Coal Mine Mining Zone 3 Heat-Flow Risk Monitoring, Early Warning and Pressure Equalization Control System and Method

Abstract

A system and method for monitoring, early warning, and pressure equalization control of thermal flow risks in three mining zones of a coal mine are proposed. The system comprises a safety monitoring unit, a safety situation early warning unit, and a pressure equalization control and intelligent scheduling unit. The safety monitoring unit collects real-time environmental indicators such as gas concentration and ventilation parameters in the coal mine. It monitors surface atmospheric pressure using a pressure sensor and integrates parameters such as the advance speed of the mining face, coal seam thickness, and inter-layer spacing. The collected parameters are transmitted to the safety situation early warning unit via a bus. The safety situation early warning unit establishes a database of thermal flow risk systems in the three zones, integrates sensor monitoring parameters, and stores the data on a surface server according to coding standards. A deep learning algorithm, combined with a thermal flow risk early warning assessment index system and model, enables dynamic risk level assessment and alarm generation. The pressure equalization control and intelligent scheduling unit generates control strategies based on the early warning level and remotely controls ventilation equipment. This invention achieves intelligent and proactive management of thermal flow risks in the three zones, improving the safety level of coal mines.

Description

Technical Field

[0001] This invention relates to a coal mine mining zone three-zone heat-flow risk monitoring, early warning, and pressure equalization control system and method, belonging to the field of coal mine safety technology. Background Technology

[0002] Under repeated intense mining, shallowly buried coal seams in close proximity are prone to collapse and connection between the goaf, adjacent goaf, and overlying goaf, forming a gas leakage channel connecting the surface, goaf, and working face. Large amounts of surface gas enter the mined coal and rock mass through surface fissures. The heat generated by the coal-oxygen reaction, along with oxidation-derived gases and low-concentration oxygen, flow across regions to the low-pressure goaf area of ​​the coal seam, eventually surging into the working face. This creates a so-called "three-zone" multi-element gas (N2-CO2-CO-O2-CH4) venting-flow-oxidation and heating effect, leading to frequent thermal-flow disasters in the "three zones," such as spontaneous combustion of coal and rock mass, low oxygen levels at the working face, and excessive CO levels.

[0003] Currently, most coal mines use equal pressure ventilation to control air leakage through fissures, which to some extent helps prevent heat flow disasters in the "three zones" (coal seams, coal seams, and coal seams). However, the existing equal pressure ventilation system is not intelligent enough and has serious deficiencies in monitoring and early warning of heat flow risks in the "three zones". This leads to inaccurate identification of heat flow risks in the "three zones" and untimely early warning and control, which seriously restricts the safety of mining shallow buried coal seams that are prone to spontaneous combustion. Summary of the Invention

[0004] The purpose of this invention is to provide a coal mine three-zone heat-flow risk monitoring, early warning and pressure equalization control system and method. This system and method can realize intelligent and advanced control of heat-flow risks in three zones: the ground, the goaf and the working face, to ensure safe production in coal mines.

[0005] To achieve the above objectives, this invention provides a coal mine mining zone three-zone heat-flow risk monitoring, early warning, and pressure equalization control system, comprising sensors, switches, an analysis host, ventilation equipment, an industrial ring network, a server, and a client; wherein the sensors, switches, analysis host, and ventilation equipment are all installed underground in the coal mine; the industrial ring network, server, and client are all located above ground in the coal mine; the server is equipped with a safety monitoring unit, a safety situation early warning unit, and a pressure equalization control and intelligent scheduling unit;

[0006] The safety monitoring unit collects real-time data on the concentration of key gases and ventilation parameters in the underground coal mine environment through sensors; it monitors the atmospheric pressure at ground level through a pressure sensor and integrates parameters such as the advance speed of the mining face, coal seam thickness, and interlayer spacing; the safety monitoring unit transmits the collected parameters to the safety situation early warning unit via a bus.

[0007] The aforementioned safety situation early warning unit is used to establish a thermal-flow risk system database for three zones: the ground, the goaf, and the working face. It integrates the monitoring parameters of the sensors in the three zones and stores the data in the surface server according to a unified coding standard. Based on the deep learning algorithm of AI neural network, combined with the thermal-flow risk early warning assessment index system and model of the three zones, it realizes dynamic assessment of risk level and provides voice early warning.

[0008] The pressure equalization and intelligent scheduling unit automatically generates control strategies based on the early warning level, remotely controls ventilation equipment, and performs pre-emptive adjustment and emergency handling.

[0009] The analysis host is used to store data fed back from the analysis host and the client;

[0010] The aforementioned switch is used to connect the underground and surface areas and to transmit information through an industrial ring network;

[0011] The client is a human-computer interaction interface used to adjust the early warning and control parameters of various systems in the server.

[0012] Furthermore, the sensors include multi-parameter sensors, air volume sensors, and differential pressure sensors; the real-time collected environmental indicators of gases in the coal mine include O2, CO, CO2, CH4, C2H4, C2H2, and C2H6; ventilation parameters include wind speed, differential pressure, and air pressure; the three-zone sensor monitoring parameters integrated in the safety situation early warning unit include ventilation fan operating status information, O2 concentration, CO concentration, CH4 concentration, C2H4 concentration, C2H2 concentration, C2H6 concentration, temperature and humidity, wind speed, differential pressure, and air pressure environmental parameters; the ventilation equipment includes fans, air ducts, and air doors / windows; the industrial ring network is a wireless WIFI or 5G network.

[0013] Furthermore, a blower, air duct, and air volume sensor are installed in the intake airway; an air door / window is installed in the return airway to regulate the working face air pressure; an air volume sensor is installed in the return airway; a differential pressure sensor is installed outside the return airway air door, with one end of the differential pressure sensor installed inside the two sealed walls of the return airway and the other end installed inside the protective pipe connecting the return airway and the deep part of the composite goaf; an intrinsically safe multi-parameter sensor for mining is installed at the corner of the return airway; and a CO sensor and a methane sensor are installed in the return airway.

[0014] A method for monitoring, early warning, and pressure equalization control of thermal-fluid risks in three mining zones of a coal mine includes the following steps:

[0015] S1. The safety monitoring unit collects the index gas concentration and ventilation parameters of the underground coal mine environment in real time through intrinsically safe wireless sensors for mining, monitors the atmospheric pressure on the ground through air pressure sensors, and integrates the mining face advance speed, coal seam thickness, and interlayer spacing mining parameters. The collected parameters are then transmitted to the safety situation early warning unit via a bus.

[0016] S2. The security situation early warning unit processes the received data through a multi-source heterogeneous data fusion model. The specific process is as follows:

[0017] S2.1. A Hampel filter is used to detect and eliminate outliers in the data. The median and absolute deviation are used to measure the dispersion of the data, thus identifying outliers. These outliers are then replaced with appropriate values. The Hampel filter formula is as follows:

[0018]

[0019] In the formula, x i It is the i-th observation in the original data, y i It is the i-th observation after filtering, m i It is x i The median of the data within the center window, s i It is x i The median of the absolute deviation of the data within the center window, i.e.:

[0020] s i =c·median(∣x j -m i ∣)

[0021] j=i-(n-1) / 2,...,i+(n-1) / 2;

[0022] In the formula, n is the window size, c is a constant, and f is a threshold parameter used to control the sensitivity of outlier detection; in this model, c = 1.4826 and f = 3. The Hampel filter can effectively detect and eliminate spike noise and impulse noise in the data while maintaining the smoothness and shape characteristics of the data.

[0023] S2.2. Lagrange interpolation is used to fill in missing values. For a given k+1 points (x0, y0), ..., (x k ,y k Find a polynomial L(x) of degree no more than k such that L(x) i )=y i For all i = 0, ..., k, the expressions for the Lagrange polynomial and the fundamental polynomial are as follows:

[0024]

[0025] In x j The value is 1 at point x, while it is 1 at other points x. i That is, the value is 0 at the position where i ≠ j;

[0026] S3. The security situation early warning unit uses a time series prediction algorithm to predict sensor monitoring parameters in advance. The specific process is as follows:

[0027] S3.1 The time series prediction algorithm model includes an encoder and a decoder. The input to the encoder is the covariate sequence x from past time points. 1:t-1 ={x1,x2,x3,…x t-1} and the target variable sequence y 1:t-1 ={y1,y2,y3,…y t-1}, where t refers to the start time of prediction; the input to the decoder is the sequence of covariates x at future times. t:T ={x t ,x t+1 ,x t+2 ,…x T}, where T refers to the prediction cutoff time; the model output is the future time sequence of the target variable y. t:T ={y t ,y t+1 ,y t+2 ,…y T};

[0028] S3.2 The encoder consists of N identical layers stacked together, each layer containing a multi-head self-attention sublayer and a feedforward neural network sublayer; the output of the multi-head attention sublayer of the I-th layer of the encoder is:

[0029]

[0030] In the formula, W Q W K and W V It is a learnable parameter matrix used to map the vector at each position in the input sequence to Q, K, V, W. O This is used to concatenate the output subvectors of multiple heads to obtain a complete output vector;

[0031] The feedforward neural network sublayer is used to perform a non-linear transformation on the representation at each position, increasing the expressive power of the model. The output of the feedforward neural network sublayer at layer I of the encoder is:

[0032] FFN(x)=max(0,xW1+b1)W2+b2;

[0033] In the formula, max(0,·) is the ReLU activation function; W1 and W2 are learnable parameter matrices; b1 and b2 are learnable bias vectors.

[0034] S3.3 The decoder is used to predict the next target sequence based on the encoder's output and the generated target sequence. The decoder consists of N identical layers stacked together. Each layer contains a multi-head self-attention sub-layer, an encoder-decoder attention sub-layer, and a feedforward neural network sub-layer. The multi-head self-attention sub-layer of the decoder uses a mask to cover the ungenerated positions. The formula for calculating the masked self-attention is:

[0035]

[0036] The output of the multi-head self-attention sublayer of layer I in the decoder is:

[0037]

[0038] The encoder-decoder attention sublayer calculates the correlation between each position in the generated target sequence and each position in the encoder output. The feedforward neural network sublayer in the decoder serves the same purpose as that in the encoder.

[0039] S4. The security situation early warning unit sets tiered early warning levels. Based on the importance of the indicators and the degree to which the monitored data exceeds limits, the early warning information is divided into four levels from low to high urgency: blue, yellow, orange, and red (R1 to R4). It automatically alarms based on real-time monitoring data. The specific process is as follows:

[0040] S4.1 Based on the experimental analysis of the production of carbon monoxide and carbon dioxide and consumption of oxygen by coal oxidation under constant temperature, normal temperature, small air volume and variable air volume conditions, it is verified that the large-scale production of carbon monoxide gas is nonlinearly related to coal temperature, and carbon monoxide and oxygen are selected as early warning indicators for coal spontaneous combustion.

[0041] S4.2. The correspondence between C2H4 and C2H2 / C2H6 and coal temperature was determined by analyzing pyrolysis gases, and C2H4, C2H6 and C2H2 were selected as early warning indicators for coal spontaneous combustion; six early warning indicators for coal spontaneous combustion were selected as CO, O2, CH4, C2H4, C2H6 and C2H2.

[0042] S4.3. Based on the requirements of the coal mine safety regulations regarding the concentration of hazardous gases such as CO and CH4 and O2 at the working face, and according to the three-zone heat-flow risk coupling mechanism, early warning levels are set. The corresponding three-zone heat-flow risk early warning levels and warning values ​​are as follows:

[0043] Blue Alert: T m >40℃; R1={{24ppm<[CO]}∪{0.5%<[CH4]}∪{20%>[O2]}}

[0044] Yellow alert: T m >60℃; R2={{50ppm<[CO]}∩{{[C2H4]>0}∪{[C2H6]>0}}}

[0045] Orange alert: T m >90℃; R3={{500ppm<[CO]}∪{[C2H4]>2ppm}}

[0046] Red Alert: T m >130℃ ； R4 = {[C2H2]>0};

[0047] S5, the pressure equalization control and intelligent scheduling unit automatically provides control strategies based on the system alarm level, enabling one-button start / stop and power adjustment of the pressure equalization ventilation equipment. It automatically adjusts the operating power based on monitoring data, changing the airflow and air pressure at the working face; the specific process is as follows:

[0048] S5.1. Pressure Equalization Control: Adjust the dampers and fans in the ventilation system according to specific circumstances to maintain air pressure balance in the goaf, reduce air leakage in the mine goaf, prevent fresh air from entering the underground spontaneous combustion zone or goaf, cut off the oxygen supply to the ignition source, and achieve asphyxiation and inerting. Therefore, the purpose of pressure equalization ventilation is to control the source of air leakage. According to the law of ventilation resistance:

[0049] Δh=RQ e ;

[0050] In the formula, Δh is the pressure difference across the leakage channel, in Pa; R is the air resistance of the leakage channel, in N·s. 2 / m 8 Q represents the leakage air volume, in meters (m³). 3 / s; e is the exponent of the leakage airflow pattern, e = 1 to 2, here e is taken as 2; that is:

[0051]

[0052] S5.2 Real-time monitoring and predictive data are transmitted to the safety situation early warning system. The system issues warnings based on the warning threshold. The pressure equalization control system automatically adjusts the fan power according to the warning level to reduce the concentration of CO and CH4 hazardous gases in the working face and return airflow, ensuring that the O2 content in the working area meets the standard. At the same time, it links the dampers and windows to control the pressure difference between 30 and 120 Pa, reducing air leakage in the three zones and preventing fires. Specifically:

[0053] The coal under blue warning is in the oxidation and heating stage, and begins to produce CO gas, accompanied by CH4, which reduces the oxygen content at the working face. At this time, the power of the blower is adjusted to increase the air volume, reduce the CO and CH4 concentrations at the working face, and ensure that the oxygen at the working face meets the standard. At the same time, the dampers and windows are adjusted to reduce the control pressure difference to less than 50Pa, thereby reducing the air leakage in the three zones and causing the remaining coal in the goaf to stop the oxidation and heating reaction due to lack of oxygen.

[0054] The coal under yellow warning is in the self-heating stage, and oxidation is further intensified, increasing the rate of oxygen consumption. The production of reaction products CO and CO2 begins to increase, with CO production more than three times higher than in the previous stage. The signs of coal spontaneous combustion, C2H4 and C2H6, begin to appear. At this time, while ensuring that the dangerous gases at the working face do not exceed the limits and the oxygen content meets the standards, the pressure difference should be adjusted to less than 20Pa, and coal spontaneous combustion should be prevented by controlling the air leakage.

[0055] When the coal is in the fission stage, the concentrations of CO, C2H4 and C2H6 increase significantly. The CO concentration reaches 500 ppm and the C2H4 concentration reaches 2 ppm. At this time, the air dampers and air windows are adjusted to make the pressure difference close to 0. By suppressing the air leakage in the "three zones", the coal in the goaf is made to spontaneously combust and suffocate due to lack of oxygen.

[0056] A red alert indicates that the coal is in the combustion stage and has emitted the coal spontaneous combustion indicator gas C2H2. At this time, it is necessary to work with the coal mine's fire prevention and extinguishing department to carry out firefighting.

[0057] The pressure equalization ventilation system adopts a dual insurance mode that combines remote control and regional linkage control. In the event of a power outage or interruption of compressed air underground, it can automatically adjust the pressure equalization ventilation to full negative pressure ventilation to ensure the safety of workers at the working face.

[0058] This invention establishes a safety monitoring unit, a safety situation early warning unit, and a pressure equalization and intelligent scheduling unit within a server. The safety monitoring unit collects real-time data on the concentration of key gases and ventilation parameters in the underground coal mine environment using sensors; it monitors the atmospheric pressure at ground level using a pressure sensor and integrates parameters such as the advance speed of the mining face, coal seam thickness, and interlayer spacing. The safety monitoring unit transmits the collected parameters to the safety situation early warning unit via a bus. The safety situation early warning unit establishes a thermal-fluid risk system database for the three zones: the surface, the goaf, and the working face. It integrates the sensor monitoring parameters from these three zones and stores the data on the surface server according to a unified coding standard. Based on a deep learning algorithm using AI neural networks, combined with the thermal-fluid risk early warning assessment index system and model for the three zones, it achieves dynamic assessment of risk levels and provides voice warnings. The pressure equalization and intelligent scheduling unit automatically generates control strategies based on the early warning level and remotely controls ventilation equipment for pre-emptive adjustment and emergency handling. Through multi-dimensional and multi-angle analysis, fire hazards were promptly identified. By changing the working face air pressure and reducing air leakage in the three zones, intelligent and proactive control of heat and flow risks in the three areas of the ground, goaf, and working face was achieved, thereby improving the level of coal mine safety production. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the system structure of the present invention;

[0060] Figure 2 This is a flowchart of the working process of the pressure equalization control and intelligent scheduling unit of the present invention.

[0061] In the diagram: 1. Pressure sensor, 2. Multi-parameter sensor, 3. Differential pressure sensor, 4. Fan, 5. Air duct, 6. Air damper / window, 7. CO sensor, 8. Air volume sensor 1, 9. Air volume sensor 2, 10. Methane sensor, 11. Bundle tube monitoring equipment. Detailed Implementation

[0062] The invention will now be further described with reference to the accompanying drawings.

[0063] like Figure 1 As shown, a coal mine three-zone heat-flow risk monitoring, early warning, and pressure equalization control system includes sensors, switches, analysis host, ventilation equipment, industrial ring network, server, and client; wherein, the sensors, switches, analysis host, and ventilation equipment are all installed underground in the coal mine; the industrial ring network, server, and client are all located above ground in the coal mine; the server is equipped with a safety monitoring unit, a safety situation early warning unit, and a pressure equalization control and intelligent scheduling unit;

[0064] The safety monitoring unit collects real-time data on the concentration of key gases and ventilation parameters in the underground coal mine environment through sensors; it monitors the atmospheric pressure at ground level through pressure sensor 1 and integrates parameters such as the advance speed of the mining face, coal seam thickness, and interlayer spacing; the safety monitoring unit transmits the collected parameters to the safety situation early warning unit via a bus.

[0065] The aforementioned safety situation early warning unit is used to establish a thermal-flow risk system database for three zones: the ground, the goaf, and the working face. It integrates the monitoring parameters of the sensors in the three zones and stores the data in the surface server according to a unified coding standard. Based on the deep learning algorithm of AI neural network, combined with the thermal-flow risk early warning assessment index system and model of the three zones, it realizes dynamic assessment of risk level and provides voice early warning.

[0066] The pressure equalization and intelligent scheduling unit automatically generates control strategies based on the early warning level, remotely controls ventilation equipment, and performs pre-emptive adjustment and emergency handling.

[0067] The analysis host is used to store data fed back from the analysis host and the client;

[0068] The aforementioned switch is used to connect the underground and surface areas and to transmit information through an industrial ring network;

[0069] The client is a human-computer interaction interface used to adjust the early warning and control parameters of various systems in the server.

[0070] Furthermore, the sensors include a multi-parameter sensor 2, an airflow sensor, and a differential pressure sensor 3; the real-time collected environmental indicators of gases in the coal mine include O2, CO, CO2, CH4, C2H4, C2H2, and C2H6; the ventilation parameters include wind speed, differential pressure, and air pressure; the three-zone sensor monitoring parameters integrated in the safety situation early warning unit include ventilation fan operating status information, O2 concentration, CO concentration, CH4 concentration, C2H4 concentration, C2H2 concentration, C2H6 concentration, temperature and humidity, wind speed, differential pressure, and air pressure environmental parameters; the ventilation equipment includes a fan 4, a duct 5, and a door / window 6; the industrial ring network is a wireless WIFI or 5G network.

[0071] In a preferred embodiment, a blower 4, an air duct 5, and an air volume sensor 8 are installed in the intake airway. The blower 4 serves as the air source for the pressure equalization control and intelligent scheduling unit, and the air volume is adjusted by regulating its power. An air door / window 6 is installed in the return airway to regulate the working face air pressure, reducing air leakage by changing the pressure difference between the goaf and the working face, thus providing parameters for pressure equalization control. The air volume sensor 8 measures the intake air volume, providing parameters for the pressure equalization control and intelligent scheduling unit. An air volume sensor 9 is installed in the return airway to measure the return air volume; the difference between the return air volume and the intake air volume represents air leakage. A differential pressure sensor 3 is installed outside the return airway door. One end of the sensor 3 is installed 10m inside the two sealed walls of the return airway, and the other end is installed inside the protective pipe connecting the return airway and the deep part of the composite goaf. The differential pressure sensor 3 measures the pressure difference between the return airway and the deep part of the composite goaf, controlling the pressure difference between 30 and 120 Pa. A mine-use intrinsically safe multi-parameter sensor 2 is installed at the corner of the return airway to monitor CO, CH4, O2, C2H4, C2H6, and C2H2 gases. A CO sensor 7 is installed in the return airway with a CO alarm concentration of 2.4 × 10⁻⁶ Pa. -6 This provides parameters for pressure equalization control; a methane sensor 10 is installed in the return airway to monitor CH4 concentration. The methane sensor 10 is installed within 0-5m of the regulating damper. When the methane concentration in the working face and return airflow is not lower than 0.5%, the methane sensor 10 alarms; when it is not lower than 1.5%, it immediately cuts off power.

[0072] A bundled tube monitoring device 11 is laid in the goaf area to observe the application effect of intelligent pressure equalization ventilation technology, monitor the degree of spontaneous combustion of coal in the goaf area after pressure adjustment, and can detect the signs of spontaneous combustion of coal in the composite goaf area in a timely manner when excessive air leakage into the composite goaf area is caused by excessive pressure in the working face.

[0073] The mine pressure equalization ventilation system is equipped with dual power supplies and dual fans, one for use and one for backup, which can ensure the system operates stably and reliably within a reasonable range. When the gas sensor data is higher than the interlock value, the main control switch of the system outputs a control signal to control the power supply switch, realizing the gas-electric interlock and wind-electric interlock functions. The mine intelligent local ventilation system is equipped with an audible and visual alarm device to prompt the staff to take appropriate measures to resolve abnormal situations.

[0074] The fan is equipped with a mining explosion-proof and intrinsically safe lithium-ion battery, which automatically charges under normal conditions. When the system experiences a power outage due to abnormal CH4 or CO levels, the fan automatically activates the lithium-ion battery to operate normally, ensuring normal ventilation at the working face.

[0075] A method for monitoring, early warning, and pressure equalization control of thermal-fluid risks in three mining zones of a coal mine includes the following steps:

[0076] S1. The safety monitoring unit collects the index gas concentration and ventilation parameters of the underground coal mine environment in real time through the intrinsically safe wireless sensor for mining, monitors the atmospheric pressure on the ground through the pressure sensor 1, and integrates the mining face advance speed, coal seam thickness, and interlayer spacing mining parameters. The collected parameters are then transmitted to the safety situation early warning unit via the bus.

[0077] S2. The security situation early warning unit processes the received data through a multi-source heterogeneous data fusion model. The specific process is as follows:

[0078] S2.1. A Hampel filter is used to detect and eliminate outliers in the data. The median and absolute deviation are used to measure the dispersion of the data, thus identifying outliers. These outliers are then replaced with appropriate values. The Hampel filter formula is as follows:

[0079]

[0080] In the formula, x i It is the i-th observation in the original data, y i It is the i-th observation after filtering, m i It is x i The median of the data within the center window, s i It is x i The median of the absolute deviation of the data within the center window, i.e.:

[0081] s i =c·median(∣x j -m i ∣)

[0082] j=i-(n-1) / 2,...,i+(n-1) / 2;

[0083] In the formula, n is the window size, c is a constant, and f is a threshold parameter used to control the sensitivity of outlier detection; in this model, c = 1.4826 and f = 3. The Hampel filter can effectively detect and eliminate spike noise and impulse noise in the data while maintaining the smoothness and shape characteristics of the data.

[0084] S2.2. Lagrange interpolation is used to fill in missing values. For a given k+1 points (x0, y0), ..., (x k ,y k Find a polynomial L(x) of degree no more than k such that L(x) i )=y i For all i = 0, ..., k, the expressions for the Lagrange polynomial and the fundamental polynomial are as follows:

[0085]

[0086] In x j The value is 1 at point x, while it is 1 at other points x. i That is, the value is 0 at the position where i ≠ j;

[0087] S3. The security situation early warning unit uses a time series prediction algorithm to predict sensor monitoring parameters in advance. The specific process is as follows:

[0088] S3.1 The time series prediction algorithm model includes an encoder and a decoder. The input to the encoder is the covariate sequence x from past time points. 1:t-1 ={x1,x2,x3,…x t-1} and the target variable sequence y 1:t-1 ={y1,y2,y3,…y t-1}, where t refers to the start time of prediction; the input to the decoder is the sequence of covariates x at future times. t:T ={x t ,x t+1 ,x t+2 ,…x T}, where T refers to the prediction cutoff time; the model output is the future time sequence of the target variable y. t:T ={y t ,y t+1 ,y t+2 ,…y T};

[0089] S3.2 The encoder consists of N identical layers stacked together, each layer containing a multi-head self-attention sublayer and a feedforward neural network sublayer; the output of the multi-head attention sublayer of the I-th layer of the encoder is:

[0090]

[0091] In the formula, W Q W K and W V It is a learnable parameter matrix used to map the vector at each position in the input sequence to Q, K, V, W. O This is used to concatenate the output subvectors of multiple heads to obtain a complete output vector;

[0092] The feedforward neural network sublayer is used to perform a non-linear transformation on the representation at each position, increasing the expressive power of the model. The output of the feedforward neural network sublayer at layer I of the encoder is:

[0093] FFN(x)=max(0,xW1+b1)W2+b2;

[0094] In the formula, max(0,·) is the ReLU activation function; W1 and W2 are learnable parameter matrices; b1 and b2 are learnable bias vectors.

[0095] S3.3 The decoder is used to predict the next target sequence based on the encoder's output and the generated target sequence. The decoder consists of N identical layers stacked together. Each layer contains a multi-head self-attention sub-layer, an encoder-decoder attention sub-layer, and a feedforward neural network sub-layer. The multi-head self-attention sub-layer of the decoder uses a mask to cover the ungenerated positions. The formula for calculating the masked self-attention is:

[0096]

[0097] The output of the multi-head self-attention sublayer of layer I in the decoder is:

[0098]

[0099] The encoder-decoder attention sublayer calculates the correlation between each position in the generated target sequence and each position in the encoder output. The feedforward neural network sublayer in the decoder serves the same purpose as that in the encoder.

[0100] S4. The security situation early warning unit sets tiered early warning levels. Based on the importance of the indicators and the degree to which the monitored data exceeds limits, the early warning information is divided into four levels from low to high urgency: blue, yellow, orange, and red (R1 to R4). It automatically alarms based on real-time monitoring data. The specific process is as follows:

[0101] S4.1 Based on the experimental analysis of the production of carbon monoxide and carbon dioxide and consumption of oxygen by coal oxidation under constant temperature, normal temperature, small air volume and variable air volume conditions, it is verified that the large-scale production of carbon monoxide gas is nonlinearly related to coal temperature, and carbon monoxide and oxygen are selected as early warning indicators for coal spontaneous combustion.

[0102] S4.2. The correspondence between C2H4 and C2H2 / C2H6 and coal temperature was determined by analyzing pyrolysis gases, and C2H4, C2H6 and C2H2 were selected as early warning indicators for coal spontaneous combustion; six early warning indicators for coal spontaneous combustion were selected as CO, O2, CH4, C2H4, C2H6 and C2H2.

[0103] S4.3. Based on the requirements of coal mine safety regulations regarding the concentration of hazardous gases such as CO and CH4 and O2 at the working face, and according to the three-zone heat-flow risk coupling mechanism, early warning levels are set:

[0104] Based on coal spontaneous combustion experiments and on-site observations, CO concentration should normally be below 24 ppm. A blue alert is triggered when CO concentration exceeds 24 ppm, with the temperature range being 40–60℃. According to laboratory temperature-programmed experiments, the earliest appearance of C2H4 and C2H6 occurs at temperatures between 70 and 80℃, which is defined as a yellow alert. An orange alert is triggered when the temperature reaches 100–130℃, at which point the concentrations of CO, C2H4, and C2H6 increase significantly, with CO reaching 500 ppm and C2H4 reaching 2 ppm. Statistics on spontaneous combustion accidents in goaf areas show that C2H6 concentration is generally about three times that of C2H4. When the coal temperature reaches 150–180℃, C2H2 will appear, and the CO concentration will increase significantly, serving as the basis for a red alert.

[0105] The corresponding heat flow risk warning levels and warning values ​​for Zone 3 are as follows:

[0106] Blue Alert: T m >40℃; R1={{24ppm<[CO]}∪{0.5%<[CH4]}∪{20%>[O2]}}

[0107] Yellow alert: T m >60℃; R2={{50ppm<[CO]}∩{{[C2H4]>0}∪{[C2H6]>0}}}

[0108] Orange alert: T m >90℃; R3={{500ppm<[CO]}∪{[C2H4]>2ppm}}

[0109] Red Alert: T m >130℃ ； R4 = {[C2H2]>0};

[0110] S5, the pressure equalization control and intelligent scheduling unit automatically provides control strategies based on the system alarm level, enabling one-button start / stop and power adjustment of the pressure equalization ventilation equipment. It automatically adjusts the operating power based on monitoring data, changing the airflow and air pressure at the working face; for example... Figure 2 As shown, the specific process is as follows:

[0111] S5.1. Pressure Equalization Control: Adjust the dampers and fans in the ventilation system according to specific circumstances to maintain air pressure balance in the goaf, reduce air leakage in the mine goaf, prevent fresh air from entering the underground spontaneous combustion zone or goaf, cut off the oxygen supply to the ignition source, and achieve asphyxiation and inerting. Therefore, the purpose of pressure equalization ventilation is to control the source of air leakage, according to the law of ventilation resistance:

[0112] Δh=RQ e ;

[0113] In the formula, Δh is the pressure difference across the leakage channel, in Pa; R is the air resistance of the leakage channel, in N·s. 2 / m 8 Q represents the leakage air volume, in meters (m³). 3 / s; k is the exponent of the leakage airflow pattern, k = 1~2; here e is taken as 2;

[0114] Right now:

[0115]

[0116] S5.2 Real-time monitoring and predictive data are transmitted to the safety situation early warning system. The system issues warnings based on the warning threshold. The pressure equalization control system automatically adjusts the fan power according to the warning level to reduce the concentration of CO and CH4 hazardous gases in the working face and return airflow, ensuring that the O2 content in the working area meets the standard. At the same time, it links the dampers and windows to control the pressure difference between 30 and 120 Pa, reducing air leakage in the three zones and preventing fires. Specifically:

[0117] The coal under blue warning is in the oxidation and heating stage, and begins to produce CO gas, accompanied by CH4, which reduces the oxygen content at the working face. At this time, the power of the blower is adjusted to increase the air volume, reduce the CO and CH4 concentrations at the working face, and ensure that the oxygen at the working face meets the standard. At the same time, the dampers and windows are adjusted to reduce the control pressure difference to less than 50Pa, thereby reducing the air leakage in the three zones and causing the remaining coal in the goaf to stop the oxidation and heating reaction due to lack of oxygen.

[0118] The coal under yellow warning is in the self-heating stage, and oxidation is further intensified, increasing the rate of oxygen consumption. The production of reaction products CO and CO2 begins to increase, with CO production more than three times higher than in the previous stage. The signs of coal spontaneous combustion, C2H4 and C2H6, begin to appear. At this time, while ensuring that the dangerous gases at the working face do not exceed the limits and the oxygen content meets the standards, the pressure difference should be adjusted to less than 20Pa, and coal spontaneous combustion should be prevented by controlling the air leakage.

[0119] When the coal is in the fission stage, the concentrations of CO, C2H4 and C2H6 increase significantly. The CO concentration reaches 500 ppm and the C2H4 concentration reaches 2 ppm. At this time, the air dampers and air windows are adjusted to make the pressure difference close to 0. By suppressing the air leakage in the "three zones", the coal in the goaf is made to spontaneously combust and suffocate due to lack of oxygen.

[0120] A red alert indicates that the coal is in the combustion stage and has emitted the coal spontaneous combustion indicator gas C2H2. At this time, it is necessary to work with the coal mine's fire prevention and extinguishing department to carry out firefighting.

Claims

1. A coal mine mining zone three-zone heat-flow risk monitoring, early warning, and pressure equalization control system, comprising sensors, switches, an analysis host, ventilation equipment, an industrial ring network, a server, and a client; wherein, Sensors, switches, analysis hosts, and ventilation equipment are all installed underground in the coal mine; the industrial ring network, server, and client are all located above ground in the coal mine; the server is characterized by having a safety monitoring unit, a safety situation early warning unit, and a pressure equalization and intelligent scheduling unit. The safety monitoring unit collects the index gas concentration and ventilation parameters of the underground coal mine environment in real time through sensors; it monitors the atmospheric pressure on the ground through a pressure sensor (1) and integrates the mining face advance speed, coal seam thickness, and interlayer spacing mining parameters; the safety monitoring unit transmits the collected parameters to the safety situation early warning unit through a bus. The aforementioned safety situation early warning unit is used to establish a thermal-flow risk system database for three zones: the ground, the goaf, and the working face. It integrates the monitoring parameters of the sensors in the three zones and stores the data in the surface server according to a unified coding standard. Based on the deep learning algorithm of AI neural network, combined with the thermal-flow risk early warning assessment index system and model of the three zones, it realizes dynamic assessment of risk level and provides voice early warning. The pressure equalization and intelligent scheduling unit automatically generates control strategies based on the early warning level, remotely controls ventilation equipment, and performs pre-emptive adjustment and emergency handling. The analysis host is used to store data fed back from the analysis host and the client; The aforementioned switch is used to connect the underground and surface areas and to transmit information through an industrial ring network; The client is a human-computer interaction interface used to adjust the early warning and control parameters of various systems in the server; The sensors include a multi-parameter sensor (2), an air volume sensor, and a differential pressure sensor (3); the real-time collected environmental indicators of coal mine gases include O2, CO, CH4, C2H4, C2H2, and C2H6; the ventilation parameters include wind speed, differential pressure, and air pressure; the three-zone sensor monitoring parameters integrated in the safety situation early warning unit include the operating status information of the ventilation fan (4), O2 concentration, CO concentration, CH4 concentration, C2H4 concentration, C2H2 concentration, C2H6 concentration, temperature and humidity, wind speed, differential pressure, and air pressure environmental parameters; the ventilation equipment includes a fan (4), a duct (5), and a door / window (6); the industrial ring network is a wireless WIFI or 5G network; A blower (4), a duct (5), and a first air volume sensor (8) are installed in the intake airway; a door / window (6) is installed in the return airway to adjust the working face air pressure; a second air volume sensor (9) is installed in the return airway; a differential pressure sensor (3) is installed outside the return airway door; one end of the differential pressure sensor (3) is installed inside the two sealed walls of the return airway, and the other end is installed in the protective pipe connecting the return airway and the deep part of the composite goaf area; an intrinsically safe multi-parameter sensor (2) is installed at the corner of the return airway; a CO sensor (7) and a methane sensor (10) are installed in the return airway.

2. A method for monitoring, early warning, and pressure equalization control of thermal-fluid risks in three mining zones of a coal mine, based on the thermal-fluid risk monitoring, early warning, and pressure equalization control system for three mining zones of a coal mine as described in claim 1, characterized in that... Includes the following steps: S1. The safety monitoring unit collects the index gas concentration and ventilation parameters of the underground environment of the coal mine in real time through the intrinsically safe wireless sensor for mining, monitors the atmospheric pressure on the ground through the air pressure sensor (1), and integrates the mining face advance speed, coal seam thickness, and interlayer spacing mining parameters, and transmits the collected parameters to the safety situation early warning unit through the bus. S2. The security situation early warning unit processes the received data through a multi-source heterogeneous data fusion model. The specific process is as follows: S2.

1. Hampel filters are used to detect and eliminate outliers in the data. The Hampel filter formula is as follows: ； In the formula, x i It is the first in the original data i One observation value, y i It is the filtered first i One observation value, m i Therefore x i The median of the data within the center window. s i Therefore x i The median of the absolute deviation of the data within the center window, i.e.: ； In the formula, n It refers to the window size. c It is a constant. f It is a threshold parameter used to control the sensitivity of outlier detection; S2.

2. Use Lagrange interpolation to fill in missing values. For a given... k +1 point ( x 0, y 0),…,( x k , y k Find a number of times that does not exceed k polynomial L ( x ), making L ( x i )= y i For all i =0,…, k The expressions for the Lagrange polynomial and the fundamental polynomial are as follows: ; ; exist x j The value is 1 at point 1, while at other points 1... x i ,Right now The value at this location is 0; S3. The security situation early warning unit uses a time series prediction algorithm to predict sensor monitoring parameters in advance. The specific process is as follows: S3.1 The time series prediction algorithm model includes an encoder and a decoder. The input to the encoder is the sequence of covariates from past time points. and target variable sequence ,in This refers to the time at which the prediction begins; the input to the decoder is a sequence of covariates for future times. ,in This refers to the prediction cutoff time; the model's output is the sequence of target variables at future times. ; S3.2 The encoder consists of N identical layers stacked together, each layer containing a multi-head self-attention sublayer and a feedforward neural network sublayer; the output of the multi-head attention sublayer of the I-th layer of the encoder is: ； In the formula, , and It is a learnable parameter matrix used to map the vector at each position in the input sequence to Q, K, V. This is used to concatenate the output subvectors of multiple heads to obtain a complete output vector; The feedforward neural network sublayer is used to perform a non-linear transformation on the representation at each position. The output of the feedforward neural network sublayer at layer I of the encoder is: ； In the formula, yes Activation function; , It is a learnable parameter matrix; , It is a learnable bias vector; S3.3 The decoder is used to predict the next target sequence based on the encoder's output and the generated target sequence. The decoder consists of N identical layers stacked together. Each layer contains a multi-head self-attention sub-layer, an encoder-decoder attention sub-layer, and a feedforward neural network sub-layer. The multi-head self-attention sub-layer of the decoder uses a mask to cover the ungenerated positions. The formula for calculating the masked self-attention is: ; The output of the multi-head self-attention sublayer of layer I in the decoder is: ； S4. The security situation early warning unit sets tiered early warning levels. Based on the importance of the indicators and the degree to which the monitored data exceeds limits, the early warning information is divided into four levels from low to high urgency: blue, yellow, orange, and red (R1 to R4). It automatically alarms based on real-time monitoring data. The specific process is as follows: S4.1 Based on the experimental analysis of the production of carbon monoxide and carbon dioxide and consumption of oxygen by coal oxidation under constant temperature, normal temperature, small air volume and variable air volume conditions, it is verified that the large-scale production of carbon monoxide gas is nonlinearly related to coal temperature, and carbon monoxide and oxygen are selected as early warning indicators for coal spontaneous combustion. S4.2, Determined through analysis of pyrolysis gases The correspondence between coal temperature and temperature was selected. As early warning indicators for spontaneous combustion of coal, six indicators were selected: CO, ... ; S4.

3. Based on the requirements of the coal mine safety regulations regarding the concentration of CO, CH4, and O2 in the working face, and according to the heat-flow risk coupling mechanism of the three zones, early warning levels are set as follows: ； ； ； ； S5, the pressure equalization control and intelligent scheduling unit automatically provides control strategies based on the system alarm level, enabling one-button start / stop and power adjustment of the pressure equalization ventilation equipment. It automatically adjusts the operating power based on monitoring data, changing the airflow and air pressure at the working face; the specific process is as follows: S5.1, Pressure equalization control: Adjust the dampers and fans (4) in the ventilation system according to the specific situation to maintain the air pressure balance in the goaf area, according to the ventilation resistance law: ； In the formula, The pressure difference between the two ends of the air leakage channel is expressed in Pa. R For the air resistance of the leakage channel, unit ; Q Leakage air volume, unit: m 3 / s; e This is an index representing the airflow pattern of the leaking air. Here we take 2; Right now: ; S5.

2. Real-time monitoring data and predictive data are transmitted to the safety situation early warning system. The system issues an early warning based on the early warning threshold. The equal pressure control system automatically adjusts the power of the fan (4) according to the early warning level to reduce the concentration of CO and CH4 dangerous gases in the working face and return airflow, ensuring that the working O2 content meets the standard. At the same time, the linkage damper and window control the pressure difference between 30 and 120 Pa to reduce the air leakage in the three zones and prevent fires. Specifically: The blue warning coal is in the oxidation and heating stage, and begins to produce CO gas, accompanied by CH4, which reduces the oxygen content of the working face. At this time, adjust the power of the blower (4) to increase the air volume, reduce the CO and CH4 concentrations of the working face, and ensure that the oxygen of the working face meets the standard. At the same time, adjust the damper and the window to reduce the control pressure difference to less than 50Pa. The coal under yellow warning is in the self-heating stage, and oxidation is further intensified, increasing the rate of oxygen consumption. The production of reaction products CO and CO2 begins to increase, with CO production more than three times higher than in the previous stage. The signs of coal spontaneous combustion, C2H4 and C2H6, begin to appear. At this time, while ensuring that the dangerous gases at the working face do not exceed the limits and the oxygen content meets the standards, the pressure difference should be adjusted to less than 20Pa, and coal spontaneous combustion should be prevented by controlling the air leakage. When the orange alert indicates that the coal is in the fission stage, the concentrations of CO, C2H4 and C2H6 increase significantly, with CO reaching 500 ppm and C2H4 reaching 2 ppm. At this time, the dampers and windows should be adjusted to make the pressure difference 0. A red alert indicates that the coal is in the combustion stage and has emitted the coal spontaneous combustion indicator gas C2H2. At this time, it is necessary to work with the coal mine's fire prevention and extinguishing department to carry out firefighting.

Images