Centralized control system and method for mine filling station, electronic equipment and storage medium

Abstract

The invention provides a mine filling station centralized control system and method, electronic equipment and a storage medium, and relates to the technical field of mine filling control, the system comprises a node control layer and a central control layer, the node control layer comprises a plurality of sub-node control modules, each sub-node control module corresponds to a process unit in a mine filling station, and the central control layer comprises a plurality of sub-node control modules; the sub-node control modules are used for acquiring filling process data of corresponding process units, processing the filling process data to obtain target data, uploading the target data to the central control layer, and receiving a global control strategy issued by the central control layer to execute local control of the process units; and the central control layer is used for receiving the target data of each sub-node control module, performing training and global optimization calculation through an artificial intelligence model according to the target data and filling constraint conditions, generating a global control strategy, and issuing the global control strategy to each sub-node control module. The automatic and intelligent level of the filling process is remarkably improved.

Description

Technical Field

[0001] This application relates to the field of mine backfilling control technology, specifically to a centralized control system, method, electronic equipment, and storage medium for a mine backfilling station. Background Technology

[0002] As a critical infrastructure in the mining process, mine backfilling stations currently operate independently, with inconsistent data collection standards and a lack of effective collaborative scheduling mechanisms, resulting in low overall system efficiency.

[0003] Meanwhile, the production scheduling process at the filling station relies excessively on manual experience, focusing only on the stability of single process parameters, resulting in delayed and biased scheduling decisions. Furthermore, the filling station suffers from significant deficiencies in fault early warning and energy management, leading to frequent unplanned equipment shutdowns, disrupting normal production rhythms, and significantly increasing operation and maintenance costs. Therefore, it is urgent to address the overall operational efficiency and stability issues of mine filling stations. Summary of the Invention

[0004] In view of the above-mentioned shortcomings of the prior art, this application provides a centralized control system, method, electronic equipment and storage medium for mine backfilling stations, which effectively solves the problem of insufficient overall operating efficiency and stability of mine backfilling stations.

[0005] In a first aspect, this application provides a centralized control system for a mine backfilling station, the system comprising a node control layer and a central control layer, wherein: The node control layer includes multiple sub-node control modules. Each sub-node control module corresponds to a process unit in the mine filling station. The sub-node control module is used to collect the filling process data of the corresponding process unit, process it to obtain target data, upload the target data to the central control layer, and receive the global control strategy issued by the central control layer to execute the local control of the process unit. The central control layer is used to receive the target data from each of the sub-node control modules, and based on the target data and filling constraints, to perform training and global optimization calculations through an artificial intelligence model, generate the global control strategy, and send it to each of the sub-node control modules.

[0006] In an optional implementation, the sub-node control module includes at least a data acquisition unit, a computing hub unit, and a control execution unit, wherein: The data acquisition unit acquires the filling process data through a multimodal sensor network; The computing hub unit performs data preprocessing and feature extraction on the filling process data, obtains the target data and sends it to the central control layer, and generates primary control commands based on the filling process data through a local optimization algorithm. The control execution unit receives the global control strategy and executes the local control of each process unit, and receives the primary control instructions and drives the process equipment in the corresponding process unit to perform adjustment operations.

[0007] In an optional implementation, the central control layer includes at least a data fusion module, a model training module, a model simulation module, an optimization feedback module, and a strategy generation module, wherein: The data fusion module performs protocol standardization processing on the target data to obtain a structured dataset; The model training module constructs and trains the artificial intelligence model based on the process parameters of the mine filling station using a digital twin engine; The model simulation module simulates the operation state of the filling process based on the structured dataset and the filling constraints, and obtains multiple initial control strategies and optimal target action parameters. The optimization feedback module uses a multi-objective genetic algorithm to iteratively optimize multiple initial control strategies to obtain the optimal control strategy; The strategy generation module generates the global control strategy based on the optimal control strategy and the optimal target action parameters.

[0008] In an optional implementation, the central control layer further includes a resource allocation module and an early warning display module, wherein: The resource allocation module dynamically adjusts the computing resources of the sub-node control module based on the structured dataset and the preset priority rules for mine filling; The early warning display module is used to visualize data and strategies and to provide early warnings for filling parameters.

[0009] Secondly, this application provides a centralized control method for a mine backfilling station, the method being applied to the centralized control system for the mine backfilling station described in the first aspect of this application, the method comprising: The filling process data of each process unit in the ore filling station is collected by the control modules of each sub-node of the node control layer. The filling process data is preprocessed and feature extracted to obtain the target data. A primary control command is generated based on the filling process data using a local optimization algorithm to drive the process equipment in the corresponding process unit to perform adjustment operations. Based on the target data, task data, and human intervention data, and combined with filling constraints, a global control strategy is generated by training and global optimization calculations using an artificial intelligence model. The global control strategy is distributed to each of the sub-node control modules to execute the local control of each of the process units.

[0010] In an optional implementation, based on the target data, task data, and human intervention data, and combined with filling constraints, an artificial intelligence model is trained and globally optimized to generate a global control strategy, including: The target data, the task data, and the human intervention data are subjected to protocol standardization processing to obtain a structured dataset; The artificial intelligence model is constructed and trained based on the process parameters of the mine filling station using a digital twin engine; Based on the structured dataset and the filling constraints, the artificial intelligence model is used to simulate the operation of the filling process and obtain multiple initial control strategies and optimal target action parameters. The optimal control strategy is obtained by iteratively optimizing multiple initial control strategies using a multi-objective genetic algorithm. The global control strategy is generated based on the optimal control strategy and the optimal target action parameters.

[0011] In an optional implementation, the local optimization algorithm is a fuzzy PID control algorithm, and the step of generating primary control commands based on the filling process data using the local optimization algorithm includes: The initial parameters of the PID controller are set based on historical process data of mine backfilling, and the input and output variables are defined. The filling process data is mapped to the corresponding input variables and then transformed into fuzzy linguistic variables through a triangular membership function. A fuzzy rule base is established based on mine backfilling process experience. The fuzzy linguistic variables are matched with the fuzzy rule base to obtain a fuzzy output set of PID parameter correction quantities. The fuzzy output set is converted into parameter correction values, and the initial PID parameters are updated according to the parameter correction values ​​to obtain the target PID parameters. Substitute the target PID parameters into the PID control formula to obtain the primary control command.

[0012] In an optional implementation, the local optimization algorithm is a model predictive control algorithm, and the step of generating primary control commands based on the filling process data using the local optimization algorithm includes: A prediction model for backfilling process parameters was constructed based on historical process data of mine backfilling. Based on the filling process data and the filling process parameter prediction model, parameter prediction is performed to obtain the optimal control sequence for multiple future control cycles; Extract the first control quantity of the optimal control sequence as the initial control command for the current cycle and issue it for execution, and collect the actual filling process parameters after execution; The actual filling process parameters and the filling process parameter prediction model are iteratively optimized until the actual filling process parameters meet the preset requirements, and the primary control command is generated based on the actual filling process parameters.

[0013] Thirdly, this application provides an electronic 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 centralized control method for a mine backfilling station as described in the second aspect of this application.

[0014] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the centralized control method for a mine backfilling station as described in the second aspect of this application.

[0015] The centralized control system, method, electronic equipment, and storage medium for mine backfilling stations provided in this application achieve distributed perception, local optimization, and global intelligent scheduling throughout the entire backfilling process by constructing a collaborative architecture of node control layer and central control layer. The node control layer integrates multimodal sensing and edge computing capabilities, combined with fuzzy PID control algorithms or model predictive control algorithms, to achieve real-time adaptive adjustment of backfilling process parameters. The central control layer generates a global control strategy through data fusion, digital twin simulation, reinforcement learning, and multi-objective iterative optimization, significantly improving the automation and intelligence level of the backfilling process, effectively reducing reliance on human intervention, and enhancing adaptability to fluctuations in incoming materials and changes in operating conditions. Simultaneously, it effectively improves the stability of backfill quality, reduces equipment downtime, accelerates production scheduling response speed, significantly improves overall operating efficiency and economic benefits, achieves predictive maintenance and closed-loop optimization, and promotes the development of mine backfilling towards high efficiency and intelligence. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a first schematic diagram of the structure of the centralized control system for a mine backfilling station provided in an embodiment of this application; Figure 2 This is a second schematic diagram of the structure of the centralized control system for a mine backfilling station provided in the embodiments of this application; Figure 3 This is a third schematic diagram of the structure of the centralized control system for a mine backfilling station provided in the embodiments of this application; Figure 4 This is a schematic diagram of the centralized control method for mine backfilling stations provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0018] Key component symbols: 10-Centralized control system for mine filling station; 100-Node control layer; 110-Sub-node control module; 111-Data acquisition unit; 112-Computation central unit; 113-Control execution unit; 200-Central control layer; 210-Data fusion module; 220-Model training module; 230-Model simulation module; 240-Optimization feedback module; 250-Strategy generation module; 260-Resource allocation module; 270-Early warning display module; 300-Electronic equipment; 310-Processor; 320-Communication interface; 330-Memory; 340-Communication bus. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be further described clearly and completely below with reference to the accompanying drawings of the embodiments. It should be noted that the described embodiments are merely some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0020] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of this application.

[0022] Currently, the subsystems of mine backfilling stations operate independently, with inconsistent data acquisition standards and a lack of effective collaborative scheduling mechanisms. This results in limited overall operational efficiency and a lack of early warning for system failures and refined energy consumption management. Furthermore, there is room for improvement in intelligent scheduling and optimization, particularly the lack of AI-based collaborative optimization, which hinders efficient collaboration between subsystems and results in insufficient intelligence in energy consumption management.

[0023] Example 1 This application provides a centralized control system for a mine backfilling station, which effectively solves the problem of insufficient overall operating efficiency and stability of the mine backfilling station. Figure 1 This is a first schematic diagram of the structure of the centralized control system for a mine backfilling station provided in an embodiment of this application, as shown below. Figure 1 As shown, the centralized control system 10 of the mine filling station includes a node control layer 100 and a central control layer 200. The node control layer 100 and the central control layer 200 realize bidirectional data interaction and command transmission through a hierarchical heterogeneous hybrid data communication network, which is adapted to the complex working conditions of multi-process unit collaboration and multi-parameter coupling in mine filling.

[0024] The node control layer 100 includes multiple sub-node control modules 110. Each sub-node control module 110 corresponds to a process unit in the mine filling station. The sub-node control module 110 is used to collect the filling process data of the corresponding process unit, process it to obtain target data, upload the target data to the central control layer 200, and receive the global control strategy issued by the central control layer 200 to execute the local control of the process unit.

[0025] It is understood that each sub-node control module 110 corresponds to a process unit of the ore filling station. The process unit includes, but is not limited to, at least two of the following systems: tailings dam system, waste rock crushing system, feeding and batching system, mixing system, pumping system, pipeline system, environmental monitoring system, power supply system, water supply system, and material balance system.

[0026] As a further implementation of the embodiments of this application, Figure 2 This is a second schematic diagram of the centralized control system structure for a mine backfilling station provided in an embodiment of this application, as shown below. Figure 2 As shown, the sub-node control module 110 includes at least a data acquisition unit 111, a computing center unit 112, and a control execution unit 113.

[0027] The data acquisition unit 111 integrates a multimodal sensor network for collecting equipment status, personnel information, environmental parameters, and material flow data, and collects filling process data in a targeted manner. This multimodal sensor network includes different types of sensors for collecting the same filling process parameters, and at least one of vibration sensors, acoustic sensors, and current sensors for collecting filling equipment status data.

[0028] The computing central unit 112 performs data preprocessing and feature extraction on the filling process data, obtains target data and sends it to the central control layer 200, and generates primary control commands based on the filling process data through a local optimization algorithm.

[0029] Optionally, the computing hub unit 112 is based on an industrial-grade node computing hardware platform and can adopt a multi-core ARM or x86 architecture processor, equipped with no less than 4GB of RAM and 32GB of storage space to meet the real-time and anti-interference requirements of edge computing at the mine filling site.

[0030] For example, data preprocessing and feature extraction can use wavelet transform algorithms to filter and denoise the raw sensor data, removing abnormal data. Then, time-domain and frequency-domain analysis methods are used to reduce the dimensionality of the raw sensor data, extracting statistical feature values ​​including but not limited to root mean square values, peak values, and kurtosis, as well as equipment health status indicators and summaries of local optimization results, and other standardized format effective data. The obtained target data is then uploaded to the central control layer 200 via a data communication network.

[0031] Understandably, the standardized format of valid data is encapsulated in JSON or Protocol Buffers format, including node identifiers (filling process unit type), timestamps, equipment health index, filling process parameter characteristic values ​​(slurry concentration and water-cement ratio, etc.) and quality indicators (filling strength prediction and proportion compliance rate), thereby adapting to the needs of mine filling data traceability and collaboration.

[0032] As an optional implementation of this application, the computing hub unit 112 can also run a lightweight machine learning model, such as a random forest model or a long short-term memory module, based on the filling equipment status data of the corresponding process unit, such as equipment vibration spectrum analysis and current harmonic analysis data, to realize fault warning and remaining life prediction of the filling process equipment.

[0033] In this embodiment, the computing central unit 112 outputs primary control instructions for the corresponding process unit based on the filling process data through a local optimization algorithm. The primary control instructions can be primary optimization adjustment instructions for the filling process parameters. The local optimization algorithm includes a fuzzy PID control algorithm or a model predictive control algorithm, which is specifically adapted to the rapid adaptive adjustment of key parameters of the filling process, and solves the technical pain points of strong parameter coupling and time-varying disturbances.

[0034] The control execution unit 113 receives the global control strategy and executes the local control of each process unit, as well as receives the primary control instructions and drives the process equipment in the corresponding process unit to perform adjustment operations.

[0035] Understandably, the control execution unit 113 receives the global control strategy and decomposes it into filling-specific control instructions for the corresponding process unit nodes. Optionally, the control execution unit 113 uses an instruction parsing engine to decompose the received global control strategy into a series of specific filling-specific control instruction sequences with timing and safety constraints for the corresponding process unit. Simultaneously, the control execution unit 113 is equipped with a separate detection and drive device to execute the filling-specific control instructions and / or primary control instructions, driving the process equipment to perform precise actions and adjustment operations, thereby achieving precise local control of the process unit nodes.

[0036] Based on this, the node control layer 100 in this embodiment significantly improves the system's real-time response capability and data processing efficiency through the deep integration of distributed architecture and edge computing technology. Each process unit node possesses autonomous perception, local decision-making, and collaborative interaction capabilities, effectively reducing dependence on the central server and minimizing network transmission latency and bandwidth pressure. The node control layer 100 supports dynamic topology adjustment and fault self-healing, enhancing system robustness and scalability, while providing stable, efficient, and intelligent data support to the upper-level central control layer 200, thereby improving the overall intelligence level and operational reliability of the system.

[0037] The central control layer 200 is used to receive target data from each sub-node control module 110, and based on the target data and filling constraints, it performs training and global optimization calculations through an artificial intelligence model to generate a global control strategy and distribute it to each sub-node control module 110.

[0038] As an optional implementation of this application, the central control layer 200 is deployed in the cloud and includes an independent server. The server's data processing method is a heterogeneous cluster big data computing architecture. For example, the server includes multiple computing nodes, all of which are configured with the same memory and CPU. Each computing node is equipped with sufficient local storage space to meet the needs of mine backfilling big data processing and AI model training.

[0039] Furthermore, the central control layer 200 communicates with all sub-node control modules 110 via a data communication network to receive target data uploaded by all sub-node control modules 110. This data communication network adopts a hierarchical heterogeneous hybrid network architecture adapted to the complex environment of mines, including intra-node sensor networks, inter-node backbone networks, and cloud access networks.

[0040] Understandably, the intra-node sensor network is deployed within each intelligent node, using a hybrid wired and wireless networking approach to connect various sensors to the computing hub unit 112. For example, the wired networking can use RS485 connections to suit fixed installation scenarios, while the wireless networking can use LoRa connections to suit mobile devices or areas with difficult cabling. The inter-node backbone network uses an industrial Ethernet ring network to connect all sub-node control modules 110, forming a self-healing redundant network to avoid data interruptions caused by single-point failures in the mine. The cloud access network connects the inter-node backbone network of the sub-node control modules 110 to the central control layer 200 via an industrial gateway device, supporting both wired and wireless dual-link backup.

[0041] In addition, the network architecture supports parallel transmission of commonly used industrial communication protocols in mining, such as Modbus, Profinet, and OPC UA, and has deterministic transmission capabilities based on time-sensitive networking, with control command latency ≤300 ms, meeting real-time scheduling requirements.

[0042] Furthermore, the central control layer 200 can also connect to the mine production scheduling system, field industrial control computers and operator terminals to obtain standardized work task data and human intervention data. The work task data includes, but is not limited to, data such as filling demand information and goaf filling plan, and the human intervention data includes, but is not limited to, data such as process adjustment instructions and emergency shutdown records.

[0043] In this embodiment, the central control layer 200 includes a data fusion module 210, a model training module 220, a model simulation module 230, an optimization feedback module 240, and a policy generation module 250, wherein: The data fusion module 210 is used to collect and uniformly encapsulate target data, task data and human intervention data, and form a structured dataset through protocol standardization processing.

[0044] Optionally, the data fusion module 210 connects to target data, task data, and human intervention data through standardized interfaces. Target data comes from the control modules 110 of each sub-node, and the protocol can be Modbus RTU, Profinet, OPC UA, etc. Task data comes from the mine production scheduling system, and the format can be JSON or Excel. Human intervention data comes from the on-site industrial control computer and operator terminal, and the protocol can be TCP / IP or manual entry forms, establishing parallel access channels to ensure the real-time performance of various types of data.

[0045] The model training module 220 uses a digital twin engine to build and train an artificial intelligence model based on the process parameters of the mine filling station.

[0046] In this embodiment, the model training module 220 integrates the core functions of a digital twin engine, a model training environment, and a remote expert interface. The digital twin engine establishes a high-fidelity artificial intelligence model based on the physical entity unit of the filling station, realizes filling process simulation, parameter inversion and reinforcement learning, and supports two-way data interaction and remote guidance functions.

[0047] Optionally, the artificial intelligence model is a reinforcement learning model, adopting an Actor-Critic architecture. The input layer is a 6-8 dimensional global state vector, used to integrate core parameters such as slurry concentration, water-cement ratio, pumping pressure, equipment health index, tailings flow rate, and energy consumption. The first hidden layer has 128 nodes, and the second hidden layer has 64 nodes, both using the ReLU activation function. The output layer is a 3-5 dimensional target action parameter and state value estimate. The target action parameters include, but are not limited to, target slurry concentration, target water-cement ratio, target pump speed, target cement content, and target tailings feed rate.

[0048] This artificial intelligence model can use the weighted sum of multiple objectives, such as the mechanical properties of the filling body, filling production efficiency, energy consumption cost, cement consumption per unit, safe production (equipment failure risk), environmental protection (dust emission), and labor intensity, as the optimization objective function, to meet the multi-objective collaborative needs of mine filling.

[0049] For example, the optimization objective function needs to focus on the core needs of mine backfilling, and the optimization objective function can be set as follows: ,in, f i Indicates the first i Several optimization targets, such as a 28-day compressive strength compliance rate of ≥95% for the filling material and an energy consumption per unit volume of filling grout ≤5kWh / m³, are also included. ω i Indicates the first i The weights corresponding to each optimization objective satisfy the following conditions: The weights can be dynamically adjusted according to the mining conditions. n This represents the total number of optimization objectives.

[0050] The model simulation module 230 simulates the operation state of the filling process based on the structured dataset and filling constraints, and obtains multiple initial control strategies and optimal target action parameters.

[0051] As an optional implementation of this application, the specific settings of the filling constraints are adapted to the mine filling process, including but not limited to process constraints, resource constraints, time constraints, and safety constraints.

[0052] For example, in terms of process constraints, the slurry flow rate needs to be maintained within a reasonable range to prevent solid particles from depositing and clogging the pipeline due to excessively low flow rate, or to cause erosion and wear of the pipe wall due to excessively high flow rate; the pumping pressure should be controlled within the range that the equipment and pipeline can withstand to avoid damage to the conveying system caused by overpressure operation; the water-cement ratio, as a core parameter affecting the performance of the filling body, must take into account the balance requirements of mechanical strength and construction fluidity; and the slurry concentration needs to meet the technical specifications of the paste filling process to ensure that it has good non-bleeding and self-supporting capabilities.

[0053] Understandably, in terms of resource constraints, the daily cement supply is limited by the mine's material reserve capacity and must match the daily filling task scale to prevent material shortages and work stoppages; the daily power supply load must meet the upper limit of the mine's power grid to ensure the stable operation of the power system; and the daily tailings processing volume is constrained by the effective capacity of the tailings dam and environmental emission permits, and must be scheduled within the scope of sustainable disposal capacity.

[0054] Another example is that, in terms of time constraints, the completion time of a single batch of filling tasks must be strictly controlled to avoid the slurry from staying for too long during transportation, which could lead to initial setting, affect the filling quality, or even cause pipe blockage accidents. Equipment maintenance windows must be included in the scheduling plan to ensure that key equipment is repaired and maintained within the specified time, extend its service life, and reduce the risk of sudden failures.

[0055] In addition, it is understandable that, in terms of safety constraints, the vibration amplitude of the equipment must comply with the relevant standards for the vibration limits of mining machinery, so as to provide timely warning of potential mechanical imbalance or bearing damage risks; the dust emission concentration should meet the prescribed limit requirements to reduce environmental pollution in the workplace and protect the occupational health of personnel; and the pipeline pressure fluctuations must be kept stable to prevent safety accidents such as loose joints or pipeline rupture caused by violent fluctuations.

[0056] Based on this, the design of filling constraints comprehensively covers the key control dimensions in the operation of mine filling stations, ensuring that the global optimization strategy achieves multi-objective collaboration under the premise of safety, stability and compliance. Various constraints together constitute the boundary conditions of the artificial intelligence model, which are strictly verified when multi-objective optimization is performed in the central control layer 200, ensuring that the generated global control strategy is both efficient and reliable, and applicable to complex and ever-changing actual mine working conditions.

[0057] The optimization feedback module 240 uses a multi-objective genetic algorithm to iteratively optimize multiple initial control strategies to obtain the optimal control strategy.

[0058] As an optional implementation of this application, the optimization feedback module 240 can optimize multiple initial control strategies based on the NSGA-II evolutionary algorithm, select control strategies with better overall performance through non-dominated sorting, ensure the diversity of schemes through crowding calculation, generate a new generation of scheme set through genetic iteration, and after 100-200 generations, extract the first non-dominated layer of the final population as the Pareto optimal solution set, i.e. the optimal control strategy.

[0059] The strategy generation module 250 generates a global control strategy based on the optimal control strategy and the optimal target action parameters.

[0060] In this embodiment, the optimal target action parameters output by the artificial intelligence model and the Pareto optimal solution set are combined, and the priority of the mine's current filling needs is comprehensively considered. For example, the efficiency of emergency filling tasks is prioritized, and the cost and intensity of routine tasks are prioritized to balance. A global control strategy containing the collaborative parameters of each process unit is generated to ensure that the strategy is adapted to the collaborative needs of the entire filling process.

[0061] As a further implementation of the embodiments of this application, Figure 3 This is a third schematic diagram of the centralized control system structure for a mine backfilling station provided in this application embodiment, as shown below. Figure 3 As shown, the central control layer 200 also includes a resource allocation module 260 and an early warning display module 270, wherein: The resource allocation module 260 dynamically adjusts the computing resources of the sub-node control module 110 based on the structured dataset and the preset priority rules for mine backfilling. Specifically, it includes the following steps: First, feature extraction is performed. The operating status features and filling task features of each process unit are extracted from the structured dataset. The operating status features include, but are not limited to, the health index of the filling equipment, the deviation coefficient of the filling process parameters, and the data processing load. The task features include, but are not limited to, the urgency of the filling task, the task constraints, and the task-related subsystems. The task constraints can be, for example, the filling strength target, the upper limit of comprehensive energy consumption, and the deadline for the operation.

[0062] Secondly, a priority stratification is implemented. Based on the safety and production needs of mine backfilling, a four-level priority rule base is set up to stratify the parameter optimization tasks of each process unit. Level P0 is the emergency guarantee level, applicable to situations with significant operational risks or emergency conditions. The grading criteria include: equipment health index below 60, indicating that the equipment is in a high-risk failure state; process parameter deviation coefficient exceeding 1.5, reflecting that key control indicators are seriously deviating from the set range; and the operation task is a critical task related to safety and production interruption, such as performing emergency backfilling underground or emergency handling of equipment failure, ensuring that the system has the highest response priority in case of emergencies. Level P1 is the core production level, targeting the core process units in the current main backfilling process, such as feeding, mixing, and pumping systems. Especially when the task is nearing its deadline and the completion progress is lagging behind, such as when the task deadline is <4 hours and the uncompleted amount is >60%, or when the process parameters show a trend of approaching the limit, such as when the process parameter deviation coefficient is ∈ [1, 1.5), a higher scheduling priority is given to ensure the efficient and stable progress of the main production tasks. Level P2 is the routine optimization level, covering non-core auxiliary systems with good equipment operation and stable process parameters, such as environmental monitoring, water supply pressure stabilization, and tailings dam level control. It performs daily optimization and control to maintain the overall system operation quality. The classification criteria include: equipment health index ∈ [80, 100], indicating good condition; and process parameter deviation coefficient < 1, indicating stable parameters. Level P3 is the low-load scheduling level, targeting subsystems in standby or near-completion states, such as equipment standby and task completion rates > 90%. Only low-frequency periodic reviews and maintenance calculations are required, such as the daily proportion review of a material balance system. At this time, resource allocation is minimal, and an idle-time scheduling strategy is adopted to improve computing power utilization efficiency.

[0063] Next, the computing power requirement is quantified. Using CPU core utilization, memory allocation, and computation time as core indicators, the computing power requirement of a single child node control module 110 is quantified. The calculation formula is as follows:

[0064] In the above formula, C This represents the computing power requirement. A This represents the algorithm complexity coefficient. For fuzzy PID control, it's set to 1.0; for model predictive control, due to the involvement of filling multivariate coupling, it can be set to 1.8. D This represents the data processing scale coefficient, calculated as data volume / 100MB. P This indicates the priority weights: P0 = 5, P1 = 3, P2 = 1, and P3 = 0.5. α , β and γ They represent the weighting coefficients, where α Take 0.4, β Take 0.2, γWe set it to 0.4 to ensure that priority is given to the core influencing factor.

[0065] Then, dynamic scheduling is implemented. The total computing power of the heterogeneous cluster servers in the central control layer 200 is allocated according to a priority-weighted distribution principle. Specifically: P0-level subsystems are allocated ≥40% of computing power to ensure real-time response under sudden operating conditions, such as parameter adjustments after a pipeline blockage warning; P1-level systems are allocated 30%-35% of computing power to support collaborative optimization of the core filling process; P2-level systems are allocated 20%-25% of computing power; and P3-level systems are allocated ≤5% of computing power for idle-time scheduling. Simultaneously, the computing power usage status is monitored in real time. For example, if the CPU utilization rate of P0-level systems is consistently <50%, excess computing power is reclaimed; if memory overflow occurs in P1-level systems, computing power is supplemented. Furthermore, the sub-node control modules 110 perform complementary functions, such as utilizing the idle computing power of the environmental monitoring system when the mixing system lacks sufficient computing power due to complex slurry proportions.

[0066] Finally, feedback optimization is performed, extracting data such as the compliance rate of filling parameters and optimization of calculation latency from the structured dataset, and regularly revising the priority rule base weight coefficient and the computing power demand quantification model to adapt to different working conditions of mine filling.

[0067] This application's embodiments dynamically adjust computing resources based on structured datasets and preset priority rules, enabling efficient allocation and real-time response of computing power. Through a four-level priority hierarchy and quantization model, it ensures that urgent tasks receive priority computing power, improving system stability and real-time scheduling, avoiding resource waste, and enhancing multi-task collaborative processing capabilities. It is particularly suitable for intelligent control needs under complex mining conditions.

[0068] The early warning display module 270 is used for data and strategy visualization display and for providing early warnings for filling parameters.

[0069] Optionally, the early warning display module 270 may include one or more devices such as an LED display screen, a touch display screen, and an audible and visual alarm to visualize filling process monitoring data and global control strategies, and trigger early warnings for filling parameters, such as concentration exceeding the standard and abnormal pumping pressure, or pipeline blockage.

[0070] Furthermore, the central control layer 200 can also export production and operation data reports of the mine filling station as a basis for mine production archiving and process review.

[0071] The central control layer 200 of this application embodiment relies on a high-fidelity digital twin model and reinforcement learning algorithm, and combines multi-source data for global collaborative optimization to generate an optimal control strategy that takes into account efficiency, energy consumption, safety and cost. Through a dynamic resource allocation mechanism, it ensures real-time response to key tasks and can realize centralized monitoring and intelligent decision-making for the entire process of the mine filling station.

[0072] The centralized control system for a mine backfilling station provided in this application embodiment achieves distributed perception, local optimization, and global intelligent scheduling throughout the entire backfilling station process by constructing a collaborative architecture of node control layer 1 and a central control layer. The node control layer integrates multimodal sensing and edge computing capabilities, combined with fuzzy PID control algorithms or model predictive control algorithms, to achieve real-time adaptive adjustment of backfilling process parameters. The central control layer generates a global control strategy through data fusion, digital twin simulation, reinforcement learning, and multi-objective iterative optimization, significantly improving the automation and intelligence level of the backfilling process, effectively reducing reliance on human intervention, and enhancing adaptability to fluctuations in incoming materials and changes in operating conditions.

[0073] Example 2 Based on the same technical concept as Embodiment 1 above, this application provides a centralized control method for a mine backfilling station, which is applied to the centralized control system 10 of the mine backfilling station in Embodiment 1. Figure 4 This is a schematic diagram of the centralized control method for mine backfilling stations provided in the embodiments of this application, such as... Figure 4 As shown, the method includes the following steps: S100: Collect filling process data of each process unit in the ore filling station through the control modules of each sub-node of the node control layer, perform data preprocessing and feature extraction on the filling process data, and obtain target data.

[0074] S200: Generate primary control commands based on filling process data using a local optimization algorithm to drive the process equipment in the corresponding process unit to perform adjustment operations.

[0075] As an optional implementation of this application, the local optimization algorithm can be a fuzzy PID control algorithm. The specific steps for generating primary control commands based on filling process data using the local optimization algorithm include the following: S201. Set the initial parameters of the PID based on the historical process data of mine backfilling, and define the input and output variables.

[0076] First, parameter initialization and variable definition are performed to clarify the control objectives and allowable deviations of key filling process parameters. Based on historical mine filling process data, the initial parameters of the PID controller are set, including the initial proportional gain. K p0 Initial integral coefficients K i0 and initial differential coefficients K d0 Meanwhile, the input variable is defined as the process parameter deviation. e and rate of change of deviation e c The output variable is defined as the proportional coefficient correction amount Δ corresponding to the PID parameters. K pΔ, the integral coefficient correction K i and differential coefficient correction Δ K d Among them, process parameter deviation e The deviation between the actual value and the target value, the rate of change of the deviation. e c This is the difference between the process parameter deviation of the current cycle and the process parameter deviation of the previous cycle.

[0077] S202. Map the filling process data and the corresponding input variables to the fuzzy domain, and transform them into fuzzy linguistic variables through triangular membership functions.

[0078] Optionally, the collected filling process data can be filtered and denoised to remove outliers caused by electromagnetic interference at the mine site and dust contamination of sensors, thus reducing process parameter deviations. e and rate of change of deviation e c After normalization, the data is mapped to a fuzzy domain and transformed into fuzzy linguistic variables through a triangular membership function. For example, negative large is NB, negative medium is NM, negative small is NS, zero is ZO, positive small is PS, positive medium is PM, and positive large is PB.

[0079] S203. Based on mine backfilling process experience, establish a fuzzy rule base, and match the fuzzy linguistic variables with the fuzzy rule base to obtain the fuzzy output set of PID parameter correction quantities.

[0080] In this embodiment, a dedicated rule base is established based on mine backfilling process experience, such as the deviation of process parameters for IF slurry concentration. e PS and the rate of change of deviation e c If ZO is the scaling factor correction Δ K p For PM, the correction amount of the integral coefficient Δ K i PS, differential coefficient correction Δ K d ZO is used. The fuzzy rules corresponding to the input variables are matched using the Mamdani inference method to generate a fuzzy output set of PID parameter correction values.

[0081] S204. Convert the fuzzy output set into parameter correction values, update the initial PID parameters according to the parameter correction values, and obtain the target PID parameters.

[0082] Optionally, the centroid method can be used to transform the fuzzy output set into precise parameter correction values, according to the formula. K p = K p0 +ΔK p , K i = K i0 +Δ K i , K d = K d0 +Δ K d Update the PID parameters to adapt to the time-varying characteristics of the filling process parameters, and obtain the target PID parameters: target proportional coefficient. K p Target integral coefficient K i and target differential coefficients K d .

[0083] S205. Substitute the target PID parameters into the PID control formula to obtain the primary control command.

[0084] In this embodiment, the updated target PID parameters are substituted into the following PID control formula:

[0085] In the above formula, u(t) This represents the output of the PID controller. K p This represents the target proportionality coefficient. e(t) express t Deviation of process parameters at any given time K i Indicates the target integral coefficient. K d This represents the objective differential coefficient.

[0086] Understandable, based on the output of the PID controller u(t) A primary control command is generated, which is then issued and executed after being processed by the safety threshold of the filling equipment. Real-time feedback data of the filling process is collected. If the deviation of the PID parameters does not meet the standard, steps S202-S205 are repeated until the filling quality requirements are met.

[0087] As an optional implementation of this application, the local optimization algorithm can be a model predictive control algorithm. The specific steps for generating primary control commands based on filling process data using the local optimization algorithm include the following: S210. Construct a prediction model for backfilling process parameters based on historical mine backfilling process data.

[0088] In this embodiment of the application, based on historical operating data of the mine backfilling system, a prediction model for backfilling process parameters is established using system identification methods such as the step response method or the least squares method, as follows:

[0089] In the above formula, y(k+ 1 ) Show k+ Predicted values ​​of filling process parameters at time 1. y(k) express k Actual values ​​of filling process parameters at the specified time. u ( k () represents the control command at time k. n and m Indicates the model order. At the same time, the control constraints and optimization objectives of the filling process parameters are clearly defined. For example, the control constraints are set as follows: upper limit of pumping pressure ≤ 8 MPa, lower limit of slurry flow velocity ≥ 1.2 m / s, and water-cement ratio range of [0.3, 0.4]). The optimization objectives are set as pressure fluctuation ≤ ±0.3 MPa and slurry concentration deviation ≤ ±1%.

[0090] S220. Based on the filling process data and the filling process parameter prediction model, parameter prediction is performed to obtain the optimal control sequence for multiple future control cycles.

[0091] In this embodiment, rolling optimization is performed to obtain the actual values ​​of the current filling process data. y(k) Predicting the future based on predictive models N The parameter sequence for each control cycle is used to construct the optimization objective function as follows:

[0092] In the above formula, J This indicates optimizing the objective function value. N Indicates the number of control cycles. y ref This indicates the target value of the filling parameters. M Indicates control of the time domain, ρ This represents the penalty coefficient, which can be set to account for the start-up and shutdown losses of the filling equipment.

[0093] Understandably, the future can be obtained by optimizing the objective function described above. N The optimal control sequence for each control cycle.

[0094] S230. Extract the first control quantity of the optimal control sequence as the initial control command for the current cycle and issue it for execution, and collect the actual filling process parameters after execution.

[0095] In this embodiment of the application, the first control quantity of the optimal control sequence is extracted as the initial control command for the current cycle and executed, and the actual filling process parameters after execution are collected.

[0096] S240. Perform iterative optimization based on the actual filling process parameters and the filling process parameter prediction model until the actual filling process parameters meet the preset requirements, and generate primary control commands based on the actual filling process parameters.

[0097] In this embodiment, the deviation between the actual backfilling process parameters and the predicted backfilling process parameters output by the backfilling process parameter prediction model is calculated and fed back to the backfilling process parameter prediction model to correct prediction deviations caused by factors such as changes in tailings properties and fluctuations in pipeline resistance. Simultaneously, iterative optimization is performed, rolling the time window forward one step and repeating steps S220-S230. Based on the latest feedback data, the backfilling process parameter prediction model and the optimal control sequence are continuously updated until the backfilling process parameters meet the backfilling design requirements.

[0098] S300: Based on target data, task data, and human intervention data, and combined with filling constraints, an artificial intelligence model is trained and globally optimized to generate a global control strategy.

[0099] In this embodiment, a high-fidelity virtual model is established based on the physical entity of the mine backfilling station using a digital twin engine. A global control strategy is generated after performing multi-scenario simulation and inversion analysis within the virtual model. This includes simulation and inversion analysis for scenarios with different tailings properties, different goaf shapes, and different pipeline lengths. The generation of the global control strategy specifically includes the following steps: S310. Standardize the target data, task data, and human intervention data according to the protocol to obtain a structured dataset.

[0100] In this embodiment, protocol parsing and conversion are first performed. The corresponding protocol parsing driver is called to extract valid data fields related to the filling process, and a unique node identifier (Node ID) is assigned to each type of valid data. For example, the Modbus RTU protocol is called to parse the tailings feed rate and cement admixture data corresponding to the function code and register address; the Profinet protocol is called to parse the pumping pressure and pump frequency data corresponding to the PDO process data; and the OPC UA protocol is called to parse the slurry concentration and water-cement ratio data corresponding to the node data. Redundant information such as check bits and frame headers and footers is removed, and all data is uniformly converted into the OPC UA protocol format to adapt to cross-equipment communication in industrial mines.

[0101] Then, data cleaning and completion are performed, with specialized data processing tailored to the characteristics of mine backfilling process data. For example, duplicate data within the same collection period is deleted using a timestamp deduplication algorithm, based on 3D... σ Abnormal data is filtered out based on principles or filling process thresholds; moving averages are used to smooth filling time series data with large fluctuations such as vibration and pressure; for single-point data loss caused by temporary sensor failures, linear interpolation is used for short-term missing data, and LSTM time series prediction is used to complete data that has been missing for more than three consecutive periods; for batch missing data such as filling strength prediction and equipment health index, historical backup data from the node computing center is called to supplement the missing data and the source of the supplementation is marked.

[0102] Next, data format and semantic standardization is carried out, defining the field format specific to mine backfilling, including: numeric types retaining 2 decimal places, timestamps in UTC+8 time zone ISO 8601 format, and equipment and process units encoded in "process unit-equipment type-serial number"; data is encapsulated in JSON format, containing 7 core fields: data ID, type, acquisition time, value, unit, source node, and data status; a domain semantic dictionary is established to unify data names and connotations.

[0103] Finally, the structured dataset is encapsulated and stored, divided into a basic layer, a related layer, and a task layer according to the mine backfilling business dimension. The basic layer contains single-node, single-parameter raw data, indexed by "collection time + source node". The related layer is associated with process flow data, and the task layer aggregates data by "backfilling task ID". A hybrid architecture of InfluxDB time-series database and MySQL relational database is adopted, where InfluxDB time-series database stores high-frequency runtime time-series data, and MySQL relational database stores structured business data. Multi-dimensional indexes for time, nodes, and tasks are established to ensure that the central control layer can retrieve any backfilling condition dataset within 1 second.

[0104] S320: Build and train an artificial intelligence model based on the process parameters of the mine backfilling station using a digital twin engine.

[0105] As an optional implementation method of this application, the training data of the artificial intelligence model comes from digital twin simulation data and historical mine filling data. The Clip mechanism is used to stabilize the training process, the policy truncation threshold is 0.2, and after completing 100,000 training steps, a stable policy is output to obtain a trained artificial intelligence model. This model has the ability to generalize to different tailings properties and different filling tasks.

[0106] S330. Based on the structured dataset and filling constraints, the artificial intelligence model is used to simulate the operation state of the filling process and obtain multiple initial control strategies and optimal target action parameters.

[0107] In this embodiment, a structured dataset is input into an artificial intelligence model, and corresponding filling constraints are set to simulate the operation of the filling process under different combinations of target action parameters. Examples include simulating slurry flowability when tailings particle size changes, the influence of goaf morphology on the strength distribution of the filling body, and the influence of pipeline length on pumping pressure. The final model outputs optimal target action parameters such as slurry performance, energy consumption, equipment load, and filling body strength distribution, and selects a set of feasible solutions that satisfy all constraints to obtain multiple initial control strategies.

[0108] S340. The optimal control strategy is obtained by iteratively optimizing multiple initial control strategies through a multi-objective genetic algorithm.

[0109] Optionally, the NSGA-II algorithm can be used to optimize the initial control strategy. This involves selecting schemes with better overall performance through non-dominated sorting, ensuring scheme diversity through crowding calculation, generating a new generation of schemes through genetic iteration, and after 100-200 generations, extracting the first non-dominated layer of the final population as the Pareto optimal solution set to obtain the optimal control strategy. Specifically, this includes the following steps: First, a multi-objective system is constructed. Combining the core needs of mine backfilling, a multi-objective optimization system is built from the dimensions of safety, efficiency, cost, and environmental protection. For example, optimization objectives include the achievement rate of the 28-day compressive strength standard for the backfill body. f Maximize 1, the objective is f 1 ≥ 95%; Slurry concentration deviation f Minimize 1, with the objective being f 1≤±2%; Energy consumption per unit volume of filling grout f 3. Minimize, with the objective being f 3≤5 kWh / m³; Cement consumption per unit f 4. Minimize, with the objective being f 4≤300 kg / m³; Equipment downtime rate f Minimize 5, with the objective being f 5≤1%; dust emission concentration f Minimize 6, with the objective being f 6≤10 mg / m³; filling volume per unit time f Maximize 7, the objective is f 7 ≥ 90% of designed capacity; equipment start-up / shutdown switching time f Minimize 8, with the objective being f 8 ≤ 10 min. Simultaneously, set specific constraints for filling. For example, set the process constraints as water-cement ratio ∈ [0.3, 0.4], slurry flow rate ∈ [1.2 m / s, 2.0 m / s], and pumping pressure ≤ 8 MPa; and set the resource constraints as daily cement supply ≤ 500 t and daily power load ≤ 800 kW. h, set the time constraint to the completion time of a single batch filling task ≤ 4 h, set the safety constraint to the equipment vibration amplitude ≤ 2.8 mm / s and the pipeline pressure fluctuation ≤ ±0.5 MPa.

[0110] Next, data preprocessing is performed. Historical mine backfilling scheduling schemes and corresponding execution data for approximately 30-60 control cycles are extracted from the structured dataset. This includes parameter combinations such as feed ratio, stirring speed, and pump frequency, as well as corresponding intensity, energy consumption, and fault data. Abnormal data caused by sensor malfunctions and human error, such as slurry concentration of 0 and negative energy consumption, are cleaned and removed. Missing data is filled using LSTM time-series prediction. Simultaneously, min-max normalization is used to map backfilling indices of different dimensions to the [0,1] interval. Each historical scheduling scheme is encoded as a real-valued chromosome, with chromosome length equal to the dimension of the backfilling decision variables, such as the number of key parameters like feed ratio, stirring speed, and pump frequency, forming the initial population for the NSGA-II algorithm.

[0111] Then, a non-dominated ranking is performed. For each filling control strategy in the initial population, a non-dominated relationship is determined. If strategy A is not inferior to strategy B on all optimization objectives, and is superior to strategy B on at least one objective, then strategy A dominates strategy B. Individuals not dominated by any strategy are selected to form the first non-dominated layer, i.e., the Pareto front layer. After removing these individuals, the second and third non-dominated layers are generated in sequence. The higher the ranking, the worse the overall performance of the strategy.

[0112] Next, crowding degree calculation is performed. For individuals within the same non-dominated layer, the crowding degree is calculated, sorted by the value of each optimization objective function, and the crowding distance of each individual in each objective dimension is calculated. The calculation formula is as follows:

[0113] In the above formula, Indicates the first m The first non-dominated layer i The degree of crowding of individuals and Indicates the first i The function values ​​of two adjacent individuals of an individual. and They represent the first m The target maximum and target minimum values ​​of each non-dominated layer.

[0114] Understandably, the total crowding is the sum of distances in all dimensions. The greater the crowding, the higher the value of individual diversity, thus preventing excellent filling control strategies from being eliminated.

[0115] Then, genetic iteration is performed to generate a better population through genetic operators. One option is to use tournament selection, randomly selecting two individuals, choosing the one with the highest rank and the highest crowding density to enter the mating pool. Crossover is performed using simulated binary crossover with a probability of 0.7-0.9, combining the best genes from the parent schemes, such as the feed ratio from strategy A plus the pump frequency parameter from strategy B. Mutation is performed using polynomial mutation with a probability of 0.01-0.05, randomly perturbing some decision variables to avoid getting trapped in local optima. Finally, the parent and offspring populations are merged, with a size of up to 2-1. N Before filtering after resorting N Individuals constitute a new generation of population.

[0116] Finally, optimal solutions are selected through iterations of 100-200 generations or until population convergence. For example, after 10 consecutive generations of Pareto front layers showing no significant changes, the first non-dominated layer is extracted as the Pareto optimal solution set. Based on the current filling requirements of the mine, weights are assigned to each optimization objective using the analytic hierarchy process (AHP). The comprehensive score of each strategy in the first non-dominated layer is calculated, and the strategy with the highest score is the optimal control strategy adapted to the current working conditions.

[0117] S350: Generate a global control strategy based on the optimal control strategy and the optimal target action parameters.

[0118] In this embodiment, the optimal target action parameters output by the artificial intelligence model are combined with the optimal control strategy generated by iterative optimization. Taking into account the priority of the mine's current filling needs, such as prioritizing efficiency for emergency filling tasks and balancing cost and intensity for routine tasks, a global control strategy containing collaborative parameters of each process unit is generated to ensure that the strategy adapts to the collaborative needs of the entire filling process.

[0119] S400: Distribute the global control strategy to each sub-node control module to execute local control of each process unit.

[0120] In this embodiment, the global control strategy is distributed to each sub-node control module. The instruction execution module of each sub-node control module decomposes the received global control strategy into executable instructions for the process unit through the instruction parsing engine. The executable instructions have a filling-specific action sequence with timing and safety constraints, driving the process equipment of each process unit to complete precise control.

[0121] The centralized control method for mine backfilling stations provided in this application effectively improves the stability of backfill quality, reduces equipment downtime, accelerates production scheduling response, significantly improves overall operating efficiency and economic benefits, realizes predictive maintenance and closed-loop optimization, and promotes the development of mine backfilling towards high efficiency and intelligence.

[0122] It is understood that the implementation method of the centralized control system for mine backfilling stations described in Embodiment 1 above is also applicable to this embodiment and can achieve the same technical effect, so it will not be described again here.

[0123] Example 3 Based on the same concept, this application also provides an electronic device. Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 5 As shown, the electronic device 300 may include a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute the steps of the centralized control method for mine backfilling stations as described in the above embodiments. For example, this includes: S100: Collect filling process data of each process unit in the ore filling station through the control modules of each sub-node of the node control layer, perform data preprocessing and feature extraction on the filling process data, and obtain target data. S200: Generate primary control commands based on filling process data using a local optimization algorithm to drive the process equipment in the corresponding process unit to perform adjustment operations; S300: Based on target data, task data, and human intervention data, and combined with filling constraints, an artificial intelligence model is trained and globally optimized to generate a global control strategy. S400: Distribute the global control strategy to each sub-node control module to execute local control of each process unit.

[0124] The processor 310 can be a central processing unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations of the above types of chips.

[0125] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] The memory 330 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created by the processor, etc. Furthermore, the memory may include high-speed random access memory and non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory may optionally include memory remotely located relative to the processor, which can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0127] Example 4 Based on the same concept, embodiments of this application also provide a computer-readable storage medium storing a computer program. This computer program includes at least one piece of code that can be executed by a master control device to control the master control device to implement the steps of the centralized control method for mine backfilling stations as described in the above embodiments. For example, it includes: S100: Collect filling process data of each process unit in the ore filling station through the control modules of each sub-node of the node control layer, perform data preprocessing and feature extraction on the filling process data, and obtain target data. S200: Generate primary control commands based on filling process data using a local optimization algorithm to drive the process equipment in the corresponding process unit to perform adjustment operations; S300: Based on target data, task data, and human intervention data, and combined with filling constraints, an artificial intelligence model is trained and globally optimized to generate a global control strategy. S400: Distribute the global control strategy to each sub-node control module to execute local control of each process unit.

[0128] Based on the same technical concept, this application also provides a computer program, which, when executed by a main control device, is used to implement the above-described method embodiments.

[0129] The computer program may be stored, in whole or in part, on a computer-readable storage medium packaged with the processor, or in part or in whole on a memory not packaged with the processor.

[0130] Based on the same technical concept, this application also provides a processor for implementing the above-described method embodiments. The processor can be a chip.

[0131] In summary, the centralized control system, method, electronic equipment, and storage medium for mine backfilling stations provided in this application achieve distributed perception, local optimization, and global intelligent scheduling throughout the entire backfilling process by constructing a collaborative architecture of node control layer and central control layer. The node control layer integrates multimodal sensing and edge computing capabilities, combined with fuzzy PID control algorithms or model predictive control algorithms, to achieve real-time adaptive adjustment of backfilling process parameters. The central control layer generates a global control strategy through data fusion, digital twin simulation, reinforcement learning, and multi-objective iterative optimization, significantly improving the automation and intelligence level of the backfilling process, effectively reducing reliance on human intervention, and enhancing adaptability to fluctuations in incoming materials and changes in operating conditions. Simultaneously, it effectively improves the stability of backfill quality, reduces equipment downtime, accelerates production scheduling response speed, significantly improves overall operating efficiency and economic benefits, achieves predictive maintenance and closed-loop optimization, and promotes the development of mine backfilling towards high efficiency and intelligence.

[0132] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0133] The embodiments described above are merely examples of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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. Such 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 this application.

Claims

1. A centralized control system for a mine backfilling station, characterized in that, The system includes a node control layer and a central control layer, wherein: The node control layer includes multiple sub-node control modules. Each sub-node control module corresponds to a process unit in the mine filling station. The sub-node control module is used to collect the filling process data of the corresponding process unit, process it to obtain target data, upload the target data to the central control layer, and receive the global control strategy issued by the central control layer to execute the local control of the process unit. The central control layer is used to receive the target data from each of the sub-node control modules, and to obtain the operation task data and human intervention data of the mine filling station. Based on the target data, the operation task data and the human intervention data, and combined with the filling constraints, the central control layer generates the global control strategy through training and global optimization calculation using an artificial intelligence model, and then distributes it to each of the sub-node control modules.

2. The centralized control system for mine backfilling stations according to claim 1, characterized in that, The sub-node control module includes at least a data acquisition unit, a computing center unit, and a control execution unit, wherein: The data acquisition unit acquires the filling process data through a multimodal sensor network; The computing hub unit performs data preprocessing and feature extraction on the filling process data, obtains the target data and sends it to the central control layer, and generates primary control commands based on the filling process data through a local optimization algorithm. The control execution unit receives the global control strategy and executes the local control of each process unit, and receives the primary control instructions and drives the process equipment in the corresponding process unit to perform adjustment operations.

3. The centralized control system for mine backfilling stations according to claim 1, characterized in that, The central control layer includes at least a data fusion module, a model training module, a model simulation module, an optimization feedback module, and a strategy generation module, wherein: The data fusion module performs protocol standardization processing on the target data, the task data, and the human intervention data to obtain a structured dataset; The model training module constructs and trains the artificial intelligence model based on the process parameters of the mine filling station using a digital twin engine; The model simulation module simulates the operation state of the filling process through the artificial intelligence model based on the structured dataset and the filling constraints, and obtains multiple initial control strategies and optimal target action parameters. The optimization feedback module uses a multi-objective genetic algorithm to iteratively optimize multiple initial control strategies to obtain the optimal control strategy; The strategy generation module generates the global control strategy based on the optimal control strategy and the optimal target action parameters.

4. The centralized control system for mine backfilling stations according to claim 3, characterized in that, The central control layer also includes a resource allocation module and an early warning display module, wherein: The resource allocation module dynamically adjusts the computing resources of the sub-node control module based on the structured dataset and the preset priority rules for mine filling; The early warning display module is used to visualize data and strategies and to provide early warnings for filling parameters.

5. A centralized control method for a mine backfilling station, characterized in that, The method is applied to the centralized control system of the mine backfilling station according to any one of claims 1-4, and the method includes: The filling process data of each process unit in the ore filling station is collected by the control modules of each sub-node of the node control layer. The filling process data is preprocessed and feature extracted to obtain the target data. A primary control command is generated based on the filling process data using a local optimization algorithm to drive the process equipment in the corresponding process unit to perform adjustment operations. Based on the target data, task data, and human intervention data, and combined with filling constraints, a global control strategy is generated by training and global optimization calculations using an artificial intelligence model. The global control strategy is distributed to each of the sub-node control modules to execute the local control of each of the process units.

6. The centralized control method for mine backfilling stations according to claim 5, characterized in that, Based on the target data, task data, and human intervention data, and combined with filling constraints, an artificial intelligence model is trained and globally optimized to generate a global control strategy, including: The target data, the task data, and the human intervention data are subjected to protocol standardization processing to obtain a structured dataset; The artificial intelligence model is constructed and trained based on the process parameters of the mine filling station using a digital twin engine; Based on the structured dataset and the filling constraints, the artificial intelligence model is used to simulate the operation of the filling process and obtain multiple initial control strategies and optimal target action parameters. The optimal control strategy is obtained by iteratively optimizing multiple initial control strategies using a multi-objective genetic algorithm. The global control strategy is generated based on the optimal control strategy and the optimal target action parameters.

7. The centralized control method for mine backfilling stations according to claim 5, characterized in that, The local optimization algorithm is a fuzzy PID control algorithm. The step of generating primary control commands based on the filling process data using the local optimization algorithm includes: The initial parameters of the PID controller are set based on historical process data of mine backfilling, and the input and output variables are defined. The filling process data is mapped to the corresponding input variables and then transformed into fuzzy linguistic variables through a triangular membership function. A fuzzy rule base is established based on mine backfilling process experience. The fuzzy linguistic variables are matched with the fuzzy rule base to obtain a fuzzy output set of PID parameter correction quantities. The fuzzy output set is converted into parameter correction values, and the initial PID parameters are updated according to the parameter correction values ​​to obtain the target PID parameters. Substitute the target PID parameters into the PID control formula to obtain the primary control command.

8. The centralized control method for mine backfilling stations according to claim 5, characterized in that, The local optimization algorithm is a model predictive control algorithm. The step of generating primary control commands based on the filling process data using the local optimization algorithm includes: A prediction model for backfilling process parameters was constructed based on historical process data of mine backfilling. Based on the filling process data and the filling process parameter prediction model, parameter prediction is performed to obtain the optimal control sequence for multiple future control cycles; Extract the first control quantity of the optimal control sequence as the initial control command for the current cycle and issue it for execution, and collect the actual filling process parameters after execution; The actual filling process parameters and the filling process parameter prediction model are iteratively optimized until the actual filling process parameters meet the preset requirements, and the primary control command is generated based on the actual filling process parameters.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the centralized control method for mine backfilling stations as described in any one of claims 4-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the centralized control method for mine backfilling stations as described in any one of claims 4-8.

Images