Coal mine mining three-area heat-flow risk monitoring and early warning and voltage-sharing prevention and control system and method

Abstract

According to the coal mine mining three-area heat-flow risk monitoring early-warning and pressure-equalizing prevention and control system and method, a safety monitoring unit, a safety situation early-warning unit and a pressure-equalizing prevention and control and intelligent scheduling unit are arranged, and the safety monitoring unit collects coal mine underground environment index gas concentration and ventilation parameters in real time; monitoring ground atmospheric pressure through an air pressure sensor, and integrating excavation parameters such as excavation working face advancing speed, coal seam thickness and interlayer spacing; the collected parameters are transmitted to a security situation early warning unit through a bus; the safety situation early warning unit establishes a three-area heat flow risk system database, integrates sensor monitoring parameters, and stores data into a ground server according to a coding standard; a deep learning algorithm is combined with a heat flow risk early warning evaluation index system and model to realize risk level dynamic evaluation and alarm; and the voltage-sharing prevention and control and intelligent scheduling unit generates a control strategy according to the early warning level, and remotely controls the ventilation equipment in a linkage manner. The three-area heat flow risk intelligent advanced management and control is realized, and the safety level of a coal mine is improved.

Description

Technical Field

[0001] The invention relates to a coal mine mining three-zone thermal-flow risk monitoring, early warning and pressure equalization prevention and control system and method, belonging to the technical field of coal mine safety. Background Art

[0002] Under the influence of multiple strong mining, the goaf, adjacent goaf and overlying goaf of shallowly buried and close-range coal seams are prone to collapse and connect, forming a connected gas leakage channel between the ground, goaf and working face. A large amount of surface gas enters the mined coal rock through ground cracks, and the heat generated by the reaction between coal and oxygen and the oxidation-derived gas and low-concentration oxygen flow across regions to the goaf area of ​​the coal seam with low negative pressure, and finally flow into the working face, forming the so-called "three-zone" multi-gas (N2-CO2-CO-O2-CH4) breathing-overflow-oxidation temperature-raising effect, resulting in frequent "three-zone" thermal-flow disasters, such as coal rock spontaneous combustion, low oxygen in the working face and CO exceeding the limit.

[0003] At present, most coal mines use equal-pressure ventilation to control air leakage from fissures, which to a certain extent plays a role in preventing thermal flow disasters in the "three zones". However, the existing equal-pressure ventilation method is not intelligent enough, and there are serious deficiencies in the monitoring and early warning of thermal-flow risks in the "three zones", resulting in inaccurate identification of thermal-flow risks in the "three zones" and untimely early warning and control, which seriously restricts the safety of mining in shallow buried and self-igniting coal seams. Summary of the invention

[0004] The purpose of the present invention is to provide a system and method for monitoring, warning and equalizing pressure control of thermal-flow risks in three mining zones of a coal mine. The system and method can realize intelligent and advanced control of thermal-flow risks in the three zones of ground, goaf and working face, and ensure safe production in coal mines.

[0005] In order to achieve the above-mentioned purpose, the present invention provides a coal mine mining three-zone heat-flow risk monitoring and early warning and pressure-equalizing control system, including sensors, switches, analysis hosts, ventilation equipment, industrial ring networks, servers and clients; wherein the sensors, switches, analysis hosts and ventilation equipment are all arranged underground in the coal mine; the industrial ring network, servers and clients are all located above the coal mine; the server is provided with a safety monitoring unit, a safety situation early warning unit, a pressure-equalizing control and intelligent scheduling unit;

[0006] The safety monitoring unit collects the index gas concentration and ventilation parameters of the underground environment of the coal mine in real time through sensors; monitors the ground atmospheric pressure through air pressure sensors, and integrates the mining working face advancement speed, coal seam thickness, and interlayer spacing mining parameters; the safety monitoring unit transmits the collected parameters to the safety situation warning unit through the bus;

[0007] The safety situation warning unit is used to establish a thermal-flow risk system database for the three zones of ground, goaf and working face, integrate the monitoring parameters of the sensors in the three zones, and store the data in the well server according to the unified coding standard; the deep learning algorithm based on AI neural network is combined with the thermal-flow risk warning evaluation index system and model of the three zones to realize dynamic evaluation of risk level and conduct voice warning;

[0008] The pressure-equalizing control and intelligent dispatching unit automatically generates control strategies and remotely controls ventilation equipment according to the warning level to perform emergency response in advance;

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

[0010] The switch is used to connect the underground and the surface of the well, and realize information transmission through the industrial ring network;

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

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

[0013] Furthermore, fans, air ducts and an air volume sensor are installed in the air intake tunnel; dampers / windows are set in the return air tunnel to adjust the wind pressure on the working face; air volume sensor 2 is installed in the return air tunnel, and a pressure difference sensor is installed outside the damper of the return air tunnel, one end of the pressure difference sensor is installed within the two closed walls of the return air tunnel, and the other end is installed in a protective pipe connecting the return air tunnel with the deep part of the composite goaf; a mining intrinsically safe multi-parameter sensor is installed at the return air corner; a CO sensor and a methane sensor are set in the return air tunnel.

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

[0015] 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 mine intrinsically safe wireless sensor, monitors the ground atmospheric pressure through the air pressure sensor, and integrates the mining parameters such as the advancement speed of the mining face, the thickness of the coal seam, and the inter-layer spacing, and transmits the collected parameters to the safety situation warning unit through the bus;

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

[0017] S2.1. Use the Hampel filter to detect and eliminate outliers in the data. Use the median and absolute deviation to measure the discreteness of the data, so as to determine which data are outliers, and then replace the outliers with reasonable values. The Hampel filter formula is as follows:

[0018]

[0019] In the formula, x i is the i-th observation in the original data, y i is the ith observation after filtering, m i Is x i The median of the data in the window centered at s i Is x i The median absolute deviation of the data within the window centered on is:

[0020] s i =c·median(|x j -m i ∣)

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

[0022] Where 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, f = 3. The Hampel filter can effectively detect and eliminate spike noise and impulse noise in the data, and maintain the smoothness and shape characteristics of the data.

[0023] S2.2, use Lagrange interpolation to fill in the missing values. For a given k+1 point (x0, y0),…, (x k ,y k ), find a polynomial L(x) of degree not exceeding k such that L(x i )=y i For all i=0,…,k, the expressions of Lagrange polynomial and basic polynomial are:

[0024]

[0025] In x j The value is 1 at the point x, and at other points x i , that is, the value of i≠j is 0;

[0026] S3, the security situation warning unit uses the time series prediction algorithm to make advance predictions on the sensor monitoring parameters. The specific process is as follows:

[0027] S3.1. The time series prediction algorithm model includes an encoder and a decoder. The input of the encoder is the covariate sequence x at the past moment. 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 is the time when prediction starts; the input of the decoder is the covariate sequence x at the future time t:T ={x t ,x t+1 ,x t+2 ,…x T}, where T is the predicted cutoff time; the output of the model is the target variable sequence y at the future time t:T ={y t ,y t+1 ,y t+2 ,…y T};

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

[0029]

[0030] Where W Q , W K and W V is a learnable parameter matrix that maps the vector at each position in the input sequence to Q, K, V, W O Used to concatenate the output sub-vectors of multiple heads to obtain a complete output vector;

[0031] The feedforward neural network sublayer is used to perform nonlinear transformation on the representation of each position to increase the expressive power of the model. The output of the feedforward neural network sublayer of layer I in the encoder is:

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

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

[0034] S3.3. The decoder is used to predict the next target sequence based on the output of the encoder and the generated target sequence. The decoder is composed of N identical layers stacked together, each of which contains a multi-head self-attention sublayer, an encoder-decoder attention sublayer, and a feedforward neural network sublayer; the multi-head self-attention sublayer of the decoder uses a mask to cover the ungenerated positions. The calculation formula for 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 computes the correlation of each position in the generated target sequence with each position in the encoder output. The feedforward neural network sublayer in the decoder does the same thing as in the encoder.

[0039] S4. The security situation warning unit sets the hierarchical warning level. According to the importance of the indicator and the degree of over-limit of the monitoring data, the warning information is divided into four levels from blue, yellow, orange and red R1 to R4 from low to high urgency, and the alarm is automatically given according to the real-time monitoring data. The specific process is as follows:

[0040] S4.1. Based on the experimental analysis of coal oxidation producing carbon monoxide and carbon dioxide consuming oxygen 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 coal spontaneous combustion warning indicators;

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

[0042] S4.3. Combined with the requirements of coal mine safety regulations on the concentration of dangerous gases such as CO, CH4 and O2 in the working face, the warning level is set according to the three-zone thermal-flow risk coupling mechanism. The corresponding three-zone thermal-flow risk warning levels and warning values ​​are as follows:

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

[0044] Yellow warning: 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-equalizing control and intelligent dispatching unit automatically provides control strategies according to the system alarm level, realizes one-button start and stop and power adjustment of the pressure-equalizing ventilation equipment, automatically adjusts the operating power according to the monitoring data, and changes the air volume and air pressure of the working surface; the specific process is as follows:

[0048] S5.1. Pressure-equalizing control: Adjust the dampers and fan facilities in the ventilation system according to the specific situation, maintain the wind pressure balance in the goaf, reduce air leakage in the goaf, prevent fresh air from entering the spontaneous combustion area or goaf, cut off the oxygen supply to the fire source, and achieve suffocation and inerting. Therefore, the purpose of pressure-equalizing ventilation is to control the air leakage source. According to the ventilation resistance law:

[0049] Δh=RQ e ;

[0050] Where Δh is the pressure difference at both ends of the air leakage channel, unit: Pa; R is the wind resistance of the air leakage channel, unit: N·s 2 / m 8 ; Q is the air leakage volume, unit is m 3 / s; e is the index of the air leakage flow state, e = 1 ~ 2, here e is 2; that is:

[0051]

[0052] S5.2. Real-time monitoring data and prediction data are transmitted to the safety situation warning system. The system issues a warning according to the warning threshold. The pressure control system automatically adjusts the fan power according to the warning level to reduce the concentration of dangerous gases CO and CH4 in the working face and return air flow to ensure that the working O2 content meets the standard. At the same time, the damper and window are linked to control the pressure difference between 30 and 120 Pa to reduce the air leakage in the three zones to prevent fire. Specifically:

[0053] 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 in the working face. At this time, the fan power is adjusted to increase the air volume, reduce the CO and CH4 concentrations in the working face, and ensure that the oxygen content in 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 residual coal in the goaf to stop the oxidation and heating reaction due to lack of oxygen.

[0054] Yellow warning coal is in the self-heating stage, oxidation is further intensified, the oxygen consumption rate increases, the production of reaction products CO and CO2 begins to increase, CO production is more than 3 times higher than the previous stage, and coal spontaneous combustion sign gases C2H4 and C2H6 begin to appear. At this time, the pressure difference should be adjusted to less than 20Pa while ensuring that the dangerous gas in the working face does not exceed the limit and the oxygen content meets the standard, and the coal spontaneous combustion should be prevented by controlling the air leakage;

[0055] The orange warning coal is in the fission stage, when the concentrations of CO, C2H4 and C2H6 increase significantly, the CO concentration reaches 500ppm, and the C2H4 concentration reaches 2ppm. At this time, adjust the dampers and windows to make the pressure difference close to 0; by suppressing the air leakage in the "three zones", the coal in the goaf will spontaneously combust due to lack of oxygen and suffocation.

[0056] The red warning coal is in the combustion stage, and the coal spontaneous combustion sign gas C2H2 appears. At this time, it is necessary to cooperate with the coal mine fire prevention and firefighting department to implement fire extinguishing.

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

[0058] The present invention sets a safety monitoring unit, a safety situation warning unit, and a pressure equalization control and intelligent scheduling unit in the server. The safety monitoring unit collects the index gas concentration and ventilation parameters of the underground environment of the coal mine in real time through sensors; monitors the ground atmospheric pressure through a pressure sensor, and integrates the mining parameters such as the advancement speed of the mining working face, the thickness of the coal seam, and the interlayer spacing; the safety monitoring unit transmits the collected parameters to the safety situation warning unit through a bus; the safety situation warning unit is used to establish a heat-flow risk system database for the three zones of the ground-goaf-working face, integrates the sensor monitoring parameters of the three zones, and stores the data in the well server according to a unified coding standard; a deep learning algorithm based on an AI neural network is combined with a three-zone heat-flow risk warning evaluation index system and model to realize a dynamic evaluation of the risk level and perform voice warning; the pressure equalization control and intelligent scheduling unit automatically generates a control strategy according to the warning level, remotely links and controls the ventilation equipment, and performs pre-adjustment emergency treatment. Through multi-dimensional and multi-angle analysis, fire hazards are discovered in a timely manner. By changing the wind pressure on the working face 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 is achieved, thereby improving the safety production level of coal mines. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0060] Figure 2 It is a flowchart of the work flow of the pressure equalization control and intelligent scheduling unit of the present invention.

[0061] In the figure: 1. Air pressure sensor, 2. Multi-parameter sensor, 3. Differential pressure sensor, 4. Fan, 5. Wind duct, 6. Air door / window, 7. CO sensor, 8. Air volume sensor 1, 9. Air volume sensor 2, 10. Methane sensor, 11. Bundle tube monitoring equipment. DETAILED DESCRIPTION

[0062] The present invention will be further described below in conjunction with the accompanying drawings.

[0063] like Figure 1 As shown, a coal mine mining three-zone heat-flow risk monitoring and early warning and pressure-equalizing control system includes sensors, switches, analysis hosts, ventilation equipment, industrial ring network, servers and clients; wherein the sensors, switches, analysis hosts and ventilation equipment are all arranged underground in the coal mine; the industrial ring network, servers and clients are all located above the coal mine; the server is provided with a safety monitoring unit, a safety situation early warning unit, a pressure-equalizing control and intelligent scheduling unit;

[0064] The safety monitoring unit collects the index gas concentration and ventilation parameters of the underground environment of the coal mine in real time through sensors; monitors the ground atmospheric pressure through the air pressure sensor 1, and integrates the mining working face advancement speed, coal seam thickness, and interlayer spacing mining parameters; the safety monitoring unit transmits the collected parameters to the safety situation warning unit through the bus;

[0065] The safety situation warning unit is used to establish a thermal-flow risk system database for the three zones of ground, goaf and working face, integrate the monitoring parameters of the sensors in the three zones, and store the data in the well server according to the unified coding standard; the deep learning algorithm based on AI neural network is combined with the thermal-flow risk warning evaluation index system and model of the three zones to realize dynamic evaluation of risk level and conduct voice warning;

[0066] The pressure-equalizing control and intelligent dispatching unit automatically generates control strategies and remotely controls ventilation equipment according to the warning level to perform emergency response in advance;

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

[0068] The switch is used to connect the underground and the surface of the well, and realize information transmission through the industrial ring network;

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

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

[0071] As a preferred embodiment, a fan 4, a wind tube 5, and an air volume sensor 8 are installed in the air intake lane; the fan 4 is the air supply source for the pressure control and intelligent scheduling unit, and the air supply volume is regulated by adjusting the power; a damper / window 6 is set in the return air lane to adjust the wind pressure of the working face, and the air leakage is reduced by changing the air pressure difference between the goaf and the working face, providing parameters for the pressure control; the air volume sensor 8 measures the air intake volume, and provides parameters for the pressure control and intelligent scheduling unit. An air volume sensor 9 is installed in the return air lane to measure the return air volume. The difference between the return air volume and the air intake volume is the air leakage. A pressure differential sensor 3 is installed outside the return air door. One end of the pressure differential sensor 3 is installed 10m inside the two closed walls of the return air lane, and the other end is installed in the protective pipe connecting the return air lane and the deep part of the composite goaf. The pressure differential sensor 3 measures the pressure difference between the return air lane and the deep part of the composite goaf, and controls the pressure difference between 30 and 120Pa. A mine-used intrinsically safe multi-parameter sensor 2 is installed at the return air corner to monitor CO, CH4, O2, C2H4, C2H6 and C2H2 gases. A CO sensor 7 is installed in the return air lane and a CO alarm concentration of 2.4×10 -6 , providing parameters for pressure equalization control; a methane sensor 10 is set in the return air lane to monitor the CH4 concentration. The methane sensor 10 is installed within 0-5m of the damper. When the methane concentration in the working face and the return air flow is not less than 0.5%, the methane sensor 10 alarms, and when it is not less than 1.5%, the power is immediately cut off;

[0072] A bundle pipe monitoring device 11 is laid in the goaf to observe the application effect of the intelligent pressure-equalizing ventilation technology and monitor the degree of spontaneous combustion of coal in the goaf after pressure regulation. It can promptly detect signs of spontaneous combustion of coal in the composite goaf when excessive air leakage to the composite goaf occurs due to excessive working face pressure.

[0073] The mine pressure-equalizing 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 lockout value, the system main control switch outputs a control signal to control the power feed switch to achieve gas-electric lockout and wind-electric lockout functions. The mine intelligent local ventilation system is equipped with an audible and visual alarm device to prompt staff to take corresponding measures to resolve abnormal situations.

[0074] The fan is equipped with a mining explosion-proof and intrinsically safe lithium-ion battery, which is automatically charged under normal circumstances. When the system loses power due to abnormal CH4 or CO, the fan automatically enables the lithium-ion battery to operate normally to ensure normal ventilation of the working face.

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

[0076] 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 mine intrinsically safe wireless sensor, monitors the ground atmospheric pressure through the air pressure sensor 1, and integrates the mining parameters such as the advancement speed of the mining face, the thickness of the coal seam, and the interlayer spacing, and transmits the collected parameters to the safety situation warning unit through the bus;

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

[0078] S2.1. Use the Hampel filter to detect and eliminate outliers in the data. Use the median and absolute deviation to measure the discreteness of the data, so as to determine which data are outliers, and then replace the outliers with reasonable values. The Hampel filter formula is as follows:

[0079]

[0080] In the formula, x i is the i-th observation in the original data, y i is the ith observation after filtering, m i Is x i The median of the data in the window centered at s i Is x i The median absolute deviation of the data within the window centered on is:

[0081] s i =c·median(|x j -m i ∣)

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

[0083] Where 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, f = 3. The Hampel filter can effectively detect and eliminate spike noise and impulse noise in the data, and maintain the smoothness and shape characteristics of the data.

[0084] S2.2, use Lagrange interpolation to fill in the missing values. For a given k+1 point (x0, y0),…, (x k ,y k ), find a polynomial L(x) of degree not exceeding k such that L(x i )=y i For all i=0,…,k, the expressions of Lagrange polynomial and basic polynomial are:

[0085]

[0086] In x j The value is 1 at the point x, and at other points x i , that is, the value of i≠j is 0;

[0087] S3, the security situation warning unit uses the time series prediction algorithm to make advance predictions on the sensor monitoring parameters. The specific process is as follows:

[0088] S3.1. The time series prediction algorithm model includes an encoder and a decoder. The input of the encoder is the covariate sequence x at the past moment. 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 is the time when prediction starts; the input of the decoder is the covariate sequence x at the future time t:T ={x t ,x t+1 ,x t+2 ,…x T}, where T is the predicted cutoff time; the output of the model is the target variable sequence y at the future time t:T ={y t ,y t+1 ,y t+2 ,…y T};

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

[0090]

[0091] Where W Q , W K and W V is a learnable parameter matrix that maps the vector at each position in the input sequence to Q, K, V, W O Used to concatenate the output sub-vectors of multiple heads to obtain a complete output vector;

[0092] The feedforward neural network sublayer is used to perform nonlinear transformation on the representation of each position to increase the expressive power of the model. The output of the feedforward neural network sublayer of layer I in the encoder is:

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

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

[0095] S3.3. The decoder is used to predict the next target sequence based on the output of the encoder and the generated target sequence. The decoder is composed of N identical layers stacked together, each of which contains a multi-head self-attention sublayer, an encoder-decoder attention sublayer, and a feedforward neural network sublayer; the multi-head self-attention sublayer of the decoder uses a mask to cover the ungenerated positions. The calculation formula for 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 computes the correlation of each position in the generated target sequence with each position in the encoder output. The feedforward neural network sublayer in the decoder does the same thing as in the encoder.

[0100] S4. The security situation warning unit sets the hierarchical warning level. According to the importance of the indicator and the degree of over-limit of the monitoring data, the warning information is divided into four levels from blue, yellow, orange and red R1 to R4 from low to high urgency, and the alarm is automatically given according to the real-time monitoring data. The specific process is as follows:

[0101] S4.1. Based on the experimental analysis of coal oxidation producing carbon monoxide and carbon dioxide consuming oxygen 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 coal spontaneous combustion warning indicators;

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

[0103] S4.3. Combined with the requirements of coal mine safety regulations on the concentration of dangerous gases such as CO, CH4 and O2 at the working face, the warning level is set according to the three-zone thermal-flow risk coupling mechanism:

[0104] According to the test of coal spontaneous combustion and on-site observation, under normal circumstances, the CO concentration should be lower than 24ppm. If it exceeds 24ppm, a blue warning will be issued, and the temperature range is 40-60℃; according to the laboratory programmed heating experiment, the temperatures at which C2H4 and C2H6 first appear are both in the range of 70-80℃, and this temperature range is defined as a yellow warning; when the temperature reaches 100-130℃, it is an orange warning, and the concentrations of CO, C2H4 and C2H6 increase significantly, reaching 500ppm and 2ppm respectively. According to statistics on spontaneous combustion accidents in goafs, the concentration of C2H6 is generally about 3 times that of C2H4; when the coal temperature reaches 150-180℃, C2H2 will appear, and the CO concentration will increase significantly, which is the basis for determining a red warning.

[0105] The corresponding three-zone heat-flow risk warning levels and warning values ​​are as follows:

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

[0107] Yellow warning: 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-equalizing control and intelligent dispatching unit automatically provides control strategies according to the system alarm level, realizes one-button start and stop and power adjustment of the pressure-equalizing ventilation equipment, automatically adjusts the operating power according to the monitoring data, and changes the air volume and air pressure of the working surface; Figure 2 As shown, the specific process is as follows:

[0111] S5.1. Pressure-equalizing control: Adjust the dampers and fan facilities in the ventilation system according to the specific situation, maintain the wind pressure balance in the goaf, reduce air leakage in the goaf, prevent fresh air from entering the spontaneous combustion area or goaf, cut off the oxygen supply to the fire source, and achieve suffocation and inerting. Therefore, the purpose of pressure-equalizing ventilation is to control the air leakage source. According to the ventilation resistance law:

[0112] Δh=RQ e ;

[0113] Where Δh is the pressure difference at both ends of the air leakage channel, unit: Pa; R is the wind resistance of the air leakage channel, unit: N·s 2 / m 8 ; Q is the air leakage volume, unit is m 3 / s; k is the index of the air leakage flow state, k = 1 ~ 2; here e is 2;

[0114] Right now:

[0115]

[0116] S5.2. Real-time monitoring data and prediction data are transmitted to the safety situation warning system. The system issues a warning according to the warning threshold. The pressure control system automatically adjusts the fan power according to the warning level to reduce the concentration of dangerous gases CO and CH4 in the working face and return air flow to ensure that the working O2 content meets the standard. At the same time, the damper and window are linked to control the pressure difference between 30 and 120 Pa to reduce the air leakage in the three zones to prevent fire. Specifically:

[0117] 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 in the working face. At this time, the fan power is adjusted to increase the air volume, reduce the CO and CH4 concentrations in the working face, and ensure that the oxygen content in 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 residual coal in the goaf to stop the oxidation and heating reaction due to lack of oxygen.

[0118] Yellow warning coal is in the self-heating stage, oxidation is further intensified, the oxygen consumption rate increases, the production of reaction products CO and CO2 begins to increase, CO production is more than 3 times higher than the previous stage, and coal spontaneous combustion sign gases C2H4 and C2H6 begin to appear. At this time, the pressure difference should be adjusted to less than 20Pa while ensuring that the dangerous gas in the working face does not exceed the limit and the oxygen content meets the standard, and the coal spontaneous combustion should be prevented by controlling the air leakage;

[0119] The orange warning coal is in the fission stage, when the concentrations of CO, C2H4 and C2H6 increase significantly, the CO concentration reaches 500ppm, and the C2H4 concentration reaches 2ppm. At this time, adjust the dampers and windows to make the pressure difference close to 0; by suppressing the air leakage in the "three zones", the coal in the goaf will spontaneously combust due to lack of oxygen and suffocation.

[0120] The red warning coal is in the combustion stage, and the coal spontaneous combustion sign gas C2H2 appears. At this time, it is necessary to cooperate with the coal mine fire prevention and firefighting department to implement fire extinguishing.

Claims

1. A coal mine mining zone heat-flow risk monitoring, early warning and pressure control system, including sensors, switches, analysis host, ventilation equipment, industrial ring network, server and client; Sensors, switches, analysis hosts, and ventilation equipment are all installed underground in the coal mine; industrial ring networks, servers, and clients are all located above the coal mine; the server is characterized in that a safety monitoring unit, a safety situation early warning unit, and a pressure-balancing prevention and control and intelligent scheduling unit are provided in the server; The safety monitoring unit collects the index gas concentration and ventilation parameters of the underground environment of the coal mine in real time through sensors; monitors the ground atmospheric pressure through a pressure sensor (1), and integrates the mining parameters such as the advancement speed of the mining face, the thickness of the coal seam, and the interlayer spacing; the safety monitoring unit transmits the collected parameters to the safety situation warning unit through a bus; The safety situation warning unit is used to establish a thermal-flow risk system database for the three zones of ground, goaf and working face, integrate the monitoring parameters of the sensors in the three zones, and store the data in the well server according to the unified coding standard; the deep learning algorithm based on AI neural network is combined with the thermal-flow risk warning evaluation index system and model of the three zones to realize dynamic evaluation of risk level and conduct voice warning; The pressure-equalizing control and intelligent dispatching unit automatically generates control strategies and remotely controls ventilation equipment according to the warning level to perform emergency response in advance; The analysis host is used to store data fed back by the analysis host and the client; The switch is used to connect the underground and the surface of the well, and realize information transmission through the industrial ring network; The client is a human-computer interaction interface used to adjust the early warning control parameters of each system in the server.

2. The coal mine mining three-zone thermal-flow risk monitoring, early warning and pressure control system according to claim 1 is characterized in that: The sensors include a multi-parameter sensor (2), an air volume sensor, and a pressure difference sensor (3); the coal mine underground environmental index gases collected in real time include O2, CO, CH4, C2H4, C2H2, and C2H6; the ventilation parameters include wind speed, pressure difference, and air pressure; the three-zone sensor monitoring parameters integrated in the safety situation warning unit include ventilator (4) operating condition information, O2 concentration, CO concentration, CH4 concentration, C2H4 concentration, C2H2 concentration, C2H6 concentration, temperature and humidity, wind speed, pressure difference, and air pressure environmental parameters; the ventilation equipment includes a fan (4), a wind tube (5), and a damper / window (6); and the industrial ring network is a wireless WIFI or 5G network.

3. The coal mine mining three-zone thermal-flow risk monitoring, early warning and pressure control system according to claim 2 is characterized in that: A fan (4), a wind tube (5), and an air volume sensor (8) are installed in the air intake tunnel; a wind door / window (6) is installed in the return air tunnel for adjusting the wind pressure of the working face; two air volume sensors (9) are installed in the return air tunnel, and a pressure difference sensor (3) is installed outside the wind door of the return air tunnel. One end of the pressure difference sensor (3) is installed inside two closed walls of the return air tunnel, and the other end is installed in a protective pipe connecting the return air tunnel and the deep part of the composite goaf; a mining intrinsically safe multi-parameter sensor (2) is installed at the return air corner; a CO sensor (7) and a methane sensor (10) are installed in the return air tunnel.

4. A method for monitoring, early warning and pressure control of thermal and flow risks in three mining zones of a coal mine, characterized in that: The steps include: 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 mine intrinsically safe wireless sensor, monitors the ground atmospheric pressure through the air pressure sensor (1), and integrates the mining parameters such as the advancement speed of the mining face, the thickness of the coal seam, and the interlayer spacing, and transmits the collected parameters to the safety situation warning unit through the bus; S2. The security situation warning unit processes the received data through a multi-heterogeneous data fusion model. The specific process is as follows: S2.

1. Use the Hampel filter to detect and eliminate outliers in the data. The Hampel filter formula is as follows: In the formula, x i is the i-th observation in the original data, y i is the i-th observation after filtering, m i Is x i The median of the data in the window centered at s i Is x i The median absolute deviation of the data within the window centered on is: s i =c·median(∣x j -m i ∣) j=i-(n-1) / 2,...,i+(n-1) / 2; Where n is the window size, c is a constant, and f is a threshold parameter used to control the sensitivity of outlier detection; S2.2, use Lagrange interpolation to fill in the missing values. For a given k+1 point (x0, y0),…, (x k ,y k ), find a polynomial L(x) of degree not exceeding k such that L(x i )=y i For all i=0,…,k, the expressions of Lagrange polynomial and basic polynomial are: In x j The value is 1 at the point x, and at other points x i , that is, the value of i≠j is 0; S3, the security situation warning unit uses the time series prediction algorithm to make advance predictions on the sensor monitoring parameters. The specific process is as follows: S3.

1. The time series prediction algorithm model includes an encoder and a decoder. The input of the encoder is the covariate sequence x at the past moment. 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 is the time when prediction starts; the input of the decoder is the covariate sequence x at the future time t:T ={x t ,x t++1 ,x t++2 ,…x T }, where T is the predicted cutoff time; the output of the model is the target variable sequence y at the future time t:T ={y t ,y t++1 ,y t++2 ,…y T }; S3.

2. The encoder is composed of N identical layers stacked together, each of which contains a multi-head self-attention sublayer and a feedforward neural network sublayer; the output of the multi-head attention sublayer of layer I in the encoder is: Where W Q , W K and W V is a learnable parameter matrix that maps the vector at each position in the input sequence to Q, K, V, W O Used to concatenate the output sub-vectors of multiple heads to obtain a complete output vector; The feedforward neural network sublayer is used to perform a nonlinear transformation on the representation of each position. The output of the feedforward neural network sublayer of layer I in the encoder is: FFN(x)=max(0,xW1+b1)W2+b2; Where max(0,·) is the ReLU activation function; W1, W2 are learnable parameter matrices; b1, b2 are learnable bias vectors; S3.

3. The decoder is used to predict the next target sequence based on the output of the encoder and the generated target sequence. The decoder is composed of N identical layers stacked together, each of which contains a multi-head self-attention sublayer, an encoder-decoder attention sublayer, and a feedforward neural network sublayer; the multi-head self-attention sublayer of the decoder uses a mask to cover the ungenerated positions. The calculation formula for 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 warning unit sets the hierarchical warning level. According to the importance of the indicator and the degree of over-limit of the monitoring data, the warning information is divided into four levels from blue, yellow, orange and red R1 to R4 from low to high urgency, and the alarm is automatically given according to the real-time monitoring data. The specific process is as follows: S4.

1. Based on the experimental analysis of coal oxidation producing carbon monoxide and carbon dioxide consuming oxygen 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 coal spontaneous combustion warning indicators; S4.

2. The corresponding relationship between C2H4 and C2H2 / C2H6 and coal temperature was determined by analyzing the pyrolysis gas, and C2H4, C2H6 and C2H2 were selected as coal spontaneous combustion warning indicators; six coal spontaneous combustion warning indicators were selected as CO, O2, CH4, C2H4, C2H6 and C2H2; S4.

3. Combined with the requirements of coal mine safety regulations on the concentration of dangerous gases such as CO, CH4 and O2 in the working face, the warning level is set according to the three-zone thermal-flow risk coupling mechanism. The corresponding three-zone thermal-flow risk warning levels and warning values ​​are as follows: Blue warning: T m >40℃; R1={{24ppm<[CO]}∪{0.5%<[CH4]}∪{20%>[O2]}} Yellow warning: T m >60℃; R2={{50ppm<[CO]}∩{{[C2H4]>0}∪{[C2H6]>0}}} Orange alert: T m >90℃; R3={{500ppm<[CO]}∪{[C2H4]>2ppm}} Red Alert: T m >130℃; R4={[C2H2]>0}; S5, the pressure-equalizing control and intelligent dispatching unit automatically provides control strategies according to the system alarm level, realizes one-button start and stop and power adjustment of the pressure-equalizing ventilation equipment, automatically adjusts the operating power according to the monitoring data, and changes the air volume and air pressure of the working surface; the specific process is as follows: S5.

1. Pressure control Adjust the dampers and fans (4) in the ventilation system according to the specific situation to maintain the wind pressure balance in the goaf. According to the ventilation resistance law: Δh=RQ e ; Where Δh is the pressure difference at both ends of the air leakage channel, unit: Pa; R is the wind resistance of the air leakage channel, unit: N·s 2 / m 8 ; Q is the air leakage volume, unit is m 3 / s; e is the index of the leakage airflow state, e = 1 ~ 2, here we take 2; Right now: S5.

2. Real-time monitoring data and prediction data are transmitted to the safety situation warning system. The system issues a warning according to the warning threshold. The pressure equalization control system automatically adjusts the power of the fan (4) according to the warning level to reduce the concentration of dangerous gases CO and CH4 in the working face and return air flow to ensure that the working O2 content meets the standard. At the same time, the damper and window are linked to control the pressure difference between 30 and 120 Pa to reduce the air leakage in the three zones to prevent fire. Specifically: The blue warning coal is in the oxidation and temperature rising stage, and begins to produce CO gas, accompanied by CH4, which reduces the oxygen content in the working face. At this time, the fan (4) power is adjusted to increase the air volume, reduce the CO and CH4 concentrations in the working face, and ensure that the oxygen in the working face meets the standard. At the same time, the damper and window are adjusted to reduce the control pressure difference to less than 50Pa. Yellow warning coal is in the self-heating stage, oxidation is further intensified, the oxygen consumption rate increases, the production of reaction products CO and CO2 begins to increase, CO production is more than 3 times higher than the previous stage, and coal spontaneous combustion sign gases C2H4 and C2H6 begin to appear. At this time, the pressure difference should be adjusted to less than 20Pa while ensuring that the dangerous gas in the working face does not exceed the limit and the oxygen content meets the standard, and the coal spontaneous combustion should be prevented by controlling the air leakage; Orange warning coal is in the fission stage. At this time, the concentrations of CO, C2H4 and C2H6 increase significantly. The CO concentration reaches 500ppm and the C2H4 concentration reaches 2ppm. At this time, the damper and window are adjusted to make the pressure difference close to 0. The red warning coal is in the combustion stage, and the coal spontaneous combustion sign gas C2H2 appears. At this time, it is necessary to cooperate with the coal mine fire prevention and firefighting department to implement fire extinguishing.

Images