alarm when it is found that the characteristic value continuously deviates from a preset threshold value; The third injection device monitoring submodule is used to analyze time series data based on the long short-term memory network fault prediction model to predict the probability of a specific fault occurring in the future and its remaining useful life. The fourth injection device monitoring submodule is used to input the prediction results into the decision rule engine, automatically match the optimal maintenance strategy according to the fault type, probability and RUL, and generate structured instructions containing maintenance objects, types, resources and operation guidelines; The fifth submodule, which injects equipment monitoring, is used to automatically distribute maintenance instructions to the responsible team or autonomous robot in the form of work orders, and collect feedback data after maintenance is completed to update the prediction model and digital twin, thereby achieving closed-loop learning and continuous optimization.

[0066] Reference Figure 3 The computer device shown in the present invention is a method for geological restoration of a mining area, and may specifically include the following: The computer device 12 is manifested as a general-purpose computing device. Components of the computer device 12 may include, but are not limited to: one or more processing units 16, system memory 28, and a bus 18 connecting different system components (including system memory 28 and processing units 16). The computer device 12 may be a device connected to the bus.

[0067] Bus 18 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MCA) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0068] Computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by computer device 12, including volatile and non-volatile media, removable and non-removable media.

[0069] System memory 28 may include computer system readable media in the form of volatile memory, such as RAM 30 (Random Access Memory) and / or cache 32. Computer device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write non-removable, non-volatile magnetic media (commonly referred to as a "hard disk drive"). Although Figure 3 As not shown, disk drives for reading and writing to removable non-volatile disks (e.g., "floppy disks") and optical disc drives for reading and writing to removable non-volatile optical discs (e.g., CD-ROMs, DVD-ROMs, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0070] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28. Such program modules 42 include—but are not limited to—an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 42 typically perform the functions and / or methods described in the embodiments of the present invention.

[0071] Computer device 12 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), and with one or more devices that enable a user to interact with the computer device 12, and / or with any device that enables the computer device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through I / O interface 22 (input / output interface). Furthermore, computer device 12 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public network (e.g., the Internet)) through network adapter 20. Figure 3 As shown, network adapter 20 communicates with other modules of computer device 12 via bus 18. It should be understood that, although... Figure 3 As not shown, it can be used in conjunction with computer device 12 with other hardware and / or software modules, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0072] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing a geological restoration method for mining areas provided in the embodiments of the present invention.

[0073] That is, when the above-mentioned processing unit 16 executes the above-mentioned program, it achieves the following: acquiring environmental parameters and reaction status data of the mining area to be repaired through a sensor network. The sensor network includes pH, Eh, humidity, temperature and pressure sensors deployed at the mining site, as well as periodic exploration data acquired through UAV remote sensing and geophysical exploration. Based on the environmental parameters and reaction status data, a digital twin is constructed and run. The digital twin is used to simulate the migration of carbon dioxide in the mining area to be restored and its reaction with minerals. Based on the simulation results of the digital twin, an injection strategy for controlling the carbon sequestration reaction process is generated; According to the injection strategy, the injection equipment is controlled to inject the reaction medium into the mining area to be repaired; The operating status of the injection equipment is monitored and faults are predicted, and maintenance instructions are generated.

[0074] Computer device 12 is merely an example and should not impose any limitation on the functionality and scope of use of embodiments of the present invention.

[0075] In this embodiment of the invention, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements a geological restoration method for a mining area as provided in all embodiments of this application: That is, when the program is executed by the processor, it achieves the following: acquiring environmental parameters and reaction status data of the mining area to be repaired through a sensor network, the sensor network including pH, Eh, humidity, temperature and pressure sensors deployed at the mining site, as well as periodic exploration data acquired through UAV remote sensing and geophysical exploration. Based on the environmental parameters and reaction status data, a digital twin is constructed and run. The digital twin is used to simulate the migration of carbon dioxide in the mining area to be restored and its reaction with minerals. Based on the simulation results of the digital twin, an injection strategy for controlling the carbon sequestration reaction process is generated; According to the injection strategy, the injection equipment is controlled to inject the reaction medium into the mining area to be repaired; The operating status of the injection equipment is monitored and faults are predicted, and maintenance instructions are generated.

[0076] Computer storage media may take the form of any combination of one or more computer-readable media. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: electrical connections having one or more wires, portable computer disks, hard disks, RAM, read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable CD-ROM, optical storage devices, magnetic storage devices, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0077] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0078] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including—but not limited to—wireless, wire, optical fiber, radio frequency (RF), etc., or any suitable combination thereof.

[0079] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including LANs or WANs—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0080] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0081] Finally, 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 terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0082] The above provides a detailed description of a geological restoration method for mining areas provided by this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and its core ideas. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.< / conclusion> < / condition>

Claims

1. A method for geological restoration of a mining area, characterized in that, Including the following steps: The environmental parameters and reaction status data of the mining area to be restored are obtained through a sensor network. The sensor network includes pH, Eh, humidity, temperature and pressure sensors deployed at the mining site, as well as periodic exploration data obtained through UAV remote sensing and geophysical exploration. Based on the environmental parameters and reaction status data, a digital twin is constructed and run. The digital twin is used to simulate the migration of carbon dioxide in the mining area to be restored and its reaction with minerals. Based on the simulation results of the digital twin, an injection strategy for controlling the carbon sequestration reaction process is generated; According to the injection strategy, the injection equipment is controlled to inject the reaction medium into the mining area to be repaired; The operating status of the injection equipment is monitored and faults are predicted, and maintenance instructions are generated.

2. The geological restoration method for a mining area according to claim 1, characterized in that, The digital twin is a multiphysics coupled model based on geology, geochemistry, and fluid dynamics. The steps of constructing and running the digital twin based on the environmental parameters and reaction state data specifically include the following steps: By integrating real-time data and periodic exploration data from the sensor network, a high-precision three-dimensional geological attribute model is constructed using a geostatistical interpolation algorithm. The three-dimensional geological attribute model includes a porosity distribution field, a permeability distribution field, a mineral composition distribution field, and an initial hydrochemical field. A multiphysics coupled mathematical model is established, and the feedback mechanism of bidirectional coupling between fluid dynamics equations, convection-diffusion-reaction equations, geochemical equations and energy conservation equations, as well as the corresponding initial and boundary conditions, are defined. The finite element numerical analysis method is used to discretize and iteratively solve the coupled nonlinear partial differential equations. The simulation is advanced in a time-step manner to dynamically simulate the entire process of plume migration, chemical reaction, temperature change and metal ion activation after carbon dioxide injection. The real-time data of the sensor network is compared with the simulation predictions at the corresponding time step and location. If the error exceeds the preset threshold, the ensemble Kalman filter is activated to back-optimize and adjust the key uncertainty parameters in the model. The calculation results of each time step are extracted to generate a dynamically evolving four-dimensional visualization field, and macroscopic performance parameters are calculated. The four-dimensional visualization field includes four-dimensional images of carbon dioxide saturation field, pH value field, secondary mineral precipitation field and metal enrichment field.

3. The geological restoration method for a mining area according to claim 1, characterized in that, The step of generating an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin specifically includes the following steps: The simulation results of the digital twin are analyzed and features are extracted to calculate the carbon sequestration rate, reaction efficiency and metal enrichment data, and a comprehensive characterization vector is generated. The reinforcement learning agent receives the comprehensive representation vector, quantifies and evaluates the long-term benefits of different control actions through a reward function, and outputs the optimal injection strategy. The control action is a set of control instructions. The optimal injection strategy is simulated in the digital twin, and the simulation results are compared with the preset security operation boundaries. When the simulation results exceed the safety boundary, the optimal injection strategy is regenerated and the simulation is performed again until an optimal injection strategy that can pass the security check is generated.

4. A geological restoration method for a mining area according to claim 3, characterized in that, The safe operating boundaries include the maximum allowable wellhead pressure, the allowable pH fluctuation range, the maximum instantaneous carbon dioxide injection rate, and the maximum formation pore pressure threshold. The step of regenerating the optimal injection strategy and performing another simulation when the simulation results exceed the safe boundaries, until an optimal injection strategy that passes the safety verification is generated, specifically includes the following steps: When the simulation results exceed the security boundary, a security verification failure signal is issued along with information indicating that the limit has been exceeded. The current injection strategy is then marked as an invalid strategy and discarded. After receiving a security check failure signal, the reinforcement learning agent applies a corresponding penalty to its decision-making logic based on the information of exceeding the limit, thereby generating a more conservative new strategy. The new strategy was simulated again in the digital twin, and the simulation results were verified for security. When a pre-run result shows that all indicators remain within the safety boundary throughout the pre-run period, a safety verification pass signal is issued, and the strategy is marked as the optimal injection strategy output.

5. A geological restoration method for a mining area according to claim 4, characterized in that, After the step of issuing a security verification pass signal and marking the strategy as the optimal injection strategy output when a certain pre-simulation result shows that all indicators remain within the safety boundary throughout the pre-simulation period, the following steps are also included: When the number of iterations of the injection strategy exceeds a preset threshold, the security verification loop is terminated and it is determined that there is no feasible solution under the current constraints. Package all rejected strategies' detailed parameters and their exceeding information to generate a diagnostic report; The diagnostic report is pushed to the engineer through the enterprise communication platform, and the engineer makes a decision and develops a specific human intervention plan.

6. A method for geological restoration of a mining area according to claim 1, characterized in that, The step of controlling the injection device to inject the reaction medium into the mining area to be repaired according to the injection strategy specifically includes the following steps: The injection strategy is compiled into standardized industrial control instructions, and an independent task instruction package containing device ID, target parameters and execution time is generated for each smart injection well. The task instruction packet is sent to the programmable logic controller of the target injection well via industrial Ethernet, and the programmable logic controller returns an acknowledgment signal after completing the instruction verification. The programmable logic controller drives the connected actuators to perform precise actions according to the instructions. During the process, the sensors integrated into the injection well body continuously collect actual operating data; The closed-loop control algorithm compares the deviation between the actual operating data and the target value, and automatically issues fine-tuning instructions to accurately match the target.

7. A geological restoration method for a mining area according to claim 1, characterized in that, The step of monitoring the operating status and predicting faults of the injection device, and generating maintenance instructions, specifically includes the following steps: Continuously collect real-time operating data such as vibration, pressure, and flow of the injection equipment, and correlate them with environmental data such as media properties. Through feature extraction, key indicators characterizing the health status of the equipment are obtained. Using the device's digital twin, the current operating characteristics are compared with the health baseline in real time. When the characteristic value is found to deviate continuously from the preset threshold, a primary anomaly alarm is triggered. Based on the Long Short-Term Memory Network Fault Prediction Model, time series data is analyzed to predict the probability of a specific fault occurring in the future and its remaining useful life. The prediction results are input into the decision rule engine, which automatically matches the optimal maintenance strategy based on the fault type, probability, and RUL, and generates structured instructions that include maintenance objects, types, resources, and operation guidelines. Maintenance instructions are automatically distributed to the responsible team or autonomous robot in the form of work orders, and feedback data is collected after maintenance is completed to update the predictive model and digital twin, thereby achieving closed-loop learning and continuous optimization.

8. A geological restoration system for mining areas, characterized in that, include: The data collection module is used to acquire environmental parameters and reaction status data of the mining area to be repaired through a sensor network. The sensor network includes pH, Eh, humidity, temperature and pressure sensors deployed at the mining site, as well as periodic exploration data acquired through UAV remote sensing and geophysical exploration. A digital twin construction module is used to construct and run a digital twin based on the environmental parameters and reaction state data. The digital twin is used to simulate the migration of carbon dioxide in the mining area to be restored and its reaction with minerals. An injection strategy generation module is used to generate an injection strategy for controlling the carbon sequestration reaction process based on the simulation results of the digital twin. An execution module is used to control the injection device to inject a reaction medium into the mining area to be repaired, according to the injection strategy. The injection equipment monitoring module is used to monitor the operating status of the injection equipment and predict faults, and generate maintenance instructions.

9. A computer device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images