Open-air sharp-dip coal seam spontaneous combustion prevention and control system based on space-air-ground multi-source fusion

Abstract

The invention relates to an open-air sharp-dip coal seam spontaneous combustion prevention and control system based on space-air-ground multi-source fusion, and belongs to the technical field of coal seam spontaneous combustion prevention and control. The system comprises a satellite remote sensing module used for obtaining original remote sensing data and determining satellite remote sensing features; the unmanned aerial vehicle monitoring module is used for collecting original unmanned aerial vehicle data and determining event camera features and infrared features; the ground monitoring module is used for monitoring CO concentration data and determining drilling characteristics; measuring a temperature signal, and determining distributed temperature characteristics; the fusion processing module is used for performing multi-modal feature fusion on all the features to obtain a fused feature vector; the risk level calculation module is used for calculating the spontaneous combustion fire risk level; and the risk assessment module is used for generating a spontaneous combustion assessment result and performing spontaneous combustion prevention and control on the target open-air sharp-dip coal seam. The spontaneous combustion risk of the target open-air sharp-dip coal seam can be accurately recognized, and maximum safety prevention and control are achieved.

Description

Technical Field

[0001] This invention relates to the field of coal seam spontaneous combustion prevention and control technology, and in particular to an open-pit, steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground. Background Technology

[0002] Spontaneous combustion fires in open-pit steeply dipping coal seams have unique drawbacks due to the combined effects of open-pit mining exposure and the dipping characteristics of the coal seam: the original fissures and the post-mining slope collapses form an open fissure network, providing ample oxygen for spontaneous combustion. The fire source is often hidden deep within the open-pit steeply dipping coal seam or below the collapse accumulation area, and is highly concealed by surface coverings, with no obvious surface signs in the early stages; in the open-pit environment without underground roadways to block the fire, the fire spreads longitudinally along the fissures much faster than in gently dipping coal seams, easily forming a three-dimensional combustion from the slope to the depths, and the range is rapidly expanded by wind; the slope stability of open-pit steeply dipping coal seams is poor, with a high risk of collapse and rockfall, and the harsh working conditions make close-range manual intervention difficult. Once the fire gets out of control, it can easily lead to slope instability, threatening the safety of the mining area.

[0003] Existing technologies have significant limitations in monitoring open-pit steeply inclined coal seam fires: A single monitoring mode is difficult to adapt to complex needs; while traditional ground-based technologies (drilling, tubular systems, fiber optics, etc.) can acquire localized data, they are limited by coverage area and environmental interference, failing to comprehensively capture the multimodal characteristics of fissure fire hazards (temperature, gas, dynamic open flames, etc.), and are prone to accuracy fluctuations due to dust and terrain disturbances; Space-based monitoring lacks microscopic support; although satellite remote sensing and UAV infrared inspections have wide coverage, they lack the ability to verify the details of concealed fire sources, and are prone to false alarms and missed detections when interfered with by strong light and dust; data fusion and response mechanisms are weak, often relying on single thresholds or simple algorithms, failing to effectively correlate multi-source data features, resulting in coarse fire hazard classification, delayed early warnings, low equipment automation, requiring significant manual intervention, and low risk assessment efficiency. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a spontaneous combustion prevention and control system for rapidly dipping open-pit coal seams based on multi-source fusion of air, space, and ground. The technical solution of this invention is as follows: The open-pit, steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground includes: The satellite remote sensing module is used to acquire raw remote sensing data of the target open-pit steeply dipping coal seam through a hyperspectral imager mounted on a satellite, and to determine the satellite remote sensing characteristics based on the raw remote sensing data. The drone monitoring module is used to collect raw drone data of the target open-pit steeply dipping coal seam through the event camera and infrared thermal imager carried by the drone, and to determine the event camera features and infrared features based on the raw drone data; The ground monitoring module is used to monitor CO concentration data of a preset number of depth segments in the target open-pit steeply dipping coal seam through drilling detection equipment, and to determine drilling characteristics based on the CO concentration data of the preset number of depth segments; it also measures the temperature signal of the target open-pit steeply dipping coal seam within a specified time period before the current time through a distributed fiber optic temperature measurement array, and determines the distributed temperature characteristics based on the temperature signal within the specified time period. The fusion processing module is used to perform multimodal feature fusion of satellite remote sensing features, event camera features, infrared features, drilling features and distributed temperature features based on an attention mechanism to obtain the fused feature vector of the target open-pit steeply dipping coal seam; The risk level calculation module is used to calculate the spontaneous combustion fire risk level of the target open-pit steeply inclined coal seam based on the fused feature vector of the target open-pit steeply inclined coal seam. The risk assessment module is used to generate spontaneous combustion assessment results based on the spontaneous combustion fire risk level and fused feature vector, and to carry out spontaneous combustion prevention and control of the target open-pit steeply inclined coal seam based on the spontaneous combustion assessment results.

[0005] Preferably, the satellite remote sensing module includes: The data scanning unit is used to acquire raw remote sensing data of the target open-pit steeply dipping coal seam during a specified time period using a hyperspectral imager mounted on a satellite. The spatial positioning unit is used to extract the coal characteristic spectral bands from the original remote sensing data, and to perform temperature inversion on the coal characteristic spectral bands to obtain inversion data. Based on the inversion data, all temperature anomaly pixels of the target open-pit steeply dipping coal seam are extracted from the coal seam distribution GIS layer of the mining area. All temperature anomaly pixels are combined to obtain the temperature anomaly zone. The center coordinates of the temperature anomaly zone are calculated, and the center coordinates of the temperature anomaly zone are used as the spatial feature of the temperature anomaly zone of the target open-pit steeply dipping coal seam. The fracture information acquisition unit extracts the fracture area of ​​the target open-pit steeply dipping coal seam from the coal seam distribution GIS layer of the mining area, and obtains the total number of fracture pixels in the fracture area according to the coal characteristic spectral band. It then calculates the percentage of the total number of fracture pixels to the total number of pixels in the coal area to obtain the fracture development intensity characteristics. The combined unit is used to combine the spatial characteristics of temperature anomaly zones and fracture development intensity characteristics of the target open-pit steeply dipping coal seam to obtain satellite remote sensing features.

[0006] Preferably, the spatial positioning unit calculates the center coordinates of the temperature anomaly zone. When, it is achieved through formula (1): (1); In formula (1), This represents the x-coordinate of the i-th temperature anomaly pixel in the temperature anomaly zone. This represents the ordinate of the i-th temperature anomaly pixel in the temperature anomaly zone. This represents the temperature of the i-th temperature anomaly pixel in the temperature anomaly zone.

[0007] Preferably, the raw UAV data includes raw event streams and raw infrared data, and the event camera and infrared thermal imager are spatiotemporally registered using a dual-modal calibration board. The UAV monitoring module includes: The event stream acquisition unit is used to acquire the original event stream of the target open-pit steeply dipping coal seam within a preset time period before the current moment using an event camera mounted on a drone. The event camera feature acquisition unit is used to extract multiple brightness abrupt change points in the original event stream, calculate the average grayscale change of all brightness abrupt change points within a unit time with the current time as the endpoint, and use the average grayscale change as the light signal intensity of the target open-pit steeply dipping coal seam at the current time; count the number of brightness abrupt change points in the original event stream within a preset time period with the current time as the endpoint, use the number of brightness abrupt change points as the light signal frequency of the target open-pit steeply dipping coal seam at the current time, and combine the light signal intensity and light signal frequency at the current time to obtain the event camera features of the target open-pit steeply dipping coal seam; The crack distribution extraction unit is used to extract crack distribution information based on multiple brightness abrupt changes in the original event stream. The infrared data acquisition unit is used to acquire raw infrared data of the target open-pit steeply dipping coal seam using an infrared thermal imager mounted on a drone. The infrared feature extraction unit divides the target open-pit steeply dipping coal seam into a predetermined number of layers along the direction perpendicular to the coal seam's dip, based on the original infrared data and fracture distribution information. Each layer corresponds to a fracture zone. The temperature gradient of each layer is calculated by taking the temperature difference between the top and bottom temperatures of the original infrared data. The temperature gradients of all layers are combined to obtain the vertical layering features of the target open-pit steeply dipping coal seam. The average temperature of all pixels in each layer is calculated based on the original infrared data. The average temperature of all layers is combined with the fracture distribution information to obtain the temperature features of the fracture zone. The infrared features of the target open-pit steeply dipping coal seam are obtained by combining the vertical layering features and the temperature features of the fracture zone.

[0008] Preferably, the crack distribution extraction unit includes: The crack zone generation sub-unit is used to cluster multiple brightness abrupt points in the original event stream according to the Hough transform, and obtain multiple crack zones and their original information. The fracture information extraction subunit is used to perform coordinate transformation on the original information of each fracture zone according to the Gauss-Kruger projection coordinate system to obtain the standard information of each fracture zone, and combine the standard information of all fracture zones to obtain the fracture distribution information.

[0009] Preferably, the ground monitoring module includes: The raw data monitoring unit is used to configure multiple sampling points in the target open-pit steeply inclined coal seam according to the preset grid size, and to configure drilling and detection equipment at each sampling point. Raw gas is extracted through all drilling and detection equipment, and raw CO concentration data of raw gas at each sampling point is analyzed by gas chromatograph. The sampling point selection unit is used to select sampling points where the original CO concentration data is greater than the preset concentration anomaly value as anomaly points, to select multiple consecutive anomaly points as anomaly zones, and to select anomaly zones where the CO concentration gradient along the target open-pit steeply dipping coal seam is greater than the preset gradient threshold as CO anomaly zones. The CO concentration data of a preset number of depth segments on the CO anomaly zone are collected by drilling detection equipment, and the CO concentration data of the preset number of depth segments are used as drilling features. The signal acquisition unit is used to acquire the temperature signal of the target open-pit steeply dipping coal seam within a specified time period before the current time through a distributed optical fiber temperature measurement array, calculate the temperature rise rate at the current time based on the temperature signal within the specified time period, and use the temperature rise rate at the current time as the distributed temperature characteristic of the target open-pit steeply dipping coal seam.

[0010] Preferably, the fusion processing module includes: The standardized unit is used to standardize and linearly map satellite remote sensing features, event camera features, infrared features, drilling features, and distributed temperature features to obtain a standard feature vector. The feature fusion unit is used to perform attention-based multimodal feature fusion on the standard feature vector to obtain the fused feature vector of the target open-pit steeply dipping coal seam. Each feature in the fused feature vector contains attention weights.

[0011] Preferably, the risk level calculation module includes: The weight adjustment unit is used to acquire historical fire data and current real-time monitoring data of the target open-pit steeply dipping coal seam, and combine the historical fire data and current real-time monitoring data to obtain actual fire data. It calculates the standard feature value of each feature of the actual fire data, and adjusts the attention weight of each feature in the fused feature vector according to the standard feature value to obtain the standard weight of each feature. The risk value calculation unit is used to calculate the spontaneous combustion fire risk level of the target open-pit steeply inclined coal seam based on the standard weight of each feature.

[0012] Preferably, when the weight adjustment unit adjusts the attention weight of each feature in the fused feature vector according to the standard feature value to obtain the standard weight of each feature, it is used to: Calculate the similarity between each feature in the fused feature vector and its corresponding standard feature value, and adjust the attention weight of each feature based on the similarity to obtain the standard weight of each feature.

[0013] Preferably, the risk assessment module includes: The instruction acquisition unit is used to acquire prevention and control instructions corresponding to the risk level of spontaneous combustion fire according to a preset risk level table. The information generation unit is used to generate spontaneous combustion risk information for each location of the target open-pit steeply dipping coal seam based on the fused feature vector, and to combine the spontaneous combustion risk information of all locations to obtain the spontaneous combustion assessment information of the target open-pit steeply dipping coal seam. The spontaneous combustion assessment result generation unit is used to combine spontaneous combustion assessment information and prevention and control instructions to generate spontaneous combustion assessment results, and to carry out spontaneous combustion prevention and control of the target open-pit steeply inclined coal seam based on the spontaneous combustion assessment results.

[0014] All of the above-mentioned optional technical solutions can be combined arbitrarily, and the present invention will not provide a detailed description of the structure after each combination.

[0015] By means of the above solution, the beneficial effects of the present invention are as follows: The system acquires satellite remote sensing features of the target open-pit steeply dipping coal seam through a satellite remote sensing module, event camera features and infrared features of the target open-pit steeply dipping coal seam through a drone monitoring module, and drilling features and distributed temperature features of the target open-pit steeply dipping coal seam through a ground monitoring module. Then, the system fuses all the above features through a fusion processing module to obtain the fused feature vector of the target open-pit steeply dipping coal seam. This achieves multi-dimensional feature analysis and fusion through air, space, and ground, solving the problems of traditional ground technology being limited by coverage and environmental interference, unable to fully capture the multimodal features of fissure fire hazards using a single sensor, and prone to accuracy fluctuations due to dust and terrain disturbances. The integration of air, space, and ground monitoring modes can simultaneously solve the problems of single air and space monitoring lacking microscopic support, single satellite remote sensing and single drone infrared inspection lacking detailed verification capabilities for hidden fire sources, and being prone to false alarms and missed detections when interfered with by strong light and dust.

[0016] The risk level calculation module calculates the spontaneous combustion risk level of the target open-pit steeply inclined coal seam based on the fused feature vector. The risk assessment module generates a spontaneous combustion assessment result based on the spontaneous combustion risk level and the fused feature vector. This solves the problem of the coarse fire hazard classification of traditional technologies and improves the accuracy of assessing the spontaneous combustion risk of open-pit steeply inclined coal seams. Combined with multi-source data fusion from air, space, and ground, it can accurately identify the spontaneous combustion risk of the target open-pit steeply inclined coal seam and achieve maximum safety control.

[0017] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0018] Figure 1This is a schematic diagram of the structure of the open-pit rapid-dip coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground provided in an embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram of the distribution of air-space-ground sensors in the open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on air-space-ground multi-source fusion provided in an embodiment of the present invention.

[0020] Figure 3 This is a schematic diagram of the process for generating fused feature vectors in an embodiment of the present invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0022] like Figure 1 As shown in the figure, the open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on air-space-ground multi-source fusion provided in this embodiment of the invention includes the following modules: The satellite remote sensing module is used to acquire raw remote sensing data of the target open-pit steeply dipping coal seam through a hyperspectral imager mounted on a satellite, and to determine the satellite remote sensing characteristics based on the raw remote sensing data. The drone monitoring module is used to collect raw drone data of the target open-pit steeply dipping coal seam through the event camera and infrared thermal imager carried by the drone, and to determine the event camera features and infrared features based on the raw drone data; The ground monitoring module is used to monitor CO concentration data of a preset number of depth segments in the target open-pit steeply dipping coal seam through drilling detection equipment, and to determine drilling characteristics based on the CO concentration data of the preset number of depth segments; it also measures the temperature signal of the target open-pit steeply dipping coal seam within a specified time period before the current time through a distributed fiber optic temperature measurement array, and determines the distributed temperature characteristics based on the temperature signal within the specified time period. The fusion processing module is used to perform multimodal feature fusion of satellite remote sensing features, event camera features, infrared features, drilling features and distributed temperature features based on an attention mechanism to obtain the fused feature vector of the target open-pit steeply dipping coal seam; The risk level calculation module is used to calculate the spontaneous combustion fire risk level of the target open-pit steeply inclined coal seam based on the fused feature vector of the target open-pit steeply inclined coal seam. The risk assessment module is used to generate spontaneous combustion assessment results based on the spontaneous combustion fire risk level and fused feature vector, and to carry out spontaneous combustion prevention and control of the target open-pit steeply inclined coal seam based on the spontaneous combustion assessment results.

[0023] Specifically, in the satellite remote sensing module, the target open-pit steeply dipping coal seam is an open-pit coal seam with a dip angle >45° used for spontaneous combustion prevention. The hyperspectral imager is an instrument capable of acquiring emission spectral data of the target open-pit steeply dipping coal seam in multiple narrow bands. The satellite remote sensing features include the spatial characteristics of the temperature anomaly zone and the intensity characteristics of fracture development in the target open-pit steeply dipping coal seam, totaling two dimensions.

[0024] In the UAV monitoring module, the event camera captures "events" in real time based on changes in illumination within the target open-pit steeply dipping coal seam, generating a raw event stream. The infrared thermal imager generates raw infrared data by detecting the infrared radiation (i.e., thermal radiation) emitted by the target open-pit steeply dipping coal seam. The raw UAV data includes the raw event stream and the raw infrared data. The event camera feature consists of the light signal intensity and frequency of the target open-pit steeply dipping coal seam at the current moment, forming a 2D feature; the infrared feature includes 8-dimensional vertical stratification features and 8-dimensional fracture zone temperature features of the target open-pit steeply dipping coal seam, for a total of 16 dimensions.

[0025] In the ground monitoring module, monitoring boreholes (10-30m deep, adapted to the thickness of the target open-pit steeply dipping coal seam) are arranged at 20m intervals along the dip direction. Drilling detection equipment (generally integrated with a gas sampling tube via armored thermocouples) is pre-embedded in the boreholes. CO concentration data at different depths are collected synchronously through the hollow channel in the drill pipe, preferably at 6 depths. The CO concentration data from these 6 depths constitute a 6-dimensional drilling feature. Distributed optical fibers are buried along the outcrop line and mining slope of the target open-pit steeply dipping coal seam. These fibers are laid through flame-retardant protective pipes to adapt to the complex terrain and open-fire risk of the target open-pit steeply dipping coal seam. A distributed optical fiber temperature measurement array is formed with a temperature measurement accuracy of ±1℃ and a spatial resolution of 1m. It acquires the temperature signal of the target open-pit steeply dipping coal seam within a specified time period prior to the current moment using optical time-domain reflectometry. The temperature rise rate at the current moment constitutes a 1-dimensional distributed temperature feature.

[0026] When fusing multimodal features based on an attention mechanism, the fusion processing module first standardizes various features to convert them into a unified dimension. Then, it extends the standardized features to the dimensions suitable for multimodal feature fusion (128 dimensions in this embodiment) through linear mapping. By fusing the 27 dimensions of features from satellite remote sensing, event camera, infrared, drilling, and distributed temperature data, a 128-dimensional fused feature vector is obtained.

[0027] In the risk level calculation module, the risk level of spontaneous combustion fire includes 5 levels: Level I (potential hazard), Level II (minor spontaneous combustion), Level III (active spontaneous combustion), Level IV (serious fire) and Level V (catastrophic fire).

[0028] In the risk assessment module, the spontaneous combustion assessment result is obtained by comprehensively analyzing the fused feature vector and spontaneous combustion fire risk level of the target open-pit steeply dipping coal seam. For example, the spontaneous combustion assessment result of a certain target open-pit steeply dipping coal seam is as follows: the spontaneous combustion fire risk level is Level II, the fracture distribution information of the fracture zone is (100, 180), the temperature gradient of the layer in which the fracture zone is located is 7, the drilling feature of the fracture zone at (100, 180) is 20; the local infrared temperature rise of the fracture zone at (100, 180) is >5℃, and there is a weak CO gas escape from the fracture area.

[0029] like Figure 2 As shown, it is a schematic diagram of the distribution of the air-ground sensors in an embodiment of the present invention.

[0030] In one specific embodiment, the satellite remote sensing module includes: The data scanning unit is used to acquire raw remote sensing data of the target open-pit steeply dipping coal seam during a specified time period using a hyperspectral imager mounted on a satellite. The spatial positioning unit is used to extract the coal characteristic spectral bands from the original remote sensing data, and to perform temperature inversion on the coal characteristic spectral bands to obtain inversion data. Based on the inversion data, all temperature anomaly pixels of the target open-pit steeply dipping coal seam are extracted from the coal seam distribution GIS layer of the mining area. All temperature anomaly pixels are combined to obtain the temperature anomaly zone. The center coordinates of the temperature anomaly zone are calculated, and the center coordinates of the temperature anomaly zone are used as the spatial feature of the temperature anomaly zone of the target open-pit steeply dipping coal seam. The fracture information acquisition unit extracts the fracture area of ​​the target open-pit steeply dipping coal seam from the coal seam distribution GIS layer of the mining area, and obtains the total number of fracture pixels in the fracture area according to the coal characteristic spectral band. It then calculates the percentage of the total number of fracture pixels to the total number of pixels in the coal area to obtain the fracture development intensity characteristics. The combined unit is used to combine the spatial characteristics of temperature anomaly zones and fracture development intensity characteristics of the target open-pit steeply dipping coal seam to obtain satellite remote sensing features.

[0031] Specifically, in the data scanning unit, the raw remote sensing data includes spectral bands and radiance. The specified time period is generally selected from the fixed time period with the least solar interference, preferably 6:00-8:00 and 18:00-20:00.

[0032] Within the spatial positioning unit, coal characteristic spectral bands are extracted from the original remote sensing data based on the characteristics of the coal's characteristic spectral bands (a reflection peak at 1.6 μm and an absorption peak at 2.2 μm). During temperature inversion, the radiative transfer model is used to perform temperature inversion on the radiance values ​​of the coal characteristic spectral bands to obtain the inversion data; the radiative transfer model is expressed by the formula: ;in, This represents the actual temperature of a given pixel at its location within the target open-pit steeply sloping coal seam. This represents the radiance value of this pixel in the characteristic spectral band of coal. , For satellite calibration coefficients. The inversion data corresponding to the coal seam outcrop and collapse areas of the target open-pit steeply dipping coal seam are used as the coal seam temperature data for the GIS layer of the mining area's coal seam distribution, excluding interference from non-coal areas (soil, vegetation, coal gangue piles). The mean value of the coal seam temperature data is calculated. with standard deviation Screening coalfield temperature data All temperature anomaly pixels are clustered using connectivity to form a continuous temperature anomaly band, and the center coordinates of the temperature anomaly band are calculated.

[0033] In the fracture information acquisition unit, the fracture region refers to cracks, fissures, or voids on the surface or inside the target open-pit steeply dipping coal seam caused by geological structure, mining, spontaneous combustion, or human factors. These are marked on the coal seam distribution GIS layer of the mining area. The total number of fracture pixels and the total number of pixels in the coal area are counted at a 30m resolution (1 pixel corresponds to a 900m² area in the target open-pit steeply dipping coal seam).

[0034] In one specific embodiment, the spatial positioning unit calculates the center coordinates of the temperature anomaly zone. When, it is achieved through formula (1): (1); In formula (1), This represents the x-coordinate of the i-th temperature anomaly pixel in the temperature anomaly zone. This represents the ordinate of the i-th temperature anomaly pixel in the temperature anomaly zone. This represents the temperature of the i-th temperature anomaly pixel in the temperature anomaly zone.

[0035] In one specific embodiment, the raw UAV data includes raw event streams and raw infrared data, and the event camera and infrared thermal imager are spatiotemporally registered using a dual-modal calibration board. The UAV monitoring module includes: The event stream acquisition unit is used to acquire the original event stream of the target open-pit steeply dipping coal seam within a preset time period before the current moment using an event camera mounted on a drone. The event camera feature acquisition unit is used to extract multiple brightness abrupt change points in the original event stream, calculate the average grayscale change of all brightness abrupt change points within a unit time with the current time as the endpoint, and use the average grayscale change as the light signal intensity of the target open-pit steeply dipping coal seam at the current time; count the number of brightness abrupt change points in the original event stream within a preset time period with the current time as the endpoint, use the number of brightness abrupt change points as the light signal frequency of the target open-pit steeply dipping coal seam at the current time, and combine the light signal intensity and light signal frequency at the current time to obtain the event camera features of the target open-pit steeply dipping coal seam; The crack distribution extraction unit is used to extract crack distribution information based on multiple brightness abrupt changes in the original event stream. The infrared data acquisition unit is used to acquire raw infrared data of the target open-pit steeply dipping coal seam using an infrared thermal imager mounted on a drone. The infrared feature extraction unit divides the target open-pit steeply dipping coal seam into a predetermined number of layers along the direction perpendicular to the coal seam's dip, based on the original infrared data and fracture distribution information. Each layer corresponds to a fracture zone. The temperature gradient of each layer is calculated by taking the temperature difference between the top and bottom temperatures of the original infrared data. The temperature gradients of all layers are combined to obtain the vertical layering features of the target open-pit steeply dipping coal seam. The average temperature of all pixels in each layer is calculated based on the original infrared data. The average temperature of all layers is combined with the fracture distribution information to obtain the temperature features of the fracture zone. The infrared features of the target open-pit steeply dipping coal seam are obtained by combining the vertical layering features and the temperature features of the fracture zone.

[0036] Specifically, in the event stream acquisition unit and the infrared data acquisition unit, the UAV is equipped with an event camera (resolution of 1280×720, frame rate ≥2MHz) and an infrared thermal imager (temperature measurement accuracy of ±0.5℃). It cruises at an altitude of 80-100m along the target open-pit steeply dipping coal seam, and takes pictures at an angle of 30~60°. It scans the target open-pit steeply dipping coal seam every 15 minutes to acquire the original event stream and the original infrared data.

[0037] In the event camera feature acquisition unit, points in the original event stream whose brightness values ​​change more than a contrast threshold are identified as brightness abrupt change points. The contrast threshold is typically set to 0.5. Then, based on the grayscale change of each point in the original event stream, the average grayscale change of all brightness abrupt change points within a unit of time (1 second) ending at the current time is calculated. The calculation formula is as follows: ;in, This represents the total number of brightness abrupt changes per unit time. The light signal intensity of the target open-pit steeply dipping coal seam at the current moment. This represents the grayscale change at the d-th brightness abrupt change point. The light signal intensity reflects the strength of the light signal released by oxidation in the crack region; higher intensity indicates more intense oxidation. The preset time period ending at the current moment is generally 1 minute.

[0038] In the fracture distribution extraction unit, the fracture distribution information includes fracture orientation, fracture density, fracture dip angle, and fracture length.

[0039] In the infrared feature extraction unit, the target open-pit steeply dipping coal seam is divided into eight 1-meter-thick layers according to the direction of the coal seam dip perpendicular to the target. Each layer corresponds to one fracture zone, which covers the shallow to deep parts of the target open-pit steeply dipping coal seam. For the j-th layer, the average temperature of the top and bottom boundaries of the layer is taken respectively. and The formula for calculating the temperature gradient of the j-th layer is: In addition, the eight temperature gradients constitute an eight-dimensional vertical stratification feature. The temperature mean of the eight strata and the fracture distribution information are combined to obtain the fracture zone temperature characteristics of the target open-pit steeply dipping coal seam.

[0040] In one specific embodiment, the crack distribution extraction unit includes: The crack zone generation sub-unit is used to cluster multiple brightness abrupt points in the original event stream according to the Hough transform, and obtain multiple crack zones and their original information. The fracture information extraction subunit is used to perform coordinate transformation on the original information of each fracture zone according to the Gauss-Kruger projection coordinate system to obtain the standard information of each fracture zone, and combine the standard information of all fracture zones to obtain the fracture distribution information.

[0041] Specifically, when the crack zone generating sub-unit clusters multiple brightness abrupt change points in the original event stream according to the Hough transform, it first maps the multiple brightness abrupt change points in the original event stream to the Hough transform space. Here, any brightness abrupt change point in the original event stream... After mapping to the Hough transform space, it is represented as , This refers to the dip angle of the fracture zone where the brightness abrupt change point is located. Multiple brightness abrupt changes with the same dip angle are clustered together. All brightness abrupt changes in a cluster are considered as all brightness abrupt changes in a fracture zone. Based on all brightness abrupt changes in a fracture zone, the corresponding fracture zone is determined in the target open-pit steeply dipping coal seam. The original information includes the dip angle and location information of the fracture zone.

[0042] In the fracture information extraction subunit, the location information of a certain fracture zone is transformed into the Gauss-Kruger projection coordinate system. The angle between the fracture zone and true north in the Gauss-Kruger projection coordinate system represents the orientation of the fracture zone. The density and length of the fracture zone can be determined based on the coal seam distribution GIS layer of the mining area.

[0043] In one specific embodiment, the ground monitoring module includes: The raw data monitoring unit is used to configure multiple sampling points in the target open-pit steeply inclined coal seam according to the preset grid size, and to configure drilling and detection equipment at each sampling point. Raw gas is extracted through all drilling and detection equipment, and raw CO concentration data of raw gas at each sampling point is analyzed by gas chromatograph. The sampling point selection unit is used to select sampling points where the original CO concentration data is greater than the preset concentration anomaly value as anomaly points, to select multiple consecutive anomaly points as anomaly zones, and to select anomaly zones where the CO concentration gradient along the target open-pit steeply dipping coal seam is greater than the preset gradient threshold as CO anomaly zones. The CO concentration data of a preset number of depth segments on the CO anomaly zone are collected by drilling detection equipment, and the CO concentration data of the preset number of depth segments are used as drilling features. The signal acquisition unit is used to acquire the temperature signal of the target open-pit steeply dipping coal seam within a specified time period before the current time through a distributed optical fiber temperature measurement array, calculate the temperature rise rate at the current time based on the temperature signal within the specified time period, and use the temperature rise rate at the current time as the distributed temperature characteristic of the target open-pit steeply dipping coal seam.

[0044] Specifically, in the raw data monitoring unit, the preset grid size is generally 5m × 5m, and it is deployed along the dip of the target open-pit steeply dipping coal seam. Bundle tubes are arranged at sampling points within the target open-pit steeply dipping coal seam according to the preset grid size, and the raw gas in the fractures is extracted using a vacuum pump. The specified time period is generally 20 seconds.

[0045] In the sampling point selection unit, the preset concentration anomaly value is generally set to 24 ppm. The preset gradient threshold is generally set to 5 ppm / m.

[0046] When calculating the temperature rise rate at the current moment, the signal acquisition unit calculates the current temperature based on the temperature signal within a specified time period. Temperature 1 hour ago The ratio of the difference to time, the formula for the rate of temperature increase at the current moment is: (Unit: ℃ / s).

[0047] In one specific embodiment, the fusion processing module includes: The standardized unit is used to standardize and linearly map satellite remote sensing features, event camera features, infrared features, drilling features, and distributed temperature features to obtain a standard feature vector. The feature fusion unit is used to perform attention-based multimodal feature fusion on the standard feature vector to obtain the fused feature vector of the target open-pit steeply dipping coal seam. Each feature in the fused feature vector contains attention weights.

[0048] In the standardized unit, for the k-th feature The formula for standardization is: ;in and These represent the mean and standard deviation of the historical features corresponding to each feature. For example, for the feature of fracture development intensity, if its historical features are 1, 2, and 3, then the mean of the fracture development intensity feature is 2, and the standard deviation is 0.82. Additionally, since temperature features are core indicators of fire risk, after standardization using the above formula, the temperature features need to be multiplied by an additional 1.4. Temperature features include infrared features and distributed temperature features. Finally, the standardized 27-dimensional features are linearly mapped to a unified 128-dimensional vector to obtain the standard feature vector.

[0049] In the feature fusion unit, multimodal feature fusion is implemented based on the attention mechanism of a transformer neural network. This mechanism optimizes the attention level of each feature in the standard feature vector through a two-layer weighting mechanism, making it more closely aligned with the evolution of fire hazard. The first layer assigns general modal weights: weights are initialized based on fire hazard sensitivity, with the general modal weights set as follows: event camera features 0.35 > distributed temperature features 0.25 > drilling features 0.2 > infrared features 0.1 > satellite remote sensing features 0.1. The second layer calculates and assigns weights to the spatiotemporal encoding vector, where the time encoding vector has a weight of 0.6 and the spatial encoding vector has a weight of 0.4. Both the time and spatial encoding vectors are 64-dimensional, incorporating the temporal and geological characteristics of the target open-pit steeply dipping coal seam.

[0050] The time-coded vector is generated based on the relative time difference of BeiDou time synchronization. The formula for generating it is: Among them, the time difference is encoded through the periodicity of the sine / cosine function. , Take 0-31 (a total of 32 groups of "sine + cosine" to generate a 64-dimensional time encoding vector). Controlling the time frequency, from Corresponding to long cycles (1 hour level) It corresponds to a short cycle (5 minutes) and covers the time changes of different stages of fire hazards.

[0051] Spatial encoding vector The formula for generating it is: ;in, The dip angle of the fracture zone, This indicates the orientation of the fracture zone.

[0052] Furthermore, when performing attention-based multimodal feature fusion on the standard feature vector, based on the universal modal weights of each feature in the standard feature vector, a dual-head attention-based multimodal feature fusion is performed by combining the temporal encoding vector and the spatial encoding vector with their respective weights (each head being one of the temporal and spatial encoding vectors), generating a 128-dimensional fused feature vector. Each feature in the fused feature vector contains attention weights. In this embodiment of the invention, the process for generating the fused feature vector is as follows: Figure 3 As shown.

[0053] In one specific embodiment, the risk level calculation module includes: The weight adjustment unit is used to acquire historical fire data and current real-time monitoring data of the target open-pit steeply dipping coal seam, and combine the historical fire data and current real-time monitoring data to obtain actual fire data. It calculates the standard feature value of each feature of the actual fire data, and adjusts the attention weight of each feature in the fused feature vector according to the standard feature value to obtain the standard weight of each feature. The risk value calculation unit is used to calculate the spontaneous combustion fire risk level of the target open-pit steeply inclined coal seam based on the standard weight of each feature.

[0054] Specifically, in the weight adjustment unit, historical fire data includes historical remote sensing data, historical UAV data, historical CO concentration data at a preset number of depth segments, and historical temperature signals. Current real-time monitoring data includes current remote sensing data, current UAV data, current CO concentration data at a preset number of depth segments, and current temperature signals. The standard feature value of each feature is obtained by extracting features from actual fire data using the feature extraction method provided in this embodiment of the invention, followed by standardization and linear mapping. Specifically, firstly, the actual fire data is processed by the satellite remote sensing module, UAV monitoring module, and ground monitoring module to obtain 27-dimensional actual satellite remote sensing features, actual event camera features, actual infrared features, actual drilling features, and actual distributed temperature features. Then, the above 27-dimensional actual satellite remote sensing features, actual event camera features, actual infrared features, actual drilling features, and actual distributed temperature features are standardized and linearly mapped to obtain a 127-dimensional actual standard feature vector, where the vector value corresponding to each feature in the actual standard feature vector is the standard feature value of each feature.

[0055] In the risk value calculation unit, the spontaneous combustion fire risk level of the target open-pit steeply inclined coal seam is obtained by weighting and summing the standard weights of all features with their standard values ​​and then rounding down.

[0056] In a specific embodiment, when the weight adjustment unit adjusts the attention weight of each feature in the fused feature vector according to the standard feature value to obtain the standard weight of each feature, it is used to: calculate the similarity between each feature in the fused feature vector and the corresponding standard feature value, adjust the attention weight of each feature according to the similarity, and obtain the standard weight of each feature.

[0057] Specifically, the similarity between each feature in the fused feature vector and its corresponding standard feature value is calculated using the cosine similarity formula. When the similarity between a feature and its corresponding standard feature value is less than 0.5, manual review is triggered. When the similarity is greater than 0.8, the corresponding feature's weight adjustment mechanism is triggered. For example, the attention weight of a feature is multiplied by its similarity to the corresponding standard feature value to obtain the adjusted weight value; this adjusted weight is then added to the feature's attention weight to obtain its standard weight. When the similarity between a feature and its corresponding standard feature value is neither less than 0.5 nor greater than 0.8, the feature's attention weight is directly used as its standard weight.

[0058] In one specific embodiment, the risk assessment module includes: The instruction acquisition unit is used to acquire prevention and control instructions corresponding to the risk level of spontaneous combustion fire according to a preset risk level table. The information generation unit is used to generate spontaneous combustion risk information for each location of the target open-pit steeply dipping coal seam based on the fused feature vector, and to combine the spontaneous combustion risk information of all locations to obtain the spontaneous combustion assessment information of the target open-pit steeply dipping coal seam. The spontaneous combustion assessment result generation unit is used to combine spontaneous combustion assessment information and prevention and control instructions to generate spontaneous combustion assessment results, and to carry out spontaneous combustion prevention and control of the target open-pit steeply inclined coal seam based on the spontaneous combustion assessment results.

[0059] Specifically, in the instruction acquisition unit, Table 1 is a schematic table of preset risk levels.

[0060] Table 1

[0061]

[0062] In the information generation unit, the spontaneous combustion risk information includes the location information of the spontaneous combustion risk, infrared features, vertical stratification features, CO concentration data of the depth segment, etc.

[0063] Based on all the above embodiments, the open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on air-space-ground multi-source fusion provided by the present invention has the following beneficial effects: After acquiring the original remote sensing data of the target open-pit steeply dipping coal seam using a satellite remote sensing module, the satellite remote sensing characteristics were determined. Original UAV data was collected using an event camera and infrared thermal imager mounted on a UAV, and the event camera characteristics and infrared characteristics were determined. CO concentration data at a preset number of depths was monitored using drilling detection equipment in the ground monitoring module to obtain drilling characteristics. Temperature signals were acquired using a distributed fiber optic temperature measurement array, and distributed temperature characteristics were determined. Multimodal feature fusion of satellite remote sensing characteristics, event camera characteristics, infrared characteristics, drilling characteristics, and distributed temperature characteristics yielded a fused feature vector of the target open-pit steeply dipping coal seam. A method for assessing the spontaneous combustion risk of open-pit steeply dipping coal seams integrating multimodal features was proposed. By fusing multimodal features, a monitoring system for open-pit steeply dipping coal seams from three perspectives (air, ground, and air) was formed, providing multifaceted and multi-dimensional data support for subsequent prediction of hidden spontaneous combustion risks.

[0064] The risk level calculation module calculates the spontaneous combustion risk level of the target open-pit steeply dipping coal seam based on the fused feature vector of the target open-pit steeply dipping coal seam. The risk assessment module generates the spontaneous combustion assessment result corresponding to the risk level and carries out spontaneous combustion prevention and control of the target open-pit steeply dipping coal seam based on the spontaneous combustion assessment result. This realizes the assessment of the spontaneous combustion risk level of the target open-pit steeply dipping coal seam based on the fused feature vector of air, space, and ground, which can accurately realize the spontaneous combustion risk assessment and significantly improve the accuracy of spontaneous combustion fire prevention and control.

[0065] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A multi-source open-pit coal seam spontaneous combustion prevention and control system based on air-space-ground fusion, characterized in that, include: The satellite remote sensing module is used to acquire raw remote sensing data of the target open-pit steeply dipping coal seam through a hyperspectral imager mounted on a satellite, and to determine the satellite remote sensing characteristics based on the raw remote sensing data. The drone monitoring module is used to collect raw drone data of the target open-pit steeply dipping coal seam through the event camera and infrared thermal imager carried by the drone, and to determine the event camera features and infrared features based on the raw drone data; The ground monitoring module is used to monitor CO concentration data at a preset number of depths in the target open-pit steeply dipping coal seam through drilling detection equipment, and to determine drilling characteristics based on the CO concentration data at the preset number of depths. The temperature signal of the target open-pit steeply dipping coal seam is measured by a distributed optical fiber temperature measurement array within a specified time period before the current time, and the distributed temperature characteristics are determined based on the temperature signal within the specified time period. The fusion processing module is used to perform multimodal feature fusion of satellite remote sensing features, event camera features, infrared features, drilling features and distributed temperature features based on an attention mechanism to obtain the fused feature vector of the target open-pit steeply dipping coal seam; The risk level calculation module is used to calculate the spontaneous combustion fire risk level of the target open-pit steeply inclined coal seam based on the fused feature vector of the target open-pit steeply inclined coal seam. The risk assessment module is used to generate spontaneous combustion assessment results based on the spontaneous combustion fire risk level and fused feature vector, and to carry out spontaneous combustion prevention and control of the target open-pit steeply inclined coal seam based on the spontaneous combustion assessment results.

2. The open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 1, is characterized in that, The satellite remote sensing module includes: The data scanning unit is used to acquire raw remote sensing data of the target open-pit steeply dipping coal seam during a specified time period using a hyperspectral imager mounted on a satellite. The spatial positioning unit is used to extract the coal characteristic spectral bands from the original remote sensing data, and to perform temperature inversion on the coal characteristic spectral bands to obtain inversion data. Based on the inversion data, all temperature anomaly pixels of the target open-pit steeply dipping coal seam are extracted from the coal seam distribution GIS layer of the mining area. All temperature anomaly pixels are combined to obtain the temperature anomaly zone. The center coordinates of the temperature anomaly zone are calculated, and the center coordinates of the temperature anomaly zone are used as the spatial feature of the temperature anomaly zone of the target open-pit steeply dipping coal seam. The fracture information acquisition unit extracts the fracture area of ​​the target open-pit steeply dipping coal seam from the coal seam distribution GIS layer of the mining area, and obtains the total number of fracture pixels in the fracture area according to the coal characteristic spectral band. It then calculates the percentage of the total number of fracture pixels to the total number of pixels in the coal area to obtain the fracture development intensity characteristics. The combined unit is used to combine the spatial characteristics of temperature anomaly zones and fracture development intensity characteristics of the target open-pit steeply dipping coal seam to obtain satellite remote sensing features.

3. The open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 2, is characterized in that, The spatial positioning unit calculates the center coordinates of the temperature anomaly zone. When, it is achieved through formula (1): （1）； In formula (1), This represents the x-coordinate of the i-th temperature anomaly pixel in the temperature anomaly zone. This represents the ordinate of the i-th temperature anomaly pixel in the temperature anomaly zone. This represents the temperature of the i-th temperature anomaly pixel in the temperature anomaly zone.

4. The open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 1, characterized in that, The raw UAV data includes raw event streams and raw infrared data, and the event camera and infrared thermal imager are spatiotemporally registered using a dual-modal calibration board. The UAV monitoring module includes: The event stream acquisition unit is used to acquire the original event stream of the target open-pit steeply dipping coal seam within a preset time period before the current moment using an event camera mounted on a drone. The event camera feature acquisition unit is used to extract multiple brightness abrupt change points in the original event stream, calculate the average grayscale change of all brightness abrupt change points within a unit time with the current time as the endpoint, and use the average grayscale change as the light signal intensity of the target open-pit steeply dipping coal seam at the current time; count the number of brightness abrupt change points in the original event stream within a preset time period with the current time as the endpoint, use the number of brightness abrupt change points as the light signal frequency of the target open-pit steeply dipping coal seam at the current time, and combine the light signal intensity and light signal frequency at the current time to obtain the event camera features of the target open-pit steeply dipping coal seam; The crack distribution extraction unit is used to extract crack distribution information based on multiple brightness abrupt changes in the original event stream. The infrared data acquisition unit is used to acquire raw infrared data of the target open-pit steeply dipping coal seam using an infrared thermal imager mounted on a drone. The infrared feature extraction unit divides the target open-pit steeply dipping coal seam into a predetermined number of layers along the direction perpendicular to the coal seam's dip, based on the original infrared data and fracture distribution information. Each layer corresponds to a fracture zone. The temperature gradient of each layer is calculated by taking the temperature difference between the top and bottom temperatures of the original infrared data. The temperature gradients of all layers are combined to obtain the vertical layering features of the target open-pit steeply dipping coal seam. The average temperature of all pixels in each layer is calculated based on the original infrared data. The average temperature of all layers is combined with the fracture distribution information to obtain the temperature features of the fracture zone. The infrared features of the target open-pit steeply dipping coal seam are obtained by combining the vertical layering features and the temperature features of the fracture zone.

5. The open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 4, is characterized in that, The crack distribution extraction unit includes: The crack zone generation sub-unit is used to cluster multiple brightness abrupt points in the original event stream according to the Hough transform, and obtain multiple crack zones and their original information. The fracture information extraction subunit is used to perform coordinate transformation on the original information of each fracture zone according to the Gauss-Kruger projection coordinate system to obtain the standard information of each fracture zone, and combine the standard information of all fracture zones to obtain the fracture distribution information.

6. The open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 1, characterized in that, The ground monitoring module includes: The raw data monitoring unit is used to configure multiple sampling points in the target open-pit steeply inclined coal seam according to the preset grid size, and to configure drilling and detection equipment at each sampling point. Raw gas is extracted through all drilling and detection equipment, and raw CO concentration data of raw gas at each sampling point is analyzed by gas chromatograph. The sampling point selection unit is used to select sampling points where the original CO concentration data is greater than the preset concentration anomaly value as anomaly points, to select multiple consecutive anomaly points as anomaly zones, and to select anomaly zones where the CO concentration gradient along the target open-pit steeply dipping coal seam is greater than the preset gradient threshold as CO anomaly zones. The CO concentration data of a preset number of depth segments on the CO anomaly zone are collected by drilling detection equipment, and the CO concentration data of the preset number of depth segments are used as drilling features. The signal acquisition unit is used to acquire the temperature signal of the target open-pit steeply dipping coal seam within a specified time period before the current time through a distributed optical fiber temperature measurement array, calculate the temperature rise rate at the current time based on the temperature signal within the specified time period, and use the temperature rise rate at the current time as the distributed temperature characteristic of the target open-pit steeply dipping coal seam.

7. The open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 1, characterized in that, The fusion processing module includes: The standardized unit is used to standardize and linearly map satellite remote sensing features, event camera features, infrared features, drilling features, and distributed temperature features to obtain a standard feature vector. The feature fusion unit is used to perform attention-based multimodal feature fusion on the standard feature vector to obtain the fused feature vector of the target open-pit steeply dipping coal seam. Each feature in the fused feature vector contains attention weights.

8. The open-pit, steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 7, is characterized in that, The risk level calculation module includes: The weight adjustment unit is used to acquire historical fire data and current real-time monitoring data of the target open-pit steeply dipping coal seam, and combine the historical fire data and current real-time monitoring data to obtain actual fire data. It calculates the standard feature value of each feature of the actual fire data, and adjusts the attention weight of each feature in the fused feature vector according to the standard feature value to obtain the standard weight of each feature. The risk value calculation unit is used to calculate the spontaneous combustion fire risk level of the target open-pit steeply inclined coal seam based on the standard weight of each feature.

9. The open-pit, steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 8, is characterized in that, When the weight adjustment unit adjusts the attention weight of each feature in the fused feature vector according to the standard feature value to obtain the standard weight of each feature, it is used for: Calculate the similarity between each feature in the fused feature vector and its corresponding standard feature value, and adjust the attention weight of each feature based on the similarity to obtain the standard weight of each feature.

10. The open-pit steeply inclined coal seam spontaneous combustion prevention and control system based on multi-source fusion of air, space, and ground as described in claim 1, characterized in that, The risk assessment module includes: The instruction acquisition unit is used to acquire prevention and control instructions corresponding to the risk level of spontaneous combustion fire according to a preset risk level table. The information generation unit is used to generate spontaneous combustion risk information for each location of the target open-pit steeply dipping coal seam based on the fused feature vector, and to combine the spontaneous combustion risk information of all locations to obtain the spontaneous combustion assessment information of the target open-pit steeply dipping coal seam. The spontaneous combustion assessment result generation unit is used to combine spontaneous combustion assessment information and prevention and control instructions to generate spontaneous combustion assessment results, and to carry out spontaneous combustion prevention and control of the target open-pit steeply inclined coal seam based on the spontaneous combustion assessment results.

Images