Coal mine fireproof monitoring and analysis method based on unmanned aerial vehicle inspection

Abstract

The invention discloses a coal mine fireproof monitoring and analysis method based on unmanned aerial vehicle inspection, which relates to the technical field of coal mine fireproof monitoring and analysis, and comprises the following steps: when an unmanned aerial vehicle arrives at a coal mine goaf area for inspection, acquiring environmental parameter information of the coal mine goaf area in real time, wherein the characteristic information comprises gas distribution characteristic information and airflow dynamic characteristic information. According to the method, the gas distribution characteristic information and the airflow dynamic characteristic information are acquired in real time, and the gas distribution index is generated by utilizing the machine learning model which is learned in advance, so that the goaf gas distribution is intelligently evaluated. According to gas distribution indexes, abnormal distribution is divided into low risk, medium risk and high risk, refined inspection strategies such as frequency increasing, path expanding, dynamic height adjusting and multi-machine collaborative operation are adopted for different risk levels, comprehensive monitoring and quick response are achieved, inspection resource distribution is optimized, and coal mine safety accidents are effectively prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine fire prevention monitoring and analysis, and in particular to a coal mine fire prevention monitoring and analysis method based on unmanned aerial vehicle inspection. Background Art

[0002] Coal mine fire prevention monitoring and analysis based on drone inspection is a modern method that uses drone technology to patrol, detect and analyze data in coal mine areas. Coal mine areas are prone to fire due to factors such as spontaneous combustion of coal seams and accumulation of gas in goafs. Drones can achieve all-round and real-time monitoring of coal mine areas by carrying infrared thermal imagers, gas sensors and high-definition cameras. Compared with traditional manual inspections, drones can quickly cover large areas, especially some areas that are difficult to reach or dangerous, reducing labor costs and potential safety risks. The core of drone inspections is to capture abnormal temperature changes, gas concentration fluctuations, etc. in real time through advanced sensing equipment and data processing capabilities, and then warn of coal mine fire risks in advance.

[0003] This fire monitoring system is not just a simple image acquisition, but also includes in-depth analysis based on drone inspection data. By conducting big data analysis on temperature, gas, video and other information collected by drones, combined with historical data and geographical environment characteristics, potential fire sources can be identified and fire risk reports can be generated. At the same time, artificial intelligence algorithms can further analyze data, predict the development trend of fires, and formulate corresponding fire prevention measures. This technology not only improves the accuracy and efficiency of coal mine fire monitoring, but also provides a scientific basis for coal mine managers, helping to optimize fire prevention decisions and reduce the threat of fire to coal mine safety production.

[0004] The prior art has the following deficiencies:

[0005] Gas sensors detect changes in the concentration of harmful gases (such as carbon monoxide and methane) in the air to determine whether there is spontaneous combustion in the coal mine area. However, due to the complex air flow in the goaf, the airflow may form eddies or stagnation zones in different areas, resulting in uneven gas distribution. If the flight route or height of the drone inspection is unreasonable, the peak or trough of the key concentration may be missed, leading to misjudgment of the gas accumulation. Especially at the edge of the goaf, the gas concentration changes dramatically and the accumulation is easily underestimated. If the accumulation of combustible gases (such as methane) is not accurately detected, the concentration in some areas may reach the lower explosion limit. There are usually potential fire sources in coal mines, such as self-igniting coal seams or sparks generated by mechanical equipment. Once monitoring is not timely, any fire source may cause an explosion, threatening the lives of miners and damaging equipment, causing serious production accidents.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0007] The purpose of the present invention is to provide a coal mine fire prevention monitoring and analysis method based on drone inspection, which can realize intelligent evaluation of gas distribution in goaf areas by acquiring gas distribution characteristic information and gas flow dynamic characteristic information in real time and using a pre-learned machine learning model to generate a gas distribution index. According to the gas distribution index, the abnormal distribution is divided into three categories: low risk, medium risk, and high risk. Refined inspection strategies are adopted for different risk levels, such as increasing frequency, expanding paths, dynamically adjusting height, and multi-machine collaborative operation, etc., to achieve comprehensive monitoring and rapid response, optimize inspection resource allocation, and effectively prevent the occurrence of coal mine safety accidents, so as to solve the problems in the above-mentioned background technology.

[0008] In order to achieve the above object, the present invention provides the following technical solution: a coal mine fire monitoring and analysis method based on drone inspection, comprising the following steps:

[0009] When the UAV arrives at the coal mine goaf area for inspection, it obtains the environmental parameter information of the coal mine goaf area in real time, including the gas distribution characteristic information and the air flow dynamic characteristic information;

[0010] Perform an abnormal analysis on the acquired gas distribution characteristic information and gas flow dynamic characteristic information, and input the gas distribution characteristic information and gas flow dynamic characteristic information after abnormal analysis into a pre-learned machine learning model, generate a gas distribution index through the machine learning model, and perform intelligent evaluation of the ambient gas in the goaf through the gas distribution index;

[0011] Based on the machine learning evaluation results, i.e., the gas distribution index, the ambient gas distribution in the goaf is divided into a normal gas distribution state or an abnormal gas distribution state, and the abnormal gas distribution in the goaf is intelligently perceived;

[0012] According to the abnormal gas distribution status, the environmental gas distribution information of the goaf is continuously obtained, and the abnormal distribution of the environmental gas in the goaf is further divided into low-risk abnormal distribution, medium-risk abnormal analysis and high-risk abnormal distribution;

[0013] Take different response measures for different risk anomaly distributions.

[0014] Preferably, the gas distribution characteristic information includes edge accumulation phenomenon, which refers to the accumulation of gas in the edge area of ​​the goaf or mining area. The airflow dynamics characteristic information includes airflow velocity difference, which describes the inconsistency of velocity between different airflows in the goaf, and is the velocity change between hierarchical airflows or regional airflows.

[0015] Preferably, an abnormal analysis is performed on the edge accumulation phenomenon and the airflow velocity difference to generate an edge effect concentration index and an airflow shear unevenness index, respectively. The edge effect concentration index and the airflow shear unevenness index generated after the abnormal analysis are input into a pre-learned machine learning model, and a gas distribution index is generated through the machine learning model. The gas distribution index is used to perform an intelligent assessment of the environmental gas in the goaf.

[0016] Preferably, it is characterized in that the edge accumulation phenomenon is analyzed abnormally, and the logic of generating the edge effect concentration index is as follows:

[0017] In the detection window, calculate the gas concentration gradient at different positions in the goaf to determine the severity of the concentration change. The calculation expression is as follows:

[0018]

[0019] , where is the gas concentration gradient vector, which represents the change in concentration per unit distance, is the rate of change of gas concentration in the x direction, that is, the concentration change per unit distance, is the rate of change of gas concentration in the y direction, is the rate of change of gas concentration in the z direction, are unit direction vectors, representing the x, y, and z directions respectively;

[0020] The airflow disturbance coefficient measures the stability and volatility of the airflow. The size of the disturbance coefficient directly affects the diffusion pattern of the gas in the goaf. The calculation expression is as follows: In the formula, Φ(x,y,z) represents the airflow disturbance coefficient, which is used to describe the degree of airflow turbulence in a local area, v = (v x ,v y ,v z ), v is the airflow velocity vector field, which is used to represent the airflow velocity in the x, y, and z directions. represents the absolute value of the airflow divergence, is the curl of the airflow, γ is the curl influence weight coefficient, which is used to adjust the influence of the curl on the airflow disturbance coefficient Φ(x,y,z);

[0021] The gas concentration gradient is combined with the airflow disturbance coefficient to determine whether the gas accumulates in the edge area. The calculation expression is as follows: Where k(x,y,z) is the edge accumulation coefficient, α is the sensitivity coefficient, and exp(·) is the exponential function;

[0022] In order to capture the dynamic changes of gas accumulation, the edge accumulation coefficient is time-weighted integrated to generate the edge effect concentration index. The calculation expression is as follows:

[0023]

[0024] , where EECI is the edge effect concentration index, t1 is the start time of the detection window, t2 is the end time of the detection window, and β(t) is the time weighting function.

[0025] Preferably, the air flow velocity difference is analyzed abnormally, and the logic of generating the air flow shear unevenness index is as follows:

[0026] Under the detection window, the entire detection area is divided into a three-dimensional space grid, and the points in the coordinate system are defined as (x, y, z). The airflow velocity components obtained by the sensor at each sampling point are combined into a velocity vector. The airflow velocity vectors of all sampling points are combined to form a three-dimensional velocity vector field covering the detection window. The expression is as follows: V(x, y, z, t) = (v x ,v y ,v z ), where V(x,y,z,t) is the three-dimensional velocity vector field, which is the airflow velocity component at the spatial coordinates x,y,z at each time t;

[0027] The local shear rate is calculated by the partial derivative of the three-dimensional velocity vector field and calibrated as the local shear rate tensor. The calculation expression is as follows:

[0028]

[0029] , where is the gradient tensor of the airflow velocity field, represents the airflow velocity component v in the x, y, and z directions x ,v y ,v z As the rate of change of the x position, represents the airflow velocity component v in the x, y, and z directions x ,v y ,v z With the rate of change of y position;

[0030] Curl is a measure of the vortex characteristics of airflow. Curl reveals the rotation phenomenon between airflows at different levels. The calculation expression of curl is as follows: The specific form after expansion is as follows:

[0031]

[0032] , where Ω(x,y,z,t) represents the curl;

[0033] Turbulence intensity measures the instability and randomness of airflow and is an important parameter for describing complex airflow. The calculation expression of local turbulence intensity is as follows:

[0034]

[0035] , where τ(x,y,z,t) is the local turbulence intensity;

[0036] In order to identify abnormal shear phenomena in the airflow, the curl and turbulence intensity are combined to construct an abnormal detection matrix. The calculation expression is as follows:

[0037] A(x,y,z,t)=Ω(x,y,z,t)·τ(x,y,z,t)

[0038] , where A(x,y,z,t) is the anomaly detection matrix;

[0039] The anomaly detection matrix is ​​integrated to generate the airflow shear unevenness index. The calculation expression is as follows:

[0040] FSII=∫ D A(x,y,z,t)dxdydz

[0041] , where FSII represents the airflow shear unevenness index and D represents the detection window.

[0042] Preferably, the edge effect concentration index EECI and the airflow shear unevenness index FSII are input into a pre-learned machine learning model, and the gas distribution index GDI is generated by the machine learning model according to the following formula:

[0043]

[0044] , where r1 and r2 are the preset proportional coefficients of the edge effect concentration index EECI and the airflow shear unevenness index FSII, respectively, and r1 and r2 are both greater than 0.

[0045] Preferably, the generated gas distribution index is compared and analyzed with a preset gas distribution index reference threshold value to divide the gas distribution state of the goaf environment, and the division results are as follows:

[0046] If the gas distribution index is greater than or equal to a preset gas distribution index reference threshold, the goaf environment gas distribution state is classified as an abnormal gas distribution state;

[0047] If the gas distribution index is less than a preset gas distribution index reference threshold, the gas distribution state of the goaf environment is classified as a normal gas distribution state.

[0048] Preferably, when the ambient gas distribution state of the goaf is classified as an abnormal gas distribution state, a plurality of gas distribution indexes subsequently generated from the ambient gas distribution of the goaf of the coal mine are continuously obtained to establish an analysis set for comprehensive analysis;

[0049] The average value and standard deviation of several gas distribution indices in the analysis set are calculated, and the calculated average value and standard deviation of the gas distribution index are compared with the preset reference threshold of the gas distribution index average value and the preset reference threshold of the gas distribution index standard deviation, respectively, to further divide the abnormal gas distribution state. The division results are as follows:

[0050] If the average value of the gas distribution index is less than the preset reference threshold value of the average value of the gas distribution index, the abnormal distribution of the ambient gas in the goaf is further classified as a low-risk abnormal distribution;

[0051] If the average value of the gas distribution index is less than the preset reference threshold of the average value of the gas distribution index and the standard deviation of the gas distribution index is greater than or equal to the preset reference threshold of the standard deviation of the gas distribution index, the abnormal distribution of the ambient gas in the goaf is further classified as a medium-risk abnormal distribution;

[0052] If the average value of the gas distribution index is greater than or equal to the preset reference threshold of the average value of the gas distribution index and the standard deviation of the gas distribution index is greater than or equal to the preset reference threshold of the standard deviation of the gas distribution index, the abnormal distribution of the ambient gas in the goaf is further divided into high-risk abnormal distribution.

[0053] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0054] The present invention realizes intelligent evaluation of gas distribution in goaf by acquiring gas distribution characteristic information and airflow dynamic characteristic information in real time, and using a pre-learned machine learning model to generate a gas distribution index. The edge effect concentration index and the airflow shear unevenness index are used as the input of the model, so that the system can accurately capture the gas accumulation in different areas, especially high-risk areas such as edges and cracks that are easily overlooked. With the help of real-time analysis and automatic evaluation of data, the problem of missed detection of concentration peaks or valleys caused by unreasonable flight altitude or path of the drone is avoided. At the same time, the system dynamically adjusts the flight path and altitude to make gas distribution monitoring more comprehensive and accurate, significantly reducing the risk of explosion or poisoning accidents caused by misjudgment of gas accumulation.

[0055] The present invention realizes refined management of inspection strategies by dividing the analysis results of the gas distribution index into three categories: low risk, medium risk, and high risk, targeting abnormal distribution states at different risk levels. For low-risk situations, the accumulation of potential risks is prevented by appropriately increasing the inspection frequency and expanding the coverage of edge areas; for medium-risk situations, multi-level flight strategies and dynamic path adjustments are adopted to ensure that areas with large changes in gas concentration are captured; for high-risk situations, high-frequency inspections and the deployment of multi-machine collaborative operations enable the system to have stronger monitoring coverage and emergency response speed. This hierarchical inspection scheme optimizes the allocation of inspection resources, improves the adaptability and reliability of drones in complex environments, and realizes rapid early warning and emergency response by uploading data to the control center in real time, greatly reducing the probability of coal mine production safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0057] Figure 1 The present invention is a method flow chart of a coal mine fire monitoring and analysis method based on drone inspection. DETAILED DESCRIPTION

[0058] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0059] The present invention provides Figure 1 A coal mine fire prevention monitoring and analysis method based on drone inspection is shown, comprising the following steps:

[0060] When the UAV arrives at the coal mine goaf area for inspection, it obtains the environmental parameter information of the coal mine goaf area in real time, including the gas distribution characteristic information and the air flow dynamic characteristic information;

[0061] Gas distribution characteristic information includes edge accumulation phenomenon, which refers to the accumulation of gas at the edge of the goaf or mining area. Airflow dynamic characteristic information includes airflow velocity difference, which describes the inconsistency of the velocity between different airflows in the goaf, and is the velocity change between hierarchical airflows or regional airflows.

[0062] Perform an abnormal analysis on the acquired gas distribution characteristic information and gas flow dynamic characteristic information, and input the gas distribution characteristic information and gas flow dynamic characteristic information after abnormal analysis into a pre-learned machine learning model, generate a gas distribution index through the machine learning model, and perform intelligent evaluation of the ambient gas in the goaf through the gas distribution index;

[0063] Anomaly analysis is performed on edge accumulation phenomenon and airflow velocity difference to generate edge effect concentration index and airflow shear unevenness index respectively. The edge effect concentration index and airflow shear unevenness index generated after anomaly analysis are input into a pre-learned machine learning model, and a gas distribution index is generated through the machine learning model. The gas distribution index is used to perform intelligent evaluation of the ambient gas in the goaf.

[0064] Changes in the edge accumulation phenomenon in coal mine goafs do cause the airflow to form eddies or stagnant areas in different areas, thus causing uneven gas distribution. This phenomenon is mainly related to the air flow characteristics in the edge areas: when the airflow speed slows down or encounters obstacles, the gas tends to be retained in the edge or crack area, forming a local high-concentration accumulation point. At the same time, edge accumulation changes the flow pattern of the overall airflow, resulting in inconsistent airflow speeds in different areas, shearing effects, and turbulence and eddies. This makes the concentration of harmful gases such as methane or carbon monoxide unstable, and high concentrations of gas may be confined to certain specific areas or levels. If the flight route or height of the drone inspection is unreasonable, these critical concentration peaks or troughs may be missed. For example, if the drone only inspects at higher or lower flight altitudes, it may not be able to fully capture the concentration differences of gases at different levels. In addition, if the inspection route planning is unscientific and fails to cover the edge or crack areas where gas is prone to accumulate, it will lead to missed inspections and missed potential risk warnings. Since the accumulation of gas concentration is directly related to major safety hazards such as fire, explosion and poisoning, any misjudgment of gas distribution may delay the implementation of prevention and control measures, posing a serious threat to miners' lives and coal mine equipment.

[0065] The edge accumulation phenomenon is analyzed abnormally, and the logic of generating edge effect concentration index is as follows:

[0066] In the detection window, calculate the gas concentration gradient at different positions in the goaf to determine the severity of the concentration change. The calculation expression is as follows:

[0067]

[0068] , where is the gas concentration gradient vector, which represents the change in concentration per unit distance, is the rate of change of gas concentration in the x direction, that is, the concentration change per unit distance, is the rate of change of gas concentration in the y direction, is the rate of change of gas concentration in the z direction, are unit direction vectors, representing the x, y, and z directions respectively;

[0069] The gas composition in the goaf is complex, and changes in concentration directly affect the safety management of the mine. The following lists the types of gases that need to be monitored in the goaf and their concentrations:

[0070] 1. Methane concentration (CH4Concentration)

[0071] Source: Coalbed methane release.

[0072] Safety risk: Methane is a flammable and explosive gas. When its concentration reaches the lower explosion limit (LEL, about 5%), it will explode when it encounters a fire source.

[0073] Detection requirements: Methane accumulation in the retained areas of the goaf should be monitored to prevent explosion risks.

[0074] 2. Carbon monoxide concentration (CO Concentration)

[0075] Source: Spontaneous combustion or incomplete combustion of coal seams.

[0076] Safety risks: Carbon monoxide is a colorless, odorless, toxic gas that can poison miners. Elevated levels of carbon monoxide are often an early sign of spontaneous combustion in coal seams.

[0077] Detection requirements: It is necessary to monitor the changes in carbon monoxide concentration in different areas in real time to warn of potential fire risks.

[0078] 3. Carbon dioxide concentration (CO2Concentration)

[0079] Source: spontaneous combustion, gas desorption, respiratory metabolism, etc.

[0080] Safety risk: High concentrations of carbon dioxide can reduce the oxygen content in the air and cause miners to suffocate.

[0081] Testing requirements: Attention should be paid to the accumulation of CO2 concentration in poorly ventilated areas to ensure air quality safety.

[0082] 4. Oxygen concentration (O2Concentration)

[0083] Source: Natural air supply.

[0084] Safety risk: Oxygen concentrations below 19.5% can cause suffocation risk. Too high oxygen concentrations can promote spontaneous combustion.

[0085] Detection Need: Proper oxygen levels need to be maintained to avoid the risk of asphyxiation and inhibit spontaneous combustion of coal seams.

[0086] 5. Hydrogen sulfide concentration (H2S Concentration)

[0087] Source: Decomposition of sulfur-containing minerals or organic matter.

[0088] Safety risks: Hydrogen sulfide is a highly toxic gas that smells like rotten eggs at low concentrations and can cause olfactory impairment and poisoning at high concentrations.

[0089] Detection needs: Focus on monitoring H2S concentration in cracks and low-lying areas to prevent poisoning accidents.

[0090] 6. Nitrogen concentration (N2Concentration)

[0091] Source: Nitrogen injection into goaf to prevent fire or changes in air composition.

[0092] Safety risk: High concentration of nitrogen will reduce the oxygen content and cause suffocation due to lack of oxygen.

[0093] Testing requirements: The nitrogen concentration needs to be reasonably controlled to ensure that nitrogen injection for fire prevention does not affect the oxygen supply in the mining area.

[0094] 7. Hydrogen concentration (H2Concentration)

[0095] Sources: Battery or equipment leakage, or byproducts of spontaneous combustion of coal seams.

[0096] Safety risks: Hydrogen is highly flammable and explosive.

[0097] Detection requirements: The hydrogen concentration needs to be monitored near the goaf detection equipment to avoid explosions caused by equipment failure.

[0098] The gas concentration monitoring in the goaf includes methane, carbon monoxide, carbon dioxide, oxygen, hydrogen sulfide, nitrogen and hydrogen. The changes in the concentration of these gases not only reflect the ventilation and combustion conditions inside the coal mine, but are also directly related to the safety of miners. The drone monitoring system needs to conduct regional and high-frequency detection of these key gases to ensure timely warning and prevention of accidents.

[0099] The airflow disturbance coefficient measures the stability and volatility of the airflow. The size of the disturbance coefficient directly affects the diffusion pattern of the gas in the goaf. The calculation expression is as follows: In the formula, Φ(x,y,z) represents the airflow disturbance coefficient, which is used to describe the degree of airflow turbulence in a local area, v = (v x ,v y ,v z), v is the airflow velocity vector field, which is used to represent the airflow velocity in the x, y, and z directions. represents the absolute value of the airflow divergence, is the curl of the airflow, γ is the curl influence weight coefficient, which is used to adjust the influence of the curl on the airflow disturbance coefficient Φ(x,y,z);

[0100] The airflow disturbance coefficient is an indicator that describes the stability and dynamic characteristics of airflow in goafs. It takes into account the divergence (divergence or contraction) and swirl (eddy strength) of the airflow, and is used to determine whether there is obvious turbulence, stagnation or gas accumulation in a certain area. A higher disturbance coefficient means that the airflow in the area is unstable, and it is more likely to form gas accumulation or eddy areas, increasing the difficulty of detection and early warning. In coal mines and goafs, the airflow disturbance coefficient is of great significance for optimizing drone inspection routes: it helps to identify potential dangerous gas accumulation points, prevent gas accumulation in local areas due to uneven airflow, and reduce the risk of fire or explosion.

[0101] The divergence of airflow is a scalar that describes how the airflow diverges or converges at a certain point. Specifically, the divergence represents the difference between the outflow rate and the inflow rate of gas mass per unit volume. When the divergence is positive, it means that the airflow diverges outward from this point and the gas concentration decreases; when the divergence is negative, it means that the airflow converges toward this point and may form a local accumulation area. Divergence is very critical in gas monitoring in goafs: areas with negative divergence are prone to gas retention and concentration accumulation, increasing the risk of fire and explosion, and require drone inspections.

[0102] The curl (Cur l) of an airflow is a vector that represents the rotation or vortex intensity of an airflow in a certain area. It reveals the relative movement between airflows in different directions and describes the rotation and turbulence of airflow in space. The size of the curl determines whether there is eddy or turbulence in a certain area, which in turn affects the diffusion pattern of gas in the area. Areas with high curl mean that airflows are prone to form vortices, which hinder the normal diffusion of gas and cause local retention or accumulation. If there are high-curl areas in the goaf, drone inspections need to cover these areas more densely to avoid missing high-concentration gas and prevent the occurrence of safety hazards.

[0103] The gas concentration gradient is combined with the airflow disturbance coefficient to determine whether the gas accumulates in the edge area. The calculation expression is as follows: Where k(x, y, z) is the edge accumulation coefficient, which indicates the degree of gas accumulation at a certain point (x, y, z) in the three-dimensional space of the goaf; α is the sensitivity coefficient, which is used to adjust the coupling strength between the concentration gradient and the airflow disturbance; exp(·) is an exponential function, which is used to amplify or attenuate the nonlinear relationship between the concentration gradient and the airflow disturbance;

[0104] In order to capture the dynamic changes of gas accumulation, the edge accumulation coefficient is time-weighted integrated to generate the edge effect concentration index. The calculation expression is as follows:

[0105]

[0106] , where EECI is the edge effect concentration index, t1 is the start time of the detection window, t2 is the end time of the detection window, and β(t) is the time weighting function, which assigns different weights to data at different time points;

[0107] From the calculation expression of the edge effect concentration index, it can be seen that the larger the edge effect concentration index performance value generated after the abnormal analysis of the edge accumulation phenomenon, the more serious the gas accumulation in the edge area of ​​the goaf, which is usually accompanied by stagnation or turbulence of airflow, and it is easy to form eddies or stagnation areas in different areas, which increases the probability of uneven gas distribution. This phenomenon will cause the concentration of harmful gases to be too high in specific areas, while other areas may not detect obvious gas changes, increasing the difficulty of inspection and the risk of accidents. On the contrary, when the edge effect concentration index performance value is smaller, it means that the gas accumulation effect is weaker, the air flow is smoother, the probability of the formation of eddies or stagnation areas is reduced, and the gas distribution is more uniform.

[0108] The difference in air velocity in the goaf of a coal mine can significantly affect the behavior of the airflow, resulting in the formation of eddies or stagnant areas in different areas, resulting in uneven gas distribution. This uneven gas accumulation may cause local high or low concentrations of gas in key areas, which are often critical for safety monitoring. If the flight route or height of the drone inspection is not designed properly, these peaks or valleys of concentration may be missed, leading to misjudgment of gas accumulation. This is because when the airflow is uneven in speed, the gas will stagnate in certain specific areas and form accumulation points. If the drone inspection only flies according to conventional paths or fixed altitudes, it may not be able to fully cover these complex areas. For example, the edge of the goaf may be enriched with methane or carbon monoxide due to airflow stagnation, but when the drone flies too high or the path is too concentrated in the main channel, it is easy to miss the accumulation of gas at these edges. In addition, the gas concentration at different heights and paths may vary significantly, and the turbulence caused by shearing the airflow may cause the gas to be stratified in the vertical direction. If the drone fails to dynamically adjust the flight altitude, it will not be able to accurately capture the true concentration of these stratified gases.

[0109] The logic of abnormal analysis of air flow velocity difference and generation of air flow shear unevenness index is as follows:

[0110] Under the detection window, the entire detection area is divided into a three-dimensional space grid, and the points in the coordinate system are defined as (x, y, z). The airflow velocity components obtained by the sensor at each sampling point are combined into a velocity vector. The airflow velocity vectors of all sampling points are combined to form a three-dimensional velocity vector field covering the detection window. The expression is as follows: V(x, y, z, t) = (v x ,v y ,v z ), where V(x,y,z,t) is the three-dimensional velocity vector field, which is the airflow velocity component at the spatial coordinates x,y,z at each time t;

[0111] The air flow velocity component v obtained at each sampling point x ,v y ,v z Represents the airflow motion state at the point in different directions. They not only provide the speed and direction of the airflow, but also convey many other information. The following are the detailed information contained in these components:

[0112] 1. Speed

[0113] Each component v x ,v y ,v z Indicates the speed of airflow in the corresponding direction (x, y, z axis), usually in meters per second (m / s).

[0114] By combining these three components, the total magnitude of the airflow velocity can be calculated.

[0115] 2. The positive and negative values ​​of each component of the velocity direction represent the flow trend of the airflow in that direction:

[0116] v z : Indicates the airflow in the east-west direction (positive value indicates eastward flow, negative value indicates westward flow).

[0117] v y : Indicates the airflow in the north-south direction (positive value indicates flow to the north, negative value to the south).

[0118] v z : Indicates vertical airflow (positive value indicates upward movement, negative value indicates downward movement).

[0119] This directional information is used to describe the airflow path in three-dimensional space, helping to understand how air flows in a goaf or mine.

[0120] 3. Time characteristics

[0121] The air velocity component is time-dependent, i.e., v x (t), v y (t), vz (t) may vary at different points in time.

[0122] Time series data can describe the dynamic characteristics of airflow, such as:

[0123] Are there any periodic changes (such as air flow changes caused by the temperature difference between day and night).

[0124] Are there any sudden changes (such as sudden changes in airflow at the beginning of a fire).

[0125] 4. Local disturbance and turbulence information

[0126] Irregular changes in air velocity components can reflect turbulence or local disturbances. For example:

[0127] In the stable airflow region, the velocity components change smoothly.

[0128] In the eddy region, the components fluctuate rapidly and change direction frequently.

[0129] Turbulence information is important for monitoring high-risk areas, such as goaf edges, where local turbulence may lead to abnormal gas accumulation or diffusion.

[0130] 5. Indirect information of pressure and density

[0131] According to Bernoulli's equation, the change in the velocity component of the air flow is related to the pressure and density of the gas:

[0132] In areas of high-speed airflow, the pressure is usually lower; in areas of low-speed airflow, the pressure may be higher.

[0133] Fluctuations in the velocity components of the air flow also reflect changes in air density (such as density changes caused by temperature differences or differences in gas concentration).

[0134] This information can be used indirectly to determine whether there are abnormal conditions, such as gas leaks or airflow blockages in local areas.

[0135] 6. Gas Diffusion Model

[0136] The velocity components reveal the diffusion pattern of the gas in space. The combination of velocity components in different directions can describe:

[0137] Convective diffusion: Gas is carried by air flow in a certain direction.

[0138] Eddy diffusion: Gas diffuses slowly in the eddy area and tends to accumulate.

[0139] This is crucial for predicting the diffusion paths of hazardous gases such as methane and carbon monoxide.

[0140] 7. Boundary conditions and environmental impact

[0141] The velocity components in different regions can reveal the effect of environmental boundaries on the airflow. For example:

[0142] At the edge of the goaf, the air velocity component may decrease rapidly, reflecting the boundary effect (i.e., air flow stagnation at the boundary).

[0143] The data collected by the sensors can also reflect the impact of external weather conditions (such as changes in internal airflow caused by increased wind speed).

[0144] 8. Energy transfer information

[0145] The velocity components of the air flow are related to the kinetic energy of the gas:

[0146]

[0147] Among them, E is the kinetic energy of the gas, ρ is the density of the gas, and V is the magnitude of the air flow velocity.

[0148] Understanding the energy transfer in airflow helps to determine potential risks, for example, increased airflow energy may mean the spread of fire.

[0149] The airflow velocity component at each sampling point contains not only the velocity magnitude and direction information of the gas in different directions, but also indirect information about time characteristics, turbulence disturbance, pressure and density, as well as gas diffusion mode and energy transfer. This information combined can fully describe the dynamic characteristics of the airflow, help analyze and predict the flow, accumulation and potential danger of the gas, and provide a reliable data basis for coal mine fire prevention and safety monitoring.

[0150] The local shear rate is calculated by the partial derivative of the three-dimensional velocity vector field and calibrated as the local shear rate tensor. The calculation expression is as follows:

[0151]

[0152] , where is the gradient tensor of the airflow velocity field, represents the airflow velocity component v in the x, y, and z directions x ,v y ,v z As the rate of change of the x position, represents the airflow velocity component v in the x, y, and z directions x ,v y ,v z With the rate of change of y position;

[0153] Curl is a measure of the vortex characteristics of airflow. Curl reveals the rotation phenomenon between airflows at different levels. The calculation expression of curl is as follows: The specific form after expansion is as follows:

[0154]

[0155] , where Ω(x,y,z,t) represents the curl, which describes the rotation characteristics of the airflow at a specific time t and spatial position x,y,z;

[0156] In airflow analysis, curl is mainly used to reveal the vortex characteristics and local rotation phenomena in the airflow, and to help identify abnormal behavior of the airflow in complex environments. It describes the rotation intensity and direction of different areas in the airflow. If the curl is not zero, it indicates that there is an eddy in the area, which may cause gas to be trapped or accumulated in certain locations, hindering normal diffusion. For coal mine goafs or other hazardous environments, the monitoring of curl is of great significance: areas with high curl may have concentrations of harmful gases such as methane and carbon monoxide, increasing the risk of fire or explosion. In addition, eddies can interfere with the inspection effect of drones and lead to missed inspections. Therefore, through curl field analysis, flight paths can be optimized, potential hot spots of hazardous gases can be accurately identified, and data support can be provided for safety monitoring and risk management.

[0157] Turbulence intensity measures the instability and randomness of airflow and is an important parameter for describing complex airflow. The calculation expression of local turbulence intensity is as follows:

[0158]

[0159] , where τ(x,y,z,t) is the local turbulence intensity;

[0160] In order to identify abnormal shear phenomena in the airflow, the curl and turbulence intensity are combined to construct an abnormal detection matrix. The calculation expression is as follows:

[0161] A(x,y,z,t)=Ω(x,y,z,t)·τ(x,y,z,t)

[0162] , where A(x,y,z,t) is the anomaly detection matrix, which indicates the degree of abnormality of the airflow at a given time t and spatial position x,y,z;

[0163] The anomaly detection matrix is ​​integrated to generate the airflow shear unevenness index. The calculation expression is as follows:

[0164] FSII=∫ D A(x,y,z,t)dxdydz

[0165] , where FSII represents the airflow shear unevenness index, and D represents the detection window;

[0166] From the calculation expression of the airflow shear unevenness index, it can be seen that the larger the airflow shear unevenness index performance value generated after the abnormal analysis of the airflow velocity difference, the more significant the airflow velocity difference is, the stronger the shear effect between the airflows is, and it is easy to form eddies or stagnation areas at the intersection, which leads to uneven gas distribution and increases the risk of local gas accumulation. On the contrary, when the airflow shear unevenness index performance value is smaller, it means that the velocity change between the airflows is smaller, the flow is more stable, and the probability of eddies and stagnation areas forming is reduced, so the gas distribution in space will be more uniform.

[0167] The edge effect concentration index EECI and the airflow shear unevenness index FSII are input into the pre-learned machine learning model, and the gas distribution index GDI is generated by the machine learning model according to the following formula:

[0168]

[0169] , where r1 and r2 are the preset proportional coefficients of the edge effect concentration index EECI and the airflow shear unevenness index FSII, respectively, and r1 and r2 are both greater than 0.

[0170] It can be seen from the calculation expression of the gas distribution index that the larger the performance value of the edge effect concentration index generated after the abnormal analysis of the edge accumulation phenomenon, the larger the performance value of the airflow shear unevenness index generated after the abnormal analysis of the airflow velocity difference. That is, the larger the performance value of the gas distribution index, the greater the probability that the airflow may form eddies or stagnation zones in different regions, resulting in uneven gas distribution. Conversely, it means that the velocity variation between the airflows is smaller, the flow is smoother, and the probability of the formation of eddies and stagnation zones is smaller.

[0171] Based on the machine learning evaluation results, i.e., the gas distribution index, the ambient gas distribution in the goaf is divided into a normal gas distribution state or an abnormal gas distribution state, and the abnormal gas distribution in the goaf is intelligently perceived;

[0172] The generated gas distribution index is compared and analyzed with the preset gas distribution index reference threshold, and the gas distribution state of the goaf environment is divided. The division results are as follows:

[0173] If the gas distribution index is greater than or equal to a preset gas distribution index reference threshold, the goaf environment gas distribution state is classified as an abnormal gas distribution state;

[0174] If the gas distribution index is less than a preset gas distribution index reference threshold, the gas distribution state of the goaf environment is classified as a normal gas distribution state;

[0175] According to the abnormal gas distribution status, the environmental gas distribution information of the goaf is continuously obtained, and the abnormal distribution of the environmental gas in the goaf is further divided into low-risk abnormal distribution, medium-risk abnormal analysis and high-risk abnormal distribution;

[0176] When the ambient gas distribution state of the goaf is classified as an abnormal gas distribution state, a number of gas distribution indices generated by the ambient gas distribution of the goaf of the coal mine are continuously obtained to establish an analysis set for comprehensive analysis;

[0177] The average value and standard deviation of several gas distribution indices in the analysis set are calculated, and the calculated average value and standard deviation of the gas distribution index are compared with the preset reference threshold of the gas distribution index average value and the preset reference threshold of the gas distribution index standard deviation, respectively, to further divide the abnormal gas distribution state. The division results are as follows:

[0178] If the average value of the gas distribution index is less than the preset reference threshold value of the average value of the gas distribution index, the abnormal distribution of the ambient gas in the goaf is further classified as a low-risk abnormal distribution;

[0179] If the average value of the gas distribution index is less than the preset reference threshold of the average value of the gas distribution index and the standard deviation of the gas distribution index is greater than or equal to the preset reference threshold of the standard deviation of the gas distribution index, the abnormal distribution of the ambient gas in the goaf is further classified as a medium-risk abnormal distribution;

[0180] If the average value of the gas distribution index is greater than or equal to the preset reference threshold value of the gas distribution index average value and the standard deviation of the gas distribution index is greater than or equal to the preset reference threshold value of the gas distribution index standard deviation, the abnormal distribution of the ambient gas in the goaf is further classified as a high-risk abnormal distribution;

[0181] Take different response measures for different risk anomaly distributions;

[0182] In the case of low-risk abnormal distribution, although the gas distribution is relatively stable overall, in order to prevent the accumulation of potential risks, it is necessary to increase the inspection frequency moderately, for example, shorten the inspection interval to 1.5 times the original plan. In addition, the flight path should be appropriately extended to the edge area of ​​the goaf to ensure that the gas accumulation trapped at the edge is not ignored. The flight altitude of the drone can be maintained at a medium level (such as 2-3 meters) to simultaneously monitor the gas conditions at different altitudes to ensure comprehensive coverage.

[0183] In the case of medium-risk abnormal distribution, due to the significant changes in gas concentrations in different areas, the inspection frequency should be increased and the flight path should be adjusted dynamically. UAVs should focus on inspecting areas with large changes and adopt a multi-level flight strategy to fully capture gas accumulation by flying alternately at low, medium and high altitudes. In addition, the inspection time of marginal areas should be appropriately increased to ensure that no key areas are missed and to avoid hidden dangers caused by local gas accumulation.

[0184] For high-risk abnormal distribution, due to the presence of high concentrations of harmful gases and drastic environmental changes, the inspection frequency needs to be significantly increased, for example, every 15-30 minutes, and the data should be uploaded to the control center in real time. The inspection path should be further refined, focusing on high-risk areas, and low-altitude flights (for example, 1-2 meters) should be used to monitor gas concentrations in cracks and hot spots. It is recommended to deploy multiple drones to work together to ensure full coverage in the area, and set up emergency stops near high-risk areas to respond to emergencies and improve the response speed and monitoring reliability of drones.

[0185] The present invention realizes intelligent evaluation of gas distribution in goaf by acquiring gas distribution characteristic information and airflow dynamic characteristic information in real time, and using a pre-learned machine learning model to generate a gas distribution index. The edge effect concentration index and the airflow shear unevenness index are used as the input of the model, so that the system can accurately capture the gas accumulation in different areas, especially high-risk areas such as edges and cracks that are easily overlooked. With the help of real-time analysis and automatic evaluation of data, the problem of missed detection of concentration peaks or valleys caused by unreasonable flight altitude or path of the drone is avoided. At the same time, the system dynamically adjusts the flight path and altitude to make gas distribution monitoring more comprehensive and accurate, significantly reducing the risk of explosion or poisoning accidents caused by misjudgment of gas accumulation.

[0186] The present invention realizes refined management of inspection strategies by dividing the analysis results of the gas distribution index into three categories: low risk, medium risk, and high risk, targeting abnormal distribution states at different risk levels. For low-risk situations, the accumulation of potential risks is prevented by appropriately increasing the inspection frequency and expanding the coverage of edge areas; for medium-risk situations, multi-level flight strategies and dynamic path adjustments are adopted to ensure that areas with large changes in gas concentration are captured; for high-risk situations, high-frequency inspections and the deployment of multi-machine collaborative operations enable the system to have stronger monitoring coverage and emergency response speed. This hierarchical inspection scheme optimizes the allocation of inspection resources, improves the adaptability and reliability of drones in complex environments, and realizes rapid early warning and emergency response by uploading data to the control center in real time, greatly reducing the probability of coal mine production safety accidents.

[0187] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A coal mine fire prevention monitoring and analysis method based on drone inspection, characterized in that: The following steps are involved: When the UAV arrives at the coal mine goaf area for inspection, it obtains the environmental parameter information of the coal mine goaf area in real time, including the gas distribution characteristic information and the air flow dynamic characteristic information; Perform an abnormal analysis on the acquired gas distribution characteristic information and gas flow dynamic characteristic information, and input the gas distribution characteristic information and gas flow dynamic characteristic information after abnormal analysis into a pre-learned machine learning model, generate a gas distribution index through the machine learning model, and perform intelligent evaluation of the ambient gas in the goaf through the gas distribution index; Based on the machine learning evaluation results, i.e., the gas distribution index, the ambient gas distribution in the goaf is divided into a normal gas distribution state or an abnormal gas distribution state, and the abnormal gas distribution in the goaf is intelligently perceived; According to the abnormal gas distribution status, the environmental gas distribution information of the goaf is continuously obtained, and the abnormal distribution of the environmental gas in the goaf is further divided into low-risk abnormal distribution, medium-risk abnormal analysis and high-risk abnormal distribution; Take different response measures for different risk anomaly distributions.

2. A coal mine fire prevention monitoring and analysis method based on drone inspection according to claim 1, characterized in that: The gas distribution characteristic information includes the edge accumulation phenomenon, which refers to the accumulation of gas in the edge area of ​​the goaf or mining area. The airflow dynamic characteristic information includes the airflow velocity difference, which describes the inconsistency of the velocity between different airflows in the goaf, and is the velocity change between hierarchical airflows or regional airflows.

3. A coal mine fire prevention monitoring and analysis method based on drone inspection according to claim 2, characterized in that: An abnormal analysis is performed on the edge accumulation phenomenon and the airflow velocity difference to generate the edge effect concentration index and the airflow shear uneven index respectively. The edge effect concentration index and the airflow shear uneven index generated after the abnormal analysis are input into the pre-learned machine learning model, and the gas distribution index is generated by the machine learning model. The gas distribution index is used to perform an intelligent assessment of the environmental gas in the goaf.

4. A coal mine fire monitoring and analysis method based on drone inspection according to claim 3, characterized in that: The edge accumulation phenomenon is analyzed abnormally, and the logic of generating edge effect concentration index is as follows: In the detection window, calculate the gas concentration gradient at different positions in the goaf to determine the severity of the concentration change. The calculation expression is as follows: , In the formula, is the gas concentration gradient vector, which represents the change in concentration per unit distance, is the rate of change of gas concentration in the x direction, that is, the concentration change per unit distance, is the rate of change of gas concentration in the y direction, is the rate of change of gas concentration in the z direction, are unit direction vectors, representing the x, y, and z directions respectively; The airflow disturbance coefficient measures the stability and volatility of the airflow. The size of the disturbance coefficient directly affects the diffusion pattern of gas in the goaf. The calculation expression is as follows: In the formula, Φ(x,y,z) represents the airflow disturbance coefficient, which is used to describe the degree of airflow turbulence in a local area, v = (v x ,v y ,v z ), v is the airflow velocity vector field, which is used to represent the airflow velocity in the x, y, and z directions. represents the absolute value of the airflow divergence, is the curl of the airflow, γ is the curl influence weight coefficient, which is used to adjust the influence of the curl on the airflow disturbance coefficient Φ(x,y,z); The gas concentration gradient is combined with the airflow disturbance coefficient to determine whether the gas accumulates in the edge area. The calculation expression is as follows: Where k(x,y,z) is the edge accumulation coefficient, α is the sensitivity coefficient, and exp(·) is the exponential function; In order to capture the dynamic changes of gas accumulation, the edge accumulation coefficient is time-weighted integrated to generate the edge effect concentration index. The calculation expression is as follows: , Where EECI is the edge effect concentration index, t1 is the start time of the detection window, t2 is the end time of the detection window, and β(t) is the time weighting function.

5. A coal mine fire monitoring and analysis method based on drone inspection according to claim 3, characterized in that: The logic of abnormal analysis of air flow velocity difference and generation of air flow shear unevenness index is as follows: Under the detection window, the entire detection area is divided into a three-dimensional space grid, and the points in the coordinate system are defined as (x, y, z). The airflow velocity components obtained by the sensor at each sampling point are combined into a velocity vector. The airflow velocity vectors of all sampling points are combined to form a three-dimensional velocity vector field covering the detection window. The expression is as follows: V(x, y, z, t) = (v x ,v y ,v z ), where V(x,y,z,t) is the three-dimensional velocity vector field, which is the airflow velocity component at the spatial coordinates x,y,z at each time t; The local shear rate is calculated by the partial derivative of the three-dimensional velocity vector field and calibrated as the local shear rate tensor. The calculation expression is as follows: , In the formula, is the gradient tensor of the airflow velocity field, represents the airflow velocity component v in the x, y, and z directions x ,v y ,v z As the rate of change of the x position, represents the airflow velocity component v in the x, y, and z directions x ,v y ,v z With the rate of change of y position; Curl is a measure of the vortex characteristics of airflow. Curl reveals the rotation phenomenon between airflows at different levels. The calculation expression of curl is as follows: The specific form after expansion is as follows: , Where Ω(x,y,z,t) represents the curl; Turbulence intensity measures the instability and randomness of airflow and is an important parameter for describing complex airflow. The calculation expression of local turbulence intensity is as follows: , Where τ(x,y,z,t) is the local turbulence intensity; In order to identify abnormal shear phenomena in the airflow, the curl and turbulence intensity are combined to construct an abnormal detection matrix. The calculation expression is as follows: A(x,y,z,t)=Ω(x,y,z,t)·τ(x,y,z,t), Where A(x,y,z,t) is the anomaly detection matrix; The anomaly detection matrix is ​​integrated to generate the airflow shear unevenness index. The calculation expression is as follows: FSII=∫ D A(x,y,z,t)dxdydz, Where FSII represents the airflow shear unevenness index, and D represents the detection window.

6. A coal mine fire prevention monitoring and analysis method based on drone inspection according to claim 3, characterized in that: The edge effect concentration index EECI and the airflow shear unevenness index FSII are input into the pre-learned machine learning model, and the gas distribution index GDI is generated by the machine learning model according to the following formula: , Where r1 and r2 are the preset proportional coefficients of the edge effect concentration index EECI and the airflow shear unevenness index FSII, respectively, and r1 and r2 are both greater than 0.

7. A coal mine fire monitoring and analysis method based on drone inspection according to claim 6, characterized in that: The generated gas distribution index is compared and analyzed with the preset gas distribution index reference threshold, and the gas distribution state of the goaf environment is divided. The division results are as follows: If the gas distribution index is greater than or equal to a preset gas distribution index reference threshold, the goaf environment gas distribution state is classified as an abnormal gas distribution state; If the gas distribution index is less than a preset gas distribution index reference threshold, the gas distribution state of the goaf environment is classified as a normal gas distribution state.

8. A coal mine fire monitoring and analysis method based on drone inspection according to claim 7, characterized in that: When the ambient gas distribution state of the goaf is classified as an abnormal gas distribution state, a number of gas distribution indices generated by the ambient gas distribution of the goaf of the coal mine are continuously obtained to establish an analysis set for comprehensive analysis; The average value and standard deviation of several gas distribution indices in the analysis set are calculated, and the calculated average value and standard deviation of the gas distribution index are compared with the preset reference threshold of the gas distribution index average value and the preset reference threshold of the gas distribution index standard deviation, respectively, to further divide the abnormal gas distribution state. The division results are as follows: If the average value of the gas distribution index is less than the preset reference threshold value of the average value of the gas distribution index, the abnormal distribution of the ambient gas in the goaf is further classified as a low-risk abnormal distribution; If the average value of the gas distribution index is less than the preset reference threshold of the average value of the gas distribution index and the standard deviation of the gas distribution index is greater than or equal to the preset reference threshold of the standard deviation of the gas distribution index, the abnormal distribution of the ambient gas in the goaf is further classified as a medium-risk abnormal distribution; If the gas distribution index average value is greater than or equal to the preset gas distribution index average value reference threshold value and the gas distribution index standard deviation is greater than or equal to the preset gas distribution index standard deviation reference threshold, The abnormal distribution of environmental gas in the goaf is further divided into high-risk abnormal distribution.

Images