Coal mine hydrological dynamic monitoring system based on Internet of Things and monitoring method thereof

Abstract

The invention relates to a coal mine hydrological dynamic monitoring system based on the Internet of Things and a monitoring method thereof, and belongs to the technical field of coal mine safety monitoring. The coal mine hydrological dynamic monitoring system comprises a sensing execution layer, a network transmission layer and an application platform layer; the sensor group is arranged in an underground coal mine excavation working face, a roadway, a hydrological observation hole and a ground observation hole and is used for collecting multi-source hydrological and geological stress data, and the execution unit is used for receiving a control instruction and executing linkage control action. According to the method, a complete water disaster risk assessment evidence chain is formed through multi-parameter cooperative monitoring of the water source, the water pressure, the water volume and the channel, the accuracy and reliability of water inrush symptom recognition are remarkably improved, misjudgment caused by information distortion in a single aspect is avoided, a heterogeneous network system suitable for the complex environment of the underground coal mine is constructed, and the method is suitable for popularization and application. The problems of incomplete coverage and poor reliability of a single communication mode are solved.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety monitoring technology, and in particular to a coal mine hydrological dynamic monitoring system and monitoring method based on the Internet of Things. Background Technology

[0002] Mine water hazards are one of the major disasters threatening the safe production of coal mines. Traditional coal mine hydrological monitoring mainly relies on manual collection of data at fixed times and locations, which has disadvantages such as high labor intensity, long monitoring cycle, discontinuous data, and delayed early warning, making it difficult to achieve early detection and rapid response to signs of water inrush.

[0003] With the development of IoT and AI technologies, various technical solutions have emerged in the field of coal mine hydrological monitoring, but each has its own limitations:

[0004] A search revealed that Chinese patent application number CN202510655734.1 discloses an intelligent safety control system and method for unmanned hydropower stations. This system adopts an architecture of perception layer, network layer, platform layer, and application layer. The network layer uses 5G+LoRa dynamic fusion networking. However, this solution is mainly designed for hydropower station environments, and its monitoring objects include equipment operating parameters, dam structural status, reservoir hydrological environment, etc., which are fundamentally different from the complex geological conditions and disaster mechanisms in coal mines. In particular, this solution does not involve key elements unique to coal mines, such as roof and floor stress monitoring and water-conducting fracture zone assessment. Its communication architecture also does not consider the special transmission environment of coal mine roadways, such as multipath effects and severe signal attenuation, and cannot be directly applied to dynamic hydrological monitoring in coal mines.

[0005] A search revealed that Chinese patent application number CN202410774329.7 discloses an intelligent prediction method and early warning system for coal mine water hazards based on artificial intelligence. This scheme predicts the water inflow at the working face by establishing a prediction model for water hazards in old mines, and has a certain predictive capability in specific scenarios. However, the monitoring elements of this scheme are relatively singular, mainly focusing on monitoring data of water in old mines. It lacks synchronous monitoring and integrated analysis of key parameters such as dynamic water level and pressure of multiple aquifers, changes in multiple water quality parameters, and stress and strain of rock mass, making it difficult to achieve a comprehensive water hazard risk assessment. In addition, this scheme lacks specific design in terms of communication network architecture and does not solve the problem of reliability of data transmission in underground coal mines.

[0006] A search revealed a Chinese patent with application number CN202111394917.0, which discloses an intelligent early warning system for coal mine water hazards. This system includes a data management component, a processing strategy selection component, and an early warning calculation component. It can collect, transmit, classify, and store large amounts of water hazard data. However, this solution is rather general in terms of data processing and analysis, lacking specific details on intelligent algorithm implementation, particularly a time-series prediction model based on deep learning. Furthermore, the system primarily focuses on early warning analysis and fails to form a complete closed loop from monitoring and early warning to coordinated control, resulting in significant shortcomings in emergency response efficiency.

[0007] A comprehensive analysis of existing technologies reveals the following pressing technical problems that still need to be addressed in current coal mine hydrological monitoring systems: 1. Insufficient monitoring elements and comprehensive analysis: Existing schemes mostly focus on monitoring a single or a few parameters, lacking comprehensive perception and collaborative analysis of multiple parameters such as water source (water quality), water quantity (water inflow), water pressure (water level), and channel (stress), making it difficult to construct a complete evidence chain for water hazard risk assessment; 2. Insufficient communication reliability: The underground environment of coal mines is complex. Existing communication solutions often use a single wireless or wired transmission method, which is difficult to adapt to complex scenarios such as changes in roadway structure and equipment movement, and cannot guarantee the continuity and integrity of monitoring data. 3. Limited accuracy of early warning models: Existing early warning models are mostly based on static thresholds or traditional statistical methods, lacking deep learning and accurate prediction of the dynamic changes in hydrology, making it difficult to achieve early warning of sudden water inrush risks; 4. Lack of closed-loop control in the system: Most systems remain at the monitoring and early warning stage, failing to achieve intelligent linkage between early warning and control, and thus failing to form a complete "perception-analysis-decision-control" closed loop, resulting in low emergency response efficiency; 5. Insufficient cross-domain technology integration: Existing solutions often focus on the application of a certain technical field, failing to deeply integrate and optimize technologies such as heterogeneous network communication, multi-source data fusion, deep learning prediction and intelligent control execution; In particular, although existing technologies have developed solutions that utilize multiple communication technologies or machine learning algorithms, they lack in-depth integration and systematic optimization tailored to the specific needs of coal mine hydrological monitoring. General heterogeneous network solutions do not consider the unique impact of underground coal mine roadway structures on wireless signals and lack network adaptation mechanisms for different hydrological monitoring parameters. Single prediction models are insufficient to effectively handle multi-source heterogeneous data and their spatiotemporal correlation characteristics in coal mine hydrology, resulting in limited early warning accuracy. Most systems fail to achieve a complete closed loop from risk prediction to automatic control, and there is a lack of effective collaboration among various technical modules.

[0008] Therefore, based on a thorough analysis of the shortcomings of existing technologies, this invention proposes a coal mine hydrological dynamic monitoring system and method that deeply integrates the Internet of Things, heterogeneous network communication, multi-source data fusion, deep learning, and intelligent control, aiming to solve the aforementioned technical problems. Summary of the Invention

[0009] This invention provides a dynamic monitoring system and method for coal mine hydrology based on the Internet of Things, which solves the problems of single monitoring elements, poor communication reliability, low level of intelligence and low system integration in existing coal mine hydrological monitoring technologies.

[0010] The solution to the above-mentioned technical problems of the present invention is as follows: a dynamic monitoring system for coal mine hydrology based on the Internet of Things, comprising a sensing and execution layer, a network transmission layer, and an application platform layer. The sensing and execution layer includes a sensor group and an execution unit. The sensor group is deployed in the underground mining face, roadway, hydrological observation hole, and surface observation hole of the coal mine to collect multi-source hydrological and geological stress data. The execution unit is used to receive control commands and execute linkage control actions. The network transmission layer adopts a converged heterogeneous network communication architecture based on LoRa, Wi-Fi, 4G, and 5G, specifically including: a. an underground access subnet consisting of LoRa gateways and Wi-Fi access points deployed in roadways and mining faces, wherein the LoRa gateways and Wi-Fi access points adaptively select transmission protocols according to the priority and bandwidth requirements of monitoring data; b. an industrial ring network serving as the underground backbone network, connected to the underground access subnet through a protocol converter to achieve data conversion and aggregation between different network protocols; c. an aboveground data transmission network using 4G / 5G mobile communication or fiber optic networks. The application platform layer includes a data middleware, a hydrological big data analysis engine, and a visualization early warning and control module. The hydrological big data analysis engine has a built-in hydrological dynamic early warning model. This model integrates real-time multi-source monitoring data from sensor groups, historical hydrological data and three-dimensional geological model data. It uses the Long Short-Term Memory (LSTM) algorithm with multiple parameters such as mine water inflow, water level, water quality and roof and floor stress data as input feature vectors for time series prediction. It also combines the Support Vector Machine (SVM) algorithm to identify water inrush sources and output the probability of water hazard risk. The visualization early warning control module is used to generate and release four levels of flood warning information (blue, yellow, orange, and red) based on the probability of flood risk and preset multi-level thresholds. It can also automatically or manually issue linkage control commands to the execution unit of the perception execution layer according to the warning level, forming a complete closed loop from monitoring to control.

[0011] Based on the above technical solution, the present invention can be further improved as follows.

[0012] Furthermore, the sensor group includes a water level and pressure sensor, a flow sensor, a multi-parameter water quality sensor, and a stress and strain sensor. The water level and pressure sensor monitors the dynamic water level and pressure in each aquifer and borehole. The flow sensor uses an ultrasonic or electromagnetic flowmeter to measure the total mine inflow and the inflow in different zones. The multi-parameter water quality sensor monitors the pH value, conductivity, temperature, and calcium and magnesium ion concentrations of the water in real time. The stress and strain sensor is buried in the roof and floor of the coal seam to monitor the deformation and stress changes of the rock mass under the influence of mining. Through multi-parameter collaborative monitoring, a complete evidence chain for water hazard risk assessment is constructed. The four sensors provide complementary data from four dimensions: water source energy, water volume change, water source type, and water inrush channel, which significantly improves the accuracy and reliability of water inrush sign identification and avoids misjudgment caused by distortion of information from a single aspect.

[0013] Furthermore, the execution unit includes a controllable drainage valve and an audible and visual alarm, used to receive instructions from the visual early warning control module and execute emergency actions such as starting the drainage pump, issuing an audible and visual alarm, or closing the waterproof gate. This establishes a direct link from early warning to control, realizing true closed-loop management. When an early warning is issued, emergency measures can be quickly activated to prevent the water tank from being flooded, buying valuable time for personnel evacuation and disaster control, and greatly improving the mine's proactive disaster prevention capabilities.

[0014] Furthermore, the LSTM prediction sub-model in the hydrological dynamic early warning model adopts a three-layer LSTM network stacked structure. The input data is the time series of multi-source monitoring data from the past 7 days, and the output is the predicted values ​​of water inflow and water level for the next 6 hours. The water source identification sub-model identifies the type of sudden water source by calculating the Euclidean distance between real-time water quality data and each standard vector in the preset water quality fingerprint database. The combination of time series prediction and water source identification considers both the trend of water volume change and the decisive characteristics of water source type, making the early warning judgment more comprehensive and accurate.

[0015] Furthermore, the hydrological dynamic early warning model also includes a decision fusion module, which integrates LSTM prediction results and water source identification results to trigger corresponding levels of early warning according to preset rules: when the LSTM predicts that the inflow volume will increase by more than 50% in the next 3 hours and the probability of the water source identification result being Ordos Grey Water is greater than 80%, an orange warning is directly triggered; when only one of the conditions is met, a yellow warning is triggered. Through quantitative decision rules, the arbitrariness of subjective judgment is reduced, and the standardization and consistency of early warning are improved. The multi-condition combination early warning mechanism considers both the severity of the risk and avoids false alarms caused by single indicator anomalies, making the early warning release more scientific and reasonable.

[0016] Furthermore, the visualization early warning control module dynamically displays the real-time status, historical data curves, and early warning information of all monitoring points on GIS and 3D geological models. When an early warning is triggered, the system automatically notifies relevant personnel via SMS and App push notifications, and highlights the risk area on the 3D visualization interface. This transforms abstract monitoring data into intuitive and three-dimensional graphic information, enabling managers to quickly grasp the overall hydrological situation and rapidly locate risk sources. The highlighted risk area in the 3D environment greatly improves situational awareness and decision-making efficiency.

[0017] A method for dynamic monitoring of hydrology in coal mines based on the Internet of Things includes the following steps: S1, through various sensors in the perception and execution layer, hydrological and geological stress data of the coal mine shaft are automatically collected at a preset frequency, wherein the water level sensor collects data once every 5 minutes, the flow meter collects data once every 1 minute, the water quality sensor collects data once every 30 minutes, and the stress sensor collects data once every 10 minutes. S2, the collected data is transmitted in real time to the application platform layer through the converged heterogeneous network in the network transmission layer. The LoRa node sends the sensor data to the nearest gateway, the Wi-Fi AP receives a large amount of information, and all data is aggregated to the wellhead through the industrial ring network and transmitted to the application platform through the ground network. S3, the application platform layer cleans, integrates and standardizes the received data, uses the isolated forest algorithm to identify abnormal data, uses the Z-Score method to unify the data units, and uses the hydrological big data analysis engine to fuse and analyze multi-source monitoring data, historical hydrological data and geological model data. S4, based on the hydrological dynamic early warning model, uses the LSTM algorithm with multiple parameters such as mine water inflow, water level, water quality and roof and floor stress data as input feature vectors to predict water hazard risk, and determines whether the current hydrological state exceeds the preset threshold or has an abnormal change trend. If it exceeds the threshold, it generates and releases water hazard early warning information of the corresponding level. S5. Based on the warning level, the corresponding emergency plan is automatically or manually activated, and the execution unit is used for linkage control. In the case of a yellow warning, the area drainage pump is automatically started; in the case of an orange warning, the waterproof gate is prepared to be closed; and in the case of a red warning, all drainage capacity is activated and an evacuation alarm is issued.

[0018] Based on the above technical solution, the present invention can be further improved as follows.

[0019] Furthermore, in step S4, the hydrological dynamic early warning model adopts a time series prediction algorithm based on Long Short-Term Memory (LSTM) network to predict mine water inflow and water level changes, and combines abrupt changes in water quality characteristics to comprehensively identify water inrush sources. Through advanced data preprocessing methods, the data quality is effectively improved, providing a reliable data foundation for subsequent model analysis. The algorithm selection considers both processing effect and computational efficiency, making it suitable for the application requirements of real-time monitoring scenarios.

[0020] Furthermore, in step S4, the warning information includes four levels: blue warning, yellow warning, orange warning, and red warning. Different levels correspond to different information notification scopes, emergency response procedures, and linkage control instructions. The tiered response mechanism ensures rapid handling in high-risk situations and avoids unnecessary production interruptions in low-risk situations, reflecting the concept of intelligent emergency management.

[0021] Furthermore, the hydrological dynamic early warning model is periodically retrained using new monitoring data and the early warning threshold is dynamically adjusted to achieve continuous optimization of the model. This enables the system to have the ability to learn and continuously optimize itself, and can continuously improve the prediction accuracy as monitoring data accumulates, adapt to changes in hydrogeological conditions during mining, and ensure the long-term effectiveness and reliability of the system.

[0022] The beneficial effects of this invention are as follows: This invention provides a dynamic monitoring system and method for coal mine hydrology based on the Internet of Things, which has the following advantages: 1. Comprehensive perception and accurate judgment: Through multi-parameter collaborative monitoring of water source, water pressure, water volume and channel, a complete evidence chain for water hazard risk assessment is formed, which significantly improves the accuracy and reliability of identifying signs of sudden water flow and avoids misjudgment caused by the distortion of information from a single aspect. 2. Reliable transmission and coverage without blind spots: Through the deep integration of LoRa, Wi-Fi, and 4G / 5G, a heterogeneous network system adapted to the complex environment of underground coal mines has been constructed, solving the problems of incomplete coverage and poor reliability of single communication methods, and ensuring the continuous and stable transmission of monitoring data. 3. The LSTM algorithm, which uses multi-source hydrological data as feature vectors, not only enables time-series prediction of water inflow, but also combines water quality change characteristics for water source identification, significantly improving the accuracy and foresight of early warning. 4. A complete technology chain has been established, from multi-parameter perception, intelligent analysis, hierarchical early warning to automatic linkage, realizing proactive prevention and control of coal mine water hazards and significantly improving emergency response efficiency.

[0023] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0024] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0025] Figure 1 A flowchart illustrating a method for a dynamic monitoring system and monitoring method for hydrological conditions in coal mines based on the Internet of Things (IoT) is provided in an embodiment of the present invention.

[0026] Figure 2 This is a system architecture diagram of a coal mine hydrological dynamic monitoring system and monitoring method based on the Internet of Things, provided in an embodiment of the present invention.

[0027] Figure 3 The present invention provides a flowchart of a hydrological dynamic early warning model for a coal mine hydrological dynamic monitoring system and monitoring method based on the Internet of Things. Detailed Implementation

[0028] The following is in conjunction with the appendix Figure 1-3 The principles and features of the present invention are described below. The examples given are for illustrative purposes only and are not intended to limit the scope of the invention. The invention is described more specifically in the following paragraphs by way of example with reference to the accompanying drawings. The advantages and features of the invention will become clearer from the following description. It should be noted that the drawings are in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the invention.

[0029] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0030] 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 invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0031] like Figure 2-3 As shown, the present invention provides a coal mine hydrological dynamic monitoring system based on the Internet of Things. This system adopts a three-layer architecture design. 1. Sensing and execution layer: Multiple types of sensors and execution units are deployed at key locations in underground coal mines; Various types of sensors include: a. Water level and pressure sensors: installed in observation wells of each aquifer to monitor changes in hydrostatic pressure; b. Ultrasonic / electromagnetic flow meter: installed in the main water tank and mining area drainage ditch to accurately measure the water inflow; c. Multi-parameter water quality sensor: Deployed at major water inflow points to monitor pH, conductivity, temperature and characteristic ion concentration in real time; d. Stress and strain sensors: embedded in the top and bottom plates of the coal seam to monitor rock deformation and stress changes; The execution unit includes: a) a controllable drain valve: which automatically adjusts the drainage capacity upon receiving instructions; b) an audible and visual alarm: which provides on-site warning signals. 2. The network transport layer constructs a heterogeneous network deeply adapted to the coal mine environment, including: a. Underground access subnet: LoRa and Wi-Fi are deployed in a complementary manner. LoRa gateways are deployed at intervals of 500-1000 meters to cover the main tunnels; Wi-Fi APs provide high-speed access in fixed areas such as chambers and parking lots. b. Underground industrial ring network: adopts gigabit industrial Ethernet, with redundancy and self-healing capabilities, and serves as the data transmission backbone; c. Ground transmission network: It adopts a combination of 4G / 5G wireless and fiber optic wired to ensure the reliability of ground data transmission; Network scheduling mechanism: The system has a built-in intelligent scheduling algorithm that automatically selects the optimal transmission path based on data priority and network status. Key alarm data is prioritized for transmission via LoRa to ensure reachability, while large amounts of monitoring data are transmitted via Wi-Fi or wired network. 3. The application platform layer includes a data middle platform and an LSTM prediction sub-model; a. The data platform adopts a distributed architecture to clean, integrate, and standardize the storage of multi-source heterogeneous data. Specifically, this includes: data cleaning using the isolated forest algorithm to identify and remove abnormal data; data standardization using the Z-Score method to unify data units; and data storage to establish a unified hydrological data warehouse. b. The hydrological big data analysis engine employs a hydrological dynamic early warning model. Details of this technology are as follows: The LSTM prediction sub-model adopts a three-layer LSTM network structure. The input features include time series data such as water level, water inflow, stress and strain, and the output is the predicted value for the next 6 hours. The water source identification sub-model uses the SVM algorithm to match and identify water quality feature vectors with a preset water quality fingerprint database. The decision fusion module integrates the prediction and identification results and triggers corresponding level warnings according to preset rules. The visualization early warning and control module realizes the visualization of the hydrological situation of the entire mining area based on GIS and 3D geological models; early warning information is automatically released via SMS and App push; it supports the issuance of control commands to the execution unit to form closed-loop control. like Figure 1 As shown, the specific working principle and usage method of this invention are as follows: The coal mine with a mining depth of -650m was selected as the implementation scenario. It is threatened by both sandstone water in the roof and Ordovician limestone water in the floor, and has complex hydrogeological conditions. System deployment; Sensing and execution layer deployment: Precisely deploying sensor networks at key locations both on the surface and underground in the mining area. Deployment of water level and pressure sensors: One set of high-precision water level and pressure sensors is deployed in each of the surface Ordovician limestone observation holes (hole depth 280m) and underground sandstone aquifer observation holes (hole depth 80m). The sensors are fixed at the middle depth of the target aquifer, with a monitoring accuracy of ±0.5%FS. They are used to monitor changes in hydrostatic pressure in real time. The sensors adopt an IP68 protection level housing to adapt to the humid underground environment. Flow sensor deployment: One DN300 ultrasonic flow meter is installed at the inlet of the main water tank of the mine, one DN150 electromagnetic flow meter is installed at the inlet of each of the three mining area drainage ditches, and one DN100 ultrasonic flow meter is installed behind each of the two tunneling faces. The flow meter measurement accuracy reaches ±1.0%, and it has an automatic temperature compensation function. It is used to continuously monitor the dynamic changes of the total water inflow and the water inflow of each zone in the mine. Deployment of multi-parameter water quality sensors: A total of 8 sets of multi-parameter water quality sensors were installed near the main water inflow points and at the water tank inlet. The probes are made of titanium alloy and are directly immersed in the water flow to monitor the pH value (measurement range 0-14, accuracy ±0.1%), conductivity (measurement range 0-2000μS / cm, accuracy ±1%), temperature (measurement range 0-50℃, accuracy ±0.1℃), and calcium and magnesium ion concentrations (measurement range 0-500mg / L, accuracy ±2%) in real time. Stress and strain sensor deployment: 20 sets of stress and strain sensors are installed by drilling holes in the roof and floor of the coal seam in the coal mining face. The roof sensors are installed at different depths of 0.5m, 2m, and 5m from the roof of the coal seam, and the floor sensors are installed at different depths of 1m, 3m, and 8m from the floor of the coal seam. The sensors adopt fiber optic grating technology, which has strong anti-electromagnetic interference ability. They are used to accurately monitor the deformation and stress changes of the rock strata during mining activities and to assess the development height of the "water-conducting fracture zone". Deploying the network transport layer to build a three-tiered heterogeneous network communication system: Deployment of underground access subnet: In the main underground roadways, 18 LoRa gateways are deployed at intervals of 600 meters. The gateway's transmission power is adjustable (5-20dBm), and the receiving sensitivity reaches -148dBm. Industrial-grade Wi-Fi APs are deployed in 12 fixed areas such as pump rooms, substations, and vehicle yards. They support the 802.11ac protocol and provide gigabit wireless access. The gateways and APs are connected via mining flame-retardant communication cables. Industrial ring network deployment: The underground ring network is constructed using gigabit industrial Ethernet. The core switch is deployed in the central substation. LoRa gateways and Wi-Fi APs in each area are connected through 8 access switches. The ring network has MS-RPR fast self-healing function, and the fault switching time is less than 50ms. Deployment of the transmission network above the well: The wellhead equipment room is connected to the ground dispatch center through single-mode fiber with a backbone bandwidth of 10Gbps. It is also equipped with two industrial-grade 5G DTUs as hot backups and uses SIM cards from different operators to achieve dual-link redundancy. Application platform layer deployment, server cluster deployment at ground dispatch center: The data platform uses a distributed architecture consisting of three servers, each with 256GB of memory and 200TB of storage capacity, to receive and store monitoring data. The analytics engine server is equipped with two NVIDIA Tesla T4 GPU cards, which are dedicated to LSTM model training and inference. The visualization server is equipped with a high-performance graphics card, supporting real-time rendering of 3D geological models; At the software level, a digital twin object is created for each physical sensor and precisely bound to its location in the three-dimensional geological model. Historical hydrological data and geological structure information of the mining area over the past three years are entered, and the hydrological dynamic early warning model is initialized. S1, Data Acquisition and Transmission: After system initialization, the system enters automatic operation mode, and all sensors acquire data at preset frequencies. Water level sensor: Collects data every 5 minutes, with a data packet size of approximately 100 bytes; Flow meter: Collects data every 1 minute, with a data packet size of approximately 150 bytes; Water quality sensor: Collects data every 30 minutes, with a data packet size of approximately 200 bytes; Stress sensor: Data is collected every 10 minutes, with a data packet size of approximately 120 bytes; S2, data transmission adopts an intelligent scheduling mechanism: regular monitoring data is transmitted through the LoRa network, which has low power consumption and wide coverage. When the sensor detects a sudden change in parameters, it automatically switches to Wi-Fi transmission to ensure the real-time performance of critical data. Alarm data enjoys the highest priority and is transmitted through multiple paths simultaneously to ensure reliability. S3, Data Processing and Intelligent Analysis 3.1 Data preprocessing: After receiving the raw data, the platform first performs data quality control: Data cleaning: The isolated forest algorithm is used to identify abnormal data. The abnormal score threshold is set to 0.65. For data points that are identified as abnormal for three consecutive sampling periods, the system automatically marks them and notifies maintenance personnel to check the sensor status. Data standardization: The Z-Score standardization method is adopted, with the formula: x' = (x - μ) / σ, where μ and σ are dynamically calculated based on the data of each sensor over the past 30 days to eliminate the influence of different sensor dimensions; 3.2 The hydrological dynamic early warning model is operated, and the model is divided into two stages: training and application; Model training phase (offline updates once a week); a. Training data preparation: Historical hydrological monitoring data from the past 18 months were extracted from the data platform as the training set, totaling approximately 500,000 records, including water levels of each aquifer, mine water inflow, water quality parameters, stress data of the roof and floor, and detailed records of 12 historical water inrush events; b. LSTM prediction sub-model training: A three-layer LSTM network structure is adopted, with 128 neurons in each layer. The dropout rate is set to 0.2 to prevent overfitting. The input time window is the past 7 days (2016 data points), and the output is the predicted values ​​of water inflow and water level for the next 6 hours (72 data points). The Adam optimizer is used, with an initial learning rate of 0.001, a batch size of 64, and a training cycle of 500 rounds. c. Water Source Identification Sub-model Training: Based on the Support Vector Machine algorithm, a water quality fingerprint database is established, encompassing five water source types: Ordovician limestone water, sandstone water, Quaternary water, etc. Feature dimensions include... , , , , 12 indicators including TDS; Model application phase (real-time operation); a. Trend prediction: Input the preprocessed real-time data sequence into the trained LSTM model, and output the predicted values ​​and confidence intervals of water inflow and water level for the next 6 hours; b. Water source identification: When the water quality data is valid, calculate its Euclidean distance with each standard vector in the water quality fingerprint database, use the K-nearest neighbor algorithm (K=3) to determine the water source type, and output the identification probability; c. Decision Fusion: The early warning decision module integrates the LSTM prediction results and the water source identification results, and triggers an early warning according to preset rules. The specific rules include: when the LSTM predicts that the water inflow will increase by more than 50% in the next 3 hours and the probability of the water source identification result being Ordos Grey Water is greater than 80%, an orange early warning is directly triggered. A yellow alert is triggered when only one of the conditions is met. When stress and strain data show that the displacement rate of the top and bottom plates exceeds 5 mm / h, the warning level will be raised by one level based on the original warning level. S4, Early Warning Issuance and Linkage Control; Based on the results of intelligent analysis, the system executes tiered early warning and coordinated control: 4.1 Level 4 Early Warning Mechanism Blue Alert (Attention Level): A single parameter slightly exceeds the limit, such as a single point fluctuation in water level exceeding the normal range by 10%. The system records the event and prompts attention on the monitoring interface. Yellow Alert (Alert Level): Multiple parameters are abnormal or LSTM predicts a medium risk in the future. Automatically send SMS notification to the technical manager and start the regional drainage pump. Orange alert (alert level): Key parameters are severely exceeded and the water source identification points to a dangerous aquifer. All management personnel will be notified via SMS and App. Automatic preparation to close the floodgates will be initiated. On-site inspection is recommended. Red Alert (Emergency Level): Signs of sudden flooding appear, or orange alert parameters continue to deteriorate; activate audible and visual alarms, automatically execute maximum capacity drainage, and prepare for personnel evacuation; 4.2 Example of linkage control: In a practical application at a certain working face, the system detected the following anomaly: 08:15: The flow meter showed that the water flow suddenly increased from the normal value of 85 m³ / h to 130 m³ / h; 08:17: The LSTM model predicts that the water inflow may reach 210 m³ / h in the next 3 hours (an increase of 147%). 08:18: The water quality sensor detected... The concentration increased from 85 mg / L to 210 mg / L. When the concentration increased from 45 mg / L to 125 mg / L, the system identified it as characteristic of Ordovician lime water with a probability of 83%. 08:19: The decision fusion module integrates various parameters and triggers an orange alert; 08:20: The system will automatically perform the following operations: S5 sent warning text messages and app push notifications to 5 managers; Two backup water pumps were activated, increasing the total drainage capacity to 350 m³ / h; Highlight risk areas in the 3D visualization interface; Notify personnel underground to prepare to close the waterproof gates to prevent water inrush accidents; System optimization and maintenance; the system possesses self-learning and continuous optimization capabilities. 5.1 Model optimization mechanism; After each warning event ends, the system automatically collects data from 24 hours before and after the event, compares and analyzes the data with the prediction results, incrementally trains the LSTM model with newly accumulated data every weekend, updates the model parameters, and evaluates and adjusts the warning threshold once a month, optimizing the threshold setting based on the data distribution characteristics of the past 30 days. 5.2 System Operation and Maintenance Support Establish a sensor health monitoring mechanism to monitor indicators such as battery power, signal strength, and data anomaly rate in real time; Real-time network status monitoring, automatic diagnosis of communication faults, and generation of maintenance work orders; Monthly system operation analysis reports are generated, including key indicators such as early warning accuracy, response time, and equipment integrity rate.

[0032] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Content not described in detail in this specification is prior art known to those skilled in the art.

[0033] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Those skilled in the art can readily implement the present invention based on the accompanying drawings and the above description. However, any modifications, alterations, or variations made by those skilled in the art without departing from the scope of the present invention, utilizing the disclosed technical content, are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, or variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. A coal mine hydrological dynamic monitoring system based on the Internet of Things, comprising a sensing and execution layer, a network transmission layer, and an application platform layer, characterized in that, The perception and execution layer includes a sensor group and an execution unit. The sensor group is deployed in the underground mining face, roadway, hydrological observation hole and surface observation hole of the coal mine to collect multi-source hydrological and geological stress data. The execution unit is used to receive control commands and execute linkage control actions. The network transmission layer adopts a converged heterogeneous network communication architecture based on LoRa, Wi-Fi, 4G, and 5G, specifically including: a. an underground access subnet consisting of LoRa gateways and Wi-Fi access points deployed in roadways and mining faces, wherein the LoRa gateways and Wi-Fi access points adaptively select transmission protocols according to the priority and bandwidth requirements of monitoring data; b. an industrial ring network serving as the underground backbone network, connected to the underground access subnet through a protocol converter to achieve data conversion and aggregation between different network protocols; c. an aboveground data transmission network using 4G / 5G mobile communication or fiber optic networks. The application platform layer includes a data middleware, a hydrological big data analysis engine, and a visualization early warning and control module. The hydrological big data analysis engine has a built-in hydrological dynamic early warning model. This model integrates real-time multi-source monitoring data from sensor groups, historical hydrological data and three-dimensional geological model data. It uses the Long Short-Term Memory (LSTM) algorithm with multiple parameters such as mine water inflow, water level, water quality and roof and floor stress data as input feature vectors for time series prediction. It also combines the Support Vector Machine (SVM) algorithm to identify water inrush sources and output the probability of water hazard risk. The visualization early warning control module is used to generate and release four levels of flood warning information (blue, yellow, orange, and red) based on the probability of flood risk and preset multi-level thresholds. It can also automatically or manually issue linkage control commands to the execution unit of the perception execution layer according to the warning level, forming a complete closed loop from monitoring to control.

2. The coal mine hydrological dynamic monitoring system based on the Internet of Things according to claim 1, characterized in that, The sensor group includes a water level and pressure sensor, a flow sensor, a multi-parameter water quality sensor, and a stress and strain sensor. The water level and pressure sensor monitors the dynamic water level and pressure in each aquifer and borehole. The flow sensor uses an ultrasonic or electromagnetic flow meter to measure the total mine inflow and the inflow in different zones. The multi-parameter water quality sensor monitors the pH value, conductivity, temperature, and calcium and magnesium ion concentrations of the water in real time. The stress and strain sensor is embedded in the roof and floor of the coal seam to monitor the deformation and stress changes of the rock mass under the influence of mining.

3. The coal mine hydrological dynamic monitoring system based on the Internet of Things according to claim 1, characterized in that, The execution unit includes a controllable drainage valve and an audible and visual alarm, which are used to receive instructions from the visual early warning control module and perform emergency actions such as starting the drainage pump, issuing an audible and visual alarm, or closing the waterproof gate.

4. The coal mine hydrological dynamic monitoring system based on the Internet of Things according to claim 1, characterized in that, The LSTM prediction sub-model in the hydrological dynamic early warning model adopts a three-layer LSTM network stacked structure. The input data is the time series of multi-source monitoring data over the past 7 days, and the output is the predicted values ​​of water inflow and water level for the next 6 hours. The water source identification sub-model identifies the type of sudden water source by calculating the Euclidean distance between real-time water quality data and each standard vector in the preset water quality fingerprint database.

5. The coal mine hydrological dynamic monitoring system based on the Internet of Things according to claim 4, characterized in that, The hydrological dynamic early warning model also includes a decision fusion module, which integrates LSTM prediction results and water source identification results, and triggers corresponding warning levels according to preset rules: when the LSTM predicts that the inflow volume will increase by more than 50% in the next 3 hours and the probability of the water source identification result being Ordos limestone water is greater than 80%, an orange warning is directly triggered; when only one of the conditions is met, a yellow warning is triggered.

6. The coal mine hydrological dynamic monitoring system based on the Internet of Things according to claim 1, characterized in that, The visualization early warning control module dynamically displays the real-time status, historical data curves, and early warning information of all monitoring points on GIS and 3D geological models. When an early warning is triggered, the system automatically notifies relevant personnel via SMS and App push, and highlights the risk area on the 3D visualization interface.

7. A method for dynamic monitoring of hydrology in coal mines based on the Internet of Things, applied to the system described in any one of claims 1-6, characterized in that, Includes the following steps: S1, through various sensors in the perception and execution layer, hydrological and geological stress data of the coal mine shaft are automatically collected at a preset frequency, wherein the water level sensor collects data once every 5 minutes, the flow meter collects data once every 1 minute, the water quality sensor collects data once every 30 minutes, and the stress sensor collects data once every 10 minutes. S2, the collected data is transmitted in real time to the application platform layer through the converged heterogeneous network in the network transmission layer. The LoRa node sends the sensor data to the nearest gateway, the Wi-Fi AP receives a large amount of information, and all data is aggregated to the wellhead through the industrial ring network and transmitted to the application platform through the ground network. S3, the application platform layer cleans, integrates and standardizes the received data, uses the isolated forest algorithm to identify abnormal data, uses the Z-Score method to unify the data units, and uses the hydrological big data analysis engine to fuse and analyze multi-source monitoring data, historical hydrological data and geological model data. S4, based on the hydrological dynamic early warning model, uses the LSTM algorithm with multiple parameters such as mine water inflow, water level, water quality and roof and floor stress data as input feature vectors to predict water hazard risk, and determines whether the current hydrological state exceeds the preset threshold or has an abnormal change trend. If it exceeds the threshold, it generates and releases water hazard early warning information of the corresponding level. S5. Based on the warning level, the corresponding emergency plan is automatically or manually activated, and the execution unit is used for linkage control. In the case of a yellow warning, the area drainage pump is automatically started; in the case of an orange warning, the waterproof gate is prepared to be closed; and in the case of a red warning, all drainage capacity is activated and an evacuation alarm is issued.

8. The method for dynamic monitoring of coal mine hydrology based on the Internet of Things according to claim 7, characterized in that, In step S4, the hydrological dynamic early warning model uses a time series prediction algorithm based on Long Short-Term Memory (LSTM) network to predict the mine water inflow and water level changes, and combines the sudden changes in water quality characteristics to comprehensively identify the water inrush source.

9. The method for dynamic monitoring of coal mine hydrology based on the Internet of Things according to claim 7, characterized in that, In step S4, the warning information includes four levels: blue warning, yellow warning, orange warning and red warning. Different levels correspond to different information notification scopes, emergency response procedures and linkage control instructions.

10. The method for dynamic monitoring of coal mine hydrology based on the Internet of Things according to claim 7, characterized in that, The hydrological dynamic early warning model is periodically retrained using new monitoring data and the early warning threshold is dynamically adjusted to achieve continuous optimization of the model.

Images