Production collaborative scheduling method based on intelligent mine comprehensive management and control platform

Abstract

The invention relates to production collaborative scheduling, in particular to a production collaborative scheduling method based on an intelligent mine comprehensive management and control platform. According to the intelligent linkage control system, a sensing-data-application three-layer framework is constructed, key links such as coal mining, transportation, storage, washing and selection and lifting are integrated, cross-equipment and cross-system intelligent linkage control is achieved, and the dispatching adaptive capacity and the cooperation efficiency under the complex working condition are remarkably improved. Comprising the following steps: S1, collecting equipment states, environmental parameters and production progress data of each production link of a mine, processing the data through an edge computing node, and uploading the data to a cloud platform; s2, constructing a digital twin mine model, and configuring an initial production scheduling rule and a resource constraint condition; s3, when an abnormal or disturbance event is detected, the system gives out an early warning and automatically responds, a scheduling instruction is issued to each equipment system, and linkage execution is realized; and S4, continuously returning a scheduling execution result and a running state to the platform for evaluation and analysis, and training and optimizing the scheduling model by using feedback data to realize self-adaptive capability enhancement.

Description

Technical Field

[0001] This invention relates to production collaborative scheduling, and more particularly to a production collaborative scheduling method based on an intelligent mine integrated management and control platform. Background Technology

[0002] With the continuous advancement of industrialization and informatization, the mining industry, as a crucial foundation of the national economy, plays an irreplaceable role in energy security, raw material supply, and strategic resource reserves. Mining production (such as coal mines, iron ore mines, non-ferrous metal mines, and non-metallic mineral mines) is generally characterized by uneven resource distribution, complex geological conditions, harsh working environments, diverse equipment types, and long and variable technological processes. For example... Figure 1 As shown, taking coal mine production as an example, any malfunction or inefficiency in any link can affect the operation of the entire production system. These factors make mine production highly systematic, dynamic, and uncertain, posing a significant challenge to traditional management and scheduling methods.

[0003] To address these issues, the concept of smart mine construction has emerged, becoming a key path to promote high-quality development in the mining industry. Smart mines, based on next-generation information technologies (such as artificial intelligence, the Internet of Things, big data, edge computing, and cloud computing), aim to achieve digital, intelligent, and automated management of the entire mining process. Their core objectives are to improve resource utilization, safety levels, and economic benefits. Among these, production collaborative scheduling, as the central system of a smart mine, is a crucial link in achieving efficient and stable mine operation. Collaborative scheduling not only involves the optimized coordination of multiple production stages such as mining, transportation, hoisting, and washing, but also requires the integration of information from multiple elements including personnel, machinery, environment, and materials. Its complexity far exceeds that of traditional single-point scheduling or local control systems.

[0004] Currently, research on intelligent mine collaborative scheduling not only needs to address the issues of system perception and information fusion, but also requires the real-time acquisition, intelligent analysis, and dynamic optimization of multi-source heterogeneous data. Some mines have begun to apply sensor networks, video surveillance, GIS systems, and SCADA systems to achieve digital monitoring of equipment status, environmental parameters, geological information, and operational behaviors. Meanwhile, artificial intelligence algorithms (such as reinforcement learning, neural networks, and swarm intelligence optimization) are gradually being applied to the construction and optimization of scheduling models. However, the high proportion of unstructured data, frequent scene changes, and multi-dimensional and conflicting optimization objectives in the mining environment mean that existing scheduling methods still primarily focus on single-objective optimization, making it difficult to achieve true multi-stage collaboration and global optimization. Summary of the Invention

[0005] This invention addresses the shortcomings of existing technologies by providing a production collaborative scheduling method based on an intelligent mine integrated management and control platform.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a production collaborative scheduling method based on an intelligent mine integrated management and control platform, characterized by comprising the following steps:

[0007] S1. Collect equipment status, environmental parameters and production progress data of each production stage in the mine, process them through edge computing nodes and then upload them to the cloud platform;

[0008] S2. Construct a digital twin mine model, configure initial production scheduling rules and resource constraints, analyze real-time data, and generate an intelligent collaborative scheduling scheme.

[0009] S3. When an abnormal or disturbance event is detected, the system issues an early warning and responds automatically, sending dispatch instructions to each equipment system to achieve coordinated execution, synchronously controlling the inflow and outflow of the buffer coal bunker, and coordinating production and transportation matching.

[0010] S4. Continuously transmit the scheduling execution results and running status back to the platform for evaluation and analysis. Use the feedback data to train and optimize the scheduling model to enhance its adaptive capabilities.

[0011] Furthermore, S1 specifically includes:

[0012] S1.1 By deploying sensors and smart terminals in the mining, tunneling, transportation, hoisting, and washing processes, real-time data on equipment status, environmental parameters, and production progress are collected. The data collection process uses a single-sensor data acquisition model to correct measurement errors and a multi-sensor data fusion model to achieve redundancy verification.

[0013] The equipment status data includes coal mining machine speed, conveyor current, and hoist temperature; environmental parameter data includes gas concentration, wind speed, and roadway displacement; and production progress data includes instantaneous output, conveyor belt coal flow rate, and washing and processing capacity.

[0014] The single-sensor data acquisition model is as follows:

[0015] x i (t,s)=f i (t,s)+ε i (t,s)

[0016] In the formula, x i (t,s) represents the raw values ​​collected by the i-th sensor at time t and spatial location s; f i (t,s) represents the target physical quantity (such as temperature, displacement, current), ε i (t,s) represents the sensor measurement error (modeled as white noise or Gaussian error);

[0017] For multiple sensor data points on the same physical quantity (such as tunnel displacement), they need to be fused according to their reliability. The multi-sensor data fusion model is as follows:

[0018]

[0019] In the formula: x is the estimated value after fusion. i (t) represents the value collected by the i-th sensor, w i The sensor reliability weight is set based on the sensor's historical error rate and signal-to-noise ratio.

[0020] Two models are used to achieve: error correction for equipment status data (such as coal mining machine vibration); multi-sensor redundancy verification for environmental parameters (such as gas concentration); and time-series consistency alignment for production progress data (such as coal flow).

[0021] S1.2, the fused data output from S1.1 Perform unified processing to output a standardized dataset;

[0022] The unification process includes data cleaning and data standardization. Data cleaning includes filling in default values ​​using linear interpolation and removing outliers using the Z-Score method. Data standardization includes dimensional normalization and aligning data timestamps of different sampling frequencies based on NTP clock synchronization.

[0023] S1.3 Deploy edge computing nodes at key locations to process the standardized dataset output from S1.2, including data filtering, data compression, and event detection; output the processed dataset D_edge.

[0024] Event detection uses an edge rule engine to detect predefined events, including:

[0025] Equipment overload: Triggered when current I > I_max threshold, where I_max represents the maximum current threshold (can be 1.1 times the rated current).

[0026] Coal bunker full: Triggered when bunker L > 90% capacity;

[0027] Abnormal vibration: Triggered when vibration amplitude A > A_threshold, where A_threshold represents the vibration amplitude threshold; (this value is taken as 5 mm / s) 2 ;)

[0028] S1.4 Upload the D_edge data output from S1.3 to the cloud platform to form a production information database covering the entire mine.

[0029] Furthermore, in S1.3, deployment at key locations specifically includes:

[0030] Mining area: coal mining machine body, hydraulic support controller;

[0031] Transportation system: Weighing terminals at the main conveyor belt inlet and coal bunker inlet / outlet;

[0032] Lifting component: PLC terminal in the hoist control room;

[0033] Coal washing and beneficiation process: Control cabinet of the heavy media separator in the coal preparation plant;

[0034] Environmental monitoring: Methane sensors in ventilation zones, temperature and humidity monitoring stations in return air tunnels.

[0035] Furthermore, S2 specifically includes:

[0036] S2.1 Construct a digital twin mine model that maps one-to-one with the actual mine production system;

[0037] S2.2 Configure initial scheduling rules: including power supply system constraint rules and production equipment scheduling rules;

[0038] S2.3. Transform the production units mapped in S2.1 into intelligent agents. Using the constraints in S2.2 as boundaries, dynamically generate scheduling strategies through distributed optimization algorithms and trigger rescheduling when parameters exceed limits.

[0039] Furthermore, S2.1 specifically includes:

[0040] Based on the real-time data uploaded by S1, a virtual mapping is established that includes the following modules:

[0041] Spatial structure mapping module: performs 3D modeling of roadways, mining areas, coal bunkers, and shafts to construct a virtual topological framework for the static physical environment of the mine;

[0042] Equipment Status Mapping Module: Dynamically binds the real-time operating parameters of the coal mining machine, conveyor, and hoist to the virtual model. The parameters include: the rotation speed, vibration amplitude, and current intensity of the coal mining machine; the operating speed, motor temperature, and load current of the conveyor; and the hoist's hoisting speed, braking status, and wire rope tension.

[0043] Material flow modeling module: Establishes a continuous dynamic model based on the input and output quantities of coal flow and / or water flow, with the dynamic equation being: Q out (t)=min(Q in (t),C max -C(t)); where Q in (t) represents the input amount per unit time; Q out (t) represents the output per unit time; C(t) represents the current inventory; C max For maximum capacity (e.g., coal bunker);

[0044] The spatial structure mapping module provides a spatial topology reference for the equipment state mapping module, and the operating parameters output by the equipment state mapping module drive the dynamic calculations of the material flow modeling module. The three are coupled through a system state space model, and the coupled system state space equation is as follows:

[0045] x(t+1)=Ax(t)+Bu(t)

[0046] y(t)=Cx(t)+Du(t)

[0047] In the formula, x(t) represents the system state, which includes three dimensions of data: spatial topological coordinates, equipment operating parameters, and material inventory; u(t) represents the input variable (scheduling instructions); y(t) represents the observed output (current system feedback); and A, B, C, and D represent the system dynamics matrices.

[0048] Furthermore, S2.2 specifically includes: setting power supply system constraint rules and production equipment scheduling rules based on the model parameters constructed in S2.1;

[0049] In the power supply system constraint rules, the mine power supply system must satisfy the following power constraint equation:

[0050]

[0051] In the formula: U i,t U j,t The voltages at nodes i and j at time t are respectively; r ij x ij Let P be the resistance and reactance of line ij; ij,t Q ij,t These represent the active and reactive power at the beginning of line ij at time t, respectively; I ij,t P is the current flowing through line ij at time t; j,t Q j,t Φ represents the active and reactive power injected into node j at time t, respectively; j Let j be the set of child nodes with node j as the parent node; and the power constraints also include node voltage constraints, distributed generation output constraints, and reactive power compensation constraints.

[0052] The production equipment scheduling rules are shown in Table 1 below;

[0053]

[0054] Furthermore, S2.3 specifically includes:

[0055] Using the state-space equations of S2.1 as the optimization object, and under the constraints of S2.2, perform the following:

[0056] S2.3.1 Based on the equipment state mapping and material flow model established in S2.1, the coal mining machine, conveyor and coal bunker unit are transformed into intelligent agents. The decision variables of each intelligent agent are limited by the equipment constraints in Table 1 of S2.2.

[0057] S2.3.2. Using the power constraint equations of S2.2 as global constraints, perform iterative optimization:

[0058] (1) Each agent solves the local optimization problem according to the following formula:

[0059]

[0060] x i ∈X i

[0061] In the formula, f i (x i Let A be the local objective function of the i-th subsystem (e.g., output, energy consumption); i x i This represents the usage of globally coupled resources (such as the total capacity of the conveyor belt); b is the upper limit of the total resources; and x i (1) The speed regulation range specified in Table 1 of S2.2 must be met; (2) The central coordinator updates the Lagrange multipliers:

[0062]

[0063] S2.3.3 When the S2.1 model detects that a parameter (such as vibration amplitude A) exceeds the threshold (such as A_threshold) set in S2.2, it triggers the local re-optimization of the affected agent (that is, it triggers dynamic rescheduling, and the relevant agent re-optimizes under the remaining resource constraints).

[0064] Furthermore, S3 specifically includes:

[0065] S3.1 Provide real-time early warning and decision support solutions when abnormal or disturbing events are detected;

[0066] S3.2 Automatically send the optimized scheduling instructions to each production system to achieve equipment linkage and production process adjustment;

[0067] S3.3 Manage the capacity and scheduling time window of the buffer coal bunker to avoid problems such as full-bunker shutdown and idling due to power outage.

[0068] Furthermore, S4 specifically includes:

[0069] S4.1, Conduct feedback evaluation: Real-time transmission of equipment status changes, task completion status, abnormal alarm information, and energy consumption indicators after scheduling execution back to the platform for:

[0070] Evaluate the scheduling effectiveness and analyze the degree of matching between equipment operating status and expected goals;

[0071] And as training data for the model, it is used to optimize the parameters of the digital twin model of S2.1 and the scheduling algorithm of S2.3;

[0072] S4.2 Model Optimization: Construct a closed-loop control system from state awareness to scheduling optimization, and achieve continuous optimization through the following methods:

[0073] When the device's operating status is detected to deviate from expectations, the constraint threshold in S2.2 is automatically adjusted;

[0074] Based on historical scheduling data, optimize the allocation of decision weights for agents in S2.3;

[0075] For frequently occurring abnormal patterns, update the device state mapping relationship in the S2.1 model.

[0076] Furthermore, S4.1 evaluates scheduling economics using the following model:

[0077]

[0078] C es,t =α es (P dis,t +P cha,t )

[0079]

[0080]

[0081] Where: T represents the scheduling duration; and These represent the operation and maintenance cost coefficients for CSP power supply and heating, respectively; α es c represents the maintenance cost coefficient of an energy storage device. ex,t The main grid purchase price is represented by α; the output power of the exhaust gas oxidation unit, water source heat pump, air source heat pump, and ground source heat pump are represented by α respectively. i P represents the corresponding cost coefficient. csp,t T represents the output electrical power of the CSP generator; dis2,t This indicates the power of the CSP thermal storage device to supply the heat load; P cha,t and P dis,t These represent the charging and discharging power of the energy storage device, respectively.

[0082] S4.2 achieves closed-loop control through the following optimization model:

[0083] S4.2.1. Based on the digital twin model state data x(t) in S2.1 and the constraints Ω in S2.2, solve for the optimal scheduling instruction u(t):

[0084]

[0085] stx(t+1)=f(x(t),u(t)),x(0)=x0

[0086] x(t)∈χ,u(t)∈μ

[0087] In the formula, γ(x,u) is the cost function, which comprehensively considers equipment energy consumption and production efficiency; χ and μ are the feasible regions of state and action (i.e., the load threshold and speed regulation range defined in Table 1);

[0088] S4.2.2 For the multi-agent system in S2.3, global resource allocation is coordinated using the following formula:

[0089]

[0090] Where: x i (t) represents the current state of the i-th agent; The desired state thresholds set for Table 1; c ij For coupling and coordination weights.

[0091] Compared with the prior art, the present invention has the following advantages.

[0092] This invention addresses the problems of traditional coal mine scheduling relying on manual experience, resulting in delayed responses and difficulty in achieving global optimization. By constructing a three-layer architecture of "sensing-data-application," it integrates key links such as coal mining, transportation, storage, washing, and hoisting, achieving intelligent linkage control across equipment and systems, significantly improving scheduling adaptability and collaborative efficiency under complex operating conditions.

[0093] This invention enables real-time fusion of multi-source heterogeneous data, dynamic simulation of scheduling status, and rapid identification and rescheduling of abnormal events. It also optimizes the allocation of production tasks under multiple objectives and constraints through algorithms.

[0094] Based on feedback learning and strategy iteration mechanisms, this invention enables the scheduling system to continuously optimize strategies using historical operational data, forming a closed-loop adaptive operating mode. It possesses the ability to become more accurate and stable with repeated use, effectively solving the technical bottlenecks of traditional scheduling in high-intensity, high-frequency collaborative control, and providing reliable support for the intelligent and efficient operation of coal mines. Attached Figure Description

[0095] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. The scope of protection of the present invention is not limited to the following description.

[0096] Figure 1 Flowchart of the main production system of a coal mine.

[0097] Figure 2 This is the overall architecture diagram of a smart mine.

[0098] Figure 3 This is the architecture diagram of the intelligent mine integrated management and control platform.

[0099] Figure 4 This is a diagram of the production collaborative scheduling system architecture based on the management and control platform.

[0100] Figure 5 This is a diagram of key technologies for integrated production collaborative management and control.

[0101] Figure 6 It is a decision-making aid and emergency control diagram for sudden events. Detailed Implementation

[0102] The technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0103] The terminology used in the embodiments of this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of this disclosure. The singular forms “a,” “the,” and “the” as used in the embodiments of this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0104] Depending on the context, words such as “if” or “suppose” used here can be interpreted as “when”, “in response to determination”, or “in response to detection”.

[0105] For ease of understanding, the embodiments of this disclosure will be described in detail first.

[0106] like Figure 1-6As shown, in the production collaborative scheduling method based on the intelligent mine integrated management and control platform, (1) sensors and intelligent equipment are used to realize the comprehensive perception of each production link of the mine. Real-time collection of multi-source data such as equipment, environment, and progress. Preprocessing is performed through edge computing, and the data is uploaded to the cloud platform for fusion and standardized governance to establish a unified data center and provide a data foundation for subsequent scheduling. In the first stage of the intelligent mine production collaborative scheduling system, the main task is to build a comprehensive perception network of multi-source data and a standardized data governance system. By deploying a large number of sensors, intelligent control terminals and edge computing nodes in core links such as coal mining, tunneling, transportation, hoisting and storage, real-time collection and local preliminary processing of information such as equipment status, environmental parameters and operation progress are realized to ensure the integrity, timeliness and reliability of the data. The collected data is converted, cleaned, labeled and fused by the edge computing unit and finally converged to the cloud platform to form a unified data asset system. This stage provides a transparent and structured high-quality data foundation for subsequent model calculation and decision inference, which is the premise of intelligent operation of the system.

[0107] (2) Construct a digital twin mine system to achieve integrated virtual and real modeling. Configure initial production scheduling rules and resource constraints. Introduce artificial intelligence algorithms to analyze real-time data and generate intelligent collaborative scheduling schemes. Based on complete data aggregation, the system enters the model-driven scheduling strategy analysis and intelligent decision-making stage. In this stage, a one-to-one mapping of the virtual and real mine model is constructed based on digital twin technology to fully reproduce the equipment layout, operation process, and resource flow characteristics of the main production system; at the same time, an initial scheduling rule framework is established by combining production constraints such as mining plans, equipment capacity, and mine storage capacity. On this basis, the system uses artificial intelligence algorithms to dynamically generate scheduling strategies, realizing intelligent modeling and deduction of key decisions such as task priority, resource allocation, and path selection, taking into account the multi-objective scheduling optimization needs of maximizing production capacity, minimizing energy consumption, and protecting equipment and ensuring stable operation.

[0108] (3) The system issues an early warning when an anomaly occurs, assisting dispatchers or responding automatically. Dispatch instructions are automatically sent to each equipment system to achieve coordinated execution. The system synchronously controls the inflow and outflow of each buffer coal bunker, coordinates production and transportation matching, and ensures continuous and efficient operation of the system.

[0109] After the scheduling strategy is transformed into a specific instruction plan through model calculations, the system automatically executes the scheduling instructions via a scheduling control bus or industrial control network. Each subsystem device (such as the coal mining machine, belt conveyor, and hoist) operates collaboratively according to the control instructions. The system simultaneously monitors key indicators such as coal bunker status, belt speed and load, and operating sequence to ensure the matching of time windows and buffer capacity during the scheduling process, avoiding problems such as "full bunker shutdown" or "flow interruption and idling." Furthermore, the system possesses linkage control and emergency strategy capabilities, automatically triggering strategy switching and process rescheduling at critical nodes (such as critical coal bunker capacity or equipment malfunctions), achieving a dynamically controllable, collaborative, and stable operating mechanism throughout the entire process.

[0110] (4) The scheduling execution results and operating status are continuously fed back for evaluation and analysis. The scheduling model is continuously trained and optimized using the feedback data to enhance its adaptive capabilities. A closed-loop control system covering the entire process is constructed to achieve continuous intelligent optimization of mine scheduling.

[0111] During production scheduling execution, the system continuously collects scheduling execution results and operational feedback data, including task completion rate, equipment status, energy consumption level, and abnormal response status. It then evaluates the scheduling effect based on a difference analysis and performance evaluation model. This feedback information is further used as training data and input into the scheduling strategy learning model (such as a reinforcement learning environment or supervised learning network) to achieve dynamic updates of scheduling strategy parameters and experience transfer optimization. This enables the scheduling system to possess continuous learning and autonomous evolution capabilities, truly forming a closed-loop control architecture from "state awareness" to "strategy optimization." Ultimately, the system can achieve efficient, stable, and intelligent collaborative operation of the mine's main production system under conditions of multi-source disturbances, high load pressure, and complex coupling.

[0112] As one possible implementation, the coordination and scheduling method specifically includes:

[0113] Step 1.1, Multi-source data sensing and acquisition: Using sensors and smart terminals, the equipment status, environmental parameters and production progress of mining, tunneling, transportation, hoisting and washing processes are collected in real time to establish a transparent production data foundation.

[0114] Multi-source data sensing and acquisition refers to the real-time acquisition of key parameters and operational status data from the main production system of a mine through sensors, smart terminals, and edge computing nodes deployed at various stages of mine production. This step aims to establish a seamless "sensing layer" for information acquisition, providing a comprehensive, real-time, and accurate raw data foundation for the intelligent scheduling system.

[0115] The sensing targets include, but are not limited to, mining, tunneling, and transportation processes. Data acquisition terminals are deployed at key nodes in industrial controllers or edge gateways for local aggregation and preliminary processing of raw data, featuring local storage and anomaly filtering capabilities. NTP clock synchronization ensures consistent data timing across all acquisition points, facilitating subsequent fusion and analysis. After local processing of the basic data, it is uploaded to the cloud or the scheduling platform's core database based on priority and importance. The sensor data acquisition model is as follows:

[0116] x i (t,s)=f i (t,s)+ε i (t,s)

[0117] In the formula: x i (t,s) represents the raw values ​​collected by the i-th sensor at time t and spatial location s; f i (t,s) represents the target physical quantity (such as temperature, displacement, current, etc.), ε i (t,s) represents the sensor measurement error (modeled as white noise or Gaussian error). This model can be further used for error correction, redundant filtering, and multi-source data fusion.

[0118] The multi-sensor data fusion model is as follows:

[0119]

[0120] In the formula: The value is the estimated value after fusion; x i (t) represents the value collected by the i-th sensor; w i The sensor reliability weight can be set based on historical error, failure rate, or signal-to-noise ratio.

[0121] Step 1.2, Data Fusion and Standardization Governance: Clean, standardize, label and merge the received multi-source heterogeneous data to build a unified data asset system and ensure data integrity and consistency.

[0122] Unifying sensor data from different sources, formats, structures, and frequencies involves processes such as cleaning, standardization, tagging, time alignment, and fusion computation to ultimately create "high-quality data assets" that can be directly accessed by system scheduling engines, analysis models, and digital twin platforms. This is one of the core prerequisites for achieving intelligent scheduling, and the overall architecture is as follows: Figure 3 As shown.

[0123] Data cleaning primarily targets noise, missing, redundant, and outlier data acquired during the data collection process, and commonly employs the following methods:

[0124] Missing values ​​can be filled using interpolation methods (linear / spline):

[0125]

[0126] Outlier removal can be done using the Z-Score method: if This is considered abnormal.

[0127] Data standardization is necessary because different sensors have different units, dimensions, and sampling frequencies, so a unified processing method is required.

[0128] Dimensional normalization: This maps all data to the [0,1] interval, facilitating model training and feature comparison.

[0129] Step 1.3, Real-time data transmission and edge preprocessing: Deploy edge computing nodes to perform preliminary analysis and preprocessing of local data, reduce platform data pressure, and improve response speed and real-time performance.

[0130] By deploying edge computing devices at key nodes in the mine site, the raw data collected by sensors is initially processed and analyzed locally, and only necessary information is uploaded to the central platform. This reduces data transmission bandwidth, alleviates the cloud computing burden, and enables rapid response and local autonomous control.

[0131] The sensors are deployed near coal mining equipment in the working face (coal mining machine, hydraulic supports), at the entrance of the main transportation system, at the coal bunker entrance and exit, in the hoisting control room, in key sections of the coal preparation plant, and at environmental monitoring points (ventilation zones, gas monitoring stations), etc. The raw data collected by the sensors is then preliminarily processed and analyzed locally, using moving averages and median filtering for real-time data de-jittering.

[0132]

[0133] Outliers are removed immediately to prevent erroneous values ​​from interfering with decision-making.

[0134] Use time-series data compression algorithms (such as Gorilla compression, Delta-encoding) to reduce the amount of data transmitted and extract key metrics (such as average, maximum, minimum, and rate of change):

[0135]

[0136] An edge rules engine is deployed to identify predefined events, such as equipment overload, near-full coal bunker, and abnormal vibration. This enables "on-site alarms," ​​supporting local buzzers, indicator lights, or voice prompts. When the network is interrupted, critical data is cached locally on edge nodes, and automatic transmission resumes upon network recovery, ensuring data integrity and no packet loss. Data upload employs a tiered upload and protocol optimization mechanism: real-time high-frequency data is processed locally and only a summary is uploaded; abnormal or alarm data is packaged and uploaded immediately; low-priority data is uploaded periodically to avoid bandwidth exhaustion.

[0137] Step 1.4, Data Center Aggregation and Cloud Storage: All processed key data is uploaded to the cloud platform data center to form a production information database covering the entire mine, supporting subsequent big data analysis and modeling.

[0138] In the production collaborative scheduling system of intelligent mines, the amount of data generated by front-end equipment, sensors, and edge computing nodes is enormous and has high real-time requirements. To enable the entire mine to have capabilities such as macro-scheduling, risk prediction, and process optimization, relying solely on local processing by edge nodes is far from sufficient. Therefore, it is necessary to build a core platform that supports "data-driven scheduling" by realizing the aggregation, storage, management, and sharing of massive amounts of data through a unified, reliable, and high-performance data center and cloud platform.

[0139] Data center aggregation and cloud storage unify the transmission of data processed by edge computing nodes to the mining company's data center or cloud platform for centralized management and storage. By building a unified cloud data architecture, it completes the archiving, indexing, retrieval, and security assurance of multi-source heterogeneous data, providing reliable global data support for subsequent scheduling optimization, intelligent analysis, and digital twin modeling.

[0140] Data centers and the cloud are not only centralized storage areas for data, but also computing and data support platforms for key applications such as decision analysis, scheduling modeling, digital twins, and visual monitoring. Their functions permeate the application layer, data layer, and perception layer of the entire smart mine. The production collaborative scheduling system architecture based on the management and control platform is as follows: Figure 4 As shown.

[0141] Step 2, Modeling, Analysis, and Decision Making: Construct a digital twin mine system to achieve integrated virtual and physical modeling. Configure initial production scheduling rules and resource constraints. Introduce artificial intelligence algorithms to analyze real-time data and generate intelligent collaborative scheduling solutions.

[0142] Step 2.1, Establish a digital twin mine model: Construct a virtual simulation model that maps one-to-one with the actual mine production system to realize the virtual mapping and monitoring of equipment status, production process and resource flow.

[0143] This involves creating a virtual model within a computing environment that closely mirrors the structure, behavior, and state of an actual mine production system. Driven by real-time data, this model synchronously and dynamically reflects the entire process of equipment operation, resource flow, environmental changes, and scheduling execution in a real mine. The aim is to facilitate the transformation of mines from "static information management" to "dynamic virtual-real integration," support predictive scheduling, simulation analysis, anomaly prediction, and collaborative control, and provide a simulation environment and training platform for AI-optimized scheduling, remote monitoring, and operation and maintenance management.

[0144] A three-dimensional mapping is established for the spatial structure of the mine (roadways, mining areas, coal bunkers, shafts), equipment (coal mining machines, conveyors, hoists), materials (coal flow, water flow), and personnel locations. Edge computing nodes and sensor data are integrated, such as coal bunker capacity, belt conveyor speed, and motor temperature. Dynamic updates to the virtual model (state changes over time) are achieved. A virtual behavioral logic chain is established using equipment control logic, scheduling rules, and work processes. Finally, the twin model is linked with scheduling platforms, MES, SCADA, ERP, and other systems to achieve visual simulation and control of the entire production chain from planning to execution. Key technologies for integrated collaborative production management include... Figure 5 As shown.

[0145] A digital twin system is a hybrid system that combines data-driven and model-driven approaches. Its mathematical model is as follows:

[0146] x(t+1)=Ax(t)+Bu(t)

[0147] y(t)=Cx(t)+Du(t)

[0148] This is a state-space model used to describe the state of a system at a given time and its evolution over time. In the formula, x(t) represents the system state (e.g., equipment speed, remaining coal in the bunker); u(t) represents the input variable (e.g., scheduling instructions); and y(t) represents the observed output (e.g., current system feedback). A, B, C, and D are the system dynamics matrices.

[0149] Continuous modeling of material transfer processes such as coal flow and water flow:

[0150] Q out (t)=min(Q in (t),C max -C(t))

[0151] In the formula Q in (t) represents the input amount per unit time; Q out (t) represents the output per unit time; C(t) represents the current inventory; C max For maximum capacity (e.g., coal bunker).

[0152] Establishing a digital twin mine model is a crucial step in realizing the transformation of intelligent scheduling systems from "data-driven" to "model-based" systems. By integrating physical modeling, time-series data, dynamic events, and virtual reality, the system can not only reflect the current state of the mine but also predict future trends, test scheduling strategies, and verify safety measures, demonstrating high practicality and technological sophistication.

[0153] Step 2.2, Initial scheduling rule modeling and parameter configuration: Based on mining plans, equipment capacity, transportation capacity, etc., set basic scheduling logic and parameter constraints to form the initial framework of the scheduling model.

[0154] Before the scheduling system is officially put into operation, a basic scheduling logic and parameter system is established based on information such as the mine's mining plan, production resources, equipment capacity, and transportation capacity. The aim is to construct a preliminary, executable scheduling model framework, providing a starting point and boundaries for subsequent intelligent dynamic scheduling and optimization algorithms.

[0155] The branch power flow of the mine power supply system must meet the following constraints:

[0156]

[0157] In the formula: U i,t U j,t The voltages at nodes i and j at time t are respectively; r ij x ij Let P be the resistance and reactance of line ij; ij,t Q ij,t These represent the active and reactive power at the beginning of line ij at time t, respectively; I ij,t P is the current flowing through line ij at time t; j,t Q j,t Φ represents the active and reactive power injected into node j at time t, respectively; j Let j be the set of child nodes with node j as the parent node. The mine power supply system also needs to meet node voltage constraints, upper / lower limit constraints of distributed power generation output, and compensation amount constraints of reactive power compensation devices.

[0158] Establish a set of static rules and parameter tables to guide the initial scheduling behavior:

[0159] Table 1 Parameter Configuration Table

[0160]

[0161] Initial scheduling rule modeling forms the "skeleton" of the intelligent scheduling system. It formalizes various resources, processes, and constraints in the mine mathematically, establishing a basic scheduling system. Based on this, the system can subsequently integrate reinforcement learning, adaptive optimization, and other methods for dynamic strategy upgrades. The scheduling rules themselves can also serve as an important point of patent protection, especially when the rules are embedded in specific scenarios (such as coal bunker protection and gradient load conveying), where they possess uniqueness.

[0162] Step 2.3 Deployment and operation of collaborative scheduling algorithm: Apply AI algorithms such as multi-agent, distributed optimization, and perturbation rescheduling to dynamically generate optimal or suboptimal scheduling strategies based on real-time status information.

[0163] Based on the initial modeling and parameter configuration of the scheduling model, an AI scheduling algorithm is introduced. Based on real-time perception data and dynamic system status, tasks, resources and paths are calculated and optimized in real time to dynamically generate globally optimal or suboptimal scheduling schemes, thereby achieving efficient collaborative operation of all aspects of the mine.

[0164] In the deployment and operation of collaborative scheduling algorithms, the main technical approach adopted is based on a multi-agent system architecture. This abstracts each operational unit in the mine (such as the coal face, conveyor section, and coal bunker) into an independent agent. Each agent possesses perception, decision-making, and execution capabilities, can autonomously formulate local scheduling strategies, and achieve global coordination through message passing. Building upon this, a distributed optimization algorithm is introduced, enabling each subsystem to achieve resource coordination and goal consistency through negotiation, information sharing, and local computation without relying on a central controller. This approach is particularly suitable for large-scale, complex, and coupled mine systems. Furthermore, when the system experiences disturbances such as equipment failures, unforeseen events, or resource conflicts, a disturbance rescheduling algorithm is used to quickly correct and partially reconstruct the plan, maintaining the overall system's operational continuity and scheduling stability. This collaborative technical approach ensures that the mine scheduling system possesses intelligence, flexibility, and high robustness, providing a solid foundation for the adaptive operation of intelligent mines.

[0165] In smart mines, multiple subsystems are highly coupled, and centralized scheduling can lead to problems such as computational bottlenecks (computational complexity increases exponentially with scale), large communication latency (especially in complex mine networks), and weak fault tolerance. To address these issues, distributed scheduling algorithms adopt a "local computation + local communication + global coordination" model, where each participating node optimizes itself and achieves near-global optimal scheduling through iterative negotiation with neighboring nodes.

[0166] The distributed Lagrange relaxation method is applicable to problems with coupling constraints, and its mathematical model is as follows:

[0167]

[0168] x i ∈X i

[0169] In the formula f i (x i Let A be the local objective function of the i-th subsystem (e.g., output, energy consumption); i x i This indicates the usage of globally coupled resources (such as the total capacity of the conveyor belt); b is the upper limit of the total resources.

[0170] Introducing the multiplier λ:

[0171]

[0172] Each sub - problem can be solved distributively:

[0173]

[0174] Update the multiplier through the dual gradient method:

[0175]

[0176] To implement each coal mining face / dispatching unit, only local resources need to be considered, and there is no need to access global variables.

[0177] Step 3, perform control and dynamic coordination: When an anomaly occurs, the system issues a warning to assist dispatchers or make an automatic response. Dispatch instructions are automatically sent to each equipment system to achieve linkage execution. Synchronously control the inflow and outflow of each buffer coal bunker, coordinate production and transportation matching, and ensure the continuous and efficient operation of the system

[0178] Step 3.1 Auxiliary decision - making and early - warning response mechanism: When an anomaly or disturbance event (such as equipment failure, full bunker and interrupted flow) is detected, the platform provides real - time early - warning and auxiliary decision - making solutions for dispatchers to refer to or make an automatic response.

[0179] In the intelligent mine collaborative dispatching system, the core objective of the auxiliary decision - making and early - warning response mechanism is that when the system detects an anomaly or disturbance event (such as equipment failure, full coal bunker, gas over - limit, transportation interruption, etc.), it can achieve automatic identification, risk assessment, and provide scientific and reasonable response suggestions. As the "early - warning center" of intelligent dispatching, this mechanism aims to enhance the self - diagnosis, self - recovery, and emergency handling capabilities of the system, and ensure the continuity and safety of the production process. The system not only supports providing early - warning prompts and auxiliary decision - making solutions to dispatchers, but also can automatically trigger a linkage response under set conditions, realizing an operation mode of "second - level identification - rapid response".

[0180] Its functional process includes five main links: First, the acquisition system monitors key parameters in real - time and makes a preliminary anomaly judgment through preset thresholds or models; Second, the system classifies and analyzes the abnormal state, identifies the risk level and possible conduction paths; Subsequently, generate a feasible emergency dispatching plan through the built - in rule base, expert experience, or AI algorithm; Then, the system submits the decision - making suggestions to the dispatcher for confirmation, or directly triggers a response instruction in scenarios without manual intervention; Finally, execute the strategy through the linkage control system (such as shutdown, path switching, load reduction operation), and feedback the execution results for model optimization, realizing response operations such as speed regulation, stop mining, path switching, early - warning broadcast, etc., and continuously obtaining execution status data through the feedback module for closed - loop optimization and model correction.

[0181] Step 3.2 Automatic execution of intelligent scheduling plan: The optimized scheduling instructions are automatically sent to each production system, and the control system realizes equipment linkage and production process adjustment to maintain efficient and coordinated operation of the system.

[0182] The intelligent scheduling plan's automatic execution mechanism transforms the scheduling optimization results generated in the previous step into specific control commands, which are then automatically distributed to various production systems in the mine, achieving closed-loop implementation of the scheduling plan. Based on the system-level scheduling plan, and considering the operating status, interface protocols, and execution capabilities of each subsystem, this mechanism logically decomposes scheduling commands into several executable operation instructions. These instructions are then sent to relevant equipment control systems via the scheduling control bus or industrial protocol communication link, including but not limited to operations such as starting and stopping coal mining equipment, adjusting conveyor operating status, controlling coal bunker unloading, and scheduling the hoisting system.

[0183] To ensure the accuracy and coordination of the execution process, this mechanism performs multi-dimensional status verification before issuing instructions to ensure that execution conditions are met and to avoid equipment conflicts or process interruptions. After the instructions are issued, the system collects execution status information through a real-time feedback mechanism, including execution success indicators, equipment response times, and abnormal interruption feedback, for use in monitoring the status of the scheduling plan execution and troubleshooting. This automatic execution module supports cascaded execution of plans and task priority control, and can coordinate strategies when scheduling instructions conflict or resource contention occurs, ensuring stable and efficient collaborative operation of the entire mining production system across different stages. Its core mathematical model is a discrete event system model.

[0184] Discrete event system model: The entire execution process can be abstracted as a discrete event system, representing the execution triggering of scheduling actions (such as "starting conveyor A" or "pausing coal mining machine B") under specific state conditions.

[0185] Set of states: S = {s0, s1, ... s} n} indicates the current operating status of the system;

[0186] Event set: E = {e1, e2, ..., e} n} represents a scheduling instruction event.

[0187] Transition state function: δ: S×E→S

[0188] s t+1 =δ(s) t ,e t If and only if gurand(s) t ,e t ) = True

[0189] Condition-triggered, this model can realize functions such as determining the timing of instruction execution, intercepting conflict states, and making linkage judgments.

[0190] Step 3.3 Buffering and Coordination Control of Key Links: Accurately manage the capacity and scheduling time window of key buffer links such as the bottom mine bunker and raw coal bunker to avoid problems such as "full-bundle shutdown" and "flow interruption and idling".

[0191] In the continuous production process of intelligent mines, there is a significant mismatch in speed and rhythm among underground coal mining, transportation, hoisting, and surface washing. To address this, the system sets up multiple "buffer zones" (such as bottom coal bunkers, surface raw coal bunkers, and commercial coal bunkers) to "break the coupling" and regulate the flow and rhythm between upstream and downstream processes.

[0192] The goals of this step are: to accurately manage the capacity, inbound rate, and outbound rate of these buffer warehouses; to consider constraints such as "stopping mining when the warehouse is full" and "stopping operation when the warehouse is empty" during scheduling; to optimize the "scheduling time window" to ensure coordinated operation of the mining, transportation, and dispatching systems; and to maintain stable system operation when fluctuations or disturbances occur, avoiding resource congestion or waste.

[0193] Dynamic material flow model: As a typical input-output system, the state changes of the ore bin can be represented by the following model:

[0194] Changes in warehouse inventory:

[0195]

[0196] In the formula, C(t) represents the amount of ore in the bin at time t (tons); Q in (t) represents the flow rate (tons / h) entering the ore bin; Q out (t) represents the outflow of ore (tons / h).

[0197] Boundary conditions:

[0198] 0≤C(t)≤C max

[0199] Where C max The model is used to determine the "full warehouse", "empty warehouse" and "safety stock zone" as the upper limit of the mine warehouse capacity, thereby triggering scheduling adjustments.

[0200] Predictive capacity management models:

[0201]

[0202] When C(t+Δt>C max This could trigger a mining slowdown or suspension of the plan;

[0203] When C(t+Δt)<C min This can trigger an upgrade plan to optimize mining or advance its mining schedule.

[0204] This model allows the scheduling system to plan "scheduling time windows" in advance.

[0205] Multi-buffered collaborative optimization model:

[0206]

[0207] The goal is to find the optimal target capacity for each buffer point, achieve dynamic balance, and avoid bottleneck propagation.

[0208] "Critical Link Buffering and Coordination Control," built upon the mathematical foundations of dynamic flow modeling, multi-level linkage optimization, and predictive scheduling strategies, is a key component enabling mine scheduling systems to possess "system-level flexible coordination capabilities." It not only improves the system's stability under dynamic fluctuations but also provides the necessary real-world constraints and response channels for intelligent scheduling algorithms.

[0209] Step 4, Feedback Learning and Optimization Closed Loop: Scheduling execution results and operational status are continuously fed back for evaluation and analysis. The feedback data is used to continuously train and optimize the scheduling model, enhancing its adaptive capabilities. A closed-loop control system covering the entire process is constructed to achieve continuous intelligent optimization of mine scheduling.

[0210] Step 4.1 Continuous Feedback and Learning of System Operation Status: Scheduling results and equipment operation status are continuously fed back to the platform to evaluate the scheduling effect and serve as training data for model iteration, enhancing the system's adaptability.

[0211] After executing scheduling instructions, the system's operational results, including changes in equipment status, task completion status, abnormal alarm information, and energy consumption indicators, are transmitted back to the scheduling platform in real time. This feedback data is not only used to evaluate scheduling effectiveness (e.g., whether tasks are completed on time or if there is overload), but also serves as "retraining data" for the scheduling model, continuously improving parameters and optimizing strategies, enabling the system to gradually develop self-learning, self-adjustment, and self-optimization capabilities. "Continuous feedback and learning of system operating status" is a key step in realizing the closed-loop control, autonomous optimization, and evolution capabilities of the intelligent scheduling system. It upgrades the scheduling system from a "static executor" to an "intelligent learning entity," enabling it to continuously accumulate experience, adapt to new scenarios, and optimize future decisions. This is the core technology supporting the transition of intelligent mines from "automation" to "brain-like scheduling."

[0212] Construction of the CMIES Interval Optimization Scheduling Model: CMIES Economic Operating Cost F1(P,P) I This includes the CSP's output electrical and thermal power and maintenance costs. and The cost of purchasing electricity for a large power grid (C) ex,t The cost of storing batteries (C) es,t Cost of output heat power of associated energy and renewable energy devices C bsny,t

[0213]

[0214] C es,t =α es (P dis,t +P cha,t )

[0215]

[0216] Where: T represents the scheduling duration; and These represent the operation and maintenance cost coefficients for CSP power supply and heating, respectively; α es c represents the maintenance cost coefficient of an energy storage device. ex，t The main grid purchase price is represented by α; the output power of the exhaust gas oxidation unit, water source heat pump, air source heat pump, and ground source heat pump are represented by α respectively. i P represents the corresponding cost coefficient. csp,t T represents the output electrical power of the CSP generator; dis2,t This indicates the power of the CSP thermal storage device to supply the heat load; P cha,t and P dis,t These represent the charging and discharging power of the energy storage device, respectively.

[0217] Step 4.2 Dynamic Optimization and Global Collaborative Closed-Loop Control: Realize a dynamic closed-loop collaborative scheduling system from perception to execution and from feedback to optimization, supporting the intelligent, refined, and highly reliable operation of the mine's main production system.

[0218] By connecting all key links in the scheduling system, forming a continuous closed loop from state awareness to intelligent decision-making, from automatic execution to feedback learning, from model updates to scheduling optimization, it can not only handle static complex tasks but also continuously optimize its strategies in the face of disturbances and changes. This enables decision support and emergency control for sudden events, such as... Figure 6 As shown in the figure. Under this mechanism, the mine scheduling system has the capabilities of full state awareness, dynamic strategy generation, execution system linkage, operation feedback collection and analysis, and optimization iteration.

[0219] Dynamic optimization model:

[0220]

[0221] stx(t+1)=f(x(t),u(t)),x(0)=x0

[0222] x(t)∈χ,u(t)∈μ

[0223] In the formula, γ(x,u) is the cost function of the system in a certain state (such as energy consumption, waiting, default risk); χ and μ are the feasible regions of the state and action.

[0224] Global collaborative optimization model:

[0225]

[0226] In the formula x i (t) represents the state of the i-th subsystem; The desired state; c ij For coupling and coordination weights.

[0227] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "preferred embodiment," "detailed description," or "preferred embodiment," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0228] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended 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 or all of the technical features therein. Therefore, these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A production collaborative scheduling method based on an intelligent mine integrated management and control platform, characterized in that, include: S1. Collect equipment status, environmental parameters and production progress data of each production stage in the mine, process them through edge computing nodes and then upload them to the cloud platform; S2. Construct a digital twin mine model, configure initial production scheduling rules and resource constraints, analyze real-time data, and generate an intelligent collaborative scheduling scheme. S3. When an abnormal or disturbance event is detected, the system issues an early warning and responds automatically, sending dispatch instructions to each equipment system to achieve coordinated execution, synchronously controlling the inflow and outflow of the buffer coal bunker, and coordinating production and transportation matching. S4. Continuously transmit the scheduling execution results and running status back to the platform for evaluation and analysis. Use the feedback data to train and optimize the scheduling model to enhance its adaptive capabilities.

2. The production collaborative scheduling method according to claim 1, characterized in that, S1 specifically includes: S1.1 Real-time acquisition of equipment status data, environmental parameter data, and production progress data; the acquisition process uses a single-sensor data acquisition model to correct measurement errors and a multi-sensor data fusion model to achieve redundancy verification. The equipment status data includes coal mining machine speed, conveyor current, and hoist temperature; environmental parameter data includes gas concentration, wind speed, and roadway displacement; and production progress data includes instantaneous output, conveyor belt coal flow rate, and washing and processing capacity. The single-sensor data acquisition model is as follows: x i (t,s)=f i (t,s)+ε i (t,s) In the formula, x i (t,s) represents the raw values ​​collected by the i-th sensor at time t and spatial location s; f i (t,s) represents the target physical quantity, ε i (t,s) represents the sensor measurement error; For multiple sensor data of the same physical quantity, they need to be fused according to their reliability. The multi-sensor data fusion model is as follows: In the formula: x is the estimated value after fusion. i (t) represents the value collected by the i-th sensor, w i The sensor reliability weight is set based on the sensor's historical error rate and signal-to-noise ratio. S1.2, the fused data output from S1.1 Perform unified processing to output a standardized dataset; The unification process includes data cleaning and data standardization. Data cleaning includes filling in default values ​​using linear interpolation and removing outliers using the Z-Score method. Data standardization includes dimensional normalization and aligning data timestamps of different sampling frequencies based on NTP clock synchronization. S1.3 Deploy edge computing nodes at key locations to process the standardized dataset output from S1.2, including data filtering, data compression, and event detection; output the processed dataset D_edge. Event detection uses an edge rule engine to detect predefined events, including: Equipment overload: Triggered when current I > I_max threshold, where I_max represents the maximum current threshold; Coal bunker full: Triggered when bunker L > 90% capacity; Abnormal vibration: Triggered when vibration amplitude A > A_threshold, where A_threshold represents the vibration amplitude threshold; S1.4 Upload the D_edge data output from S1.3 to the cloud platform to form a production information database covering the entire mine.

3. The production collaborative scheduling method according to claim 2, characterized in that, In S1.3, the deployment of key locations specifically includes: Mining area: coal mining machine body, hydraulic support controller; Transportation system: Weighing terminals at the main conveyor belt inlet and coal bunker inlet / outlet; Lifting component: PLC terminal in the hoist control room; Coal washing and beneficiation process: Control cabinet of the heavy media separator in the coal preparation plant; Environmental monitoring: Methane sensors in ventilation zones, temperature and humidity monitoring stations in return air tunnels.

4. The production collaborative scheduling method according to claim 1, characterized in that, S2 specifically includes: S2.1 Construct a digital twin mine model that maps one-to-one with the actual mine production system; S2.2 Configure initial scheduling rules: including power supply system constraint rules and production equipment scheduling rules; S2.

3. Transform the production units mapped in S2.1 into intelligent agents. Using the constraints in S2.2 as boundaries, dynamically generate scheduling strategies through distributed optimization algorithms and trigger rescheduling when parameters exceed limits.

5. The production collaborative scheduling method according to claim 4, characterized in that: S2.1 specifically includes: Based on the real-time data uploaded by S1, a virtual mapping is established that includes the following modules: Spatial structure mapping module: performs 3D modeling of roadways, mining areas, coal bunkers, and shafts to construct a virtual topological framework for the static physical environment of the mine; Equipment Status Mapping Module: Dynamically binds the real-time operating parameters of the coal mining machine, conveyor, and hoist to the virtual model. The parameters include: the rotation speed, vibration amplitude, and current intensity of the coal mining machine; the operating speed, motor temperature, and load current of the conveyor; and the hoist's hoisting speed, braking status, and wire rope tension. Material flow modeling module: Establishes a continuous dynamic model based on the input and output quantities of coal flow and / or water flow, with the dynamic equation being: Q out (t)=min(Q in (t),C max -C(t)); where Q in (t) represents the input amount per unit time; Q out (t) represents the output per unit time; C(t) represents the current inventory; C max Maximum capacity; The spatial structure mapping module provides a spatial topology reference for the equipment state mapping module, and the operating parameters output by the equipment state mapping module drive the dynamic calculations of the material flow modeling module. The three are coupled through a system state space model, and the coupled system state space equation is as follows: x(t+1)=Ax(t)+Bu(t) y(t)=Cx(t)+Du(t) In the formula, x(t) represents the system state, which includes three dimensions of data: spatial topological coordinates, equipment operating parameters, and material inventory; u(t) represents the input variable; y(t) represents the observed output; and A, B, C, and D represent the system dynamics matrices.

6. The production collaborative scheduling method according to claim 5, characterized in that: Specifically, S2.2 includes: setting power supply system constraint rules and production equipment scheduling rules based on the model parameters constructed in S2.1; In the power supply system constraint rules, the mine power supply system must satisfy the following power constraint equation: In the formula: U i,t U j,t The voltages at nodes i and j at time t are respectively; r ij x ij Let P be the resistance and reactance of line ij; ij,t Q ij,t These represent the active and reactive power at the beginning of line ij at time t, respectively; I ij,t P is the current flowing through line ij at time t; j,t Q j,t Φ represents the active and reactive power injected into node j at time t, respectively; j Let j be the set of child nodes with node j as the parent node; and the power constraints also include node voltage constraints, distributed generation output constraints, and reactive power compensation constraints. The production equipment scheduling rules are shown in Table 1 below; 7. The production collaborative scheduling method according to claim 6, characterized in that: Specifically, S2.3 includes: Using the state-space equations of S2.1 as the optimization object, and under the constraints of S2.2, perform the following: S2.3.1 Based on the equipment state mapping and material flow model established in S2.1, the coal mining machine, conveyor and coal bunker unit are transformed into intelligent agents. The decision variables of each intelligent agent are limited by the equipment constraints in Table 1 of S2.

2. S2.3.

2. Using the power constraint equations of S2.2 as global constraints, perform iterative optimization: (1) Each agent solves the local optimization problem according to the following formula: x i ∈X i In the formula, f i (x i Let A be the local objective function of the i-th subsystem; i x i This represents the usage of globally coupled resources; b is the total resource limit; and x i The speed regulation range specified in Table 1 of S2.2 must be met; (2) The central coordinator updates the Lagrange multipliers: S2.3.3 When the S2.1 model detects that the parameters exceed the threshold set in S2.2, it triggers local re-optimization of the affected agent.

8. The production collaborative scheduling method according to claim 1, characterized in that, S3 specifically includes: S3.1 Provide real-time early warning and decision support solutions when abnormal or disturbing events are detected; S3.2 Automatically send the optimized scheduling instructions to each production system to achieve equipment linkage and production process adjustment; S3.3 Manage the capacity and scheduling time window of the buffer coal bunker to avoid problems such as full-bunker shutdown and idling due to power outage.

9. The production collaborative scheduling method according to claim 1, characterized in that, S4 specifically includes: S4.1, performing feedback evaluation: Real-time transmission of equipment status changes, task completion status, abnormal alarm information, and energy consumption indicators after scheduling execution back to the platform for: Evaluate the scheduling effectiveness and analyze the degree of matching between equipment operating status and expected goals; And as training data for the model, it is used to optimize the parameters of the digital twin model of S2.1 and the scheduling algorithm of S2.3; S4.2 Model Optimization: Construct a closed-loop control system from state awareness to scheduling optimization, and achieve continuous optimization through the following methods: When the device's operating status is detected to deviate from expectations, the constraint threshold in S2.2 is automatically adjusted; Based on historical scheduling data, optimize the allocation of decision weights for agents in S2.3; For frequently occurring abnormal patterns, update the device state mapping relationship in the S2.1 model.

10. The production collaborative scheduling method according to claim 9, characterized in that, In S4.1, the scheduling economy is evaluated using the following model: C es,t =α es (P dis,t +P cha,t ) Where: T represents the scheduling duration; and These represent the operation and maintenance cost coefficients for CSP power supply and heating, respectively; α es c represents the maintenance cost coefficient of an energy storage device. ex,t The main grid purchase price is represented by α; the output power of the exhaust gas oxidation unit, water source heat pump, air source heat pump, and ground source heat pump are represented by α respectively. i P represents the corresponding cost coefficient. csp,t T represents the output electrical power of the CSP generator; dis2,t This indicates the power of the CSP thermal storage device to supply the heat load; P cha,t and P dis,t These represent the charging and discharging power of the energy storage device, respectively. S4.2 achieves closed-loop control through the following optimization model: S4.2.

1. Based on the digital twin model state data x(t) in S2.1 and the constraints Ω in S2.2, solve for the optimal scheduling instruction u(t): stx(t+1)=f(x(t),u(t)),x(0)=x0 x(t)∈χ,u(t)∈μ In the formula, γ(x,u) is the cost function, which comprehensively considers equipment energy consumption and production efficiency; χ and μ are the feasible regions of state and action. S4.2.2 For the multi-agent system in S2.3, global resource allocation is coordinated using the following formula: Where: x i (t) represents the current state of the i-th agent; The desired state thresholds set for Table 1; c ij For coupling and coordination weights.

Images