A shallow coal seam hidden fire full field monitoring and early warning method and system

Abstract

The present application provides a kind of shallow coal seam hidden fire source full field monitoring and early warning method and system, belong to coal spontaneous combustion fire monitoring and early warning technical field, comprising: S1: obtain thermal infrared image and visible light image data;S2: eliminate redundant thermal imaging frame and fuzzy thermal imaging frame;S3: get preliminary ground temperature field distribution;S4: distinguish open fire source and suspicious area, and mark position;S5: obtain the ground temperature gradient distribution and indicating gas concentration data of suspicious area, obtain the ground temperature data, underground temperature data, indicating gas data and radioactive radon content data escaping from detection hole;S6: establish the identification index of hidden fire source;S7: establish hidden fire source grading risk standard;S8: output early warning result.This method realizes full-time, full field, three-dimensional monitoring, improves accuracy and reliability, realizes intelligent grading early warning and decision support, enhances environmental adaptability and automation level.

Description

Technical Field

[0001] This invention relates to the field of coal spontaneous combustion fire monitoring and early warning technology, specifically to a method and system for full-field monitoring and early warning of concealed fire sources in shallow buried coal seams. Background Technology

[0002] In the mining industry, coal fires caused by spontaneous combustion are a common and serious hidden fire hazard, typically occurring underground, in open-pit mines, and in coal piles. Therefore, timely monitoring and effective early warning of spontaneous combustion of coal are of great importance.

[0003] Current methods for detecting coal fires have many shortcomings: traditional methods, including geochemical, electrical, and magnetic methods, while highly accurate, require in-depth fieldwork, which is dangerous and inefficient; satellite thermal infrared remote sensing has limited accuracy. In recent years, infrared thermal imaging technology and inspection robots have been increasingly applied to coal fire hazard inspections. Inspection robots can enter high-risk areas that are difficult for humans to reach, improving the coverage of hidden fire source inspections and achieving comprehensive perception of coal fire information; low-altitude remote sensing using UAVs equipped with thermal imagers has the advantages of fast data acquisition and high measurement accuracy, providing new means for detecting hidden fire sources in shallow coal seams. However, these detection and analysis methods are limited, lack quantitative analysis, and are independent and unconnected.

[0004] Therefore, we propose a method and system for full-field monitoring and early warning of concealed fire sources in shallow coal seams. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for full-field monitoring and early warning of concealed fire sources in shallow coal seams, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] This invention provides a method for full-field monitoring and early warning of concealed fire sources in shallow coal seams, comprising the following steps:

[0008] S1: Use the wide-area preliminary screening unmanned detection and early warning flight device to cruise and scan the monitoring area, acquire thermal infrared and visible light image data of the monitoring area, and record the auxiliary positioning information data of the wide-area preliminary screening unmanned detection and early warning flight device in real time. The auxiliary positioning information data includes flight time data and spatial coordinate data.

[0009] S2: Remove redundant and blurred thermal imaging frames from the thermal infrared images obtained in step S1.

[0010] S3: Perform temperature compensation on the thermal infrared image processed in step S2 to obtain a preliminary surface temperature field distribution map;

[0011] S4: Based on the preliminary surface temperature field distribution map obtained in step S3, identify open flame sources and suspicious areas, and mark the locations of open flame sources and suspicious areas;

[0012] S5: For marked suspicious areas, use a fixed-point detailed survey unmanned detection and early warning flight device to conduct cruise scanning, obtain the surface temperature gradient distribution and indicator gas concentration data of the suspicious area, and record the flight time data and spatial coordinate data of the fixed-point detailed survey unmanned detection and early warning flight device in real time. Use a ground tracked walking detection and early warning device to conduct detection. The ground tracked walking detection and early warning device measures the temperature at the designated verification point in the suspicious area and drills a detection hole to obtain the surface temperature data, underground temperature data, indicator gas data and radioactive radon content data at the verification point.

[0013] S6: The monitoring data obtained in step S5, the flight time data and spatial coordinate data of the fixed-point detailed investigation unmanned detection and early warning flight device, the auxiliary positioning information data of the wide-area preliminary screening unmanned detection and early warning flight device in step S1, and the preliminary surface temperature field distribution map data in step S4 are fused and analyzed to establish the identification index of dark fire source.

[0014] S7: Establish a risk classification standard for hidden fire sources based on the identification indicators of hidden fire sources;

[0015] S8: Generate early warning information and output the early warning results.

[0016] Furthermore, in step S1, the wide-area initial screening unmanned detection and early warning flight device adopts a partitioned adaptive coverage route planning strategy when performing cruise scanning in the monitoring area. Specifically, it includes: forming an outer closed cruise trajectory along the outer boundary of the monitoring area, and generating several inner cruise sub-paths within the area. The sub-paths are connected by radial connecting segments to form a progressive ring-radial composite route from the outside to the inside.

[0017] In step S5, the fixed-point detailed investigation unmanned detection and early warning flight device adopts a multi-scale local flight path planning strategy centered on the suspicious area marker point. Specifically, it includes: a closed loop trajectory with the marker point as the center, and a fan-shaped scanning trajectory that diverges radially outward or inward. Each trajectory corresponds to a different flight altitude.

[0018] Furthermore, step S2 includes:

[0019] The process of removing redundant thermal imaging frames is as follows: the geographic overlap area is calculated using POS data of consecutive frames. When the overlap rate exceeds 85% and the average temperature difference of the overlap area is less than 0.5℃ and the structural similarity is higher than 0.98, it is determined to be a redundant frame and removed.

[0020] The process of removing blurry thermal imaging frames is as follows: the image sharpness of each frame is evaluated, and the violent motion at the moment of acquisition is detected simultaneously in combination with IMU data. If the frame sharpness is lower than 60% of the median value of the flight segment, or if there is violent motion and the sharpness is lower than 80%, it is determined to be an invalid blurry frame and is removed.

[0021] Furthermore, step S3 includes:

[0022] S31: A continuous correction function is introduced to correct temperature data at different altitudes. The formula is as follows:

[0023]

[0024] in, This indicates the temperature received by the thermal imager as the altitude approaches infinity; This represents a reference temperature value when the altitude is close to the Earth's surface. This is the compensation coefficient for topographic and environmental impacts, with a value range of 1.0-1.2; The attenuation coefficient is obtained by fitting the atmospheric extinction characteristics. The current flight altitude of the wide-area initial screening unmanned detection and early warning flight device; This is the corrected equivalent temperature measurement value;

[0025] S32: Utilizing a multi-layer adaptive network algorithm to compensate for the output temperature of the compensation model. The formula is:

[0026]

[0027] in, For the input secondary parameter vector; For the first One radial base center; For the corresponding width parameter; For output weights;

[0028] S33: Output the compensated temperature measurement value The formula is:

[0029]

[0030] in, This represents the equivalent temperature measurement value calculated using a continuous correction function based on flight altitude. It is a multi-layer adaptive network algorithm compensation model based on the compensation temperature output of multiple secondary parameters; 、 Adaptive weighting coefficients for interaction terms; It is a temperature scale normalization factor used to control the nonlinear range of the cross term; It is the hyperbolic tangent function, used to introduce nonlinear interaction terms;

[0031] S34: Generate a preliminary surface temperature field distribution map .

[0032] Furthermore, step S4 includes:

[0033] S41: Based on the preliminary surface temperature field distribution map, the temperature value exceeding the preset open flame detection threshold will be considered. The pixel regions were extracted as high-temperature candidate regions. An image recognition model was used to perform visual feature analysis on the corresponding high-temperature candidate regions in the simultaneously acquired visible light images to identify flames, smoke, and severe heat burns on the ground. Then, spatial connectivity analysis and clustering were performed on the high-temperature candidate regions to remove isolated noise points. Regions simultaneously meeting the criteria of "temperature exceeding..." were then selected. The continuous area with "positive visible light characteristics" and the area with significant high temperature accumulation were identified as open flame sources;

[0034] S42: Keep the temperature below the threshold. Areas that are negative or uncertain in visible light verification are identified as suspicious areas; areas with local disturbances in texture features in infrared images are identified as suspicious areas; and areas with normal temperature but located near historical fire zones, crack zones, or gas escape channels are identified as suspicious areas.

[0035] S43: Mark the location of fire sources and suspected areas using GIS electronic maps.

[0036] Furthermore, step S5 includes:

[0037] S51: Generate a set of suspicious regions for the marked suspicious regions. Each region Preliminary center coordinates provided by GIS electronic map ;

[0038] S52: The flight path planning of the fixed-point detailed survey unmanned detection and early warning flight device is based on... The polar coordinate grid centered on the location of the sampling points in the suspicious area is as follows: ,in, Radial distance, It is the azimuth angle. At the flight altitude, surface temperature was measured at each sampling point to establish a surface temperature gradient distribution; at each sampling point, the indicator gas concentration vector was measured. , construct regions Indicator gas concentration field in the air Complete the acquisition of three-dimensional profiles indicating gas distribution, among which, This indicates the sampling location coordinates of the fixed-point detailed survey unmanned detection and early warning flight device;

[0039] S53: Ground-tracked detection and early warning device in the target area Designated verification points within Surface temperature was measured at the location to obtain surface temperature data; at the verification point... Drill a probe hole to depth Measure the temperature profile inside the probe hole To obtain underground temperature data and measure the radioactive radon content in the probe borehole. To obtain radon content data at a height above the Earth's surface At the location, the concentration of the indicator gas escaping from the probe hole is measured. .

[0040] Furthermore, step S6 includes:

[0041] S61: For temperature measurement values The optimized formula is as follows:

[0042]

[0043] in, The optimized temperature value; The temperature readings before optimization; Indicates altitude; The transmittance correction factor for infrared radiation in the atmosphere is calculated from the indicator gas concentration data measured by the fixed-point detailed survey unmanned detection and early warning flight device.

[0044] Based on temperature measurement value and temperature value To obtain local temperature difference ;

[0045] S62: Based on the surface temperature gradient distribution from step S52, the surface temperature data from step S53, and the optimized temperature values. The auxiliary positioning information data of the wide-area initial screening unmanned detection and early warning flight device and the flight time data and spatial coordinate data of the fixed-point detailed survey unmanned detection and early warning flight device are used to target the area. Preliminary surface temperature field distribution map Calibration and refinement are performed to generate a surface temperature field distribution map of the target area. ;

[0046] S63: The concentration of indicator gas escaping from the detection hole measured by the ground tracked movement detection and early warning device. The indicator gas concentration field measured vertically above the same location by a fixed-point detailed survey unmanned detection and early warning flight device. The formula for superposition is:

[0047]

[0048] in, Indicates source strength, representing the rate at which characteristic gases are released into the atmosphere; Indicates wind speed; , These are the horizontal and vertical diffusion parameters, respectively; Indicates source height; Indicates the lateral distance from the central axis; Vertical height represents the height of the calculation point above the ground; This indicates the sampling location coordinates of the fixed-point detailed survey unmanned detection and early warning flight device;

[0049] S64: Based on temperature profile within the probe hole Calculate the geothermal gradient anomaly index The formula is:

[0050]

[0051] in, This represents the normal geothermal gradient for the region.

[0052] S65: Calculate the radioactive radon anomaly ratio based on radon content data. The formula is:

[0053]

[0054] in, This indicates the radon content data measured by the ground-based tracked detection and early warning device; The background radon content of the region;

[0055] S66: Based on the integration of all data, identification indicators for dark fire sources are obtained. These indicators include temperature-related indicators and gas-related indicators. The temperature indicators include optimized temperature values. Temperature profile inside the probe hole Local temperature difference Geothermal gradient anomaly index Optimized verification points exist Size of the thermal anomaly region in the temperature field ,correspond Average temperature in the temperature field With the highest temperature Gas-related indicators include the indicator gas concentration field measured by the fixed-point detailed survey unmanned detection and early warning flight device. Concentration of indicator gas escaping from the detection hole Gas source strength Ratio to radioactive radon anomaly ;

[0056] S67: Based on the identification indicators obtained in step S66, select the highest temperature. Geothermal gradient anomaly index Local temperature difference Thermal anomaly area Average temperature Radioactive radon anomaly ratio Gas source strength , concentration, concentration, Concentration serves as an indicator for identifying smoldering fire sources.

[0057] Furthermore, step S7 includes:

[0058] S71: Based on the identification indicators determined in step S67, determine the safety baseline value, critical outlier value, and danger threshold of each indicator, and calculate the standardized score according to the following mapping rules. ,in, Indicates the index sequence number;

[0059] The mapping rules are as follows:

[0060] When the parameter value is less than or equal to the upper limit of the safety baseline value point;

[0061] When the upper limit of the safety benchmark value < the parameter value ≤ the lower limit of the critical outlier value ;

[0062] When the lower limit of the critical outlier is less than the parameter value and ≤ the upper limit of the critical outlier... point;

[0063] When the upper limit of the critical outlier value is less than the parameter value and less than the lower limit of the danger threshold... ;

[0064] When the parameter value is greater than the lower limit of the danger threshold point;

[0065] S72: The hazard level of smoldering fire sources is divided into five levels. Correlation analysis is used to calculate the correlation coefficient between each indicator and the hazard level of smoldering fire sources, and then the weights are obtained by normalization. ;

[0066] S73: Based on standardized scores and weight Calculate the comprehensive risk score of hidden fire sources. The formula is as follows:

[0067]

[0068] in, The higher the score, the greater the level of danger.

[0069] S74: To avoid misjudgment based on a single parameter, a correction rule is set:

[0070] If the highest temperature ℃ and concentration , Add 1.5 points directly, but No more than 10 points;

[0071] If the radon content is abnormally high... And thermal anomaly area , Add 1.0 point directly, but No more than 10 points;

[0072] If all parameter values ​​are below the safety baseline value The score was directly determined to be the background fluctuation area;

[0073] like Points, verification points The suspicious area was identified as a source of smoldering fire.

[0074] S75: Based on comprehensive risk assessment of hidden fire sources Establish a rating system The correspondence between the danger levels of dark fire sources and their corresponding danger levels is as follows:

[0075] when At that time, the danger level of the hidden fire source was Level 1 represents the background fluctuation region;

[0076] when The hazard level of the hidden fire source is Level IV, which is a low-risk area;

[0077] when The hazard level of the hidden fire source is Level III, which is a medium-risk area;

[0078] when The hazard level of the hidden fire source is Level II, which is a high-risk area;

[0079] when The ignition source is classified as Level I, indicating an emergency risk zone.

[0080] Furthermore, step S8 includes:

[0081] S81: Based on the classification results, generate a warning message containing the following:

[0082] Spatial information: The precise location and extent of the ignition source corresponding to the verification point on the GIS electronic map;

[0083] Attribute information: identified hazard level, indicator value, risk score ;

[0084] Time information: discovery time, data collection time;

[0085] Recommended measures: Emergency response or monitoring plans recommended based on the risk level;

[0086] S82: Based on the warning information in step S81, generate a warning map in the form of an electronic map, and mark the location, range and danger level of the smoldering fire source on the warning map with different colors or symbols;

[0087] S83: Based on the warning information in step S81, generate a color risk heat map and a warning report containing the warning information;

[0088] S84: Output early warning map, color risk heat map and early warning report.

[0089] This invention also provides a full-field monitoring and early warning system for concealed fire sources in shallow coal seams, used to realize the full-field monitoring and early warning method for concealed fire sources in shallow coal seams described above, including a full-field detection module, a low-altitude-ground collaborative scheduling and task triggering module, a real-time intelligent identification module, and an early warning output module;

[0090] The full-field detection module includes a low-altitude monitoring unit and a ground-based verification unit. The low-altitude monitoring unit includes a wide-area initial screening unmanned detection and early warning flight device and a fixed-point detailed investigation unmanned detection and early warning flight device. The ground-based verification unit includes a ground-based tracked detection and early warning device. The wide-area initial screening unmanned detection and early warning flight device is used to perform cruise scanning of the monitoring area and acquire thermal infrared and visible light image data of the monitoring area. The fixed-point detailed investigation unmanned detection and early warning flight device is used to perform cruise scanning of suspicious areas and acquire surface temperature gradient distribution and indicator gas concentration profile detection data of the suspicious areas. The ground-based tracked detection and early warning device is used to detect designated verification points in the suspicious areas and acquire surface temperature data, underground temperature data, indicator gas data escaping from the detection borehole, and radioactive radon content data.

[0091] The low-altitude-ground coordinated scheduling and task triggering module is used to organize, allocate, and switch the detection tasks of the low-altitude monitoring unit and the ground verification unit according to the detection results during the detection of shallow buried concealed fire sources.

[0092] The all-time intelligent identification module includes a ground analysis terminal, which is used to receive and process detection data from the low-altitude monitoring unit and the ground verification unit, determine suspicious areas and open flame sources in the monitoring area, and identify smoldering flame sources and their degree of danger in the suspicious areas.

[0093] The warning output module is used to output the warning results.

[0094] Compared with the prior art, the present invention has the following technical effects:

[0095] 1. This method enables real-time, all-area, and three-dimensional monitoring of concealed fire sources in shallow coal seams. Through air-ground collaborative operations, a monitoring network of "wide-area initial screening in the air, detailed investigation at fixed points in the air, and close-range verification detection on the ground" is constructed. This overcomes the contradictions of "inaccurate coarse detection and narrow coverage of fine detection" in traditional methods, significantly improving the spatial coverage and temporal continuity of fire source detection, and achieving blind-spot-free and uninterrupted monitoring.

[0096] 2. This method improves the accuracy and reliability of fire source detection. By fusing and analyzing multiple data such as temperature field (full-field thermal infrared image data), indicator gas concentration data, radioactive radon anomaly ratio, and surface temperature, and combining high-altitude temperature correction and multi-layer adaptive network algorithm compensation model temperature correction, environmental interference is effectively eliminated, achieving high-precision temperature identification and quantitative characterization of open and hidden fire sources, and significantly reducing the risk of false alarms and missed alarms.

[0097] 3. This method realizes intelligent hierarchical early warning and decision support based on data fusion analysis. By integrating multiple parameters such as thermal infrared temperature measurement, indicator gas concentration and radioactive radon content, a comprehensive risk scoring standard for smoldering fire sources is constructed. Based on the five-level classification standard, the method automatically classifies the hazard level and outputs a visual early warning map and disposal suggestions. This provides timely and scientific hierarchical response basis for mine safety management personnel, enhancing the pertinence and timeliness of fire prevention and control.

[0098] 4. This method enhances environmental adaptability and operational automation. It adopts a differentiated unmanned detection and early warning flight device platform and intelligent route planning strategy, combined with an air-ground collaborative task scheduling mechanism. It can adapt to complex terrain and weather conditions, realize autonomous rotation of detection equipment, real-time data transmission and dynamic adjustment of tasks, reduce the intensity of human intervention, and improve the long-term stable operation capability in harsh environments such as mining areas and spoil heaps. Attached Figure Description

[0099] Figure 1 This is a flowchart illustrating the monitoring and early warning method according to an embodiment of the present invention;

[0100] Figure 2 This is a schematic diagram of the air-ground collaborative detection method for monitoring and early warning in an embodiment of the present invention;

[0101] Figure 3 This is a flight path planning diagram for the wide-area initial screening unmanned detection and early warning flight device according to an embodiment of the present invention;

[0102] Figure 4 This is a flight path planning diagram for the fixed-point detailed survey unmanned detection and early warning flight device according to an embodiment of the present invention;

[0103] Figure 5 This is a flowchart illustrating the identification process for open flame sources and suspicious areas according to an embodiment of the present invention.

[0104] Figure 6 This is a flowchart illustrating the process of identifying the level of hidden fire sources in a suspected area according to an embodiment of the present invention.

[0105] Figure 7 This is a warning diagram of a hidden fire source according to an embodiment of the present invention;

[0106] Figure 8 This is a schematic diagram of the monitoring and early warning system according to an embodiment of the present invention. Detailed Implementation

[0107] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present invention.

[0108] In this article, terms such as "left," "right," "up," "down," "front," and "back" are established based on the positional relationships shown in the attached drawings. Depending on the attached drawings, the corresponding positional relationships may also change. Therefore, they should not be interpreted as an absolute limitation on the scope of protection.

[0109] Please see Figure 1 This invention provides a method for full-field monitoring and early warning of concealed fire sources in shallow coal seams, comprising the following steps:

[0110] S1: The wide-area preliminary screening unmanned detection and early warning flight device is used to cruise and scan the monitoring area to obtain thermal infrared and visible light image data of the monitoring area, and to record the auxiliary positioning information data of the wide-area preliminary screening unmanned detection and early warning flight device in real time. The auxiliary positioning information data includes flight time data and spatial coordinate data.

[0111] Specifically, before the wide-area initial screening unmanned detection and early warning aerial vehicle (UAV) conducts formal cruise scanning of the monitoring area, on-site adaptability preparations must be carried out. First, a field survey and preliminary adaptability tests should be conducted on the area to be monitored, including: assessing the topography, climate, and known fire source distribution of the target area; identifying key environmental factors affecting the flight path planning of the UAV and the path design of the ground-based tracked detection and early warning vehicle; and conducting pre-flight and pre-operation tests on the deployed detection and early warning vehicle to verify its operational stability and mission adaptability under current site conditions.

[0112] Specifically, in step S1, the wide-area initial screening unmanned detection and early warning flight device, based on the terrain boundaries, historical fire zone distribution information, and prior remote sensing data of the test area obtained from the on-site adaptability preparation test, divides the monitoring area into irregular partitions. During the cruise scan (i.e., wide-area initial screening) of the monitoring area, it does not perform equidistant zigzag flight in a single direction, but adopts a partitioned adaptive coverage route planning strategy, such as... Figure 3 As shown, the process specifically includes: forming an outer closed cruise trajectory (i.e., an outer boundary loop) along the outer boundary of the monitoring area, generating several inner cruise sub-paths within the area, performing zonal scanning on these inner cruise sub-paths, and conducting dense scanning of each inner cruise sub-path. The sub-paths are connected by radial connecting segments (i.e., radial transition paths), forming a progressive circular-radial composite flight path from the outside in. During the flight path execution, the wide-area initial screening unmanned detection and early warning flight device hovers briefly or flies at low speed at preset key sampling nodes, simultaneously acquiring thermal infrared images and location information. When significant temperature anomalies or potential ignition source characteristics are detected, a suspicious area marker is automatically generated at that location, providing spatial constraints for the subsequent path generation of the fixed-point detailed investigation unmanned detection and early warning flight device.

[0113] Compared to traditional equidistant S-shaped routes, the above flight path can significantly increase the spatial sampling density of suspected abnormal areas while ensuring wide coverage, reduce repeated flights to non-critical areas during the wide-area initial screening stage, and improve overall inspection efficiency.

[0114] S2: Remove redundant and blurred thermal imaging frames from the thermal infrared images obtained in step S1.

[0115] Specifically, step S2 includes:

[0116] Redundant thermal imaging frames were removed by analyzing spatial overlap and comparing temperature field similarity. Specifically, the geographical overlap area was calculated using POS data from consecutive frames. When the overlap rate exceeded 85%, the average temperature difference in the overlap area was less than 0.5℃, and the structural similarity was higher than 0.98, the frames were identified as redundant and removed.

[0117] The system automatically identifies and removes blurry thermal imaging frames based on a sharpness evaluation function and motion state detection. The specific process is as follows: the sharpness of each frame is evaluated, and the system simultaneously detects violent motion at the moment of acquisition by combining IMU data. If the frame sharpness is lower than 60% of the median value of the flight segment, or if there is violent motion and the sharpness is lower than 80%, it is determined to be an invalid blurry frame and is removed.

[0118] S3: Perform temperature compensation on the thermal infrared image processed in step S2 to obtain a preliminary surface temperature field distribution map.

[0119] Specifically, step S3 includes:

[0120] S31: A continuous correction function is introduced to correct temperature data at different altitudes. The formula is as follows:

[0121]

[0122] in, This indicates the temperature received by the thermal imager (i.e., the ambient background temperature) as the altitude approaches infinity. This represents a reference temperature value when the altitude is close to the Earth's surface. The topography and environmental impact compensation coefficient has a value range of 1.0-1.2 and is used to correct temperature measurement deviations caused by non-uniform environmental factors such as topographic relief, differences in surface emissivity, and atmospheric turbulence. The attenuation coefficient is obtained by fitting the atmospheric extinction characteristics. The current flight altitude of the wide-area initial screening unmanned detection and early warning flight device; This is the corrected equivalent temperature measurement value.

[0123] This formula characterizes the attenuation of infrared radiation with flight altitude: as altitude... As the temperature increases, the measured temperature will gradually approach the ambient temperature. Approximation. Using this formula, unmanned detection and early warning aircraft can be positioned at any altitude. Measured temperature Equivalent temperature value at reference height This enables the unification of data across different flight altitudes.

[0124] S32: Utilizing a multi-layer adaptive network algorithm to compensate for the output temperature of the compensation model. The formula is:

[0125]

[0126] in, For the input secondary parameter vector, ambient temperature, primary indicator gas concentration ( , The model input includes indicators such as the actual temperature (indicator gases), humidity, terrain slope, wind speed, and infrared image texture features. For the first One radial base center; For the corresponding width parameter; For output weights.

[0127] Specifically, the multilayer adaptive network algorithm compensation model is a radial basis function neural network (RBFNN), whose hidden layer consists of multiple radial basis function units. The Gaussian kernel function is used to perform nonlinear transformation on the input minor parameters to achieve fitting between the machine temperature measurement data and the actual surface temperature.

[0128] S33: Output the compensated temperature measurement value The formula is:

[0129]

[0130] in, This represents the equivalent temperature measurement value calculated using a continuous correction function based on flight altitude. It is a multi-layer adaptive network algorithm compensation model based on the compensation temperature output of multiple secondary parameters; 、 The adaptive weight coefficients of the interaction terms are dynamically generated by the neural network based on the current environmental parameters; It is a temperature scale normalization factor used to control the nonlinear range of the cross term; It is the hyperbolic tangent function, used to introduce nonlinear interaction terms to enhance the expressive power of the model.

[0131] S34: Generate a preliminary surface temperature field distribution map .

[0132] S4: Based on the preliminary surface temperature field distribution map obtained in step S3, identify open flame sources and suspicious areas, and mark the locations of open flame sources and suspicious areas, such as... Figure 5 As shown.

[0133] Specifically, step S4 includes:

[0134] S41: The process for identifying open flame sources in shallow coal seams includes initial temperature field anomaly extraction, visible light feature co-verification, and spatial clustering and confirmation. Specifically, it involves: based on the preliminary surface temperature field distribution map... , temperature value Exceeding the preset open flame threshold The pixel regions were extracted as high-temperature candidate regions. An image recognition model was used to perform visual feature analysis on the corresponding high-temperature candidate regions in the simultaneously acquired visible light images to identify flames, smoke, and severe heat burns on the ground. Then, spatial connectivity analysis and clustering were performed on the high-temperature candidate regions to remove isolated noise points. Regions simultaneously meeting the criteria of "temperature exceeding..." were then selected. The continuous region and temperature of "visible light characteristics verified as positive" far exceed Significantly high-temperature clusters are identified as open flame sources. The geographical location, outline, area, and highest temperature of the open flame source are automatically recorded and marked as the highest hazard level.

[0135] S42: Keep the temperature below the threshold. Areas with negative or uncertain visible light verification results are classified as suspicious areas. Areas with localized textural disturbances or uneven thermal fields in infrared images are also classified as suspicious areas. Areas with normal temperatures but located near historical fire zones, crack zones, or gas escape channels are also classified as suspicious areas. Corresponding spatial markers are generated at the center of each suspicious area. All other situations are classified as safe zones.

[0136] S43: Mark the location of fire sources and suspicious areas using GIS electronic maps. Suspicious areas pose potential risks and should be the focus of subsequent detailed investigation and verification of hidden fire sources in shallow coal seams.

[0137] S5: For marked suspicious areas, use a fixed-point detailed survey unmanned detection and early warning flight device to conduct cruise scanning, obtain the surface temperature gradient distribution and indicator gas concentration data of the suspicious area, and record the flight time data and spatial coordinate data of the fixed-point detailed survey unmanned detection and early warning flight device in real time. Use a ground tracked walking detection and early warning device to conduct detection. The ground tracked walking detection and early warning device measures the temperature at the designated verification point in the suspicious area and drills a detection hole to obtain the surface temperature data, underground temperature data, indicator gas data and radioactive radon content data at the verification point.

[0138] Specifically, step S5 includes:

[0139] S51: Generate a set of suspicious regions for the marked suspicious regions. Each region Preliminary center coordinates provided by GIS electronic map .

[0140] S52: The flight path planning of the fixed-point detailed survey unmanned detection and early warning flight device is based on... The polar coordinate grid centered on the location of the sampling points in the suspicious area is as follows: ,in, Radial distance, It is the azimuth angle. At the flight altitude, surface temperature was measured at each sampling point to establish a surface temperature gradient distribution; at each sampling point, the indicator gas concentration vector was measured. , construct regions Indicator gas concentration field in the air Complete the acquisition of three-dimensional profiles indicating gas distribution, among which, This indicates the sampling location coordinates of the fixed-point detailed investigation unmanned detection and early warning flight device.

[0141] S53: Ground-tracked detection and early warning device in the target area Designated verification points within Surface temperature was measured at the location to obtain surface temperature data; at the verification point... Drill a probe hole to depth Measure the temperature profile inside the probe hole To obtain underground temperature data and measure the radioactive radon content in the probe borehole. To obtain radon content data at a height above the Earth's surface (0.1-0.5) At point ), measure the concentration of the indicator gas escaping from the probe hole. .

[0142] Specifically, in step S5, after completing the regional screening and generating several suspicious area markers in the wide-area initial screening stage, the fixed-point detailed investigation stage constructs a local fine-scale detection area based on the location of the markers. The fixed-point detailed investigation unmanned detection and early warning flight device adopts a multi-scale local flight path planning strategy centered on the suspicious area markers, such as... Figure 4 As shown, it specifically includes: a closed loop path with the suspicious area marker as the center, and radial fan-shaped scanning paths that diverge outward or inward along the radial direction. Each path corresponds to a different flight altitude, realizing the three-dimensional profile acquisition of the indicator gas distribution above the target area.

[0143] During its orbital and sector scanning operations, the fixed-point detailed survey unmanned detection and early warning flight device dynamically adjusts its flight path density and hovering point spacing based on changes in the indicator gas concentration gradient. It also feeds back indicator gas information collected at multiple altitudes and locations to a ground analysis terminal for processing temperature measurements. The optimization process avoids the redundant flight issues associated with using regular grids or broken lines within small areas, thus improving detection sensitivity and spatial resolution in complex fire source environments during the detailed point survey phase.

[0144] S6: The monitoring data obtained in step S5, the flight time data and spatial coordinate data of the fixed-point detailed investigation unmanned detection and early warning flight device, the auxiliary positioning information data of the wide-area preliminary screening unmanned detection and early warning flight device in step S1, and the preliminary surface temperature field distribution map data in step S4 are fused and analyzed to establish the identification index of dark fire source.

[0145] Specifically, step S6 includes:

[0146] S61: For temperature measurement values The optimized formula is as follows:

[0147]

[0148] in, The optimized temperature value; The temperature readings before optimization; Indicates altitude; The transmittance correction factor for infrared radiation in the atmosphere is calculated from the indicator gas concentration data measured by the fixed-point detailed survey unmanned detection and early warning flight device.

[0149] Based on temperature measurement value and temperature value To obtain local temperature difference .

[0150] S62: Based on the surface temperature gradient distribution from step S52, the surface temperature data from step S53, and the optimized temperature values. The auxiliary positioning information data of the wide-area initial screening unmanned detection and early warning flight device and the flight time data and spatial coordinate data of the fixed-point detailed survey unmanned detection and early warning flight device are used to target the area. Preliminary surface temperature field distribution map Calibration and refinement are performed to generate a surface temperature field distribution map of the target area. .

[0151] S63: The concentration of indicator gas escaping from the detection hole measured by the ground tracked movement detection and early warning device. The indicator gas concentration field measured vertically above the same location by a fixed-point detailed survey unmanned detection and early warning flight device. The formula for superposition is:

[0152]

[0153] in, The source strength represents the rate at which characteristic gases are released into the atmosphere. A significant non-zero source strength can indirectly indicate the presence of a hidden underground ignition source (i.e., a smoldering ignition source). Indicates wind speed; , These are the horizontal and vertical diffusion parameters, respectively; The source height indicates the effective release altitude at which the gas from the release source finally escapes and enters the near-surface atmosphere. Indicates the lateral distance from the central axis; Vertical height represents the height of the calculation point above the ground; This indicates the sampling location coordinates of the fixed-point detailed investigation unmanned detection and early warning flight device.

[0154] S64: Based on temperature profile within the probe hole Calculate the geothermal gradient anomaly index The formula is:

[0155]

[0156] in, This represents the normal geothermal gradient for the region.

[0157] S65: Calculate the radioactive radon anomaly ratio based on radon content data. The formula is:

[0158]

[0159] in, This indicates the radon content data measured by the ground-based tracked detection and early warning device; The background radon content is for the region.

[0160] S66: Based on the integration of all data, identification indicators for dark fire sources are obtained. These indicators include temperature-related indicators and gas-related indicators. The temperature indicators include optimized temperature values. Temperature profile inside the probe hole Local temperature difference Geothermal gradient anomaly index Optimized verification points exist Size of the thermal anomaly region in the temperature field ,correspond Average temperature in the temperature field With the highest temperature Gas-related indicators include the indicator gas concentration field measured by the fixed-point detailed survey unmanned detection and early warning flight device. Concentration of indicator gas escaping from the detection hole Gas source strength Ratio to radioactive radon anomaly .

[0161] S67: Based on the identification indicators obtained in step S66, select the highest temperature. Geothermal gradient anomaly index Local temperature difference Thermal anomaly area Average temperature Radioactive radon anomaly ratio Gas source strength , concentration, concentration, Concentration serves as an indicator for identifying smoldering fire sources.

[0162] S7: Establish a risk classification standard for hidden fire sources based on the identification indicators of hidden fire sources.

[0163] Specifically, step S7 includes:

[0164] S71: Based on the identification indicators determined in step S67, determine the safety benchmark value, critical anomaly value and danger threshold of each indicator. The numerical table of each indicator for identifying hidden fire sources is shown in Table 1.

[0165] Table 1 Threshold Table for Various Parameters in Dark Fire Source Identification

[0166]

[0167] To eliminate the differences in the dimensions of different parameters, the measured values ​​of each indicator are mapped to a standardized score of 0-10 according to the following mapping rules. ,in, This represents the index number. The mapping rule is designed based on the normal distribution characteristics of the laboratory data.

[0168] The mapping rules are as follows:

[0169] When the parameter value is less than or equal to the upper limit of the safety baseline value point;

[0170] When the upper limit of the safety benchmark value < the parameter value ≤ the lower limit of the critical outlier value ;

[0171] When the lower limit of the critical outlier is less than the parameter value and ≤ the upper limit of the critical outlier... point;

[0172] When the upper limit of the critical outlier value is less than the parameter value and less than the lower limit of the danger threshold... ;

[0173] When the parameter value is greater than the lower limit of the danger threshold point.

[0174] S72: The hazard level of smoldering fire sources is divided into five levels. Correlation analysis is used to calculate the correlation coefficient between each indicator and the hazard level of smoldering fire sources, and then the weights are obtained by normalization. This ensures that key indicators play a dominant role in the scoring. The weights of the parameters for each indicator in the dark fire source identification are shown in Table 2.

[0175] Table 2 Weighting of Indicators for Identifying Hidden Fire Sources

[0176]

[0177] The weights were optimized and verified to ensure stability and error under different burial depths and ventilation conditions. .

[0178] S73: Based on standardized scores and weight Calculate the comprehensive risk score of hidden fire sources. Dark Fire Source Comprehensive Risk Score The formula for the weighted sum of the standardized scores and corresponding weights of each indicator is as follows:

[0179]

[0180] in, The higher the score, the greater the level of danger.

[0181] S74: To avoid misjudgment based on a single parameter, a correction rule is set:

[0182] If the highest temperature ℃ (critical temperature for spontaneous combustion of coal) and concentration , Add 1.5 points directly, but No more than 10 points.

[0183] If the radon content is abnormally high... And thermal anomaly area , Add 1.0 point directly, but No more than 10 points.

[0184] If all parameter values ​​are below the safety baseline value The score was directly determined to be the background fluctuation area.

[0185] like Points, verification points The suspicious area was identified as a source of smoldering fire.

[0186] S75: Based on comprehensive risk assessment of hidden fire sources Establish a rating system The correspondence between the danger levels of dark fire sources and their corresponding danger levels is as follows:

[0187] when At that time, the danger level of the hidden fire source was Level 1 represents the background fluctuation region;

[0188] when The hazard level of the hidden fire source is Level IV, which is a low-risk area;

[0189] when The hazard level of the hidden fire source is Level III, which is a medium-risk area;

[0190] when The hazard level of the hidden fire source is Level II, which is a high-risk area;

[0191] when The ignition source is classified as Level I, indicating an emergency risk zone.

[0192] The risk classification is shown in Table 3.

[0193] Table 3 Risk Level 5 Classification Identification Table

[0194]

[0195] S8: Generate early warning information and output early warning results. The entire process for identifying the level of ignition sources in suspected areas of shallow coal seams is as follows: Figure 6 As shown.

[0196] Specifically, step S8 includes:

[0197] S81: Based on the classification results, generate a warning message containing the following:

[0198] Spatial information: The precise location and extent of the corresponding hidden fire source on the GIS electronic map.

[0199] Attribute information: identified hazard level, indicator value, risk score .

[0200] Time information: discovery time, data collection time.

[0201] Recommended measures: Emergency response or monitoring plans recommended based on the risk level.

[0202] S82: Based on the warning information in step S81, generate a warning map in the form of an electronic map, and mark the location, range and danger level of the smoldering fire source on the warning map with different colors or symbols.

[0203] S83: Based on the warning information in step S81, generate a color risk heat map (e.g., Figure 7 (as shown) and warning reports containing warning information.

[0204] S84: Outputs early warning maps, color-coded risk heat maps, and early warning reports. Managers can quickly understand the severity and precise location of a fire and take timely measures such as cooling, injecting inert gas, or excavating firebreaks. If only general signs of temperature rise are detected and the combustion criteria are not met, it is classified as a low-level hazard and monitored continuously. However, if relevant parameters rise further and exceed thresholds during subsequent inspections, the alarm level will be automatically upgraded to ensure early detection and early warning, preventing potential hazards from escalating into disasters.

[0205] Specifically, this method achieves real-time, all-area, and three-dimensional monitoring of concealed fire sources in shallow coal seams. Through air-ground collaborative operations, it constructs a monitoring network of "wide-area initial screening in the air, detailed investigation at fixed points in the air, and close-range verification detection on the ground," overcoming the contradictions of "inaccurate coarse detection and narrow coverage of fine detection" in traditional methods. It significantly improves the spatial coverage and temporal continuity of fire source detection, achieving blind-spot-free and uninterrupted monitoring. It also enhances the accuracy and reliability of fire source detection. Through the fusion analysis of multiple data sources, including temperature field (full-area thermal infrared image data), indicator gas concentration data, radioactive radon anomaly ratio, and surface temperature, combined with flight altitude temperature correction and multi-layer adaptive network algorithm compensation model temperature correction, it effectively eliminates environmental interference, achieving high-precision temperature identification and quantitative characterization of open and hidden fire sources, significantly reducing the risk of false alarms and missed alarms. It achieves intelligent hierarchical early warning and decision support based on data fusion analysis. By integrating multiple parameters such as thermal infrared thermography, indicator gas concentration, and radioactive radon content, a comprehensive risk scoring standard for smoldering fire sources is constructed. Based on a five-level classification standard, it automatically classifies hazard levels and outputs visualized early warning maps and disposal suggestions, providing timely and scientific hierarchical response basis for mine safety management personnel, enhancing the pertinence and timeliness of fire prevention and control. It also enhances environmental adaptability and operational automation. Employing a differentiated unmanned detection and early warning flight device platform and intelligent flight path planning strategy, combined with an air-ground collaborative task scheduling mechanism, it can adapt to complex terrain and weather conditions, enabling autonomous rotation of detection equipment, real-time data transmission, and dynamic task adjustment. This reduces the intensity of manual intervention and improves long-term stable operation in harsh environments such as mining areas and spoil heaps.

[0206] Please see Figure 2 and Figure 8 This invention provides a full-field monitoring and early warning system for concealed fire sources in shallow coal seams, used to realize the full-field monitoring and early warning method for concealed fire sources in shallow coal seams described above. The system includes a full-field detection module, a low-altitude-ground collaborative scheduling and task triggering module, a real-time intelligent identification module, and an early warning output module.

[0207] The full-field detection module includes a low-altitude monitoring unit and a ground-based verification unit. The low-altitude monitoring unit comprises a wide-area initial screening unmanned detection and early warning flight device and a fixed-point detailed survey unmanned detection and early warning flight device. The ground-based verification unit includes a ground-based tracked detection and early warning device. The wide-area initial screening unmanned detection and early warning flight device and the fixed-point detailed survey unmanned detection and early warning flight device employ powered flight modules with different rotor configurations. The wide-area initial screening unmanned detection and early warning flight device uses a 6-rotor powered flight module, which has higher system redundancy, flight stability, and payload capacity. It can stably carry a high-sensitivity infrared dual-view thermal imager and low-light imaging equipment for long-term cruise under large-scale and complex weather conditions, ensuring thermal infrared imaging quality and temperature measurement accuracy. The wide-area initial screening unmanned detection and early warning flight device is used to monitor the monitoring area (in the wide-area initial screening layer, 50-100...). It conducts cruise scanning at low altitudes to acquire thermal infrared and visible light image data of the monitored area.

[0208] The fixed-point detailed survey unmanned detection and early warning flight device adopts a four-rotor low-altitude powered flight module, which is lightweight, maneuverable, and has minimal airflow disturbance, making it suitable for low-altitude (fixed-point detailed survey layer, <50) operations. The low-altitude (low-speed hovering) unmanned aerial vehicle (UAV) performs precise indicator gas detection, equipped with an indicator gas acquisition unit and an infrared thermal imager. This allows for accurate acquisition of gas concentration data and surface temperature gradient distribution, while avoiding interference from rotor downwash airflow on indicator gas diffusion. The fixed-point detailed survey UAV is used for cruise scanning of suspicious areas, acquiring surface temperature gradient distribution and indicator gas concentration profile data.

[0209] The ground-based tracked detection and early warning device, equipped with multi-angle temperature probe units, multiple indicator gas detectors, and a radioactive radon information acquisition instrument, receives suspicious area marker information sent by the fixed-point detailed survey unmanned detection and early warning aerial device. It works in conjunction with the device to conduct point-by-point temperature probes, indicator gas concentration detection, and underground temperature and radon content sampling verification in suspicious areas, assisting in the identification of hidden fire sources in shallow coal seams. The ground-based tracked detection and early warning device is used to probe designated verification points in suspicious areas, acquiring surface temperature data, underground temperature data, indicator gas data escaping from probe holes, and radioactive radon content data.

[0210] During actual detection, the system follows a collaborative monitoring strategy of "sparse initial screening followed by dense, layered and step-by-step" approach. This involves unmanned detection and early warning flight devices employing differentiated flight altitudes at different stages to achieve a progressive detection process, moving from large-scale initial screening to detailed local investigation. This strategy is specifically manifested in the altitude configuration of three stages:

[0211] On-site adaptation preparation phase: Using a cruising altitude of 150m, a rapid and extensive preliminary survey of the monitoring area is conducted, completing the first macroscopic inspection of the entire mining area. This altitude provides a wide field of view, suitable for assessing topography, identifying obvious obstacles and the distribution of macroscopic thermal anomalies, providing an area overview and environmental adaptation basis for subsequent detailed exploration.

[0212] Wide-area initial screening stage: The cruising altitude of the wide-area initial screening unmanned detection and early warning flight device is set at 100m. This altitude ensures a large coverage area while maintaining high infrared imaging resolution and gas detection effectiveness, making it suitable for systematic thermal anomaly scanning and open flame source identification across the entire field. The unmanned detection and early warning flight device flies along a preset route, acquiring surface temperature field data in real time, and combining temperature threshold and spatial continuity analysis to extract areas with potential hazards.

[0213] Detailed Targeted Investigation Phase: For marked suspicious areas, the unmanned aerial vehicle (UAV) for detailed targeted investigation descends to an altitude of 50m to perform low-altitude precision detection. This altitude offers extremely high spatial resolution (detecting thermal anomalies at the <1m level) and simultaneously positions the onboard indicator gas collection unit at its optimal detection altitude, enabling accurate acquisition of concentration profile data for anomalous indicator gases. A fan-shaped scan is performed above the target area, combined with close-range verification by a ground-based tracked detection and early warning device, achieving multi-angle, high-precision detection of potential ignition sources.

[0214] To achieve full-area detection, a grid route was used during the on-site adaptation preparation phase to cover the ground from multiple angles in different directions, avoiding missed detections due to stepped terrain. For the wide-area initial screening and fixed-point detailed investigation phases, a circular route was used for meticulous inspection. The grid route is suitable for full-coverage census tasks, while the circular route facilitates hotspot re-examination.

[0215] The low-altitude-ground coordinated scheduling and task triggering module is used to organize, allocate and switch the detection tasks of the low-altitude monitoring unit and the ground verification unit according to the detection results during the detection of hidden fire sources (i.e., dark fire sources) in shallow coal seams.

[0216] The all-time intelligent identification module includes a ground analysis terminal, which is used to receive and process detection data from the low-altitude monitoring unit and the ground verification unit, determine suspicious areas and open flame sources in the monitoring area, and identify smoldering flame sources and their degree of danger in the suspicious areas.

[0217] The early warning output module is used to output early warning results.

[0218] The system possesses continuous monitoring capabilities over time and full-coverage, blind-spot-free detection capabilities in space, achieved through the following methods:

[0219] The low-altitude monitoring unit and the ground-based verification unit can operate alternately or collaboratively to achieve uninterrupted operation. The wide-area preliminary screening unmanned detection and early warning flight device is equipped with thermal infrared imaging and low-light imaging equipment to adapt to nighttime, smoke, or low-visibility conditions; multiple unmanned detection and early warning flight devices can take turns on duty or automatically replace batteries to maintain continuous patrol; when the fixed-point detailed survey unmanned detection and early warning flight device is unable to operate, the ground-based tracked detection and early warning device takes over the near-ground monitoring task; monitoring data is transmitted back in real time through wireless communication links to achieve continuous data updates and immediate early warning of anomalies.

[0220] The system achieves comprehensive, blind-spot-free detection of the monitoring area through air-to-ground collaboration. The wide-area initial screening unmanned detection and early warning flight device conducts extensive cruise scans based on preset routes planned by GIS, dynamically adjusting flight altitude, speed, and track spacing according to terrain undulations. It performs full-field temperature correction on the wide-area initial screening temperature measurement results based on subsequent test parameters to eliminate the influence of different ground object emissivity. When the fixed-point detailed investigation unmanned detection and early warning flight device identifies suspected anomalies, it transmits spatial location information to the ground-based tracked detection and early warning device. The ground-based tracked detection and early warning device autonomously plans its path based on the transmitted information and enters areas difficult for the unmanned detection and early warning flight device to conduct close-range verification. The system prioritizes multiple suspicious points based on their degree of anomaly and spatial distribution, coordinating multiple air-to-ground detection and early warning devices to operate in parallel. Wireless self-organizing network communication is used between the air-to-ground platforms to achieve synchronous transmission of commands and data, avoiding monitoring blind spots.

[0221] The low-altitude-ground coordinated scheduling and task triggering module realizes the issuance of task commands and data synchronization between air-to-ground detection platforms through wireless communication links. Based on the surface temperature anomaly information obtained in the wide-area preliminary screening stage, it automatically identifies suspected fire sources and suspicious areas, and selectively triggers at least one of the following detection tasks according to preset detection process rules:

[0222] (1) When the surface heat anomaly is significant, the low-altitude end monitoring unit is prioritized to perform fixed-point detailed investigation and detection tasks, and the target area is scanned for suspicious area indicator gas profile detection.

[0223] (2) When the surface thermal anomaly is not significant but there are continuous anomaly distribution characteristics, a verification task instruction is issued to the ground-end verification unit to guide it to approach the target area along the preset safe path and perform near-field fine detection such as surface touch temperature detection, indicator gas detection and borehole radon detection.

[0224] (3) During the exploration process, the execution order, exploration platform type and task priority of subsequent exploration tasks are adjusted based on the real-time monitoring data transmitted back by the low-altitude end monitoring unit and the ground end verification unit.

[0225] The above embodiments merely illustrate the basic principles and characteristics of the present invention, but are not limited to the above implementation schemes. It should be understood that those skilled in the art can make various changes and modifications to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for full-field monitoring and early warning of concealed fire sources in shallow coal seams, characterized in that, Includes the following steps: S1: Use the wide-area preliminary screening unmanned detection and early warning flight device to cruise and scan the monitoring area, obtain thermal infrared and visible light image data of the monitoring area, and record the auxiliary positioning information data of the wide-area preliminary screening unmanned detection and early warning flight device in real time. The auxiliary positioning information data includes flight time data and spatial coordinate data. In step S1, the wide-area initial screening unmanned detection and early warning flight device adopts a partitioned adaptive coverage route planning strategy when it performs cruise scanning in the monitoring area. Specifically, it includes: forming an outer closed cruise trajectory along the outer boundary of the monitoring area, and generating several inner cruise sub-paths within the area. The sub-paths are connected by radial connecting segments to form a progressive ring-radial composite route from the outside to the inside. S2: Remove redundant and blurred thermal imaging frames from the thermal infrared images obtained in step S1. S3: Perform temperature compensation on the thermal infrared image processed in step S2 to obtain a preliminary surface temperature field distribution map; S4: Based on the preliminary surface temperature field distribution map obtained in step S3, identify open flame sources and suspicious areas, and mark the locations of open flame sources and suspicious areas; S5: For marked suspicious areas, use a fixed-point detailed survey unmanned detection and early warning flight device to conduct cruise scanning, obtain the surface temperature gradient distribution and indicator gas concentration data of the suspicious area, and record the flight time data and spatial coordinate data of the fixed-point detailed survey unmanned detection and early warning flight device in real time. Use a ground tracked walking detection and early warning device to conduct detection. The ground tracked walking detection and early warning device measures the temperature at the designated verification point in the suspicious area and drills a detection hole to obtain the surface temperature data, underground temperature data, indicator gas data and radioactive radon content data at the verification point. In step S5, the fixed-point detailed investigation unmanned detection and early warning flight device adopts a multi-scale local flight path planning strategy centered on the suspicious area marker point. Specifically, it includes: a closed loop trajectory with the marker point as the center, and a fan-shaped scanning trajectory that diverges radially outward or inward. Each trajectory corresponds to a different flight altitude. S6: The monitoring data obtained in step S5, the flight time data and spatial coordinate data of the fixed-point detailed investigation unmanned detection and early warning flight device, the auxiliary positioning information data of the wide-area preliminary screening unmanned detection and early warning flight device in step S1, and the preliminary surface temperature field distribution map data in step S4 are fused and analyzed to establish the identification index of dark fire source. S7: Establish a risk classification standard for hidden fire sources based on the identification indicators of hidden fire sources; S8: Generate early warning information and output the early warning results.

2. The method for full-field monitoring and early warning of concealed fire sources in shallow coal seams according to claim 1, characterized in that, Step S2 includes: The process of removing redundant thermal imaging frames is as follows: the geographic overlap area is calculated using POS data of consecutive frames. When the overlap rate exceeds 85% and the average temperature difference of the overlap area is less than 0.5℃ and the structural similarity is higher than 0.98, it is determined to be a redundant frame and removed. The process of removing blurry thermal imaging frames is as follows: the image sharpness of each frame is evaluated, and the violent motion at the moment of acquisition is detected simultaneously in combination with IMU data. If the frame sharpness is lower than 60% of the median value of the flight segment, or if there is violent motion and the sharpness is lower than 80%, it is determined to be an invalid blurry frame and is removed.

3. The method for full-field monitoring and early warning of concealed fire sources in shallow coal seams according to claim 1, characterized in that, Step S3 includes: S31: A continuous correction function is introduced to correct temperature data at different altitudes. The formula is as follows: in, This indicates the temperature received by the thermal imager as the altitude approaches infinity; This represents a reference temperature value when the altitude is close to the Earth's surface. This is the compensation coefficient for topographic and environmental impacts, with a value range of 1.0-1.2; The attenuation coefficient is obtained by fitting the atmospheric extinction characteristics. The current flight altitude of the wide-area initial screening unmanned detection and early warning flight device; This is the corrected equivalent temperature measurement value; S32: Utilizing a multi-layer adaptive network algorithm to compensate for the output temperature of the compensation model. The formula is: in, For the input secondary parameter vector; For the first One radial base center; For the corresponding width parameter; For output weights; S33: Output the compensated temperature measurement value The formula is: in, This represents the equivalent temperature measurement value calculated using a continuous correction function based on flight altitude. It is a multi-layer adaptive network algorithm compensation model based on the compensation temperature output of multiple secondary parameters; 、 Adaptive weighting coefficients for interaction terms; It is a temperature scale normalization factor used to control the nonlinear range of the cross term; It is the hyperbolic tangent function, used to introduce nonlinear interaction terms; S34: Generate a preliminary surface temperature field distribution map .

4. The method for full-field monitoring and early warning of concealed fire sources in shallow coal seams according to claim 3, characterized in that, Step S4 includes: S41: Based on the preliminary surface temperature field distribution map, the temperature value exceeding the preset open flame detection threshold will be considered. The pixel regions were extracted as high-temperature candidate regions. An image recognition model was used to perform visual feature analysis on the corresponding high-temperature candidate regions in the synchronously acquired visible light images to identify flames, smoke, and severe heat burns on the ground. Then, spatial connectivity analysis and clustering were performed on the high-temperature candidate regions to remove isolated noise points, and regions that simultaneously met the criteria of "temperature exceeding..." The continuous area with "positive visible light characteristics" and the area with significant high temperature accumulation were identified as open flame sources; S42: Keep the temperature below the threshold. Areas that are negative or uncertain in visible light verification are identified as suspicious areas; areas with local disturbances in texture features in infrared images are identified as suspicious areas; and areas with normal temperature but located near historical fire zones, crack zones, or gas escape channels are identified as suspicious areas. S43: Mark the location of fire sources and suspected areas using GIS electronic maps.

5. The method for full-field monitoring and early warning of concealed fire sources in shallow coal seams according to claim 4, characterized in that, Step S5 includes: S51: Generate a set of suspicious regions for the marked suspicious regions. Each region Preliminary center coordinates provided by GIS electronic map ; S52: The flight path planning of the fixed-point detailed survey unmanned detection and early warning flight device is based on... The polar coordinate grid centered on the suspicious area has the sampling point locations as follows: ,in, Radial distance, It is the azimuth angle. At the flight altitude, surface temperature was measured at each sampling point to establish a surface temperature gradient distribution; at each sampling point, the indicator gas concentration vector was measured. , construct regions Indicator gas concentration field in the air Complete the acquisition of three-dimensional profiles indicating gas distribution, among which, This indicates the sampling location coordinates of the fixed-point detailed survey unmanned detection and early warning flight device; S53: Ground-tracked detection and early warning device in the target area Designated verification points within Surface temperature was measured at the location to obtain surface temperature data; at the verification point... Drill a probe hole to depth Measure the temperature profile inside the probe hole To obtain underground temperature data and measure the radioactive radon content in the probe borehole. To obtain radon content data at a height above the Earth's surface At the location, the concentration of the indicator gas escaping from the probe hole is measured. .

6. The method for full-field monitoring and early warning of concealed fire sources in shallow coal seams according to claim 5, characterized in that, Step S6 includes: S61: For temperature measurement values Optimization was performed to obtain the temperature value. ; Based on temperature measurement value and temperature value To obtain local temperature difference ; S62: Based on the surface temperature gradient distribution from step S52, the surface temperature data from step S53, and the optimized temperature values. The auxiliary positioning information data of the wide-area initial screening unmanned detection and early warning flight device and the flight time data and spatial coordinate data of the fixed-point detailed survey unmanned detection and early warning flight device are used to target the area. Preliminary surface temperature field distribution map Calibration and refinement are performed to generate a surface temperature field distribution map of the target area. ; S63: The concentration of indicator gas escaping from the detection hole measured by the ground tracked movement detection and early warning device. The indicator gas concentration field measured vertically above the same location by a fixed-point detailed survey unmanned detection and early warning flight device. The formula for superposition is: in, Indicates source strength, representing the rate at which characteristic gases are released into the atmosphere; Indicates wind speed; , These are the horizontal and vertical diffusion parameters, respectively; Indicates source height; Indicates the lateral distance from the central axis; Vertical height represents the height of the calculation point above the ground; This indicates the sampling location coordinates of the fixed-point detailed survey unmanned detection and early warning flight device; S64: Based on temperature profile within the probe hole Calculate the geothermal gradient anomaly index The formula is: in, This represents the normal geothermal gradient for the region. S65: Calculate the radioactive radon anomaly ratio based on radon content data. The formula is: in, This indicates the radon content data measured by the ground-based tracked detection and early warning device; The background radon content of the region; S66: Based on the integration of all data, identification indicators for dark fire sources are obtained. These indicators include temperature-related indicators and gas-related indicators. The temperature indicators include optimized temperature values. Temperature profile inside the probe hole Local temperature difference Geothermal gradient anomaly index Optimized verification points exist Size of the thermal anomaly region in the temperature field ,correspond Average temperature in the temperature field With the highest temperature Gas-related indicators include the indicator gas concentration field measured by the fixed-point detailed survey unmanned detection and early warning flight device. Concentration of indicator gas escaping from the detection hole Gas source strength Ratio to radioactive radon anomaly ; S67: Based on the identification indicators obtained in step S66, select the highest temperature. Geothermal gradient anomaly index Local temperature difference Thermal anomaly area Average temperature Radioactive radon anomaly ratio Gas source strength , concentration, concentration, Concentration serves as an indicator for identifying smoldering fire sources.

7. The method for full-field monitoring and early warning of concealed fire sources in shallow coal seams according to claim 6, characterized in that, Step S7 includes: S71: Based on the identification indicators determined in step S67, determine the safety baseline value, critical outlier value, and danger threshold of each indicator, and calculate the standardized score according to the following mapping rules. ,in, Indicates the index sequence number; The mapping rules are as follows: When the parameter value is less than or equal to the upper limit of the safety baseline value point; When the upper limit of the safety benchmark value < the parameter value ≤ the lower limit of the critical outlier value ; When the lower limit of the critical outlier is less than the parameter value and the upper limit of the critical outlier is less than or equal to the critical outlier. point; When the upper limit of the critical outlier value is less than the parameter value and less than the lower limit of the danger threshold... ; When the parameter value is greater than the lower limit of the danger threshold point; S72: The hazard level of smoldering fire sources is divided into five levels. Correlation analysis is used to calculate the correlation coefficient between each indicator and the hazard level of smoldering fire sources, and then the weights are obtained by normalization. ; S73: Based on standardized scores and weight Calculate the comprehensive risk score of hidden fire sources. The formula is as follows: in, The higher the score, the greater the level of danger. S74: To avoid misjudgment based on a single parameter, a correction rule is set: If the highest temperature ℃ and concentration , Add 1.5 points directly, but No more than 10 points; If the radon content is abnormally high... And thermal anomaly area , Add 1.0 point directly, but No more than 10 points; If all parameter values ​​are below the safety baseline value The score is directly determined to be the background fluctuation area; like Points, verification points The suspicious area was identified as a source of smoldering fire. S75: Based on comprehensive risk assessment of hidden fire sources Establish a rating system The correspondence between the danger levels of dark fire sources and their corresponding danger levels is as follows: when At that time, the danger level of the hidden fire source was Level 1 represents the background fluctuation region; when The hazard level of the hidden fire source is Level IV, which is a low-risk area; when The hazard level of the hidden fire source is Level III, which is a medium-risk area; when The hazard level of the hidden fire source is Level II, which is a high-risk area; when The ignition source is classified as Level I, indicating an emergency risk zone.

8. The method for full-field monitoring and early warning of concealed fire sources in shallow coal seams according to claim 7, characterized in that, Step S8 includes: S81: Based on the classification results, generate a warning message containing the following: Spatial information: The precise location and extent of the ignition source corresponding to the verification point on the GIS electronic map; Attribute information: identified hazard level, indicator value, risk score ; Time information: discovery time, data collection time; Recommended measures: Emergency response or monitoring plans recommended based on the risk level; S82: Based on the warning information in step S81, generate a warning map in the form of an electronic map, and mark the location, range and danger level of the smoldering fire source on the warning map with different colors or symbols; S83: Based on the warning information in step S81, generate a color risk heat map and a warning report containing the warning information; S84: Output early warning map, color risk heat map and early warning report.

9. A full-field monitoring and early warning system for concealed fire sources in shallow coal seams, used to implement the full-field monitoring and early warning method for concealed fire sources in shallow coal seams as described in any one of claims 1-8, characterized in that, It includes a full-field detection module, a low-altitude-ground coordinated scheduling and mission triggering module, a real-time intelligent identification module, and an early warning output module; The full-field detection module includes a low-altitude monitoring unit and a ground-based verification unit. The low-altitude monitoring unit includes a wide-area initial screening unmanned detection and early warning flight device and a fixed-point detailed investigation unmanned detection and early warning flight device. The ground-based verification unit includes a ground-based tracked detection and early warning device. The wide-area initial screening unmanned detection and early warning flight device is used to perform cruise scanning of the monitoring area and acquire thermal infrared and visible light image data of the monitoring area. The fixed-point detailed investigation unmanned detection and early warning flight device is used to perform cruise scanning of suspicious areas and acquire surface temperature gradient distribution and indicator gas concentration profile detection data of the suspicious areas. The ground-based tracked detection and early warning device is used to detect designated verification points in the suspicious areas and acquire surface temperature data, underground temperature data, indicator gas data escaping from the detection borehole, and radioactive radon content data. The low-altitude-ground coordinated scheduling and task triggering module is used to organize, allocate, and switch the detection tasks of the low-altitude monitoring unit and the ground verification unit according to the detection results during the detection of shallow buried concealed fire sources. The all-time intelligent identification module includes a ground analysis terminal, which is used to receive and process detection data from the low-altitude monitoring unit and the ground verification unit, determine suspicious areas and open flame sources in the monitoring area, and identify smoldering flame sources and their degree of danger in the suspicious areas. The warning output module is used to output the warning results.

Images