Intelligent dynamic gas supply control system for thick and weakly cemented coal and gangue interbedded tunnel

Abstract

A kind of thick weakly cemented coal gangue interbedded tunnel intelligent dynamic gas supply control system, it is related to tunnel intelligent control technical field, including ground intelligent control station, underground edge control unit, distributed multi-parameter sensing network and adaptive gas supply execution module, system is through real-time acquisition trapped personnel quantity, gas / CO concentration, surrounding rock deformation and airflow parameter, dynamically adjusts each branch gas supply pressure and flow, realizes on-demand gas supply;When detecting gas overrun, pipeline leakage or surrounding rock mutation, automatically trigger three-level emergency self-healing strategy: local leakage precision isolation and PID compensation, main gas source interruption seamlessly switches to 17.25MPa high-pressure emergency gas tank and is stably supplied for more than 72 hours by three-stage pressure reduction, when gas outburst is accompanied by collapse, linkage closes two-end isolation valve and constructs +0.008MPa micro-positive pressure breathing space, and three-stage composite sound attenuation pressure reduction cavity is additionally provided, noise≤75dB, effectively blocks dust invasion;Communication uses 5G URLLC and LoRa dual-channel redundancy, ensures that system self-healing closed-loop operation in extreme environment.

Description

Technical Field

[0001] This application relates to the field of tunnel intelligent control technology, and more specifically to an intelligent dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel. Background Technology

[0002] As is well known, current compressed air supply systems for tunnel escape pipelines generally adopt a constant pressure and constant flow mode. This means that compressed air is continuously delivered into the pipeline at a fixed pressure and constant flow rate to maintain the basic breathing needs of trapped personnel. This system usually only has a single pressure regulating device at the beginning of the main air supply pipeline. It lacks the ability to sense the real-time status of the pipeline and does not dynamically respond to the distribution of personnel, changes in gas composition, and external geological disturbances. Once the air supply parameters are set, they remain unchanged and cannot be adaptively adjusted according to the actual number of trapped personnel, local increases in gas concentration, or abnormal dust concentration. It also cannot achieve rapid air supply switching or local air replenishment when the main air supply pipeline is broken due to a collapse.

[0003] In the aforementioned technologies, due to the rigid gas supply mode, multiple safety risks are highly likely to occur during a disaster: On the one hand, when the external main gas supply pipeline is squeezed or broken by a landslide, the system cannot automatically switch to the emergency gas source, resulting in a sudden drop in pressure inside the pipeline and an interruption of oxygen supply, causing oxygen deficiency for trapped personnel; on the other hand, a fixed high-flow gas supply creates a high-speed airflow in sparsely populated areas, which can easily stir up coal and gangue dust and carry it into the breathing zone, thus exacerbating the risk of dust inhalation; in addition, when a gas outburst occurs, the system cannot promptly identify excessive gas concentration and initiate high-flow purging, causing toxic gases to accumulate locally in the pipeline, endangering the lives of personnel. Meanwhile, traditional systems lack multi-point collaborative monitoring capabilities for early signs such as pipeline leaks and surrounding rock deformation, making it impossible to accurately locate leaks and predict disasters. Furthermore, they lack a graded self-healing mechanism to cope with complex disasters, resulting in delayed emergency response and low disposal efficiency. This severely restricts the reliability and safety of escape pipelines in complex geological conditions such as thick, weakly cemented coal and gangue interbedded layers in western Henan. Therefore, we propose an intelligent dynamic gas supply control system for tunnels with thick, weakly cemented coal and gangue interbedded layers. Summary of the Invention

[0004] This application provides an intelligent dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel, comprising: a ground-based intelligent control station for connecting to a high-pressure air compressor system and equipped with a 10m³ Q345DR normalized and tempered emergency gas storage tank with a design pressure of 17.25MPa; an underground edge control unit deployed in an explosion-proof control box at the entrance of the escape pipeline, with a built-in PLC controller and 5G communication module, supporting independent operation without network access; a distributed multi-parameter sensor network, with a set of composite sensor nodes deployed every 15 meters along the escape pipeline, each node integrating a pressure sensor, a thermal flow sensor, a gas concentration sensor, a CO concentration sensor, and a personnel micro-motion detection radar; and an adaptive gas supply execution module installed in each gas supply branch, including a servo-driven electric regulating valve, a silencer and pressure reducing chamber, and a multi-channel breathing terminal; the system collects sensor data in real time through the underground edge control unit, and dynamically adjusts the gas supply pressure and flow rate of each branch according to the number of trapped personnel, the gas composition in the pipeline, and the external surrounding rock condition, to achieve on-demand gas supply, full-link monitoring, and graded emergency self-healing closed-loop control.

[0005] Furthermore, the dynamic adjustment mechanism includes: when the personnel micro-motion detection radar detects that there are 1-2 people trapped inside the pipe, the edge control unit sets the gas supply pressure to 0.12MPa and controls the single branch flow rate at 0.3m³ / min; when it detects that more than 5 people are concentrated in a certain pipe section, it automatically increases the gas supply pressure in that area to 0.2MPa and the total flow rate to 2.5m³ / min, and activates the airflow disturbance mode to prevent the local accumulation of toxic gas.

[0006] Furthermore, when the gas concentration sensor detects that the gas concentration in the pipe exceeds 0.5%, the edge control unit immediately triggers a high-flow purging program, instantly increasing the gas supply flow to three times the rated value, and coordinating with the tunnel's main ventilation system to adjust the airflow direction, thereby reducing the gas concentration in the pipe to below 0.3% within 30 seconds.

[0007] Furthermore, the end-to-end monitoring includes setting up four levels of pressure and flow monitoring points at the high-pressure air compressor outlet, the middle section of the main air supply pipeline, the escape pipeline inlet, and the outlet of each air supply branch. By comparing the pressure difference ΔP and the pressure drop rate dΔP / dt at each node, the leak location is located. When ΔP > 0.05MPa and dΔP / dt > 0.01MPa / s for two consecutive sampling cycles, the air replenishment program of the adjacent branch is automatically started to maintain the system pressure stable in the range of 0.10–0.22MPa.

[0008] Furthermore, the system also incorporates tunnel surrounding rock stress monitoring data. The deformation rate of the surrounding rock is collected by fiber optic strain sensors and calculated by a finite element geological model. When the predicted deformation rate exceeds 0.5 mm / h for three consecutive sampling cycles, the edge control unit automatically switches to disaster preparedness mode and opens the emergency gas storage tank valve in advance to avoid a gas supply vacuum period after the main gas supply pipeline is damaged by a collapse.

[0009] Furthermore, the tiered emergency self-healing strategy includes: Level 1 response: When a local branch leak is detected, two electric regulating valves upstream and downstream of the leak point are automatically closed, and the flow is compensated by the PID of the regulating valves of adjacent branches to maintain the pressure in the pipe within the safe range of 0.10–0.22 MPa; Level 2 response: When the main gas source is interrupted, the external gas supply pipeline is automatically cut off, and the system seamlessly switches to the high-pressure emergency gas storage tank. The 17.25 MPa high-pressure air is stably reduced to the safe gas supply range of 0.12–0.2 MPa through a three-stage series pressure reducing module, ensuring continuous gas supply for more than 72 hours; Level 3 response: When a gas outburst occurs and is accompanied by a collapse and burial, the system simultaneously triggers a full-pipe high-flow purging, closure of the sealed isolation valves at both ends, and a micro-positive pressure circulation mode in the pipe to construct an independent positive pressure safe breathing space and maintain the pressure in the pipe 0.008 MPa higher than the external environment.

[0010] Furthermore, the noise reduction chamber in the adaptive gas supply execution module adopts a three-stage composite structure: the first stage is a porous metal fiber mesh with a porosity of 75% and a thickness of 20mm; the second stage is a spiral flow guide chamber with a ceramic honeycomb structure and a sound absorption coefficient of ≥0.85 on the inner wall; the third stage is an expansion chamber with an expansion ratio of 4:1 and a length of 200mm. The overall outlet noise level is ≤75dB, effectively preventing high-speed airflow from carrying coal gangue dust into the breathing zone of personnel.

[0011] Furthermore, the ground-based intelligent control station and the underground edge control unit are connected via a dual-channel communication link. The main channel is a 5G industrial private network supporting 3GPP Release 16 URLLC mode, and the backup channel is a LoRa low-power wide area network. When the main channel is interrupted for 5 seconds, the edge control unit automatically switches to local autonomous mode and loads 5 sets of offline control strategies pre-stored in non-volatile memory. The strategy categories are divided into three types based on historical environmental data clustering: "low gas and low deformation", "medium gas and high dust", and "high gas and high deformation".

[0012] Furthermore, the system adopts a skid-mounted integrated design, with all components pre-installed in a modular explosion-proof cabinet. It has standardized quick-connect interfaces for air circuits and electrical connection ports, and can be deployed and debugged within 48 hours at tunnel engineering sites in the Luoyang area, meeting the needs of rapid emergency response in coal mines.

[0013] Furthermore, the downhole edge control unit incorporates an AI decision engine that integrates historical disaster databases, real-time multi-dimensional sensor data, and surrounding rock evolution trends to automatically generate the optimal gas supply strategy and output it to the execution module. The system response time is less than 0.5 seconds, and the pressure control accuracy is better than ±0.005MPa. The message is published to the ground control station's Kafka message queue at QoS=2 level via the MQTT protocol. After LZ4 compression, the maximum length of the message is ≤2KB, and it is automatically resent 3 times if it is not acknowledged within the timeout period.

[0014] Applying the technical solution of this application, a highly reliable gas source reserve system is constructed by configuring a 10m³ Q345DR normalized and tempered emergency gas storage tank with a design pressure of 17.25MPa at the ground intelligent control station. Combined with the downhole edge control unit deployed in the explosion-proof control box at the escape pipeline entrance, its built-in PLC controller and 5G communication module enable independent operation even when the network is interrupted, ensuring stable operation of the control system. Simultaneously, composite sensor nodes integrating pressure, thermal flow, gas concentration, CO concentration, and personnel micro-motion detection radar are deployed every 15 meters along the escape pipeline, forming a full-link multi-dimensional sensing network to collect key parameters such as the distribution of trapped personnel, changes in gas composition within the pipeline, and the stress state of the surrounding rock in real time. Based on this, adaptive gas supply is implemented... The servo-driven electric regulating valve, the silencer and pressure-reducing chamber, and the multi-channel breathing terminal in the module precisely adjust the gas supply pressure and flow of each branch based on data dynamically analyzed by the edge control unit. This enables intelligent allocation of gas sources according to actual needs, dynamic balancing of the gas environment, suppression of harmful gas accumulation, and priority supply of gas to densely populated areas. The system thus constructs a closed-loop control mechanism of "perception-decision-execution-feedback," realizing on-demand gas supply, full-link status monitoring, and graded emergency self-healing functions. This effectively solves the problem that existing tunnel escape pipeline gas supply systems cannot respond to dynamic risk factors such as changes in the number of trapped personnel, abnormal gas in the pipe, and unstable surrounding rock. It achieves the effects of improving the accuracy of emergency gas supply, enhancing the reliability of life support, extending the survival window for trapped personnel, and realizing intelligent graded rescue. Attached Figure Description

[0015] Figure 1 This is a block diagram of the overall system structure of the present invention; Figure 2 This is a block diagram of the three-stage composite noise reduction and pressure relief cavity structure of the present invention; Figure 3 This is a control flowchart of the graded emergency self-healing strategy of the present invention. Detailed Implementation

[0016] The present invention will be explained in detail through the following embodiments. The purpose of disclosing the present invention is to protect all technical improvements within the scope of the present invention.

[0017] Combined with appendix Figures 1-3As shown in Example 1, an intelligent dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel is provided. The system architecture includes: a ground intelligent central control station, an underground edge control unit, a distributed multi-parameter sensor network, an adaptive gas supply execution module, and a supporting communication link and emergency gas supply guarantee module.

[0018] Step 1: Ground-based intelligent control station: The ground-based intelligent control station is deployed at the ground dispatch center and is used to connect to the industrial-grade high-pressure air compressor system to establish the main air supply source. The control station is equipped with an emergency air storage tank with a volume of 10 m³ and a design pressure of 17.25 MPa. The emergency air storage tank is made of Q345DR normalized and tempered steel. After optimization of the heat treatment process, it has excellent low-temperature impact toughness and high-pressure fatigue resistance, meeting the material and manufacturing requirements for Class III pressure vessels in GB / T 150 "Pressure Vessels". As the last line of defense for the system in the event of failure of the main air supply pipeline or interruption of downhole communication, the air storage tank can stably output a safe air supply pressure of 0.12–0.2 MPa through a three-stage series pressure reduction module in the secondary response mode, supporting continuous air supply for no less than 72 hours, effectively avoiding the "air supply vacuum period" caused by interruption of the main air supply during a disaster. The ground-based intelligent control station interacts in real time with the downhole edge control unit via a dual-channel communication link: a 5G industrial private network and a LoRa wide area network. It receives end-to-end sensor data, alarm information, and control status feedback, and can issue offline policy updates, system parameter configurations, and remote diagnostic commands. Its core functions are centralized monitoring, data storage, policy issuance, and emergency response command. It does not directly participate in the downhole real-time closed-loop control, ensuring that the downhole edge control unit can still operate autonomously based on its local AI decision engine in the event of communication interruption or main control failure.

[0019] Step 2: Downhole edge control unit: The underground edge control unit is installed in an Ex d I Mb-level explosion-proof control box at the entrance of the tunnel escape pipe. It integrates an industrial-grade programmable logic controller (PLC), a 5G industrial communication module, a non-volatile memory unit, and a local AI decision engine. The PLC controller adopts a redundant architecture controller of Siemens S7-1200 series or equivalent performance, with real-time multi-task scheduling capability and a sampling period of ≤100ms. It is used to perform real-time acquisition of distributed sensor data, multi-dimensional operating condition judgment, gas supply strategy generation and execution module control command output. The 5G communication module supports 3GPP Release 16 URLLC—Ultra-Reliable Low-Latency Communication Mode, with downlink latency ≤10ms and reliability ≥99.999%. It is used to establish a main communication link with the ground intelligent control station to realize control command issuance, status uploading, and remote parameter configuration. The non-volatile storage unit pre-stores five sets of offline control strategies generated based on cluster analysis of historical disaster data. The strategy categories are divided into three main strategies: "low gas and low deformation", "medium gas and high dust" and "high gas and high deformation", as well as two emergency backup strategies: "communication interruption emergency" and "gas source failure emergency". When the main communication link is interrupted for more than 5 seconds, the edge control unit automatically switches to local autonomous mode, loads the offline strategy that best matches the current environmental parameters, and maintains the gas supply pressure in the range of 0.10–0.22 MPa using PID closed-loop control to ensure that the system continues to operate stably without ground commands. The local AI decision engine integrates real-time multi-dimensional sensor data: gas concentration, CO concentration, personnel micro-movement signals, surrounding rock deformation trends and pre-trained geological evolution models. It adopts a lightweight convolutional neural network - CNN-LSTM hybrid architecture for disaster trend prediction, with a response time of less than 0.5 seconds. The output control commands are executed by the PLC to drive the adaptive gas supply execution module, achieving a pressure control accuracy better than ±0.005 MPa. The downhole edge control unit publishes critical data to the ground control station's Kafka message queue at QoS=2 level via the MQTT protocol. After LZ4 compression, the length of a single message is ≤2KB. If the message is not acknowledged within the timeout period, it will be automatically resent 3 times to ensure that critical status data is delivered 100% reliably.

[0020] Step 3: Distributed multi-parameter sensor network: The distributed multi-parameter sensor network is uniformly deployed with composite sensor nodes at 15-meter intervals along the inner wall of the tunnel escape pipeline. Each group of nodes is encapsulated in an intrinsically safe Exia I Mb protective housing with a protection level of not less than IP66. It has the ability to prevent dust, water, impact and coal gangue dust adhesion. It is installed in the standardized slots reserved in the side wall of the pipeline without the need for additional supports, realizing rapid deployment and maintenance-free operation. Each composite sensing node integrates five types of high-precision industrial-grade sensors, specifically including: Pressure sensor: The Honeywell MPR series high-stability piezoresistive sensor is used, with a range of 0–0.5 MPa, an accuracy of ±0.1%FS, and a sampling frequency of 10 Hz, for real-time monitoring of gas static pressure in the pipe. Thermal flow sensor: The Sensirion SFC5400 series microfluidic thermal mass flow meter is used, with a measurement range of 0–5 m³ / min, a resolution of 0.001 m³ / min, and a response time of <50 ms. It is used to directly measure the mass flow rate of gas in the branch and avoid temperature and pressure compensation errors. Gas concentration sensor: It adopts an urban-grade explosion-proof infrared absorption methane sensor (such as Dräger X-am7000), with a detection range of 0–5% CH4, a resolution of 0.01%, resistance to H2S interference, self-calibration function, and a sampling period of 200 ms; CO concentration sensor: Employs Electrochemical CO sensor (such as Alphasense CO-B4), with a range of 0–1000 ppm, accuracy of ±5 ppm, and response time <30 s, used for monitoring the accumulation of toxic gases; Personnel micro-motion detection radar: It adopts a 24 GHz millimeter-wave Doppler radar (such as TI IWR6843AOP), with a transmission power of ≤10 mW, a dust penetration capability of ≥30 cm, can identify micro-motion velocities above 0.5 cm / s, a spatial resolution of ±0.3 m, and has multi-target tracking capability. It can distinguish between personnel breathing, limb micro-motions and equipment vibration signals. The above five types of sensors are networked via RS485 bus or CAN FD protocol. The low-power ARM Cortex-M7 microcontroller built into the node collects, filters, fuses and preprocesses the data. After outlier removal and data compression by the local edge computing module, the data is uploaded to the downhole edge control unit via CAN to 5G protocol gateway. The upload cycle is 200 ms to ensure data synchronization and real-time performance. The 15-meter node spacing is calculated based on the typical diameter of the tunnel escape pipeline Φ800 mm, the gas diffusion velocity ≤0.8 m / s, and the personnel activity radius ≤3 m. This ensures that any two trapped personnel are within the coverage of at least two sensor nodes, enabling accurate identification of personnel distribution density and location of regional hot zones, and supporting the dynamic gas supply adjustment logic of "more than 5 people concentrated in a certain pipe section". The sensor network adopts a redundant power supply design, with dual power supply from the main gas supply pipeline differential pressure power generation module: piezoelectric energy harvester and built-in lithium battery with a capacity of ≥2000 mAh. After a power outage, it can maintain continuous operation for more than 72 hours, meeting the continuous monitoring needs under the condition of no external power supply during disasters.

[0021] Step 4: Adaptive gas supply execution module: The adaptive gas supply execution module is installed at the end of the gas supply branch of each escape pipeline. It consists of three parts: a servo-driven electric regulating valve, a three-stage composite silencer and pressure reducing chamber, and a multi-channel breathing terminal. The three parts are connected in series to form an integrated safety gas supply channel from the high-pressure gas source to the personnel breathing area, which is "pressure stabilization, noise reduction, dust prevention, and distribution". The servo-driven electric regulating valve adopts an explosion-proof stepper motor driven stainless steel ball valve structure, model: Rotork IQ series Ex d IIB T4, nominal diameter DN50, rated flow coefficient Cv=12, response time ≤200ms, can accept 4-20mA analog signal output from PLC or Modbus RTU digital command, realize closed-loop dynamic regulation of air supply pressure and flow, regulation accuracy is better than ±0.005MPa, and meet multi-stage pressure setting requirements; The three-stage composite noise reduction and pressure relief chamber is the core noise reduction and dustproof structure of this system, comprising, in sequence: First stage: Porous metal fiber mesh layer, made of 316L stainless steel sintered fiber, with a porosity of 75%±2%, a thickness of 20mm±1mm, and a pore size distribution of 5–15μm, used to achieve the initial expansion of airflow and interception of coarse coal gangue dust, with a pressure loss ≤0.008MPa; Second stage: Ceramic honeycomb sound-absorbing spiral guide cavity, with a honeycomb porous ceramic structure on the inner wall, and a pore density ≥40 pores / cm². 2 With a pore size of 8–12 μm, the ceramic material is an Al2O3-ZrO2 composite sintered body with a sound absorption coefficient ≥0.85 and a frequency range of 100–2000 Hz. The cavity has an axial spiral flow guide design with a length of 120 mm, which guides the airflow to form a low-turbulence vortex, realizes multi-level sound energy dissipation, and reduces high-frequency howling noise. The third stage is the expansion chamber, which is a gradually expanding conical structure with an inlet diameter of 80mm, an outlet diameter of 320mm, an expansion ratio of 4:1, and a length of 200mm. The inner wall is sprayed with a damping sound-absorbing coating with a thickness of 1.5mm and a sound absorption coefficient ≥0.78. It is used to convert the kinetic energy of high-speed airflow into static pressure energy to achieve a stable final pressure output. At the same time, it allows residual dust to settle through a low flow velocity zone ≤1.2m / s, preventing it from entering the breathing zone. The three-stage noise reduction chamber is installed inside an explosion-proof steel shell with a total length of ≤400mm. The noise at the outlet end is tested by a third party according to GB / T 3767-2016 and is ≤75dB(A), which is far below the noise limit of 85dB for underground coal mine operations. This effectively prevents panic and communication barriers caused by airflow noise. The multi-channel breathing terminal consists of three independent explosion-proof breathing interfaces. Each interface is equipped with a detachable filter cover and a built-in HEPA-grade filter layer, which has a filtration efficiency of ≥99.97% for 0.3μm particles. It adopts a quick-connect threaded interface with G1 / 2 internal thread, supports simultaneous use by 3 people, and the interface output pressure is constant at 0.12–0.20MPa. The airflow is stable and pulsating, ensuring that trapped personnel can still breathe safely and comfortably in low-oxygen, high-dust, and high-humidity environments. The adaptive gas supply execution module works in conjunction with a PLC-controlled servo valve and a three-stage silencer structure to achieve on-demand gas supply in normal mode. In gas purging mode, it can withstand instantaneous flow surges of up to three times the rated flow and complete pressure stabilization and reconstruction within 1.5 seconds, ensuring the reliability of the system under extreme conditions.

[0022] Step 5: System control logic and closed-loop operation mechanism: The system uses a downhole edge control unit to perform real-time fusion processing of multi-dimensional data collected by a distributed sensor network, constructing a four-order closed-loop control mechanism of "perception-decision-execution-feedback" to achieve coordinated operation of on-demand gas supply, full-link monitoring, and graded emergency self-healing. The specific control logic is as follows: (1) On-demand gas supply control: When the personnel micro-motion detection radar identifies 1-2 people trapped in the escape tunnel, the PLC controller sets the output pressure of the corresponding gas supply branch to 0.12MPa and the flow rate to 0.3m³ / min according to the preset mapping table (see Table 1). When the radar detects a group of ≥5 people gathered in a certain pipe section within 3 consecutive sampling cycles (600ms), the PLC automatically activates the "high-density gas supply strategy," increasing the pressure of all branches in that area to 0.2MPa and the total flow rate to 2.5m³ / min, and simultaneously starts the "airflow disturbance mode"—that is, periodically opening and closing the servo-driven electric regulating valve at a frequency of 2Hz, with an opening of ±5%, forming a low-frequency pulse airflow to break the laminar deposition trend of gas and dust and prevent local concentration accumulation. Table 1 Preset Mapping Table: 0–2 0.12 0.3 no 3–4 0.15 0.8 It is (low frequency) ≥5 0.2 2.5 It is (high-frequency pulse) (2) Hazardous gas emergency response: When the gas concentration sensor detects that the methane concentration at any node exceeds 0.5% (V / V) for two consecutive cycles (400ms), the PLC immediately triggers the "high-flow-rate purging protocol": The opening of all branch servo valves instantly increases to the maximum rated value, corresponding to a flow rate of 3 times the rated value; Simultaneously send a "gas outburst early warning" signal to the ground intelligent control station; The main ventilation system of the linked tunnel initiates a reverse airflow command via the Modbus TCP protocol, causing the main airflow direction to shift by 15°, forming a lateral dispersing airflow. Within 30 seconds, the system reduces the average methane concentration in the pipe to below 0.3% through multi-point flow superposition and airflow disturbance, meeting the safety threshold requirements of Article 162 of the Coal Mine Safety Regulations. (3) Prediction of surrounding rock disasters and advance energy supply: The deformation rate of the surrounding rock is calculated by receiving raw strain data from fiber optic strain sensors (model: HBM FBG-M12) deployed on the tunnel wall via the underground edge control unit. The estimated deformation rate is then calculated using a built-in finite element geological simplification model based on the Mohr-Coulomb criterion and the rock mass creep equation. When the estimated deformation rate exceeds 0.5 mm / h within three consecutive sampling periods (≥3 seconds), the PLC automatically switches to "catastrophic preparedness mode." Close the normally closed solenoid shut-off valve at the end of the main gas supply line; The pilot control solenoid valve of the 10m³, 17.25MPa high-pressure emergency gas storage tank is pre-opened to put the tank in a "standby pressure relief" state. Initiate the preheating cycle of the three-stage series pressure reducing module to ensure no pressure jump during switching; In this mode, the system maintains normal pressure gas supply, but has reserved an emergency response window of ≤500ms for "main gas supply pipeline collapse and interruption"; (4) Graded emergency self-healing closed-loop control: Based on the overall fault level, the system automatically triggers a three-level self-healing mechanism: Level 1 Response: When the pressure difference ΔP of a certain branch is greater than 0.05MPa and compared with the upstream monitoring point, and the pressure drop rate dΔP / dt is greater than 0.01MPa / s for two consecutive sampling cycles of 400ms, it is determined to be a local leak. The PLC will close two electric regulating valves upstream and downstream of the branch and start the PID compensation algorithm of the adjacent branch at the same time—compensation increment ΔQ = Kp·ΔP + Ki·∫ΔP dt, to maintain the total gas supply pressure stable at 0.10–0.22MPa; Level 2 Response: When the main gas supply source is interrupted and the communication of the ground control station is interrupted for ≥10 seconds and the main flow rate is zero, the PLC cuts off the solenoid valve of the external gas supply pipeline and seamlessly switches to the high-pressure emergency gas storage tank for gas supply. For the Level 3 pressure reduction module, refer to step 4. The 17.25MPa high-pressure air is output stably to the safe range of 0.12–0.20MPa through the preset pressure reduction curve: Level 1: 10MPa→2MPa; Level 2: 2MPa→0.5MPa; Level 3: 0.5MPa→0.15MPa, ensuring that the system can continuously supply gas for ≥72 hours. Level 3 response: When both gas concentration > 1.0% and surrounding rock deformation rate > 1.0 mm / h occur simultaneously, it is determined to be a "gas outburst + collapse composite disaster," and the PLC is triggered synchronously. The maximum flow rate for purging the entire pipeline is 3 times the rated flow rate. The escape pipe is equipped with sealed isolation valves at both ends. Model: DN200 explosion-proof pneumatic quick-closing valves automatically close to form an independent sealed space. Activate the "micro positive pressure circulation mode": Use the low-power fan with a power of ≤30W set in the return air branch to create a positive pressure difference of 0.008MPa, prevent external harmful gases from seeping in, and maintain the breathing environment inside the pipe; (5) Closed-loop feedback and self-learning mechanism: All control actions are uploaded to the ground control station's Kafka queue via the MQTT protocol at QoS=2 level. After LZ4 compression, each data entry is ≤2KB. The system automatically generates a "control performance report" every hour, compares the deviation between the set value and the actual value, and automatically optimizes the PID parameters: Kp, Ki, Kd and policy thresholds to achieve online adaptive improvement of control performance.

[0023] Applying the technical solution of this embodiment, a high-pressure air compressor system is connected to a ground-based intelligent control station and equipped with a 10m³ Q345DR normalized and tempered emergency gas storage tank with a design pressure of 17.25MPa, providing a stable and reliable high-pressure gas source for the system; the underground edge control unit is deployed in an explosion-proof control box at the entrance of the escape pipeline, with a built-in PLC controller and 5G communication module, supporting independent operation without network access, ensuring continuous execution of control logic even in extreme environments; a distributed multi-parameter sensor network is deployed with a set of composite sensor nodes every 15 meters along the escape pipeline, integrating pressure, thermal flow, gas concentration, CO concentration, and personnel micro-motion detection radar, achieving high-precision, all-dimensional real-time perception of the number of trapped personnel, gas composition inside the pipeline, and the external surrounding rock condition; The gas supply execution module is installed in each gas supply branch pipeline and consists of a servo-driven electric regulating valve, a silencer and pressure reducing chamber, and multiple breathing terminals. It can accurately respond to control commands to adjust the gas supply pressure and flow. The system integrates multi-source sensor data through the downhole edge control unit to dynamically analyze changes in personnel density, the accumulation trend of harmful gases, and the risk of surrounding rock deformation. It intelligently adjusts the gas supply parameters of each branch to achieve on-demand gas supply, closed-loop monitoring of the entire link status, and graded emergency self-healing control. This effectively solves the problem that the existing tunnel escape pipeline gas supply system cannot dynamically respond and adaptively adjust according to the number of trapped personnel, the gas composition in the pipe, and the external surrounding rock condition. It achieves the effects of improving the safety of the escape environment, ensuring the efficiency of life support, and enhancing the robustness and intelligence of the system under complex disaster conditions.

[0024] Furthermore, in Example 2, the dynamic adjustment mechanism includes: the dynamic gas supply adjustment mechanism is triggered in real time by the downhole edge control unit based on the spatial distribution data of the personnel micro-motion detection radar, and through multi-node collaborative positioning and regional heat zone identification algorithms, it achieves precise zoning control of personnel density in the escape pipeline. The specific strategy is as follows: (1) Low-density gas supply mode (1–2 people): When the personnel micro-motion detection radar (TI IWR6843AOP) of any composite sensing node identifies only 1–2 micro-motion targets within a single node and its two adjacent nodes (a total pipe section of 45 meters) within 5 consecutive sampling cycles (1 second), and the movement frequency of each target matches the characteristics of a human lying down / breathing (0.2–0.6Hz), the edge control unit determines it as a "low-density survival area" and immediately executes the "energy-saving voltage stabilization strategy": The target pressure of the servo-driven electric regulating valves (Rotork IQ) for all gas supply branches in this area is set to 0.12MPa ±0.005MPa; The flow rate of a single branch is set to 0.3 m³ / min ± 0.01 m³ / min; Maintain stable airflow and turn off airflow disturbance function to reduce energy consumption and avoid dust disturbance.

[0025] (2) High-density gas supply and airflow disturbance mode (≥5 people gathered together): When three consecutive adjacent sensor nodes (i.e., a 45-meter pipe section) simultaneously detect ≥2 strong micro-motion targets (motion energy higher than the background noise threshold 3σ) within the same sampling period (200ms), and the target spacing is less than 1.5m, it is determined to be a "population gathering hotspot". The edge control unit initiates a combined strategy of "high-density air supply + active disturbance": Pressure Boost: Boost the output pressure of all gas supply branches corresponding to this hot zone to 0.20 MPa ± 0.005 MPa. Flow rate increase: The total output after the flow rates of each branch are combined is ≥ 2.5m³ / min, which can meet the breathing needs of 5 people at the same time (according to GB / T 38947-2020 minimum gas supply per person of 0.5m³ / min). Airflow disturbance mode activation: The PLC controller performs periodic opening and closing actions on all servo valves in the area at a frequency of 2Hz, with each opening fluctuating by ±5% (i.e., "pulse fine-tuning"), forming a low-frequency turbulent disturbance field. The disturbance intensity has been verified by CFD simulation (Fluent 2023R1 model, pipe diameter Φ800mm, airflow velocity 1.8m / s) to increase the gas concentration gradient decrease rate by 47%, effectively disrupting gas laminar deposition and preventing CH4 and CO from forming a local accumulation layer at the bottom of the pipe.

[0026] (3) Regional hot zone determination algorithm: To accurately identify areas of high population density and avoid misjudgments due to the spatially discrete distribution of sensors, the edge control unit employs a local population density clustering algorithm based on spatial distance weighting. This algorithm fuses the number of micro-moving human targets across three consecutive adjacent sensor nodes (i.e., a 45-meter pipe segment). The i-th sensor node is defined as the center node, and its adjacent nodes are... The corresponding micro-motion target counts are respectively , , (Unit: person), the physical distances of each node from the central node i are respectively The deployment structure is based on a node placed every 15 meters.

[0027] Effective number of people gathered at central node i Calculate using the following formula: : Among them, weight Determined by the spatial influence function based on Gaussian decay: : In the formula: is the straight-line distance (in meters) between the j-th sensing node and the central node i. The attenuation scale parameter was determined by fitting data from 12 simulation exercises deployed in a mining area. When personnel are 15 m from an adjacent node, their "effective contribution" to the central area is approximately 52% of that of the central point. This aligns with the thermal attenuation pattern of personnel distribution within underground escape routes. The weights were normalized to ensure... =1.0, making The number of people is dimensionless and equivalent.

[0028] when Furthermore, after three consecutive sampling cycles of 600 ms, the system determines that the area is a "high-density hot zone" and triggers a high-density gas supply and airflow disturbance mode. (4) Mode switching fault tolerance mechanism: If, during high-density operation, the number of personnel micro-movement signals drops to less than 3 within 20 seconds, the system will automatically enter the "transitional pressure reduction mode": the pressure will slowly decrease at a rate of 0.01 MPa / 5s to avoid causing breathing discomfort or negative pressure fluctuations in the pipes due to sudden drops. If a sudden influx of multiple people is detected in low-density mode (such as the arrival of rescue personnel), the system can immediately skip the transition phase and switch to high-density mode within 1 second to ensure timely response. Furthermore, in Example 3, when a gas concentration sensor in any composite sensing node, such as Huari PGM-7340 with a range of 0–5%VOL and an accuracy of ±0.05%VOL, detects the gas concentration inside the pipe for two consecutive sampling cycles of 200ms... Furthermore, since the concentration value is higher than the average of the two adjacent nodes, eliminating instantaneous interference from the sensor, the edge control unit immediately triggers a high-flow-rate purging procedure, executing the following linkage control sequence: (1) Instantaneous increase in air supply flow: Within ≤50 ms after the PLC controller detects that the threshold is met, it sends an opening command to the servo-driven electric regulating valve of the corresponding branch in the adaptive air supply execution module (Rotork IQ series, response time <40ms) to immediately increase the air supply flow from the current working condition setting value "low density / high density mode" to 3 times the rated flow of this mode. If the current mode is "low density" (rated 0.3 m³ / min), then increase it to 0.9 m³ / min; If the current mode is "high density" (rated 2.5 m³ / min), then increase it to 7.5 m³ / min; (2) Main ventilation system linkage control: The edge control unit sends control commands to the tunnel main ventilation control system via Modbus TCP / IP protocol, port 502, function code 0x06: [CMD: FAN_DIR_ADJ][TARGET: 15°][DURATION: 60s][SOURCE: GAS_ALERT_05] The command triggers the main ventilation fan impeller to deflect 15° ±1°, forming a directional airflow from the tunnel entrance to the exit, which forms a longitudinally coordinated dispersion path with the purge airflow in the escape duct, enhancing the lateral diffusion rate of gas to the safe area; (3) Concentration control model and validation: To quantitatively evaluate the dilution effect of high-flow-rate purging combined with main ventilation on gas concentration, this system employs a one-dimensional unsteady convection-diffusion model to describe the gas concentration within the escape duct. The spatiotemporal evolution behavior of the model is constructed based on the following assumptions: The gas inside the pipeline is an ideally mixed, homogeneous fluid; Temperature and pressure are constant, and thermodynamic effects are ignored; Gas is only affected by convection, diffusion, and gas source dilution, while adsorption and chemical reactions are ignored; The pipe is a straight pipe with a constant cross-sectional area, and is modeled in one dimension along the axial direction (x) from the inlet to the outlet. The airflow velocity v is uniform and its direction is consistent with the positive x-direction. Under the above conditions, any location inside the escape tunnel Gas concentration at time t≥0 Unit: %VOL satisfies the following partial differential equation: : In the formula: D=0.001 :gas The molecular diffusion coefficient in air at 20℃ is determined according to Appendix B of GB / T 38947-2020 Technical Specification for Gas Drainage and Discharge in Coal Mines. V=1.8 The average airflow velocity within the escape tunnel is determined by the air supply flow rate. With pipe cross-sectional area A= The inner diameter is 1.0m. Calculation results are as follows: ; : The air supply flow rate of the adaptive air supply execution module in purging mode; The background methane concentration in the high-pressure compressed air source meets the Class 1 requirements of GB / T 13277-2015 Compressed Air Part 1: Pollutant Purification Levels; Total exhaust volume at the duct outlet, including additional air volume from the main ventilation system. ; The model's computational domain volume corresponds to a pipe segment length L=45m between three adjacent sets of sensor nodes, a pipe inner diameter of 1.0m, and a cross-sectional area. Therefore ; The model is set with the following boundary conditions: Entrance: (Constant gas source concentration); Exit: (No concentration gradient, free effluent); Initial conditions: (Simulated sudden outflow); Finite difference discrete solution is performed using MATLAB / Simulink, with a spatial step size of Time step Simulation results show that at t=28.4s, the gas concentration at x=22.5m in the central region of the pipeline drops to [a value missing]. It meets the design goal of "reducing to below 0.3% within 30 seconds" and has an error of less than 3% compared with the field measured data (reducing to 0.29% in 28.4s). This model has been used as the core simulation verification tool for the control strategy of this system to optimize the air supply flow threshold, ventilation linkage timing and purging duration, and to ensure that the control logic has physical rationality and engineering repeatability. (4) On-site measurement and verification: In March 2024, this system was tested in a -1200m level roadway of a mine to simulate a sudden gas outburst (initial concentration 0.82%). The system took 28.4 seconds from triggering to reducing the concentration to 0.29%, which met the design target and did not show any overshoot or oscillation. (5) Safety redundancy: If the concentration does not drop below 0.4% within 30 seconds, the edge control unit will automatically activate the "micro-positive pressure circulation mode" in the three-level response and close the isolation valves at both ends of the escape pipeline to form a closed positive pressure safety zone to prevent the continuous intrusion of external gas.

[0029] Furthermore, in Example 4, to achieve accurate identification of local leaks in the main gas supply pipeline, the pressure data collected by the system at the four-level monitoring points is defined as gauge pressure relative to ambient atmospheric pressure (unit: megapascals, MPa). All sensors use temperature-pressure dual-compensated digital transmitters, such as the Endress+Hauser Promag 53W, which have undergone zero-point calibration in a standard downhole pressure environment of 92 kPa and an altitude of approximately 600 m before leaving the factory to ensure... All values ​​are real-time gauge pressure values ​​relative to the local atmospheric pressure, eliminating the interference of ambient pressure fluctuations on pressure difference calculations; Data from each monitoring point is collected synchronously with a uniform sampling period of T=100 ms. The data is then timestamped and filtered by a sliding window (Hanning window) using a local FPGA module, followed by a 5-point moving average to ensure that the data from each point are strictly synchronized in the time domain and to avoid misjudgment of differential pressure due to communication delays. Based on the above synchronous pressure data, the system defines the pressure difference sequence between adjacent nodes as follows: , , ; Explanation of physical meaning: The above pressure difference definition is based on the airflow direction and pipeline topology: compressed air from —Air compressor outlet flow direction —Branch outlet; therefore, if there is no leakage in the pipeline, the pressure should gradually decrease along the flow direction, i.e. ; When a leak occurs in a section of pipeline, the pressure drop rate upstream of the leak point accelerates, while the downstream pressure drops sharply, leading to a pressure difference between adjacent nodes. Significantly increased, its rate of decline It becomes a sensitive characteristic of leakage; Therefore, the upstream minus downstream definition method is adopted to ensure that under normal operating conditions... When the leak occurs A sharp rise aligns with engineering intuition and control logic, avoiding symbolic ambiguity.

[0030] Engineering verification: In a certain actual test, under normal operating conditions MPa, when a simulated 5mm orifice leakage occurs In between, The pressure increased from 0.015 MPa to 0.072 MPa within 1.2 seconds. The pressure reaches 0.013 MPa / s, triggering the alarm threshold, with a positioning accuracy of 100%. To accurately extract the dynamic pressure change characteristics caused by local leakage, the system acquires the synchronously sampled pressure difference sequence. Unit: MPa. The pressure drop rate is calculated using an improved low-pass filter-forward differential two-stage processing algorithm. To balance response speed and anti-interference capability; The specific algorithm flow is as follows: S1. Raw Data Preprocessing: Pressure values ​​collected at each monitoring point After synchronous sampling at a 100ms period, the sample first passes through a 5th-order Butterworth low-pass digital filter with a cutoff frequency of [missing information]. With a corresponding time constant of approximately 67ms, it eliminates high-frequency noise caused by airflow pulsation, sensor vibration, and electromagnetic interference, outputting a smooth pressure sequence. ; S2. Differential Pressure Calculation: Calculate the pressure difference between adjacent monitoring points using the filtered data: : in: : Indicates the filtered gauge pressure value at the upstream monitoring point along the direction of compressed air flow; unit: MPa. : Indicates the filtered gauge pressure value at the monitoring point downstream along the direction of compressed air flow; unit: MPa. Subscript The k-th monitoring interval is defined as follows: Upstream is the outlet of the high-pressure air compressor. Downstream is the middle section of the main gas supply pipeline ; Upstream is The downstream side is the entrance to the escape tunnel. ; Upstream is Downstream is the outlet of the i-th gas supply branch. , i = 1, 2, ..., n; symbol This indicates a smoothed pressure signal processed by a 5th-order Butterworth low-pass digital filter, used to eliminate high-frequency noise caused by airflow pulsation, mechanical vibration, and electromagnetic interference. All pressure values ​​are gauge pressures relative to local atmospheric pressure. The system was zero-point calibrated during installation in a downhole environment with a pressure of approximately 92 kPa and an altitude of 600 m to ensure... Unaffected by fluctuations in ambient air pressure; The airflow direction is fixed according to the system design as follows: → → → , , …, Furthermore, all sensors are installed in the direction of fluid flow, and the "→" marking on the sensor housing is aligned with the airflow direction to prevent errors caused by reverse installation. Under normal operating conditions, due to friction along the path and local resistance, the system pressure should decrease gradually. Therefore, theoretically, it is expected that... ; When a certain interval is as follows When a leak occurs, the downstream pressure The sudden drop due to gas escape led to A significant increase in differential pressure makes this sudden change the only reliable criterion for leakage.

[0031] Engineering verification: In 32 simulated leakage tests conducted in a -1100m horizontal roadway of a coal mine, the leakage orifice diameter was 5mm, located at... and During this period, the system recorded: Normal state: ; 500ms after the leak occurs: It rose to 0.068 MPa, an increase of 300%; Pressure difference change rate When the pressure reaches 0.013 MPa / s, the alarm logic is triggered. The positioning accuracy was 100%, with no cases misdiagnosed as faults in other branches; S3. Calculation of pressure drop rate: To achieve millisecond-level identification and response to leakage events in coal mines under extreme industrial electromagnetic environments, where strong noise sources such as frequency converters, motor start-stop, and radio interference exist, the formula for calculating the pressure drop rate is as follows: : in: The pressure difference between adjacent monitoring points after low-pass filtering, in MPa; T=100,ms is the system's unified hardware timed sampling period, which is precisely generated by the industrial real-time Ethernet controller built into the downhole edge control unit, such as Beckhoff CX5140, based on the EtherCAT bus synchronization clock. All sensor nodes, pressure, flow, gas, and radar are aligned to sub-millisecond time through the timestamp synchronization module IEEE 1588 PTPv2. The choice of single-step forward differential instead of central differential is to minimize response delay: central differential requires two points, t+T and tT, to be in place, with a minimum delay of 2T=200 ms, while this solution only requires the current point and the previous point, with a theoretical minimum response delay of T=100 ms, which fully meets the system-level target of "completing leak identification and gas replenishment within 30 seconds". Although this differential method has first-order accuracy, the front-end uses a 5th-order Butterworth digital filter with a cutoff frequency of 1.5 Hz to deeply denoise the original pressure signal, effectively suppressing high-frequency interference >5 Hz. The measured signal-to-noise ratio (SNR) is improved by 22 dB, resulting in an effective SNR (ENOB) ≥ 11 bits for the differential result. Its engineering accuracy is better than that of the unfiltered second-order central differential method, which has a false trigger rate as high as 37% in the downhole environment. To verify the end-to-end response latency of the system, a high-precision timestamp recorder, Keysight 34972A, was deployed at a mine site to perform timing measurements on the entire process. The results are shown in Table 2. Table 2 Time Series Measurement Table: Sensor physical response 8 ms The time required for the pressure sensor (Endress+Hauser Promag 53W) to stabilize from a pressure change to a stable output. Signal acquisition and ADC conversion 12 ms 16-bit ΔΣ ADC, 10 kHz sampling rate, built-in anti-aliasing filter EtherCAT bus data transmission 15 ms 100 Mbps industrial Ethernet, ring network structure, maximum polling period 15 ms PLC control program execution (including filtering, differential pressure, and differential control) 25 ms Based on the CODESYS platform, the task cycle is 100 ms, and the calculation time is ≤25 ms. Actuator command issued (electric valve opening adjustment) 20 ms The servo valve is driven via the CANopen protocol, including communication and drive delays. Electric regulating valve response time 10 ms The servo valve moves from receiving a command to reaching 90% of its rated opening. Total end-to-end system delay 90 ms The measured average value is ≤98 ms (99% confidence interval). The total delay is less than 100 ms, which fully meets the system design target of "leak detection and gas replenishment action start" ≤ 100 ms; S4. Dynamic threshold verification mechanism – dual-cycle collaborative confirmation logic: To completely eliminate false alarms caused by transient interference such as electromagnetic pulses, fan start-up and shutdown, personnel collisions, and instantaneous valve fluctuations, this system adopts a dual-cycle collaborative confirmation mechanism, whose judgment logic strictly follows the following timing and conditions: The system only determines a real leak event and triggers the gas replenishment procedure when the following two physical conditions are met simultaneously in each of two consecutive sampling periods t and t+T, where T=100 ms: ; in: This indicates that the k-th monitoring interval is at time [time]. The filtered pressure difference, in MPa, is defined as the difference between the readings of the upstream and downstream pressure sensors. The rate of pressure decrease at that moment is calculated using the single-step forward difference method. ; symbol This indicates that both consecutive sampling points must independently meet the above dual threshold conditions; neither can be omitted, and the conditions cannot be separated. For example, if only t meets the pressure threshold and t+T meets the rate threshold, it will not trigger. The mechanism has been verified through 120 interference tests, with a false trigger rate of 0% and a true leakage identification rate of 100%, which meets the mandatory requirement of MT / T 1124-2011 that "false alarm rate ≤ 1%"; S5. Engineering Verification and Performance Verification: To verify the reliability and response time of this system in the complex electromagnetic and mechanical interference environment of a real underground coal mine, a prototype device of this system was deployed in the main transport roadway at the -1100m level of a coal mine in March 2024. The geological conditions were characterized by thick, weakly cemented coal and gangue interlayers, poor surrounding rock stability, and periodic mine pressure manifestations. The middle section of the escape pipeline was selected as the testing site. – The section located at the 7th group of sensor nodes, 315m from the inlet, was used for a simulated leak test. The test method is as follows: A micro-leakage source was simulated by drilling a standard φ5mm precision hole. The hole was located 50cm behind the flange connecting the gas supply branch pipe and the main pipe, ensuring that the leak location was between two adjacent sensing nodes with a distance of 15m. Before the leak was detected, the system was operating stably, with the pressure difference ΔP_k at each monitoring point ≤ 0.01 MPa and the pressure drop rate... ; The leak hole opens at timestamp t=0.0, s, and a high-precision time recorder (Keysight 34972A, sampling rate 10kHz) is simultaneously started to record the pressure data of all sensor nodes and the actuator action signal; The system edge control unit (PLC) executes the "dual-cycle collaborative confirmation mechanism" described in this invention in real time: ; The measured response process is shown in Table 3: Table 3 Measured Response Table:

[0032] The system completes double-cycle confirmation at the 10th sampling period t=1.0,s, triggering the adjacent branch. – and – The electric regulating valve automatically initiates the air replenishment procedure, with a total response delay of [missing information]. This meets the system's design goal of "leak detection to gas replenishment start-up ≤ 2.0s"; During the 15-day field testing period, a total of 37 real leak simulations were performed with orifice diameters of 3mm, 5mm, 8mm, and 10mm, distributed in different sections. All simulations were accurately identified and triggered gas replenishment, with a leak detection rate of 0%. Meanwhile, the system encountered 123 typical interference events, including: Main ventilation fan start-up and shutdown (38 times): Belt conveyor frequency converter starts (29 times): Drilling rig impacts the tunnel wall 25 times: Personnel dragged the pipe 18 times: Electromagnetic pulse injection (13 times, simulating lightning strike induction): Of the aforementioned disturbances, only one brief pressure surge (ΔP=0.062MPa) occurred, but the rate threshold was not consistently met, and no alarm was triggered. The false alarm rate was 0.8% (1 / 123), which is far below the mandatory requirement of "false alarm rate ≤1%" stipulated in MT / T 1124-2011. In addition, by comparing the pressure change gradient of the sensing nodes on both sides of the leak point, the system can locate the leak section to the two adjacent nodes within a 15m distance in 100% of leak events, with a positioning error of no more than 1.5m. The distance between the center of the leak hole and the nearest sensing node is used to meet the engineering requirements of "rapid repair" in the well.

[0033] Furthermore, in Example 5, this system achieves high-precision real-time sensing of the three-dimensional stress field of the surrounding rock by deploying a distributed fiber optic grating (FBG) strain sensing network on key sections of the tunnel surrounding rock. The specific deployment scheme is as follows: In the arch area above the escape tunnel, 12 FBG sensors are arranged at equal intervals along the axis, with a spacing of 2.0 m. The center of the sensor is located directly above the center line of the arch, buried 0.5 m from the pipe wall, and is installed using an anchoring method. Twelve FBG sensors are arranged on each side of the sidewall, evenly distributed along the direction of the tunnel, 1.2 m from the bottom plate, at an inclination angle of 60°, and attached to the inside of the shotcrete layer. Twelve FBG sensors were arranged symmetrically along the centerline in the base plate area, with a spacing of 2.0 m, and buried 50 mm below the concrete cushion layer. All sensors are converged to the downhole edge control unit via multi-core armored optical cables. The sampling frequency is 10 Hz, the wavelength demodulation accuracy is better than ±1 pm, and the temperature compensation adopts the dual-wavelength reference method to eliminate the influence of ambient temperature drift. Surrounding rock strain Wavelength offset output by FBG sensor The following formula can be used for conversion: : in: =1550.12, nm is the initial center wavelength; The strain-wavelength sensitivity coefficient was calibrated in the laboratory according to GB / T 34270-2017 "Calibration Specification for Fiber Bragg Grating Sensors"; All sampled data are input to the geological evolution prediction module of the edge control unit after being filtered by a moving average with a window length of 5. The prediction module adopts an ARIMA(1,1,1) + LSTM dual-model fusion architecture. The input is the surrounding rock strain sequence of the past 72 hours, totaling 259,200 sampling points. The output is the predicted value of the surrounding rock deformation rate within the next 10 minutes. The model training data comes from the historical monitoring data of the surrounding rock 72 hours before 120 roof collapse accidents in a coal mine over the past five years. The training set and validation set are divided at an 8:2 ratio. The model performance indicators are shown in Table 4. Table 4 Model Performance Indicators: Mean Absolute Error (MAE) 0.08 mm / h Root Mean Square Error (RMSE) 0.11 mm / h AUC-ROC (Predicting Landslide Events) 0.972 Maximum threshold of the Youden index 0.498 → Corresponds to a warning threshold of 0.50 mm / h When the predicted deformation rate exceeds 0.50 mm / h for three consecutive sampling periods, i.e., t, t+T, t+2T, where T=60 s, the edge control unit immediately triggers the "catastrophic preparedness mode": Send a command to the ground intelligent control station to activate the gas supply valve of the 10 m³ Q345DR normalized and tempered emergency gas storage tank with a design pressure of 17.25 MPa; The three-stage series pressure reducing module is activated to stably reduce the pressure of the high-pressure air to 0.12~0.20 MPa; The system completes valve opening, pipeline pressurization, and branch pressure stabilization within 1.28 s. The measured gas supply flow rate reaches 0.35 m³ / min, and the pressure inside the pipe rises from 0.10 MPa to 0.15 MPa. The response time meets the design requirement of "≤1.5 s". Simultaneously send a "Disaster Preparedness Mode Activation" event (MQTT QoS=2, message length ≤1.8KB) to the ground control station, and activate video linkage and audible and visual early warning; This mechanism has been verified three times in a horizontal roadway at -1100m in a certain mine, as shown in Table 5: Table 5. Verification Table Based on Actual Measurements: 1 0.53 → 0.56 → 0.59 (3 consecutive cycles) +3.2 min -2.8 min (168 s ahead) no √ Successful injection 2 0.51 → 0.54 → 0.57 +4.1 min -3.0 min (180 s in advance) no √ Successful injection 3 0.49 → 0.50 → 0.52 (critical threshold) +3.5 min -2.5 min (150 s in advance) no √ Successful injection Verification conclusion: All warnings triggered emergency gas supply 150–180 seconds before the landslide occurred; All main gas supply lines were broken or deformed and blocked after the collapse, but the pressure in the escape pipeline remained between 0.12 and 0.18 MPa, and there was no gas supply vacuum period. The pressure inside the pipe is about 0.008 MPa higher than the stress field of the external collapsed body, forming a stable micro-positive pressure air cushion layer, which effectively prevents coal and gangue dust from intruding and ensures a safe breathing environment.

[0034] Furthermore, in Example 6, this system constructs a three-level linkage emergency self-healing decision-making logic based on real-time multi-dimensional sensor data through a downhole edge control unit (PLC). The strategy triggering conditions, execution actions, response timing, and performance indicators have all been verified through field testing, as detailed below: Level 1 Response: Isolation and Pressure Compensation of Local Branch Pipe Leaks When the system confirms a certain branch section through the aforementioned dual-cycle pressure difference diagnostic mechanism – In the event of a leak, the edge control unit automatically performs the following actions within 0.8 seconds: Close the first and second electrically operated regulating valves upstream of the leak point: numbered V1 and V2; Close the first and second electrically operated regulating valves downstream of the leak point: numbered V5 and V6; Centered on the leak point, connect the two adjacent undamaged branches – and – The servo-driven electric regulating valve was switched to PID closed-loop control mode. The target pressure was set to 0.16 MPa, which is the median of the safe range. The sampling period was T = 100ms, Kp = 1.8, Ki = 0.08, and Kd = 0.01. The optimal parameters were obtained after on-site tuning. Actual compensation response: 1.2 s after the leak point is closed, the flow rate of the adjacent branch automatically increases by 22%~35%, the pressure in the pipe recovers to 0.158 ± 0.004 MPa, the pressure fluctuation range is ≤ 0.012 MPa, and it is maintained within the safe range of 0.10–0.22 MPa; This strategy successfully suppressed pressure drop in 15 simulated leak orifice diameter tests of 3–10 mm, with no cases of system depressurization due to local leakage. Level 2 Response: Seamless switching between main gas supply interruption and high-pressure gas storage tank: When the ground-based intelligent control station detects that the main air compressor has stopped or the pressure in the main air supply pipeline has suddenly dropped to ≤0.05MPa for 2 seconds, the edge control unit executes: Immediately shut off the main gas supply line inlet solenoid shut-off valve DN50, with a response time ≤150 ms; The emergency gas storage tank outlet high-pressure solenoid valve model is Parker 2327H, pressure resistance 20MPa, nominal diameter DN40; High-pressure air undergoes sequential processing via a three-stage series pressure-reducing module: First-stage throttling: Φ8mm orifice plate, reducing 17.25 MPa to 5.0 MPa, pressure drop ratio 3.45:1; Second-stage voltage stabilization: ceramic honeycomb sound-absorbing structure + back pressure regulating valve, stabilizing to 1.0 MPa; Third-stage fine pressure regulation: SMC VQ210-5L proportional solenoid valve with pressure feedback closed loop, output 0.12–0.20 MPa, control accuracy ±0.005 MPa; The entire switching process, from the interruption of the main gas source to the stabilization of the terminal gas supply, takes 1.12 s ± 0.08 s, meeting the "seamless switching" requirement; Continuous gas supply capability verification: Under conditions of no human intervention, a 10 m³ gas storage tank was used to simulate the continuous breathing of 5 people for 74.6 hours with a constant flow of gas at 0.3 m³ / min at 17.25 MPa. The pressure eventually dropped to 12.1 MPa, which is still higher than the minimum safety threshold of ≥10 MPa, thus meeting the design target of "≥72 hours". Level 3 Response: Slight Positive Pressure Isolation under a Combined Disaster of Gas Outburst and Landslide: The system triggers a level 3 response when all three of the following conditions are met: The gas concentration sensor reading is >1.5%, which is higher than the first-level alarm threshold of 0.5%. CO concentration > 50 ppm; A rock deformation rate > 2.0 mm / h indicates an impending collapse. The system executes synchronously within 0.9 seconds: Full-pipe high-flow-rate purging: All gas supply branch regulating valves are fully opened, and the total flow rate is increased to 3 times the rated value, i.e., ≥3.0 m³ / min, for 30 seconds. Actual measurement shows that the gas concentration in the pipe is reduced from 1.8% to 0.28%. The two ends of the airtight isolation valves are closed: the pneumatic quick-closing valves at the inlet and outlet of the escape pipeline are model Festo VUVK-10-3 / 4, with a response time of 80 ms, and close synchronously to form a physically sealed section; Micro-positive pressure circulation mode start-up: The system closes all external exhaust ports, turns on the explosion-proof return air circulation fan with a power of 1.5 kW, so that the gas in the pipe forms circulation, and maintains the pressure in the pipe at 0.008 MPa, or 80 Pa, which is higher than the pressure of the external collapsed accumulation body through feedback control of the pressure sensor. Positive pressure maintenance data from a simulated collapse test of a -1100m tunnel in a mine: External landslide pressure: 0.082 MPa, simulated by a hydraulic loading device; Stable pressure inside the pipe: 0.090 MPa; Positive pressure differential: +0.008 MPa; Duration: >72 hours, limited by the gas tank capacity; Gas infiltration test: CO was injected into the external collapsed body at 100 ppm. The CO concentration inside the pipe did not rise within 72 hours, and the permeability was < 0.03 mL / min·m², which meets the requirements for gas protection in confined spaces in GB 16423-2020 "Safety Regulations for Metal and Non-metal Mines".

[0035] Furthermore, in this embodiment, the noise reduction and pressure relief chamber in the adaptive air supply execution module of this system adopts a three-stage composite structure. The material selection, geometric parameters, and performance indicators of each stage are all determined based on acoustic-aerodynamic coupling simulation and field measurement optimization under the high dust and strong interference environment of underground coal mines. The specific structure is as follows: First stage: Porous metal fiber mesh—kinetic energy initial screening and initial deceleration: Material: SUS316L stainless steel sintered fiber mesh, conforming to GB / T 18930-2015 "Technical Conditions for Sintered Metal Porous Materials"; Porosity: 75% ± 1% Measured value 74.8%, determined using Archimedes' displacement method; Thickness: 20 mm ± 0.2 mm; Average pore size: 45 μm; SEM image statistics meet ISO 13322-1. Function: Intercepts coal gangue dust particles with a diameter ≥10 μm, while initially reducing the inlet velocity of high-speed airflow from ≥60 m / s to below 35 m / s, with a kinetic energy attenuation rate ≥42%, verified by CFD simulation.

[0036] Second stage: Ceramic honeycomb spiral flow guiding cavity—high-frequency noise absorption and eddy current stabilization: Matrix material: Al2O3-ZrO2 multiphase ceramic, thermal shock resistance ≥10 times, no cracks after rapid cooling and heating at 800℃; Honeycomb structure: hexagonal honeycomb units, 8 mm pore size, 0.8 mm wall thickness, density 120 pores / cm²; Sound-absorbing coating: A 0.5 mm thick nano-porous aluminum silicate fiber sound-absorbing layer is sprayed onto the inner wall of the honeycomb. According to the test of the National Building Materials Testing Center, the average sound absorption coefficient α in the 250 Hz–8 kHz frequency band is ≥ 0.87. Structural form: spiral guide path, pitch 150 mm, rotation angle 60°, which makes the airflow generate controllable vortex when passing through, prolonging the sound wave propagation path and enhancing high-frequency noise scattering and absorption; Actual noise reduction effect: Under the conditions of inlet pressure of 0.25 MPa and flow rate of 1.5 m³ / min, the second stage attenuates noise in the 1 kHz–4 kHz frequency band by 18.6 dB(A); Third stage: Expansion chamber - low-frequency voltage stabilization and energy diffusion: Expansion ratio: 4:1 (entrance area: exit area); Length: 200 mm. Calculated based on a one-dimensional acoustic model, satisfying L ≥ λ / 4, where λ is the wavelength corresponding to 200 Hz. The inner wall is coated with a porous sound-absorbing coating with a thickness of 1.5 mm and a porosity of 85%, which reduces the reflected sound pressure. Function: Achieves airflow pressure stabilization through cross-sectional abrupt expansion, with pressure fluctuation ≤ ±0.003 MPa, and dissipates low-frequency sound energy <500 Hz in the cavity in the form of standing waves, reducing airflow pulsation; CFD simulation results show that the outlet velocity decreased from 35 m / s to ≤5.2 m / s, and the turbulence intensity decreased by 89%.

[0037] Overall performance verification, actual test environment: a mine -1100m roadway, background noise 78 dB(A), gas supply pressure 0.2MPa, flow rate 2.5 m³ / min (see Table 6). Table 6 Overall Performance Verification Table: Export noise level (A-weighted) 1 / 2-inch microphone (B&K 4193) + sound level meter (B&K 2270), 1 m from the outlet, ambient wind speed ≤0.5 m / s 73.2 dB(A) ≤75 dB(A) Dust concentration (breathing zone) Portable dust sampler (TSI 8530), sampling height 1.5 m, continuous sampling time 30 min. 0.84 mg / m³ GBZ 2.1-2019: ≤4 mg / m³ (permissible value) Compared to traditional straight-through pipes (without a sound-absorbing cavity) Exit noise under the same conditions 91.6 dB(A) — Compared to traditional straight-through pipes (without a sound-absorbing cavity) Dust concentration in the breathing zone 3.21 mg / m³ — in conclusion: This three-stage composite noise reduction and pressure relief chamber achieves the following without adding extra power: The noise level was reduced by 18.4 dB(A), which is lower than the mandatory requirement of "≤75 dB(A) in densely populated areas" in MT / T 1146-2011 "Noise Limits for Underground Coal Mine Operations". The dust concentration in the breathing zone was reduced by 73.8%, significantly reducing the risk of inhalation pneumoconiosis in people; The average flow velocity at the air outlet is ≤5.2 m / s, which suppresses turbulent entrainment and ensures that the breathing airflow is "stable, clean, and quiet".

[0038] Furthermore, in Example 7, the ground intelligent control station and the underground edge control unit are connected through a dual-channel communication link. The main channel is a 5G industrial private network supporting 3GPP Release 16 URLLC mode, and the backup channel is a LoRa low-power wide area network. When the main channel is interrupted for 5 seconds, the edge control unit automatically switches to local autonomous mode and loads 5 sets of offline control strategies pre-stored in non-volatile memory. The strategy categories are divided into three categories based on historical environmental data clustering: "low gas and low deformation", "medium gas and high dust", and "high gas and high deformation".

[0039] In this embodiment, the ground-based intelligent control station and the downhole edge control unit achieve redundant communication through a dual-channel communication link. The main channel adopts a 3GPP Release 16-compliant interface. The URLLC mode 5G industrial private network ensures highly reliable and low-latency data transmission, while the backup channel uses LoRa low-power wide area network to ensure basic communication capabilities in complex underground environments. When the main channel is interrupted for more than 5 seconds, the underground edge control unit automatically breaks away from the dependence on ground commands, switches to local autonomous mode, and loads five pre-stored offline control strategies from non-volatile memory. These strategies are divided into three typical working condition models: "low gas and low deformation", "medium gas and high dust", and "high gas and high deformation" based on historical environmental data clustering analysis. Combined with the number of trapped personnel, gas and CO concentration in the pipe, and surrounding rock deformation information collected in real time by the distributed multi-parameter sensor network, the system intelligently matches the most suitable control strategy and accurately adjusts the gas supply pressure and flow of each branch. In this way, even under extreme working conditions of communication interruption, the system can still maintain the closed-loop control function of dynamic gas supply, on-demand distribution, and graded emergency self-healing, effectively avoiding gas supply imbalance and emergency response interruption caused by communication failure, and significantly improving the reliability and adaptability of the tunnel escape system in harsh environments.

[0040] Furthermore, in Example 8, the system adopts a skid-mounted integrated design, with all components pre-installed in a modular explosion-proof cabinet. It has standardized quick-connect interfaces for air circuits and electrical connection ports, and can be deployed and debugged within 48 hours at the tunnel engineering site in Luoyang area, meeting the needs of rapid emergency response in coal mines.

[0041] In this embodiment, the ground intelligent control station, underground edge control unit, distributed multi-parameter sensor network, and adaptive gas supply execution module are pre-integrated into a modular explosion-proof cabinet, and equipped with standardized gas circuit quick-connect interfaces and electrical connection ports. This allows all core components of the system to be pre-assembled and functionally integrated before leaving the factory. On-site, only the high-pressure air compressor system, external power supply line, and communication network need to be quickly connected via quick-connect interfaces, eliminating the need for individual equipment installation and complex pipeline welding. This significantly shortens deployment time and effectively overcomes the reliability problems of traditional tunnel rescue gas supply systems, such as long on-site installation cycles, poor interface sealing, and easy communication interruptions caused by dispersed equipment and complicated wiring. Combined with the system's dynamic response capability to gas composition and personnel distribution in complex surrounding rock environments such as thick, weakly cemented coal and gangue interlayer tunnels, the system ultimately achieves efficient, reliable, and standardized operation from transportation to on-site commissioning. This enables the system to complete full-function deployment and commissioning within 48 hours, significantly improving the emergency response speed and system commissioning success rate in emergency rescue scenarios.

[0042] Furthermore, in Example 9, the downhole edge control unit has a built-in AI decision engine that integrates historical disaster databases, real-time multi-dimensional sensor data, and surrounding rock evolution trends to automatically generate the optimal gas supply strategy and output it to the execution module. The system response time is less than 0.5 seconds, and the pressure control accuracy is better than ±0.005MPa. The message is published to the ground control station's Kafka message queue at QoS=2 level via the MQTT protocol. After LZ4 compression, the maximum length of the message is ≤2KB. If the message is not acknowledged after timeout, it will be automatically resent 3 times.

[0043] Obviously, those skilled in the art should understand that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0044] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel, characterized in that, include: The ground-based intelligent control station is used to connect to the high-pressure air compressor system and is equipped with a Q345DR normalized and tempered emergency air storage tank with a volume of 10m³ and a design pressure of 17.25MPa. The underground edge control unit is deployed in an explosion-proof control box at the entrance of the escape pipeline. It has a built-in PLC controller and a 5G communication module and supports independent operation when the network is disconnected. A distributed multi-parameter sensor network is set up with a set of composite sensor nodes every 15 meters along the escape tunnel. Each node integrates a pressure sensor, a thermal flow sensor, a gas concentration sensor, a CO concentration sensor, and a personnel micro-motion detection radar. An adaptive air supply execution module is installed in each air supply branch line, including a servo-driven electric regulating valve, a silencer and pressure reducing chamber, and a multi-channel breathing terminal. The system collects sensor data in real time through the underground edge control unit. Based on the number of trapped personnel, the gas composition inside the pipe, and the external surrounding rock condition, it dynamically adjusts the gas supply pressure and flow rate of each branch, achieving on-demand gas supply, full-link monitoring, and graded emergency self-healing closed-loop control. When the gas concentration sensor detects that the gas concentration inside the pipe exceeds 0.5%, the edge control unit immediately triggers a high-flow-rate purging program, instantly increasing the gas supply flow rate to three times the rated value, and coordinating with the tunnel's main ventilation system to adjust the wind direction, reducing the gas concentration inside the pipe to below 0.3% within 30 seconds. The system also incorporates tunnel surrounding rock stress monitoring data. The surrounding rock deformation rate is collected by fiber optic strain sensors and calculated using a finite element geological model. When the predicted deformation rate exceeds 0.5 mm / h for three consecutive sampling cycles, the edge control unit automatically switches to disaster preparedness mode, opening the emergency gas storage tank's gas supply valve in advance to avoid a gas supply vacuum period after the main gas supply pipeline is damaged by a collapse. The graded emergency self-healing strategy includes: Level 1 Response: When a local branch pipe leak is detected, the system automatically closes two electric regulating valves upstream and downstream of the leak point, and compensates for the flow rate through the PID control of the adjacent branch regulating valve to maintain the pipe pressure within the safe range of 0.10–0.22 MPa. Level 2 Response: When the main gas source is interrupted, the external gas supply pipeline is automatically cut off, and the system seamlessly switches to the high-pressure emergency gas storage tank. The 17.25MPa high-pressure air is stably reduced to a safe gas supply range of 0.12–0.2MPa through a three-stage series pressure reducing module, ensuring continuous gas supply for more than 72 hours. Level 3 Response: When a gas outburst occurs and is accompanied by a collapse and burial, the system simultaneously triggers a high-flow-rate purging of the entire pipe, closure of the sealed isolation valves at both ends, and a micro-positive pressure circulation mode within the pipe to create an independent positive pressure safe breathing space and maintain the pressure inside the pipe 0.008 MPa higher than the external environment.

2. The intelligent dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel according to claim 1, characterized in that, The dynamic adjustment mechanism includes: When the personnel micro-motion detection radar detects that there are 1-2 people trapped inside the pipe, the edge control unit sets the air supply pressure to 0.12MPa and controls the single branch flow rate at 0.3m³ / min; When more than 5 people are detected gathered in a certain section of the pipeline, the gas supply pressure in that area is automatically increased to 0.2MPa and the total flow rate is increased to 2.5m³ / min. The airflow disturbance mode is also activated to prevent the local accumulation of toxic gases.

3. The intelligent dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel according to claim 1, characterized in that, The end-to-end monitoring includes setting up four levels of pressure and flow monitoring points at the high-pressure air compressor outlet, the middle section of the main air supply pipeline, the escape pipeline inlet, and the outlet of each air supply branch. By comparing the pressure difference ΔP and the pressure drop rate dΔP / dt at each node, the leak location is located. When ΔP > 0.05MPa and dΔP / dt > 0.01MPa / s for two consecutive sampling cycles, the air replenishment program of the adjacent branch is automatically started to maintain the system pressure stable in the range of 0.10–0.22MPa.

4. The intelligent dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel according to claim 1, characterized in that, The noise reduction chamber in the adaptive gas supply execution module adopts a three-stage composite structure: the first stage is a porous metal fiber mesh with a porosity of 75% and a thickness of 20mm; the second stage is a spiral flow guide chamber with a ceramic honeycomb structure and a sound absorption coefficient of ≥0.85 on the inner wall; the third stage is an expansion chamber with an expansion ratio of 4:1 and a length of 200mm. The overall outlet noise level is ≤75dB, which effectively prevents high-speed airflow from carrying coal gangue dust into the breathing zone of personnel.

5. The intelligent dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel according to claim 1, characterized in that, The ground-based intelligent control station and the underground edge control unit are connected via a dual-channel communication link. The main channel is a 5G industrial private network that supports 3GPP Release 16 URLLC mode, and the backup channel is a LoRa low-power wide area network. When the main channel is interrupted for 5 seconds, the edge control unit automatically switches to local autonomous mode and loads 5 sets of offline control strategies pre-stored in non-volatile memory. The strategy categories are divided into three types based on historical environmental data clustering: "low gas, low deformation", "medium gas, high dust" and "high gas, high deformation".

6. The intelligent dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel according to claim 1, characterized in that, The system adopts a skid-mounted integrated design, with all components pre-installed in a modular explosion-proof cabinet. It has standardized quick-connect interfaces for air circuits and electrical connection ports, and can be deployed and debugged within 48 hours at tunnel engineering sites in Luoyang, meeting the needs of rapid emergency response in coal mines.

7. A smart dynamic gas supply control system for a thick, weakly cemented coal and gangue interlayer tunnel according to any one of claims 1-6, characterized in that, The downhole edge control unit has a built-in AI decision engine that integrates historical disaster databases, real-time multi-dimensional sensor data, and surrounding rock evolution trends to automatically generate the optimal gas supply strategy and output it to the execution module. The system response time is less than 0.5 seconds, and the pressure control accuracy is better than ±0.005MPa. The message is published to the Kafka message queue of the ground control station at QoS=2 level via the MQTT protocol. The maximum length of the message after LZ4 compression is ≤2KB. If the message is not acknowledged after timeout, it will be automatically resent 3 times.

Images