An in-situ leaching uranium mining precise liquid injection regulation system, method, device and medium

Abstract

The application relates to the technical field of automatic control of uranium mining. A ground leaching uranium precise liquid injection regulation system, method, equipment and medium are disclosed. The system comprises a multi-source sensing module for collecting operation data of each branch liquid injection well; a digital twin generation module for simulating and predicting the concentration field, temperature field, pressure field and uranium leaching rate of acid liquid in the underground ore bed according to the operation data of each branch liquid injection well; an intelligent liquid injection decision module for generating concentration control instructions and flow control instructions of each branch liquid injection well; and an edge liquid injection execution module for regulating and controlling the liquid injection concentration and liquid injection flow of each branch liquid injection well according to the concentration control instructions and flow control instructions of each branch liquid injection well. The application can accurately depict the ore bed response characteristics, realize the decoupling of liquid injection concentration and liquid injection flow, and realize the "one hole one strategy" precise liquid injection strategy for heterogeneous ore beds.

Description

Technical Field

[0001] This application relates to the field of automated control technology for uranium mining, and more specifically, to a precise liquid injection control system, method, equipment, and medium for in-situ leaching uranium mining. Background Technology

[0002] In-situ leaching is the core method for developing natural uranium resources in my country, and its leaching efficiency directly depends on the precise matching of leaching agent injection parameters. Currently, acid leaching generally adopts a centralized injection mode, that is, a fixed concentration of leaching solution is uniformly prepared in the main pipeline, and the flow rate of each branch well is controlled by regulating valves. This results in all injection wells using the same injection concentration and pressure strategy, which is difficult to respond to the spatial heterogeneity of the ore layer, causing problems such as insufficient leaching in high-grade areas and excessive acidification in low-grade areas. The reagent utilization rate is low, and it is easy to cause local ore layer blockage or groundwater pollution.

[0003] Existing technologies mainly rely on two types of control mechanisms: one type is based on preset empirical parameters for timed and fixed-flow injection, which lacks the ability to sense real-time formation response; the other type introduces online pH or conductivity monitoring to achieve local closed-loop feedback, but its regulation is still limited by the centralized liquid supply mode of the main system and cannot decouple the coupling relationship between injection concentration and injection flow rate.

[0004] Therefore, in on-site process control, how to decouple the coupling relationship between injection concentration and injection flow rate to realize the injection strategy response to the spatial heterogeneity of the ore layer has always been a difficult problem in the industry. Summary of the Invention

[0005] In view of the above situation, this application provides a precise liquid injection control system, method, equipment and medium for in-situ leaching uranium mining, which aims to solve the above problems or at least partially solve the above problems.

[0006] Firstly, this application provides a precise injection control system for in-situ uranium leaching, comprising: The multi-source sensing module is used to collect operational data from each branch injection well; The digital twin generation module is used to simulate and predict the concentration field, temperature field, pressure field, and uranium leaching rate of acid in underground mineral formations based on the pre-built digital twin model and the operating data of each branch injection well. The intelligent injection decision module is used to generate concentration control commands and flow control commands for each branch injection well based on the pre-set single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model, according to the predicted concentration field, temperature field, pressure field and uranium leaching rate of acid in the underground mineral layer. The edge injection execution module is used to regulate the injection concentration and injection flow rate of each branch injection well based on the operating data, concentration control instructions and flow control instructions of each branch injection well.

[0007] Secondly, this application provides a method for precise injection control in in-situ leaching uranium mining, comprising: Collect operational data from each branch injection well; Based on a pre-built digital twin model, the concentration field, temperature field, pressure field, and uranium leaching rate of acid in the underground mineral layer are simulated and predicted according to the operating data of each branch injection well. Based on the pre-set single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model, the concentration control command and flow control command of each branch injection well are generated according to the predicted concentration field, temperature field, pressure field and uranium leaching rate of acid in the underground mineral layer. Based on the operating data of each branch injection well, the concentration control instructions and flow control instructions of each branch injection well, the injection concentration and injection flow rate of each branch injection well are adjusted separately.

[0008] Thirdly, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the precise injection control method for in-situ leaching uranium mining as described in the second aspect.

[0009] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the precise injection control method for in-situ leaching uranium mining as described in the second aspect.

[0010] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects: This application collects operational data from each branch injection well in real time, performs simulation and prediction through a digital twin generation module to accurately characterize the response characteristics of the ore layer, and generates concentration control and flow control commands through a single-well multi-objective optimization sub-model and a multi-well collaborative optimization sub-model based on the prediction results. While ensuring the maximization of leaching efficiency of a single well, it achieves coordinated control among multiple wells. Finally, the concentration control and flow control commands are executed through the edge injection execution module, realizing the decoupling of injection concentration and injection flow rate, and achieving a precise injection strategy of "one policy per well" for heterogeneous ore layers. Attached Figure Description

[0011] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of a precise liquid injection control system for uranium leaching in an embodiment of the present invention; Figure 2This is a schematic diagram of an application environment for a precise liquid injection control method for uranium mining by in-situ leaching, according to one embodiment of the present invention; Figure 3 This is a schematic flowchart of a method for precise injection control of uranium mining by in-situ leaching in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0013] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such use can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the term "comprising" and its variations should be interpreted as open-ended terms meaning "including but not limited to."

[0014] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0015] As mentioned earlier, current acid leaching methods generally employ a centralized injection mode, which cannot decouple the injection concentration and flow rate, making it difficult to respond to the spatial heterogeneity of the ore layer. To address this technical problem, this application provides a precise injection control system for uranium mining through in-situ leaching.

[0016] In one embodiment, a precise liquid injection control system for in-situ leaching uranium mining is provided, the system being as follows: Figure 1As shown, the precise injection control system for uranium leaching in this field includes a multi-source sensing module 101, a digital twin generation module 102, an intelligent injection decision module 103, and an edge injection execution module 104. The detailed descriptions of each functional module are as follows: The multi-source sensing module 101 is used to collect operational data from each branch injection well; the digital twin generation module 102 is used to simulate and predict the concentration field, temperature field, pressure field, and uranium leaching rate of the acid in the underground ore layer based on a pre-built digital twin model and the operational data of each branch injection well; the intelligent injection decision module 103 is used to generate concentration control commands and flow control commands for each branch injection well based on a pre-set single-well multi-objective optimization sub-model and a multi-well collaborative optimization sub-model, according to the predicted concentration field, temperature field, pressure field, and uranium leaching rate of the acid in the underground ore layer; the edge injection execution module 104 is used to regulate the injection concentration and flow rate of each branch injection well according to the operational data, concentration control commands, and flow control commands of each branch injection well. This application collects operational data from each branch injection well in real time, performs simulation and prediction through a digital twin generation module to accurately characterize the response characteristics of the ore layer, and generates concentration control and flow control commands through a single-well multi-objective optimization sub-model and a multi-well collaborative optimization sub-model based on the prediction results. While ensuring the maximization of leaching efficiency of a single well, it achieves coordinated control among multiple wells. Finally, the concentration control and flow control commands are executed through the edge injection execution module, realizing the decoupling of injection concentration and injection flow rate, and achieving a precise injection strategy of "one policy per well" for heterogeneous ore layers.

[0017] In one embodiment, the hardware deployment of the system is described below: (1) Deployment of the edge injection execution module 104: An independent edge injection execution module 104 is configured on each branch injection well. The edge injection execution module 104 is installed upstream of the injection pipeline, about 15m-20m from the wellhead. The reason for choosing this installation location is that the liquid flow in the pipeline here has been fully mixed, the concentration field and temperature field distribution are relatively uniform, and the sensor measurement results are representative; at the same time, this location is far enough from the wellhead to facilitate timely adjustment of injection parameters in case of abnormal situations, so that the adjusted liquid flow reaches a stable state before reaching the wellhead. Based on the average flow rate of a conventional injection pump of 5m³ / h and the flow velocity in the pipe of 1.5m / s, the transmission time of the liquid flow from the edge injection execution unit to the wellhead is about 10s, which meets the time delay requirements of real-time control.

[0018] (2) Deployment of the central monitoring station: A central monitoring station is built on the surface of the mining area, and an industrial-grade server cluster is configured to run the digital twin generation module 102, parameter correction module 105, intelligent injection decision module 103, and anomaly early warning module 106. The central monitoring station is connected to the edge injection execution modules 104 of each edge node via fiber optic industrial Ethernet.

[0019] (3) Deployment of the sensor network: An electromagnetic flowmeter (accuracy ±0.5%FS), an online pH meter (accuracy ±0.1pH), and a conductivity meter (accuracy ±1%FS) are installed at the inlet of each edge injection execution module 104 to monitor the flow rate, pH, and ion concentration of the injected fluid in real time. A high-precision pressure sensor (accuracy ±0.1%FS) and a temperature sensor (accuracy ±0.1°C) are installed at the outlet of the edge injection execution module 104 (i.e., near the wellhead) to monitor the injection pressure and liquid temperature rise. An online uranium concentration analyzer (accuracy ±2%FS) is installed at the pumping wellhead to monitor the uranium concentration in the leaching fluid in real time, serving as a feedback signal for leaching efficiency.

[0020] In one embodiment, the three-layer communication network topology of the system is described as follows: This embodiment adopts a layered communication network architecture, which is divided into three layers from bottom to top: the device layer, the edge layer, and the monitoring layer. Each layer uses different communication protocols and technical standards to meet the different requirements of each layer for real-time performance, reliability, and data throughput.

[0021] (1) Device Layer (Sensing and Execution Layer): The device layer is responsible for data acquisition from each sensor in the multi-source sensing module 101 and transmission of device control commands. In this embodiment, the device layer adopts a dual-mode communication scheme combining RS-485 bus and 4~20mA analog signal. Digital sensors (such as flow meters and pressure sensors) communicate with the multi-source sensing module 101 via RS-485 bus using the Modbus-RTU protocol. The communication baud rate is set to 115200bps, the single frame data length is 8 bytes, and the single polling cycle is approximately 10ms. Analog sensors (such as temperature sensors) output signals through a 4~20mA standard current loop, which are then converted into digital signals by the A / D conversion module (16-bit resolution, 1kHz sampling rate) on the multi-source sensing module 101. The communication cycle of the device layer is in the 10ms range, with a typical value of 10~20ms, which can meet the high-frequency sampling requirements for rapidly changing processes such as injection pressure and flow rate.

[0022] The data frame format at the device layer is defined as follows: Frame header (2 bytes, 0xAA0x55) + Device address (1 byte) + Function code (1 byte) + Data length (1 byte) + Data field (N bytes) + CRC checksum (2 bytes). Function code 0x03 indicates reading the holding register (used for sensor data reading), and function code 0x06 indicates writing to a single register (used for issuing control commands).

[0023] (2) Edge Layer (Inter-node Communication Layer): The edge layer is responsible for data interaction and collaborative control information transmission between the edge injection execution modules 104. Information collected by the multi-source sensing module 101 is transmitted to the monitoring layer via the edge layer network, and control commands received by each edge injection execution module 104 are transmitted to the device layer to realize device control. In this embodiment, the edge layer uses industrial Ethernet (100Mbps) as the main communication link, and a wireless Mesh network (2.4GHz band, IEEE 802.15.4 standard) is used as a supplementary link between some edge nodes where cabling is difficult. The industrial Ethernet uses the EtherNet / IP protocol, the network topology is a ring redundant structure, and the MRP (Media Redundancy Protocol) is used to realize fast switching in case of link failure, with a switching time of less than 50ms.

[0024] The communication content at the edge layer mainly includes: the local optimization solutions of each edge node. State estimation vector The system includes constraint violation information and the original and multiplier variables required for collaborative optimization using the Alternating Direction Method of Multipliers (ADMM). The communication cycle of the edge layer is on the order of 1 second, typically 1 second, meaning that each edge node broadcasts its local state information to its neighboring nodes once per second and receives state information from other nodes.

[0025] The wireless mesh network serves as a backup link, activated during the initial network deployment or in case of primary link failure. The mesh network employs the AODV (Ad-hoc On-demand Distance Vector) routing protocol, with a maximum hop count of no more than 4 between nodes and end-to-end latency controlled within 100ms.

[0026] (3) Monitoring Layer (Management Layer): The monitoring layer is responsible for the transmission of large amounts of data between the central monitoring station and the edge nodes of each edge injection execution module 104, including the release of boundary conditions by the digital twin generation module 102, the uploading of calculation results (concentration field, temperature field, pressure field) by the digital twin generation module 102, the release of parameter correction results by the parameter correction module 105, the transmission of global coordination concentration control commands and flow control commands by the intelligent injection decision module 103, and the early warning by the anomaly early warning module 106. In this embodiment, the monitoring layer adopts TCP / IP Ethernet (1000Mbps, Gigabit Ethernet), the network topology is a star structure, the central monitoring station is the core node, and each edge node is connected to the central monitoring station through a fiber optic switch.

[0027] The monitoring layer uses the OPCUA (Open Platform Communications Unified Architecture) protocol for data interaction. This protocol has good cross-platform compatibility and security, supporting encrypted transmission and digital signatures. The communication cycle of the monitoring layer is 1-5 minutes, with a typical value of 2 minutes, meaning that a complete closed loop from data upload to decision issuance is completed every 2 minutes. The specific timing is as follows: ① 0-30s: Each multi-source sensing module 101 edge node uploads collected data and status information; ② 30-60s: The central monitoring station runs the digital twin generation module 102 and parameter correction module 105; ③ 60-90s: The intelligent injection decision module 103 executes Sequential Quadratic Programming (SQP) and Alternating Direction Multiplier Method (ADMM) optimization calculations to obtain concentration control commands and flow control commands; ④ 90-120s: Concentration control commands and flow control commands are issued to each edge injection execution module 104 edge node to control the execution of equipment actions in the equipment layer.

[0028] In one embodiment, the edge layer may be provided with an independent communication module to enable data collected by the multi-source sensing module 101 to be transmitted to the monitoring layer through the independent communication module of the edge layer. In other embodiments, the edge layer may couple the communication module into the edge injection execution module 104, and the data collected by the multi-source sensing module 101 may be transmitted to the monitoring layer through the communication module in the edge injection execution module 104 of the edge layer.

[0029] In one embodiment, the three-layer communication timing and data flow of the system are described as follows: Figure 1As shown, the multi-source sensing module 101 at the equipment layer performs high-frequency data acquisition with a period of 10ms. Data is packaged every 100 acquisitions (i.e., every 1 second) and sent to adjacent nodes and the central monitoring station at the monitoring layer via the edge layer network. After receiving data from all edge nodes of the multi-source sensing module 101, the central monitoring station initiates the calculation of EKF (Extended Kalman Filter) parameters by the digital twin generation module 102 and the parameter correction module 105. This calculation process typically completes within 30 seconds. Subsequently, the intelligent injection decision module 103 performs single-well SQP optimization and multi-well ADMM collaboration. This optimization calculation typically converges within 20-30 seconds. Finally, the optimized target concentrations for each branch injection well are determined. and target traffic The control loop is completed by sending data from the monitoring layer network to each edge injection execution module 104 in the edge layer.

[0030] In one embodiment, under abnormal circumstances, the abnormal warning module 106 can directly send a shutdown command to all edge nodes through the emergency broadcast channel (reserved VLAN channel) of the edge layer. The transmission delay of this command is no more than 100ms, ensuring a rapid response in the event of a serious abnormality.

[0031] In one embodiment, the power supply and redundancy design of the system is described as follows: (1) In this embodiment, each edge injection execution module 104 adopts a dual power supply design: the main power supply is the 220VAC mains power in the mining area, which is converted to 24VDC by UPS (uninterruptible power supply) to power the edge controller, sensors and actuators of the edge injection execution module 104; the backup power supply is a solar + battery system, which can maintain the edge node to work continuously for more than 4 hours when the mains is powered off. (2) The key sensors (pressure sensor, flow meter) adopt a dual redundancy configuration. When the main sensor fails, it automatically switches to the backup sensor, and the switching time is less than 1 second. The communication link also adopts a redundancy design. When the Ethernet link fails, it automatically switches to the wireless Mesh backup link to ensure that the communication reliability of the control system is not less than 99.9%.

[0032] In one embodiment, the multi-source sensing module 101 serves as the system's information entry point, responsible for collecting multi-source heterogeneous data during the uranium leaching process, preprocessing and synchronizing the raw data, and finally distributing standardized time-series data to subsequent modules. The multi-source sensing module 101 includes various sensors deployed at the wellhead and downhole area of ​​each branch injection well to collect real-time operational data from each branch injection well. This operational data includes, but is not limited to, uranium concentration, injection flow rate, acid injection volume, permeability, acid pH value, outlet pressure, and mixing temperature data. The operational data is also synchronously accessed through geological exploration data and historical databases.

[0033] In one embodiment, a uranium concentration sensor is installed at the outlet of the extraction tube to detect the uranium concentration in the extracted liquid. (Unit: mg / L), the uranium concentration in the leaching solution is measured in real time by an online uranium concentration analyzer (accuracy ±2%FS) based on fluorescence or ICP-MS; the injection flow meter is installed at the inlet of the injection pipe to measure the actual injection flow rate. (Unit: m³ / h); Liquid flow meter, installed at the outlet of the liquid extraction pipe, is used to measure the liquid extraction flow rate. (Unit: m³ / h); Acid flow meter, installed on the outlet pipeline of the acid storage tank, to monitor the acid injection rate; Permeability detection device, to obtain formation permeability through downhole pressure recovery tests. The outlet pressure sensor (accuracy ±0.1%FS) is installed in the wellhead annulus to monitor the injection pressure. (Unit: MPa); Temperature sensor array (accuracy ±0.1°C), installed at the outlet of the acid mixing pipe and the downhole monitoring area, respectively, to monitor the mixing temperature. An online pH meter (accuracy ±0.1pH) is used to monitor the pH value of the acid solution in the injection fluid in real time.

[0034] In one embodiment, the multi-source sensing module 101 collects the following three types of data through a device-layer communication interface at a period of 10ms: (1) Injection process data: including the injection flow rate of each branch injection well. (Unit: m³ / h), Injection Pressure (Unit: MPa), Injection temperature (Unit: °C), Injection concentration (Unit: mol / L), acid concentration (Unit: mol / L) and pH value of the acid solution. These data are obtained in real time by corresponding sensors installed at the inlet and outlet of the edge injection execution module 104.

[0035] (2) Pumping process data: including the pumping flow rate of each pumping well. (Unit: m³ / h), pumping pressure (Unit: MPa) Uranium concentration in leachate (Unit: mg / L) and pH value of the leachate. The uranium concentration in the leachate was measured in real time by an online uranium concentration analyzer based on fluorescence or ICP-MS.

[0036] (3) Environmental and condition data: including reservoir temperature (Unit: °C) Ambient temperature (Unit: °C), Operating status of peripheral equipment (metering pump speed) Injection pump frequency (e.g., valve opening) and the self-test status codes of each sensor.

[0037] In one embodiment, the multi-source sensing module 101 is further configured to preprocess the collected raw operating data to obtain the operating data.

[0038] In one embodiment, the preprocessing of raw operational data by the multi-source sensing module 101 includes: (1) Outlier removal: using a 3-based method The sliding window filtering method of the criterion identifies and removes outliers. Specifically, for each sensor signal, a sliding window filter of length is maintained. A sliding window is used to calculate the mean of the data within the window. and standard deviation If the current sample value satisfy If a value is found to be outlier, it is removed and replaced with the window mean. The running data after outlier removal... This can be expressed as formula (1):

[0039] (2) Low-pass filtering: The data after removing outliers still needs to be low-pass filtered to suppress high-frequency noise. In this embodiment, a second-order Butterworth digital low-pass filter is used, and the cutoff frequency is set to 1 / 10 of the sampling frequency (i.e., 5Hz, corresponding to the time constant). ms). Filter transfer function As shown in formula (2):

[0040] Where z is the variable of the transformation; , , These are the numerator coefficients of the filter, also called feedforward coefficients; , These are the denominator coefficients of the filter, also called feedback coefficients.

[0041] The filter coefficients were obtained by discretizing the analog prototype filter using the bilinear transform method. The running data after low-pass filtering is denoted as... .

[0042] (3) Time Synchronization: Due to differences in response time and communication delay among different sensors, time synchronization of multiple data streams is required. This embodiment adopts a time-stamp-based minimum mean square error synchronization method. The central monitoring station acts as the time master station, broadcasting a synchronization timestamp to all edge nodes every 1 second; after receiving the timestamp, each edge node uses that time as a reference and employs linear interpolation to align each data stream to a unified time grid. Let sensor A be at time... and The sampled values ​​are respectively and The target synchronization time is The value after synchronization for:

[0043] In one embodiment, the standardized data, after preprocessing and time synchronization, is sent in parallel to different functional modules of the system via four paths through the edge layer and monitoring layer communication network: (1) The first path is distributed to the digital twin generation module 102: This path includes the injection flow rate, injection concentration, injection pressure of each branch injection well and the pumping flow rate and uranium concentration of each pumping well, which serve as the boundary conditions and verification data of the digital twin generation module 102.

[0044] (2) The second path is distributed to the anomaly warning module 106: This path includes the real-time measurement values ​​of all sensors and the device status code, which is used for anomaly detection and fault diagnosis.

[0045] (3) The third path is distributed to the parameter correction module 105: This path includes the measured values ​​of key monitoring points (injection concentration, mixing temperature, wellhead pressure, etc.), which are used to compare with the predicted values ​​of the digital twin generation module 102 to drive the EKF parameter correction.

[0046] (4) The fourth route is distributed to the edge injection execution module 104: This route includes the injection flow rate, injection concentration, injection pressure of each branch injection well and the pumping flow rate and uranium concentration of each pumping well, which serve as the control parameters for the injection concentration and injection flow rate of each branch injection well.

[0047] The above four-way distribution is implemented through a message queue mechanism to ensure that each functional module can receive and process data independently and in parallel without affecting each other.

[0048] In one embodiment, the digital twin generation module includes a fluid flow sub-model, a solute transport sub-model, an energy conservation sub-model, and a leaching reaction sub-model. The fluid flow sub-model is used to simulate and predict the pressure field of the acid solution in the underground mineral layer, the solute transport sub-model is used to simulate and predict the concentration field of the acid solution in the underground mineral layer, the energy conservation sub-model is used to simulate and predict the temperature field of the acid solution in the underground mineral layer, and the leaching reaction sub-model is used to simulate and predict the uranium leaching rate in the underground mineral layer.

[0049] In one embodiment, the digital twin generation module 102 is responsible for constructing and solving the fully coupled mathematical model of the uranium leaching process in real time, and outputting the predicted results of the concentration field, temperature field, pressure field and uranium leaching rate.

[0050] In one embodiment, the sub-models in the digital twin generation module 102 are described as follows: (1) Fluid flow sub-model (pressure field equation): Based on the law of conservation of mass and Darcy's law, it describes the multiphase fluid flow process in the reservoir:

[0051] in, This refers to relative penetration rate; For fluid dynamic viscosity; Porosity; For fluid density; For source and sink items; The permeability coefficient is dimensionless and is updated online by parameter correction module 105 (initial value). ).

[0052] (2) Solute transport model (concentration field equation): describes the convection-dispersion-reaction transport process of acid and uranium ions in the reservoir:

[0053] in, This refers to the solute concentration (injection solution concentration or uranium ion concentration). The molecular diffusion coefficient; Darcy velocity; For temperature-dependent chemical reaction source and sink terms; Porosity; The scalar dispersion coefficient is updated online by parameter correction module 105; The reaction rate is the baseline value and is updated online by the parameter correction module 105.

[0054] (3) Energy conservation sub-model (temperature field equation): describes the heat transfer and convective heat transfer during the injection process:

[0055] in, and Specific heat capacities of fluid and rock, respectively; and The densities of the fluid and the rock are respectively. Porosity; T is the local temperature; It is the equivalent thermal conductivity; Darcy velocity; It is the heat of a chemical reaction.

[0056] (4) Leaching reaction sub-model (uranium leaching rate equation): describes the acid leaching kinetics of uranium ore:

[0057] in, This refers to the concentration of uranium in the liquid phase. This is the leaching rate constant; The reaction order (usually) ); This refers to the solid uranium content; The reaction rate is the baseline value and is updated online by the parameter correction module 105.

[0058] In one embodiment, the digital twin generation module 102 receives the following input data through a monitoring layer communication network: (1) Boundary condition data: including the injection flow rate of each branch injection well. Injection concentration and injection temperature ( ), and the pumping flow rate of each pumping well. ( The above boundary conditions are collected by the multi-source sensing module 101 and provided after preprocessing.

[0059] (2) Model parameters: including key physical property parameters corrected by the parameter correction module 105EKF—corrected permeability coefficient Scalar dispersion coefficient and reaction rate baseline These parameters are updated and written back by the Kalman filter parameter correction module 105 at each decision cycle.

[0060] (3) Geometric model data: including the reservoir three-dimensional mesh model (used in this embodiment). The structured hexahedral mesh (25,000 mesh elements), well location coordinates, ore layer thickness distribution, and porosity field distribution.

[0061] In one embodiment, the digital twin generation module 102 uses the fully implicit finite volume method to numerically solve the above-mentioned coupled equations (Equations 4-7). The specific steps are as follows: (1) Spatial discretization: Spatial discretization is performed on the structured grid of the reservoir three-dimensional grid model using the finite volume method. The convection term adopts the second-order upwind scheme, and the diffusion term adopts the central difference scheme.

[0062] (2) Time progression: Backward Euler method is used for time discretization, and the time step size is... The value is adaptively adjusted according to CFL conditions, with a typical value of 1~10s.

[0063] (3) Coupled solution: The strong coupling relationship between the equations is handled by the Sequential Iterative Approach, that is, the pressure field, concentration field and temperature field are solved sequentially in each time step until the residual between each field is less than the convergence threshold. .

[0064] (4) Parallel computing: Using MPI (Message Passing Interface), computing tasks are distributed to the multi-core processors of the central monitoring station for parallel execution.

[0065] (5) Output prediction results: After the solution is completed, the digital twin generation module 102 outputs the following prediction results: reservoir three-dimensional concentration field This includes the spatial distribution of acid concentration field and uranium ion concentration field in underground mineral formations; and the three-dimensional temperature field of the reservoir. This includes the temperature diffusion and distribution of acid in underground reservoirs; and the three-dimensional pressure field of the reservoir. This includes the pressure field of the acid solution in the underground ore layer, the pressure funnel near the branch injection well, and the pressure reduction zone near the pumping well; leaching rate prediction. The volume-weighted average leaching rate of the entire field is calculated based on the concentration field, and the uranium production of each pumping well is predicted to obtain the uranium leaching rate in the underground ore layer. The above prediction results are sent to the intelligent injection decision module 103 and parameter correction module 105 through the monitoring layer communication network as the basis for optimization and parameter correction.

[0066] In one embodiment, the system further includes a parameter correction module 105; the parameter correction module 105 is used to update the parameters in the digital twin generation module 102 based on the operating data of each branch injection well and the predicted concentration field, temperature field, pressure field and uranium leaching rate of the acid in the underground ore layer.

[0067] In one embodiment, the parameter correction module 105 is responsible for comparing the prediction results of the digital twin generation module 102 with the actual measurement values ​​of the multi-source sensing module 101, and using the extended Kalman filter (EKF) algorithm to correct the key uncertain parameters in the digital twin generation module 102 in real time. This enables dynamic adaptive calibration of model parameters, improving the prediction accuracy of the digital twin generation module 102.

[0068] In one embodiment, the parameter correction module 105 receives two input data streams: (1) Model prediction values: The prediction results from the digital twin generation module 102 at the current moment, including the predicted values ​​of the injection concentration at each monitoring point. Predicted mixing temperature and wellhead pressure prediction .

[0069] (2) Measured values: Measured data from the corresponding measuring points provided by the multi-source sensing module 101 after preprocessing, including the measured values ​​of the injection concentration. Measured values ​​of mixing temperature and measured wellhead pressure .

[0070] In one embodiment, the parameter correction module 105 performs a five-step recursive process of the extended Kalman filter algorithm to correct the state estimates of reservoir physical parameters. The state vector in the parameter correction module 105 is defined as shown in formula (8):

[0071] in, For the first EKF state vector at time t; For the first Permeability correction factor at time, dimensionless, initial value EKF online estimates are dynamically updated; For the first Scalar dispersion coefficient correction value at time (m) 2 / d); For the first The reaction rate baseline correction factor at time t, dimensionless, initial value .

[0072] In one embodiment, the state transition equation (prediction step) in the parameter correction module 105 is shown in equations (9) and (10): A random walk model is used to describe the slowly varying characteristics of the parameters, and the state transition matrix is... Approximately the identity matrix:

[0073] In the formula: For the first State prior estimate (predicted value) at time 1; For the first Posterior state estimate (correction value) at time 1; This is the state transition matrix (3×3 identity matrix). Let be the process noise vector, assumed to be zero-mean Gaussian white noise, with the covariance matrix being... Covariance matrix The inclusion of reflects the fact that model uncertainties and parameters may drift slowly.

[0074] In one embodiment, the prior error covariance prediction in parameter correction module 105 is as shown in formula (11):

[0075] In the formula: For the first The prior error covariance matrix at time (3×3); For the first The posterior error covariance matrix at time (3×3); The process noise covariance matrix is ​​a 3×3 diagonal matrix with diagonal elements as follows: , , , representing the process noise variance of each parameter.

[0076] In one embodiment, the observation vector in the parameter correction module 105 is as shown in formula (12):

[0077] In the formula: For the first The observation vector at time; For the first Real-time measured value (mol / L) of acid concentration sensor; For the first Mixing temperature at the outlet of the L-shaped mixed acid exothermic tube (°C); For the first The wellhead outlet pressure (MPa) at any given time.

[0078] In one embodiment, the observation vector in the parameter correction module 105 can be expressed as the observation equation as shown in formula (13):

[0079] In the formula: The nonlinear observation function is estimated by the digital twin generation module 102 based on the prior state. Calculate model predictions get; The observed noise vector is assumed to be zero-mean Gaussian white noise, with the covariance matrix being... Observation function This represents the mapping relationship from model parameters to observations, i.e., given parameter estimates. The corresponding predicted observation values ​​are obtained by running the digital twin generation module 102.

[0080] In one embodiment, the parameter correction module 105 performs the following recursive steps: Input: The posterior state estimate from the previous time step The posterior covariance matrix of the previous time step The observation vector at the current moment .

[0081] Parameter: State transition matrix Process noise covariance matrix Observation noise covariance matrix Observation function .

[0082] Output: Posterior state estimate at the current time step. (i.e., the corrected model parameters), and the updated covariance matrix. .

[0083] (1) State prediction (time update): Based on the parameter estimate of the previous time step using formula (9) and state transition model Predicting the prior estimates of parameters at the current moment .

[0084] (2) Covariance prediction (time update): The error covariance matrix of the prior estimate is calculated using formula (11). .

[0085] (3) Gain calculation (measurement update): Calculate the Kalman gain matrix and determine the optimal weight between the model prediction and the measured value.

[0086] Calculate the Jacobian matrix (observation matrix) of the observation function:

[0087] In the formula: For the first The observation Jacobian matrix (3×3) at time t is obtained by the digital twin model through numerical difference method or adjoint equation method; This is the predicted value for the injection concentration; This is a predicted value for the mixing temperature; This is the predicted wellhead pressure value.

[0088] Kalman gain matrix calculation:

[0089] In the formula: For the first The Kalman gain matrix at time t (3×3).

[0090] Kalman gain matrix This determines the optimal weight allocation between model predictions and observed data during the state update step. (When observation noise...) When the noise level is high, the gain is low, and the model prediction is trusted more; when the observation noise is low, the gain is high, and the measured values ​​are trusted more.

[0091] (4) State update (measurement update): Using the residuals (innovation) between the measured values ​​and the model predictions, the prior state estimate is corrected to obtain the posterior state estimate.

[0092] In the formula: For the first Posterior state estimate (correction value) at time 1; This is the innovation vector, representing the deviation between the measured values ​​and the model predictions. If the innovation is zero, the state estimate is not corrected; the larger the innovation, the larger the state correction. The corrected state vector. This is the optimal parameter estimate for the current moment.

[0093] Perform a physical rationality constraint check on the revised parameters:

[0094] (5) Covariance update (measurement update): Calculate the error covariance matrix of the posterior estimate.

[0095]

[0096] In the formula: For the first The posterior error covariance matrix at time (3×3); It is a 3×3 identity matrix.

[0097] To ensure numerical stability, the Joseph stable form is often used in practical implementations:

[0098] The above five steps are performed in each sampling period. (recommend This process executes once within a minute (min), enabling dynamic, adaptive, online correction of model parameters. The corrected parameters... The monitoring layer communication network writes back to the digital twin generation module 102 to update its internal physical parameters. In the next decision cycle T... d The digital twin generation module 102 will use the updated parameters for calculation. The prior covariance matrix for the next decision cycle The initial value is used to continuously optimize the model's accuracy.

[0099] In one embodiment, the parameter correction module 105 is further configured to determine whether the parameters have converged based on a pre-set parameter correction threshold and the parameter correction amounts over multiple consecutive decision periods. Specifically, if the correction amounts of all three parameters are less than the threshold for five consecutive decision periods... If the parameters have converged to a stable value, the update interval of EKF can be extended in subsequent decision cycles (from once per decision cycle to once every 5 decision cycles) to reduce the computational load.

[0100] In one embodiment, the single-well multi-objective optimization sub-model is used to obtain initial concentration control commands for each branch injection well based on the predicted concentration field, temperature field, pressure field, and uranium leaching rate of the acid in the underground ore layer. and initial flow control command The multi-well collaborative optimization sub-model is used to obtain the concentration control command for each branch injection well based on pre-set global constraints and the initial concentration control command and the initial flow control command for each branch injection well. and the flow control command .

[0101] In one embodiment, the intelligent injection decision module 103 serves as an intelligent hub connecting the digital twin generation module 102 and the edge injection execution module 104. The intelligent injection decision module 103 receives the state simulation prediction results from the digital twin generation module 102 and generates concentration control commands and flow control commands for each branch injection well using multi-objective optimization and collaborative optimization algorithms. The concentration control command is the target injection concentration. The flow control command is the target injection flow rate. The intelligent injection decision module 103 outputs the optimal injection parameter settings. The command is then sent to the edge injection execution module 104 for execution.

[0102] In one embodiment, for the first For branch injection wells, a multi-objective optimization sub-model for a single well comprehensively considers three objectives: maximizing leaching efficiency, minimizing cost, and stability, and constructs a normalized weighted objective function:

[0103] In the formula: , , The weighting coefficients are dimensionless and satisfy the following conditions: The default value is set by the operator based on production priority. , , .

[0104] In the formula: for This is used to measure the contribution of injection parameters to uranium leaching efficiency, and is defined as the square of the difference between the predicted leaching rate and the target leaching rate: The negative sign is used because maximizing the leaching rate is equivalent to minimizing its negative value. For the first Injection parameters of branch injection wells The dissolution efficiency is as follows.

[0105] In the formula: The acid consumption cost item is defined as follows: It is used to measure the amount of acid consumed during the injection process. This is the unit acid cost coefficient; In the formula: The flow stabilization term is a penalty term used to prevent drastic fluctuations in injection flow rate from adversely affecting the reservoir. It is defined as follows: , This is the stability weighting coefficient; For the first Injection concentration of the branch injection well; For the first Injection flow rate of the branch injection well.

[0106] Single-well optimization constraints:

[0107] in, and These are the minimum and maximum permissible injection concentrations, respectively. and These are the minimum and maximum permissible injection flow rates, respectively; This refers to the reservoir fracture pressure. This is the measured value of the injection pressure.

[0108] In one embodiment, the single-well optimization problem is solved using a sequential quadratic programming (SQP) algorithm, with the predicted output of the digital twin generation module 102 serving as the evaluation basis for the objective function and constraints in each decision cycle. (recommend Complete one optimization solution within (min), and output the initial concentration control command and the initial flow rate control command, i.e., the initial injection parameters. .

[0109] In one embodiment, Sequential Quadratic Programming (SQP) is an efficient algorithm for solving constrained nonlinear optimization problems. This application uses it to solve for the local optimum of injection parameters in a single well. The detailed execution steps of single-well SQP optimization are as follows: Input: The prediction results of the current digital twin generation module 102, including the spatial distribution of the concentration field, temperature field, and pressure field; optimized weighting coefficients. The upper and lower limits of each constraint.

[0110] Control parameters: Convergence threshold Maximum number of iterations Line search parameters (Armijo constant), .

[0111] Optimize variables: That is, the first Injection concentration and flow rate of the branch injection well.

[0112] The objective function is shown in formula (20), and the constraints in formula (21) are transformed into:

[0113] Step S1: Initialization Set the initial values ​​for the optimization variables:

[0114] The initial value can be taken as the injection parameter corresponding to the prediction result of the current digital twin generation module 102. The initial values ​​can also be estimated based on the prediction results of the current digital twin generation module 102, using a pre-set steady-state model to approximate an initial optimization variable. Alternatively, the initial values ​​can use the optimization solution from the previous decision cycle as a reference, but this must be verified and corrected using the current field data.

[0115] Step S2: Calculate the objective function value and gradient. Call the fast prediction interface of the digital twin generation module 102 at the current iteration point. Calculate the objective function value at [location]. and gradient :

[0116] The gradient components are calculated numerically using the central difference method:

[0117] Simultaneously calculate the values ​​of each constraint function. and constrained gradient ( ).

[0118] In the first iteration, the objective function value is calculated using the prediction result of the current digital twin generation module 102 from the input parameters. and gradient Starting from the second iteration, the digital twin generation module 102 is invoked based on the optimized variables updated in the iteration. Perform a simulation, obtain the simulation results, and calculate the objective function value. and gradient .

[0119] Step S3: Construct a quadratic programming subproblem At the current iteration point Here, the original nonlinear optimization problem is approximated as a quadratic programming (QP) subproblem. The QP subproblem has the following form:

[0120] in, Indicating the search direction; To approximate the Hessian matrix of the objective function, the BFGS (Broyden-Fletcher-Goldfarb-Shanno) quasi-Newton method can be used for updating. This is to ensure its positive definiteness.

[0121] The BFGS update formula is:

[0122] in:

[0123] in, To constrain the corresponding Lagrange multipliers.

[0124] Step S4: Solve the QP subproblem The above QP subproblem is solved using either the ActiveSet Method or the InteriorPoint Method to obtain the search direction. and the corresponding Lagrange multipliers The solution process must satisfy the KKT (Karush-Kuhn-Tucker) conditions:

[0125] in, For an effective constraint set.

[0126] like If the algorithm fails to converge, proceed to step S5. like If the algorithm converges, proceed to step S7 to output the result.

[0127] Step S5: Armijo line search Along the search direction Perform a line search to determine the optimal step size. This allows the objective function to decrease sufficiently. The Armijo backtracking search criterion is adopted:

[0128] Initial step size of line search If the Armijo condition is not met, then proceed as follows: ( Reduce the step size until the condition is met or the minimum step size is reached. .

[0129] At the same time, it is required that the iteration points after the line search satisfy all constraints (feasibility):

[0130] If the feasibility conditions are not met, the step size should also be reduced.

[0131] Step S6: Update the iteration point Update and optimize variables based on search direction and step size:

[0132] Update the Hessian approximation matrix using the BFGS formula. ,make .like If the iteration continues, return to step S2; otherwise, output the current iteration point as the approximate optimal solution.

[0133] Step S7: Convergence Judgment and Output The algorithm terminates when any of the following conditions are met: (1) Gradient norm condition: ( ); (2) Search direction norm condition: ; (3) Conditions for the change of the objective function: ; (4) Reaching the maximum number of iterations .

[0134] Output: The Local optimal injection parameters for branch injection wells The initial concentration control commands for each branch injection well were obtained. and initial flow control command .

[0135] In one embodiment, the multi-well collaborative optimization sub-model receives the SQP optimization solutions from each branch injection well generated by the single-well multi-objective optimization sub-model. In one embodiment, there are a total of 8 branch injection wells. The SQP optimization solution for each branch injection well only considers the optimization objective of each branch injection well, without taking into account inter-well interference and global constraints. The multi-well collaborative optimization sub-model uses the Alternating Direction Multiplier Method (ADMM) to optimize the SQP solution for each branch injection well. Perform global coordination to achieve multi-well collaborative optimization while satisfying global constraints.

[0136] In one embodiment, the execution steps for the multi-well collaborative optimization sub-model are described below: Input: Initial local solution - SQP optimized solution for each branch injection well Global consensus variables Initial Lagrange multipliers ( , (Number of injection wells), penalty parameters .

[0137] Parameter: Original residual threshold Dual residual threshold Maximum number of iterations Penalty parameters (Adaptive adjustment).

[0138] Step D1: Local optimization is performed in parallel on each branch injection well. In the In each iteration, each branch injection well Solve the following subproblem of minimizing the augmented Lagrange function independently and in parallel:

[0139] This subproblem is formally a single-well optimization problem with a quadratic regularization term, and it can also be solved using the SQP algorithm in the single-well multi-objective optimization sub-model mentioned above, which will not be elaborated here.

[0140] Quadratic regularization term Its function is to penalize the degree to which the local solutions of each injection well deviate from the global consensus variables; the penalty strength is determined by the parameters. control.

[0141] Specifically, after expanding the regularization term, the objective function of the local subproblem becomes:

[0142] in, It is an auxiliary variable.

[0143] Step D2: The global coordinator updates the consensus variables. The global coordinator collects updated local solutions from each branch injection well. Calculate global consensus variables :

[0144] The physical meaning of this formula is as follows: the global consensus variable is the arithmetic mean of all local solutions (after multiplier offset), representing the "collective will" of the injection parameters of each branch injection well. Through the global consensus variable, implicit coordination is achieved between the injection wells of each branch, avoiding inter-well interference and local congestion caused by independent optimization.

[0145] Step D3: Renew the Lagrange multipliers in each branch injection well. Each injection well in each branch updates its Lagrange multiplier based on the deviation between the local solution and the global consensus variables:

[0146] The multiplier update reflects the "dissatisfaction" of each branch injection well with the global consensus: if the local solution of a well is much higher than the global consensus, its multiplier increases, and the well will be subject to a stronger penalty in the next iteration, thus tending to reduce the local solution; conversely, the opposite is also true. Through this mechanism, the ADMM algorithm achieves adaptive coordination among the injection wells in each branch.

[0147] Step D4: Calculate the original residual and the dual residual. Calculate the original residual. And dual residuals :

[0148] Original residual The degree of consistency between the local solutions of each branch injection well and the global consensus is measured. If The small value indicates that the local solutions of the injection wells in each branch are becoming consistent, and global coordination has been basically achieved.

[0149] Dual residuals This measures the change in the global consensus variable between two consecutive iterations. If The value is very small, indicating that the global consensus has stabilized and further iterations are unlikely to significantly improve the results.

[0150] Step D5: Convergence Judgment and Iterative Control Determine if the ADMM algorithm has converged. The convergence condition is that the following conditions must be met simultaneously: (1) Original residual condition: The original residual is less than or equal to the original residual threshold, i.e. ; (2) Dual residual condition: The dual residual is less than or equal to the dual residual threshold, i.e. .

[0151] The original residual threshold is defined as:

[0152] The dual residual threshold is defined as follows:

[0153] If both of the above conditions are met, the algorithm terminates and outputs the final globally coordinated solution. Otherwise, let Return to step D1 and continue iterating. If the maximum number of iterations is reached... If convergence is still not achieved, output the current iteration result and record a warning message.

[0154] Output: Concentration control commands for each branch injection well after global coordination. and the flow control command The final output is the set value of the injection parameters. ,in This represents the number of iterations required for ADMM convergence. The concentration control commands for each branch injection well. and the flow control command The target value for the next control cycle is sent to each edge injection execution module 104 through the monitoring layer communication network.

[0155] In one embodiment, the edge injection execution module 104 includes an injection concentration control submodule and an injection flow rate control submodule. The injection concentration control submodule is used to regulate the injection concentration of each branch injection well based on a pre-set feedforward compensation strategy and a feedback correction strategy, according to the operating data and concentration control commands of each branch injection well. The injection flow rate control submodule is used to regulate the injection flow rate of each branch injection well based on a pre-set adaptive fuzzy PID control strategy, according to the operating data and flow rate control commands of each branch injection well.

[0156] In one embodiment, the edge injection execution module 104 is responsible for implementing concentration control commands issued by the central monitoring station of the monitoring layer. and flow control commands It independently and in real-time performs precise control of injection parameters. Its workflow consists of two parallel paths: a concentration control loop and a flow control loop. The concentration control loop runs the injection concentration control submodule, and the flow control loop runs the injection flow control submodule.

[0157] In one embodiment, the feedforward compensation strategy is based on the concentration control command. Estimate the base pump speed and use the feedback correction strategy based on the concentration control command. Measured values ​​of injection concentration The deviation between the two values ​​is used to correct the estimated base pump speed to obtain the final pump speed corresponding to the injection concentration. The pump speed corresponding to the injection concentration is the pump speed used to pump the acid solution in the concentrated acid storage tank, so that the acid solution and dilution water in the concentrated acid storage tank are mixed to obtain the final pump speed. .

[0158] The injection concentration control submodule is described below: Input: Concentration control command (Unit: mol / L), dilution water flow rate (Unit: m³ / h), concentrated acid concentration (Unit: mol / L), Metering pump flow coefficient (Unit: L / r).

[0159] Parameter: Proportional Gain (Unit: rpm / (mol / L)), integration time constant (Unit: seconds), control cycle (Unit: s).

[0160] variable: For concentration deviation; For feedforward base pump speed; This is the PI feedback correction amount; For final pump speed output; integral cumulative quantity .

[0161] Step A1: Calculate the base pump speed based on the feedforward compensation strategy.

[0162] According to concentration control instructions and dilution water flow rate The feedforward base speed of the metering pump is calculated using the material balance relationship:

[0163] The physical meaning of this formula is: in order to achieve the target injection concentration of the injection well. A certain amount of concentrated acid (concentration of ) needs to be disposed of within a unit of time. Inject a certain amount of dilution water (flow rate is...) The mixture is then mixed in the following manner. The speed of the metering pump is directly proportional to the volume of acid to be delivered, with a proportionality constant of [value missing]. Feedforward control can pre-adjust the actuator according to changes in concentration control commands, effectively reducing system response delay.

[0164] Step A2: Calculate the concentration deviation.

[0165] The measured value of the injected liquid concentration fed back by the concentration sensor is obtained from the multi-source sensing module 101: the measured value of the injected liquid concentration at the current moment is read from the concentration sensor installed at the outlet of the injection pipeline. The concentration sensor output signal is transmitted via a 4~20mA current loop and converted into a digital value by the A / D conversion module on the edge controller.

[0166] Assume the sensor range is The A / D conversion result is (12 bits, range 0~4095), then: .

[0167] Calculate concentration control instructions Measured values ​​of injection concentration Deviation between:

[0168] like ( Let be the dead zone threshold (usually 0.1% of the range), then let This is to avoid frequent controller adjustments caused by measurement noise during steady-state operation.

[0169] Step A3: Calculate the feedback correction amount using a PI controller based on the feedback correction strategy.

[0170] The pump speed feedback correction is calculated using an incremental PI control algorithm. Incremental PI control has the following advantages over positional PI control: ① No integral saturation problem; ② Automatic anti-integral saturation when the controller output is limited; ③ No disturbance when switching from manual to automatic.

[0171] The discrete form of the incremental PI control algorithm is:

[0172] in, For proportional gain, The integral time constant is... To control the sampling period (in this embodiment) ms).

[0173] Accumulated points are updated using the following formula:

[0174] Step A4: Calculate the final pump speed.

[0175] feedforward base pump speed With PI feedback correction amount By superimposing these values, the final speed command for the metering pump is obtained:

[0176] right Perform upper and lower limit processing: If the final pump speed is less than the preset minimum pump speed, adjust the final pump speed to the minimum pump speed and then transmit it to the next step; if the final pump speed is greater than the preset maximum pump speed, adjust the final pump speed to the maximum pump speed and then transmit it to the next step; if the final pump speed is between the minimum pump speed and the maximum pump speed, the final pump speed is transmitted to the next step.

[0177]

[0178] Step A5: Output to metering pump frequency converter.

[0179] After the limit Frequency commands converted to metering pump inverters :

[0180] in, For metering purposes. Pump rated speed (300 rpm). The frequency is the inverter's rated frequency (50Hz). The frequency command is sent to the metering pump inverter via the device layer RS-485 bus (Modbus-RTU protocol) to drive the metering pump to operate at the final pump speed.

[0181] When the intelligent injection decision module 103 issues a new concentration control command If necessary, repeat steps A1 to A5. The expected performance index of this concentration control loop is: steady-state error. Adjusting time s, overshoot .

[0182] In one embodiment, the adaptive fuzzy PID control strategy adjusts the PID parameters in real time based on the injection flow rate deviation and its rate of change, in order to take into account reservoir heterogeneity and inter-well interference effects and adapt to the time-varying characteristics of reservoir parameters.

[0183] The following is a description of the injection flow control submodule: Input: Flow control commands for each branch injection well ; Pre-constructed fuzzy mapping relationship between flow deviation, deviation change rate, and PID parameters; Measured value of injection flow rate ; Step B1: Receive the target injection flow rate setting.

[0184] Edge injection execution module 104 receives flow control commands from each branch injection well. (Unit: m³ / h). The results were determined through ADMM multi-well collaborative optimization, taking into account reservoir heterogeneity and inter-well interference effects.

[0185] Step B2: Adaptive fuzzy PID control calculation.

[0186] Calculate the flow deviation and the rate of change of deviation:

[0187] Fuzzification: Flow deviation is fuzzified based on triangular membership functions. and rate of change of deviation These are mapped to a pre-constructed fuzzy domain to obtain the fuzzy states of flow deviation and deviation change rate. Seven fuzzy linguistic variables are defined in the fuzzy domain: {NB (negative large), NM (negative medium), NS (negative small), ZO (zero), PS (positive small), PM (positive medium), PB (positive large)}.

[0188] Fuzzy reasoning: Based on 49 pre-built fuzzy control rules, according to From the fuzzy state, the PID parameter correction amount is derived. , , For example: If For NB and If it is NB, then For PB, For NB, For PS.

[0189] Defuzzing and Parameter Update: The centroid method is used for defuzzing to obtain accurate parameter corrections, and the PID parameters are then updated.

[0190] PID control quantity calculate:

[0191] Step B3: Drive the injection pump.

[0192] Control quantity based on pre-set amplitude limiting rules After amplitude limiting and rate limiting, the frequency command is converted into that of the injection pump inverter. (Unit: Hz), output to the injection pump driver. The injection pump is a variable frequency centrifugal pump with a frequency adjustment range of 20~50Hz, corresponding to a flow rate adjustment range of 2~8m³ / h. The injection pump injects the prepared leachate into the ore layer through the injection pipeline, completing the closed-loop control of the injection process.

[0193] In one embodiment, the system includes an injection concentration control subsystem and an injection flow rate subsystem. The injection concentration control subsystem includes an acid storage tank, a metering pump, a dilution water source, an L-shaped mixed acid exothermic pipe, an online pH sensor, and an online concentration sensor. The acid storage tank is used to store the raw acid solution (typically industrial sulfuric acid or hydrochloric acid solution) required for uranium leaching. The metering pump is a frequency-controlled diaphragm or plunger metering pump, operating according to the final speed command output by a feedforward-feedback composite controller. Precise adjustment of the concentrated acid injection rate enables closed-loop control of the mixed acid concentration. The L-shaped mixed acid exothermic tube is a key component in the mixing reaction of acid and dilution water. Its L-shaped design extends the mixing path, allowing for uniform release of the heat of chemical reaction during acid dilution and preventing localized overheating. The tube body is made of Hastelloy alloy or PTFE-lined steel, offering corrosion resistance and excellent thermal conductivity. An online pH sensor, installed at the outlet of the L-shaped mixed acid exothermic tube, monitors the pH value of the mixed acid solution in real time. The online pH sensor uses a glass electrode industrial pH meter with automatic temperature compensation (ATC). The measurement value from the online pH sensor serves as a cross-validation signal for concentration control; when the deviation between the pH value and the theoretical pH value corresponding to the concentration measured by the concentration sensor exceeds a threshold, a calibration prompt is triggered. An online concentration sensor, installed alongside the pH sensor at the outlet of the L-shaped mixed acid exothermic tube, measures the actual concentration of the mixed acid solution in real time. The online concentration sensor measures acid concentration using either the conductivity method or the ultrasonic velocity method. The measured value from the online concentration sensor is the core feedback signal of the PI feedback controller, used to calculate the concentration deviation. During operation, the metering pump pumps the acid from the acid storage tank to the L-shaped mixed acid exothermic pipe, where it is mixed with dilution water and then injected into the underground mineral layer through the injection flow subsystem.

[0194] The injection flow rate subsystem includes an injection pump, a speed sensor, and a pressure sensor. The injection pump is the core actuator of the independent injection pump subsystem. It is a variable frequency driven centrifugal pump or plunger pump. Its inlet connects to the outlet of an L-shaped mixed acid exothermic pipe (after pH and concentration detection), and its outlet connects to the wellhead injection pipeline, injecting the pre-mixed acid solution into the underground mineral formation. The injection pump adjusts the injection flow rate according to the speed command output by an adaptive fuzzy PID controller. The speed sensor, mounted on the pump shaft, is used to detect the actual speed of the injection pump in real time. The speed sensor employs either a photoelectric encoder or a Hall effect sensor. The feedback signal from the speed sensor is used for closed-loop speed control of the injection pump and equipment status monitoring (determining whether speed deviation exceeds limits). A pressure sensor, installed on the injection pump's outlet pipeline, is used to monitor the pump's outlet pressure and wellhead back pressure in real time. The pressure sensor employs a piezoresistive industrial pressure transmitter, outputting a standard 4-20 mA signal. The pressure sensor's measurements are used for pump pressure coupling function calculations and equipment condition monitoring.

[0195] In one embodiment, the system further includes an anomaly warning module 106; the anomaly warning module 106 is used to compare the operating data of each branch injection well with its corresponding normal threshold based on a pre-set normal threshold, and determine a first anomaly index based on the comparison result; the anomaly warning module is used to compare the operating data of each branch injection well with the simulated prediction value of its corresponding digital twin generation module, and determine a second anomaly index based on the comparison result; the anomaly warning module is used to generate a Kalman filter innovation sequence based on the operating data of each branch injection well and the simulated prediction value of its corresponding digital twin generation module, analyze the root mean square error of the Kalman filter innovation sequence within multiple consecutive preset decision periods, and determine a third anomaly index based on the analysis result; the anomaly warning module is used to issue an anomaly warning based on the first anomaly index, the second anomaly index, and the third anomaly index using a pre-set weighted voting mechanism. It should be noted that the operating data refers to a specific measured value and is not specifically limited; the operating data used in the first anomaly index, the second anomaly index, and the third anomaly index can be the same or different.

[0196] In one embodiment, the anomaly warning module 106 is deployed at the central monitoring station and is responsible for monitoring the operating status of the entire system and providing anomaly warnings.

[0197] In one embodiment, the anomaly warning module 106 simultaneously receives three input data streams, performs a comprehensive anomaly judgment on the three input data streams, and adopts a multi-source information fusion strategy to improve the accuracy and reliability of anomaly detection: (1) The first stream is real-time sensor data from the multi-source sensing module 101, including the measured values ​​of each sensor and the device status code; used to calculate the first anomaly index. (2) The second stream is model prediction data from the digital twin generation module 102, including the normal operating condition prediction values ​​of each monitoring point; used to calculate the second anomaly index. (3) The third stream is filtered residual data from the parameter correction module 105, including the deviation statistics between the model prediction index and the measured value, used to calculate the third anomaly index.

[0198] In one embodiment, the anomaly warning module 106 determines whether there is an anomaly in the system. The anomalies in the anomaly warning module 106 include mixing uniformity anomalies, thermal safety anomalies, and equipment status anomalies. The first anomaly index in the anomaly warning module 106 comprehensively considers mixing uniformity anomalies, thermal safety anomalies, and equipment status anomalies; the second anomaly index considers reservoir anomalies; and the third anomaly index provides an early warning of system anomalies from the perspective of model parameters.

[0199] In one embodiment, the anomaly warning module 106 is further configured to: collect operating data including pipeline friction temperature gradient, mixing temperature, rotational speed deviation, and flow rate deviation; compare the pipeline friction temperature gradient, mixing temperature, rotational speed deviation, and flow rate deviation with their corresponding normal thresholds, and determine whether there are any anomalies in mixing uniformity, thermal safety, or equipment status based on the comparison results; determine the number of anomaly types in mixing uniformity, thermal safety, and equipment status, with a first anomaly index representing the number of anomaly types. For example, if there are no anomalies in mixing uniformity, thermal safety, or equipment status, the first anomaly index is 0; if any one of these anomalies exists, the first anomaly index is 0.33; if two of these anomalies exist, the first anomaly index is 0.66; and if all three anomalies exist, the first anomaly index is 1.

[0200] In one embodiment, the mixing uniformity is abnormal: through a temperature gradient. To assess the homogeneity of the mixed acid, an abnormality is identified if the temperature exceeds a preset normal threshold minus a temperature gradient threshold. Thermal safety abnormality: When the mixing temperature... (Normal threshold - safe temperature threshold) If an abnormality is detected, the system will automatically increase the dilution water flow rate or suspend acid injection. Abnormal equipment status: Monitor rotational speed deviation. and flow deviation If the value continuously exceeds its corresponding normal threshold (assuming 10%), the device is considered abnormal.

[0201] In one embodiment, when judging the abnormality of mixing uniformity, the inlet, middle and outlet temperatures of the L-shaped mixed acid exothermic pipe fed back by the multi-source sensing module 101 are obtained in real time, and the temperature gradient along the pipe is calculated:

[0202] In the formula: The friction temperature gradient along the pipeline (°C / m); , Temperature (°C) at adjacent measuring points along the pipeline; , The location (m) of adjacent measuring points.

[0203] When the temperature gradient between adjacent measuring points along any pipeline exceeds the threshold (recommend When the concentration of concentrated sulfuric acid is less than 1 / m, the system determines that the mixing uniformity is abnormal and triggers a mixing unevenness alarm, indicating that there may be a risk of insufficient mixing between concentrated sulfuric acid and dilution water.

[0204] In one embodiment, the mixing temperature at the outlet of the L-shaped mixed acid exothermic tube is acquired in real time from the feedback of the multi-source sensing module 101. When the mixing temperature at the outlet of the L-type mixed acid exothermic tube Exceeding the safe temperature threshold (recommend When a thermal safety anomaly is detected, a thermal safety anomaly alarm is triggered, and the following protective measures are automatically implemented: Prioritize increasing the dilution water flow rate. This reduces the concentration and temperature of the mixed acid solution. If the dilution water flow rate is increased within 5 seconds... It has not yet dropped to The injection of concentrated sulfuric acid shall then be suspended (i.e., the injection shall be stopped). (This continues) until the mixing temperature returns to a safe range.

[0205] In one embodiment, during equipment anomaly detection, the measured value of the metering pump speed fed back by the multi-source sensing module 101 is acquired in real time. Measured value of injection flow rate Calculate the measured value of the metering pump speed. With final speed (See formula (48)) deviation and measured values ​​of injection flow rate With flow control commands deviation :

[0206] when or Continue to exceed When the time reaches min, it is determined that the equipment is abnormal, triggering an equipment abnormality warning and automatically switching to the backup pump or executing a safety shutdown procedure.

[0207] In one embodiment, the anomaly warning module 106 is further configured to: determine that an anomaly exists if the running data fed back at multiple consecutive sampling times all exceed their corresponding normal thresholds.

[0208] In one embodiment, the anomaly warning module 106 is further configured to: calculate the standardized residual between the measured value of the running data and its corresponding model prediction value; if the standardized residual is greater than a preset residual threshold, an anomaly is determined; and a second anomaly index is determined based on the number of types of running data with anomalies.

[0209] Specifically, the measured values ​​of the operating data Its corresponding model prediction value Standardized residuals between:

[0210] in, The standard deviation of the model for the running data is known. It can be the measured value of reservoir temperature, the measured value of reservoir acid concentration, the measured value of reservoir pressure, or the measured value of uranium leaching rate at the measuring point. The measured value of uranium leaching rate can be calculated based on the measured value of uranium concentration in the leaching solution. It is obtained from the concentration field, temperature field, pressure field, and uranium leaching rate obtained from the simulation.

[0211] Specifically, the standardized residuals corresponding to reservoir temperature, reservoir acid concentration, reservoir pressure, and uranium leaching rate are calculated respectively. If the standardized residuals corresponding to reservoir temperature, reservoir acid concentration, reservoir pressure, and uranium leaching rate are within a preset time period... None of them exceeded the preset residual threshold, and the second anomaly index was 0; if the standardized residuals corresponding to reservoir temperature, reservoir acid concentration, reservoir pressure, and uranium leaching rate within the preset time period are within the preset time period, then the index is 0. If any of the following exceeds a preset residual threshold, the second anomaly index is 0.25; if the standardized residuals corresponding to reservoir temperature, reservoir acid concentration, reservoir pressure, and uranium leaching rate exceed a preset time limit, the second anomaly index is 0.25. If any two of the parameters exceed the preset residual threshold, the second anomaly index is 0.5; if the standardized residuals corresponding to reservoir temperature, reservoir acid concentration, reservoir pressure, and uranium leaching rate exceed the preset residual threshold within the preset time period, the second anomaly index is 0.5. If any three exceed the preset residual threshold, the second anomaly index is 0.75; if the standardized residuals corresponding to reservoir temperature, reservoir acid concentration, reservoir pressure, and uranium leaching rate are within the preset time period... All exceeded the preset residual threshold, and the second anomaly index was 1. 0, 0.25, 0.5, 0.75, and 1 are just examples and not specific limitations. The second anomaly index can also be a weighted sum of the number of data types.

[0212] For example, if the preset residual threshold is 3 and the preset duration is 10 seconds, then... If the condition persists for more than 10 seconds, the reservoir state is considered to be abnormal.

[0213] In one embodiment, the operational data of each branch injection well when calculating the third anomaly index includes the measured values ​​of injection concentration from the multi-source sensing module 101. Measured values ​​of mixing temperature and measured wellhead pressure And the predicted injection concentration from the digital twin generation module 102 Predicted mixing temperature and wellhead pressure prediction .

[0214] The abnormal warning module 106 is also used to predict the value of the injection concentration. Predicted mixing temperature and wellhead pressure prediction The nonlinear observation function is obtained (as shown in Equation 13), based on the measured value of the injection concentration. Measured values ​​of mixing temperature and measured wellhead pressure The observation vector is obtained (as shown in Equation 12). Based on the nonlinear observation function and the observation vector, the Kalman filter innovation sequence for each decision period is obtained. If the root mean square error of the Kalman filter innovation sequence monotonically increases over multiple consecutive preset decision periods, and the actual increase exceeds a preset increase threshold (e.g., 20%), then the third anomaly index is the actual increase. For example, if the actual increase is 60%, the third anomaly index is 0.6. If the actual increase does not exceed the preset increase threshold, the third anomaly index is 0. The third anomaly index indicates that the model parameters may drift or that reservoir conditions may change.

[0215] In one embodiment, a multi-source fusion decision is employed: the detection results of the three detection methods are combined using a weighted voting mechanism. Each detection method is assigned a different weight based on its historical accuracy, and the final anomaly confidence level is:

[0216] in, For the first The weights of the detection methods, m=1, 2, 3; This is the anomaly index of the method.

[0217] In one embodiment, the anomaly warning module 106 is further configured to generate an anomaly risk value based on a pre-set weighted voting mechanism, according to the first anomaly index, the second anomaly index, and the third anomaly index. The anomaly warning module 106 is also used to compare the anomaly risk value with multiple pre-set anomaly risk level thresholds to determine the anomaly warning level and warning instruction.

[0218] Multiple abnormal risk level thresholds constitute three early warning ranges: the first early warning range Second warning range Third warning range Based on the confidence level and severity of the anomaly, the anomaly risk level threshold and the anomaly early warning module 106 implement a three-level early warning mechanism: (1) Level 1 Warning (Yellow Warning): When The alert is triggered automatically. The alert includes issuing an audible and visual alarm to the operator and highlighting abnormal parameters on the monitoring interface. The alert also suggests that the operator check and confirm within 30 minutes. The system continues to operate normally, but the decision-making cycle is shortened to 1 minute to enhance monitoring.

[0219] (2) Level II Warning (Orange Warning): When The alert is triggered automatically. The alert commands include automatically adjusting the injection parameters of the relevant branch injection wells (reducing the injection concentration by 10% and the injection flow rate by 20%), and sending a detailed anomaly diagnosis report to the operator, recommending that the operator conduct an on-site inspection within 10 minutes. If the anomaly is not resolved within 10 minutes, the alert is automatically escalated to a level three alert.

[0220] (3) Level III Warning (Red Warning): When The warning is triggered on a timely basis. The warning instructions include immediately executing the shutdown protection procedure: sequentially shutting down the injection pump and metering pump, closing relevant valves, and switching the system to a safe state. The warning instructions also include sending an emergency shutdown notification to the central monitoring station and on-site operators, and recording complete data snapshots before and after the shutdown for subsequent fault analysis.

[0221] In one embodiment, each warning event (regardless of level) is fully recorded in the warning log database, including: warning time, triggering parameters, warning level, system response measures, operator handling measures, and final result. The anomaly warning module 106 periodically (weekly) performs statistical analysis on the warning logs and automatically adjusts the weights of each detection method. The threshold parameters within the warning range enable the anomaly detection capability of the anomaly warning module 106 to continuously improve with the increase of running time.

[0222] This invention provides a precise injection control system for in-situ leaching uranium mining. By collecting operational data from each branch injection well in real time, the system uses a digital twin generation module to simulate and predict the ore layer response characteristics to accurately characterize the ore layer. Based on the prediction results, concentration control commands and flow control commands are generated through a single-well multi-objective optimization sub-model and a multi-well collaborative optimization sub-model. This ensures maximum leaching efficiency for a single well while achieving coordinated control among multiple wells. Finally, the concentration control commands and flow control commands are executed through an edge injection execution module, achieving decoupling of injection concentration and injection flow rate, and realizing a precise injection strategy of "one policy per well" for heterogeneous ore layers.

[0223] Specifically, the technical effects of the present invention include: (1) This invention solves the technical bottleneck of traditional "flood irrigation" injection mode, which cannot implement differentiated injection strategies for ore layers with different permeabilities and grades, by deploying an edge injection execution module 104 with independent acid mixing and pumping control commands at the front end of the injection pipeline of each branch injection well. Each branch injection well can independently adjust the acid concentration and injection flow rate, and customize an exclusive injection plan for each injection well according to the response characteristics of the ore layer, realizing true "one well, one policy" precision injection, significantly improving reagent utilization efficiency and uranium recovery rate.

[0224] (2) This invention achieves decoupling between injection concentration and injection flow rate by independently controlling the acid ratio using a metering pump and independently controlling the injection flow rate to the formation using an injection pump. This allows the injection concentration to be controlled within the specified range. ~ The injection flow rate can be continuously adjusted within a certain range. ~ The system allows for continuous adjustment within a given range, with each adjustment independent of the others, fundamentally eliminating control nonlinearity caused by the operating limits of the pumping system. Especially during the later stages of mining when formation pressure increases, the independent injection pump can provide stable and adjustable pumping power, effectively overcoming the "injection failure" problem caused by insufficient output pressure head of traditional valves, and expanding the system's applicable operating conditions.

[0225] (3) This invention utilizes an intelligent injection decision module 103 to achieve a coordinated strategy for dynamically optimizing acid concentration and injection flow rate based on the prediction results and measured results of the digital twin generation module 102. This module, based on the current state and trend prediction of the ore layer, effectively avoids ore layer damage and reagent waste caused by localized over-acidification while maximizing the leaching efficiency of a single well. This achieves a balance between economic efficiency and high efficiency in the leaching process, promoting the transformation of in-situ uranium leaching from experience-driven to data and model-driven approaches.

[0226] (4) Based on multi-source sensing data and geological prior knowledge, this invention constructs a dynamic digital twin generation module 102 for each injection well, realizing three-dimensional dynamic simulation of the in-situ leaching process. The digital twin generation module 102 not only includes mass transfer processes such as acid transport and uranium dissolution, but also couples reaction heat effects and temperature field evolution. Compared with simplified models that only focus on equipment status or single parameter adjustment, it has higher simulation fidelity and prediction accuracy.

[0227] (5) This invention achieves online adaptive calibration of the parameters of the digital twin generation module 102 through the Kalman filter parameter correction module 105. The closed-loop feedback optimization mechanism continuously compares the model prediction output with the actual sensed data, dynamically correcting key time-varying parameters in the model such as the penetration coefficient and reaction rate. Compared with static parameter models, the dynamic parameter correction mechanism of this invention can reduce the model prediction error under simulation experimental conditions. The above measures effectively address time-varying factors such as changes in ore layer permeability and decay of reactivity, continuously improving the system's self-adaptive capabilities.

[0228] (6) This invention achieves coordinated control among multiple wells while maximizing the leaching efficiency of a single well through a multi-objective collaborative optimization mechanism. The introduction of the ADMM algorithm allows each well to reach the global optimum through lightweight coordinated iteration based on local independent decision-making, avoiding main pipe pressure conflicts and inter-well interference. Compared with a centralized optimization architecture, this invention can reduce the communication load under simulation experimental conditions. The above significantly improves the system scalability and real-time performance in large-scale well cluster scenarios.

[0229] (7) The present invention covers three key risk dimensions—mixing uniformity, thermal safety, and equipment status—through the multi-dimensional monitoring system of the abnormal early warning module 106, forming a full-process safety guarantee from pre-event prevention, in-event monitoring to post-event handling.

[0230] (8) By configuring independent injection pumps for each edge injection execution module 104, this invention enables precise and independent control of injection parameters, while each pump can operate at its optimal efficiency point according to actual working conditions. This avoids the energy waste caused by using high-power pumps for full-area pressurization to overcome pipeline resistance in centralized liquid supply architectures. The distributed pumping strategy of supplying liquid on demand effectively reduces ineffective pipeline losses and improves the overall system's energy utilization efficiency and operational economy.

[0231] (9) This invention constructs a complete intelligent control chain through deep collaboration of multi-source sensing, dynamic modeling, intelligent decision-making, precise execution, and closed-loop feedback, realizing autonomous sensing, autonomous analysis, and autonomous optimization of the entire in-situ leaching uranium mining process. This closed-loop architecture significantly improves reagent utilization efficiency, uranium recovery rate, and process safety, enabling the in-situ leaching uranium mining process to transform from a traditional experience-driven model to a data and model-driven model, and has broad engineering applicability and promotion prospects.

[0232] The precise injection control method for uranium mining provided in this invention can be applied to, for example... Figure 2 In this application environment, the device communicates with the server via a network. The server can collect operational data from each branch injection well through the device; based on a pre-built digital twin model, it simulates and predicts the concentration field, temperature field, pressure field, and uranium leaching rate of the acid in the underground ore layer according to the operational data of each branch injection well; based on a pre-set single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model, it generates concentration control commands and flow control commands for each branch injection well according to the predicted concentration field, temperature field, pressure field, and uranium leaching rate of the acid in the underground ore layer; and it regulates the injection concentration and flow rate of each branch injection well according to the operational data, concentration control commands, and flow control commands. The device can be, but is not limited to, various sensors and operating equipment. The server can be implemented using a standalone server or a server cluster composed of multiple servers. The invention will be described in detail below through specific embodiments.

[0233] Please see Figure 3 As shown, Figure 3 A flowchart illustrating the precise injection control method for in-situ uranium leaching provided in this embodiment of the invention includes the following steps: S100: Collect operational data from each branch injection well; S200: Based on a pre-built digital twin model, it simulates and predicts the concentration field, temperature field, pressure field, and uranium leaching rate of acid in underground mineral formations according to the operational data of each branch injection well; S300: Based on the pre-set single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model, the concentration control command and flow control command of each branch injection well are generated according to the predicted concentration field, temperature field, pressure field and uranium leaching rate of acid in the underground mineral layer. S400: Based on the operating data of each branch injection well, the concentration control command and the flow control command of each branch injection well, the injection concentration and injection flow rate of each branch injection well are regulated separately.

[0234] Specifically, step S100 also includes collecting the original operating data of each branch injection well, and preprocessing the original operating data to obtain the operating data. The preprocessing includes outlier removal, low-pass filtering, and time synchronization.

[0235] Specifically, the digital twin generation module includes a fluid flow sub-model, a solute transport sub-model, an energy conservation sub-model, and a leaching reaction sub-model. Step S200 further includes simulating and predicting the pressure field of the acid solution in the underground ore layer based on the fluid flow sub-model, simulating and predicting the concentration field of the acid solution in the underground ore layer based on the solute transport sub-model, simulating and predicting the temperature field of the acid solution in the underground ore layer based on the energy conservation sub-model, and simulating and predicting the uranium leaching rate in the underground ore layer based on the leaching reaction sub-model.

[0236] Specifically, the method further includes updating the parameters in the digital twin generation module 102 based on the operating data of each branch injection well and the predicted concentration field, temperature field, pressure field and uranium leaching rate of the acid in the underground mineral layer.

[0237] Specifically, step S300 also includes obtaining initial concentration control commands for each branch injection well based on the predicted concentration field, temperature field, pressure field, and uranium leaching rate of the acid in the underground ore layer. and initial flow control command Based on pre-set global constraints, the concentration control command for each branch injection well is obtained according to the initial concentration control command and the initial flow control command for each branch injection well. and the flow control command .

[0238] Specifically, step S400 also includes adjusting the injection concentration of each branch injection well based on the pre-set feedforward compensation strategy and feedback correction strategy, according to the operating data of each branch injection well and the concentration control command of each branch injection well; and adjusting the injection flow rate of each branch injection well based on the pre-set adaptive fuzzy PID control strategy, according to the operating data of each branch injection well and the flow control command of each branch injection well.

[0239] Specifically, the method further includes comparing the operating data of each branch injection well with its corresponding normal threshold based on a pre-set normal threshold, and determining a first anomaly index based on the comparison result; comparing the operating data of each branch injection well with the simulated prediction value of the corresponding digital twin generation module, and determining a second anomaly index based on the comparison result; generating a Kalman filter innovation sequence based on the operating data of each branch injection well and the simulated prediction value of the corresponding digital twin generation module, analyzing the root mean square error of the Kalman filter innovation sequence within multiple consecutive preset decision periods, and determining a third anomaly index based on the analysis result; and issuing an anomaly warning based on the first anomaly index, the second anomaly index, and the third anomaly index based on a pre-set weighted voting mechanism.

[0240] Specifically, the method further includes generating an abnormal risk value based on a pre-set weighted voting mechanism, according to the first abnormal index, the second abnormal index, and the third abnormal index; and comparing the abnormal risk value with multiple pre-set abnormal risk level thresholds to determine the abnormal warning level and warning instruction.

[0241] As can be seen, in the above scheme, this application collects the operating data of each branch injection well in real time, performs simulation and prediction through the digital twin generation module to accurately characterize the response characteristics of the ore layer, and generates concentration control commands and flow control commands through single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model based on the prediction results. While ensuring the maximization of the leaching efficiency of a single well, it realizes the coordinated control between multiple wells. Finally, the concentration control commands and flow control commands are executed through the edge injection execution module, realizing the decoupling of injection concentration and injection flow rate, and realizing the "one well, one policy" precise injection strategy for heterogeneous ore layers.

[0242] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0243] Specific limitations regarding the precise injection control method for in-situ leaching uranium mining can be found in the limitations of the precise injection control system for in-situ leaching uranium mining described above, and will not be repeated here. The aforementioned precise injection control method for in-situ leaching uranium mining can be implemented entirely or partially through software, hardware, or a combination thereof. The modules in the above system can be embedded in hardware or independent of the processor in the computer device, or stored in software in the memory of the computer device, so that the processor can call and execute the corresponding operations of each module.

[0244] In one embodiment, a computer device is provided, which may be a device terminal, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a precise injection control method for uranium leaching in in-situ mining on the device side.

[0245] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Collect operational data from each branch injection well; Based on a pre-built digital twin model, the concentration field, temperature field, pressure field, and uranium leaching rate of acid in the underground mineral layer are simulated and predicted according to the operating data of each branch injection well. Based on the pre-set single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model, the concentration control command and flow control command of each branch injection well are generated according to the predicted concentration field, temperature field, pressure field and uranium leaching rate of acid in the underground mineral layer. Based on the operating data of each branch injection well, the concentration control instructions and flow control instructions of each branch injection well, the injection concentration and injection flow rate of each branch injection well are adjusted separately.

[0246] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Collect operational data from each branch injection well; Based on a pre-built digital twin model, the concentration field, temperature field, pressure field, and uranium leaching rate of acid in the underground mineral layer are simulated and predicted according to the operating data of each branch injection well. Based on the pre-set single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model, the concentration control command and flow control command of each branch injection well are generated according to the predicted concentration field, temperature field, pressure field and uranium leaching rate of acid in the underground mineral layer. Based on the operating data of each branch injection well, the concentration control instructions and flow control instructions of each branch injection well, the injection concentration and injection flow rate of each branch injection well are adjusted separately.

[0247] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and device side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0248] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0249] If any AI models or software tools not belonging to this company appear in the embodiments of this application, they are merely illustrative examples and do not represent actual use. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In actual applications, the above functions can be assigned to different functional units or modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A precise liquid injection control system for in-situ uranium leaching, characterized in that, include: The multi-source sensing module is used to collect operational data from each branch injection well; The digital twin generation module is used to simulate and predict the concentration field, temperature field, pressure field, and uranium leaching rate of acid in underground mineral formations based on the pre-built digital twin model and the operating data of each branch injection well. The intelligent injection decision module is used to generate concentration control commands and flow control commands for each branch injection well based on the pre-set single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model, according to the predicted concentration field, temperature field, pressure field and uranium leaching rate of acid in the underground mineral layer. The edge injection execution module is used to regulate the injection concentration and injection flow rate of each branch injection well based on the operating data, concentration control instructions and flow control instructions of each branch injection well.

2. The precise injection control system for in-situ uranium leaching mining according to claim 1, characterized in that, The digital twin generation module includes a fluid flow sub-model, a solute transport sub-model, an energy conservation sub-model, and a leaching reaction sub-model. The fluid flow sub-model is used to simulate and predict the pressure field of the acid in the underground ore layer, the solute transport sub-model is used to simulate and predict the concentration field of the acid in the underground ore layer, the energy conservation sub-model is used to simulate and predict the temperature field of the acid in the underground ore layer, and the leaching reaction sub-model is used to simulate and predict the uranium leaching rate in the underground ore layer.

3. The precise liquid injection control system for in-situ leaching uranium mining according to any one of claims 1 or 2, characterized in that, The system also includes a parameter correction module; The parameter correction module is used to update the parameters in the digital twin generation module based on the operating data of each branch injection well and the predicted concentration field, temperature field, pressure field and uranium leaching rate of the acid in the underground ore layer.

4. The precise injection control system for in-situ uranium leaching mining according to claim 1, characterized in that, The single-well multi-objective optimization sub-model is used to obtain the initial concentration control command and initial flow control command for each branch injection well based on the predicted concentration field, temperature field, pressure field and uranium leaching rate of the acid in the underground ore layer. The multi-well collaborative optimization sub-model is used to obtain the concentration control command and the flow control command of each branch injection well based on the pre-set global constraints and the initial concentration control command and the initial flow control command of each branch injection well.

5. The precise injection control system for in-situ uranium leaching mining according to claim 1, characterized in that, The edge injection execution module includes an injection concentration control submodule and an injection flow rate control submodule; The injection concentration control submodule is used to adjust the injection concentration of each branch injection well based on the pre-set feedforward compensation strategy and feedback correction strategy, according to the operating data of each branch injection well and the concentration control command of each branch injection well. The injection flow control submodule is used to regulate the injection flow of each branch injection well based on a pre-set adaptive fuzzy PID control strategy, according to the operating data and flow control commands of each branch injection well.

6. The precise injection control system for in-situ uranium leaching mining according to claim 1, characterized in that, The system also includes an anomaly warning module; The anomaly warning module is used to compare the operating data of each branch injection well with its corresponding normal threshold based on a pre-set normal threshold, and determine the first anomaly index based on the comparison result. The anomaly early warning module is used to compare the operating data of each branch injection well with the simulated and predicted values ​​of its corresponding digital twin generation module, and determine the second anomaly index based on the comparison results. The anomaly early warning module is used to generate a Kalman filter innovation sequence based on the operating data of each branch injection well and the simulated prediction value of the corresponding digital twin generation module, analyze the root mean square error of the Kalman filter innovation sequence within multiple consecutive preset decision periods, and determine the third anomaly index based on the analysis results. The anomaly warning module is used to issue anomaly warnings based on a pre-set weighted voting mechanism, according to the first anomaly index, the second anomaly index, and the third anomaly index.

7. The precise injection control system for in-situ leaching uranium mining according to claim 6, characterized in that, The anomaly warning module is also used to generate an anomaly risk value based on the first anomaly index, the second anomaly index, and the third anomaly index, according to a pre-set weighted voting mechanism. The anomaly warning module is also used to compare the anomaly risk value with multiple pre-set anomaly risk level thresholds to determine the anomaly warning level and warning instruction.

8. A method for precise injection control in in-situ leaching uranium mining, characterized in that, The method employs the precise liquid injection control system for in-situ leaching uranium mining as described in any one of claims 1 to 7, the method comprising: Collect operational data from each branch injection well; Based on a pre-built digital twin model, the concentration field, temperature field, pressure field, and uranium leaching rate of acid in the underground mineral layer are simulated and predicted according to the operating data of each branch injection well. Based on the pre-set single-well multi-objective optimization sub-model and multi-well collaborative optimization sub-model, the concentration control command and flow control command of each branch injection well are generated according to the predicted concentration field, temperature field, pressure field and uranium leaching rate of acid in the underground mineral layer. Based on the operating data of each branch injection well, the concentration control instructions and flow control instructions of each branch injection well, the injection concentration and injection flow rate of each branch injection well are adjusted separately.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the precise liquid injection control method for in-situ leaching uranium mining as described in claim 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the precise injection control method for in-situ leaching uranium mining as described in claim 8.

Images