Mine fire early warning management system based on multi-source data fusion analysis

Abstract

The present application relates to the technical field of mine monitoring, and particularly relates to a mine fire early warning management system based on multi-source data fusion analysis. The present application retrieves the wind speed and direction, pressure difference and gas concentration data of the alarm monitoring point and its upstream and downstream, reverses the air leakage path in combination with the three-dimensional ventilation network model to infer the fire source area, generates the fire spread prediction trajectory and determines the early warning level according to the time sequence change of the gas components on the downwind side of the fire source area, and then matches the fire extinguishing medium and control mode according to the early warning level and the fire prevention level of the fire source area, so as to realize the closed-loop management from fire source positioning, spread prediction to differentiated fire extinguishing disposal.

Description

Technical Field

[0001] This invention relates to the field of mine monitoring technology, specifically to a mine fire early warning management system based on multi-source data fusion analysis. Background Technology

[0002] With the expansion of coal mine production scale and the improvement of electrification and intelligence, the equipment in key electrical locations such as underground substations and main distribution rooms is becoming increasingly dense and powerful, leading to a continuous increase in the risk of electrical fires. Moreover, hidden fire sources in goaf areas are difficult to detect due to complex ventilation channels. Once a fire occurs, it can spread rapidly, be difficult to extinguish, and can easily cause significant casualties and property damage.

[0003] Existing technologies, such as Chinese invention patent publication number CN120426089A, disclose a dynamic monitoring method and system for fire prevention and extinguishing in enclosed goaf areas. This method collects downhole gas concentration and temperature data and sets fixed thresholds for alarm judgment; an alarm is triggered when the monitored data exceeds the threshold. However, it lacks fusion analysis of multi-source data such as thermal imaging, pressure difference, wind speed, and wind direction, making it difficult to rule out false alarms caused by sensor malfunctions or environmental interference, and it cannot spatially locate the fire.

[0004] For example, Chinese invention patent CN119294832B discloses a method, equipment, and medium for coal mine risk analysis based on IoT monitoring data. It continuously collects thermal images of the goaf using a thermal imager, divides the thermal images into blocks and processes the temperature data, and predicts goaf temperature changes based on a multilayer perceptron model to predict and analyze spontaneous combustion risks. However, it does not address the construction of the underground ventilation network topology, nor does it combine airflow direction and upstream / downstream monitoring data to invert and locate the fire source, making it difficult to identify and handle fire spread paths under complex ventilation conditions. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and solve the problems of high false alarm rate of fire, difficulty in locating fire source area, lack of fire spread prediction and inability to differentiate and match fire extinguishing methods due to the lack of multi-source data fusion analysis.

[0006] The technical solution adopted by the present invention to solve its technical problem is: a mine fire early warning management system based on multi-source data fusion analysis, including: a model building module, used to build a three-dimensional ventilation network model and mark the location of monitoring points based on the parameters of mine roadway engineering, ventilation facility parameters and air leakage channel detection data of goaf.

[0007] The rule building module is used to time-align the gas composition, point temperature data, thermal imaging video, goaf pressure difference and wind speed and direction data of the monitoring points, and establish association rules between each parameter and fire characteristic parameters in combination with the location of the monitoring points.

[0008] The verification and positioning module is used to compare thermal imaging videos with point temperature data according to association rules to confirm the fire. After confirming the fire, it retrieves the wind speed, wind direction, pressure difference and gas concentration data of the alarm monitoring point and its upstream and downstream, and combines them with the three-dimensional ventilation network model to invert the air leakage path to infer the fire source area.

[0009] The analysis and early warning module is used to generate a fire spread prediction trajectory based on the temporal changes of gas composition on the downwind side of the fire source area, and to determine the early warning level by combining it with point temperature data on the downwind side.

[0010] The linkage control module is used to match the extinguishing medium and control method according to the warning level and the fire protection level of the fire source area.

[0011] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention constructs a three-dimensional ventilation network model based on the parameters of the tunnel engineering, the parameters of the ventilation facilities and the detection data of the air leakage channels in the goaf, thereby realizing the digital reconstruction of the underground ventilation network and the air leakage channels, providing a topological basis for the subsequent inversion of the air leakage path to infer the fire source area.

[0012] 2. Based on gas composition, point temperature, thermal imaging video, pressure difference, and wind speed and direction data, this invention establishes correlation rules between various parameters and fire characteristic parameters, thereby realizing multi-source data fusion analysis and feature quantification. By cross-checking thermal imaging video and point temperature data, the fire situation is confirmed, reducing the false alarm rate and providing a quantifiable basis for fire confirmation.

[0013] 3. This invention retrieves wind speed, direction, pressure difference, and gas concentration data from alarm monitoring points and their upstream and downstream areas, combines this data with a three-dimensional ventilation network model to infer the air leakage path to deduce the fire source area, generates a fire spread prediction trajectory based on the temporal changes in gas composition on the downwind side of the fire source area, and determines the warning level. Then, based on the warning level and the fire protection level of the fire source area, it matches the extinguishing medium and control method, thus realizing closed-loop management from fire source location and spread prediction to differentiated fire extinguishing and disposal. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic diagram of the system module connections of the present invention.

[0016] Figure 2 This is a schematic diagram illustrating the process of constructing a three-dimensional ventilation network model for this invention.

[0017] Figure 3 This is a flowchart illustrating the process of establishing association rules for this invention.

[0018] Figure 4 This is a schematic diagram of the process for inverting the air leakage path according to the present invention. Detailed Implementation

[0019] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention. Furthermore, it should be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale.

[0020] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification.

[0021] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] The specific solution of the mine fire early warning management system based on multi-source data fusion analysis provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Please see Figure 1 The diagram illustrates the module connection of the mine fire early warning management system based on multi-source data fusion analysis provided by the present invention, which specifically includes: a model building module, a rule building module, a verification and positioning module, an analysis and early warning module, and a linkage control module.

[0025] The output of the model building module is connected to the rule building module, the output of the rule building module is connected to the verification and positioning module, the output of the verification and positioning module is connected to the analysis and early warning module, and the output of the analysis and early warning module is connected to the linkage control module.

[0026] Please see Figure 2 The model building module is used to construct a three-dimensional ventilation network model based on tunnel engineering parameters, ventilation facility parameters, and air leakage channel detection data in the goaf.

[0027] The tunnel engineering parameters include the tunnel's cross-sectional dimensions, length, frictional resistance, and start and end node numbers. The cross-sectional dimensions include the width and height; the length is the actual distance from the start to the end point of the tunnel; and the frictional resistance characterizes the tunnel wall's obstruction of airflow (dimensionless). These tunnel engineering parameters can be obtained from mine survey drawings, geological exploration reports, and on-site measurements.

[0028] The ventilation facility parameters include the facility's location coordinates, frictional resistance, local resistance, and dynamic air pressure. The facility's location coordinates are obtained from mine survey drawings or marked using the X, Y, and Z axes in the mine's spatial coordinate system. Frictional resistance and local resistance quantify the facility's obstruction effect on airflow, while dynamic air pressure (such as fan pressure) characterizes the facility's ability to drive airflow. All three parameters can be obtained from the facility's factory specifications.

[0029] The data for detecting air leakage channels in the goaf includes the coordinates of the starting and ending points of the leakage channel, its equivalent diameter, leakage air volume, and leakage air velocity. An air leakage channel refers to the airflow path formed between the goaf and adjacent roadways due to fissures or voids.

[0030] The starting and ending position coordinates are obtained through tracer gas method or borehole detection. The leakage air volume and leakage air velocity quantify the leakage intensity, and both can be obtained through an anemometer installed in the leakage channel or indirectly calculated through pressure difference.

[0031] The equivalent diameter is obtained by drilling or 3D laser scanning to obtain the cross-sectional profile of the air leakage channel, calculating the actual area of ​​the cross-section, dividing it by pi, taking the square root, and then multiplying by 2.

[0032] The construction process of the three-dimensional ventilation network model is as follows: the start and end node numbers are used as nodes, each node is assigned a unique coordinate, and each set of corresponding start and end node numbers is used as a branch connecting the corresponding nodes to form a ventilation network topology. The ventilation network topology is used to describe the underground airflow path and branch connection relationship.

[0033] Then, based on the cross-sectional dimensions, shaft length, and shaft friction resistance associated with the node number, the shaft friction resistance is multiplied by the shaft length and then divided by the cube of the product of the cross-sectional width and height to obtain the shaft ventilation resistance value of each branch. The shaft ventilation resistance value is then stored as an additional attribute of the corresponding branch.

[0034] Based on the facility's frictional resistance, local resistance, and dynamic wind pressure associated with its location coordinates, all nodes are traversed, and nodes with coordinates matching the facility's location coordinates are selected. If no node with matching coordinates exists, the facility is skipped. If multiple nodes with matching coordinates exist, the facility's frictional resistance, local resistance, and dynamic wind pressure are stored as additional attributes of these nodes. If a node is associated with multiple facilities, the node's frictional resistance and local resistance are the sum of the values ​​of all facilities, and the dynamic wind pressure is the vector sum of the values ​​of all facilities.

[0035] Simultaneously, based on the equivalent diameter, leakage air volume, and leakage air velocity associated with the coordinates of the start and end points of the leakage channel, using the start and end point coordinates as indexes, nodes with the same coordinates as the start point are identified as goaf nodes, and nodes with the same coordinates as the end point are identified as roadway nodes. If no nodes with the same coordinates exist, new nodes are created at the start and end point coordinates, respectively serving as goaf nodes and roadway nodes. If multiple nodes with the same coordinates exist, the node with the smallest Euclidean distance from the start point is identified as the goaf node, and the node with the smallest Euclidean distance from the end point is identified as the roadway node.

[0036] Since the starting and ending points of the air leakage channels in the goaf are usually located at different nodes, and it is rare for multiple air leakage channels to have multiple starting and ending points at the same location, the case of multiple air leakage channels associated with the same node is not considered.

[0037] If a direct connection already exists between the nodes of the goaf and the roadway, this branch is taken as the target branch; otherwise, the node connecting the goaf and the roadway forms the target branch. The starting node of the target branch is the goaf node, and the ending node is the roadway node. Furthermore, the equivalent diameter, leakage air volume, and leakage air velocity are stored as additional attributes of the target branch.

[0038] Subsequently, the ventilation resistance values ​​of each branch, the facility friction resistance, facility local resistance, facility dynamic wind pressure, and the equivalent diameter, leakage air volume, and leakage wind velocity of the target branch are integrated into the ventilation network topology to form a three-dimensional ventilation network model that also includes node coordinates and branch connection relationships.

[0039] After obtaining the three-dimensional ventilation network model, the locations of monitoring points are marked in the model. Monitoring points refer to the physical locations where gas sensors, point temperature sensors, thermal imaging cameras, differential pressure sensors, and anemometers are deployed. The arrangement principle is as follows: monitoring points are set every 50 to 100 meters in locations such as the intake and return airways of the mining face, electromechanical chambers, belt conveyor heads and tails, and goaf connecting roadways. At the same time, at least one monitoring point is set at the upstream and downstream of each branch. The number of monitoring points depends on the scale of the mine.

[0040] Please see Figure 3The rule building module is used to establish association rules between various parameters and fire characteristic parameters based on the gas composition, point temperature data, thermal imaging video, goaf pressure difference, wind speed and direction data, and location of the monitoring points.

[0041] Specifically, the gas components refer to the concentration set of fire-indicating gases such as carbon monoxide, carbon dioxide, methane, ethylene, and acetylene.

[0042] Specifically, firstly, for each monitoring point, the thermal imaging video is decomposed into a continuous sequence of frame images using a decoder; then, the pixel grayscale values ​​are converted into radiation temperature values ​​according to the calibration parameters of the thermal imaging camera. Based on the pixel coordinates of the monitoring point in the thermal imaging image, the radiation temperature value of that coordinate is read from each frame, and the radiation temperature values ​​of all frames are sorted according to the acquisition time to obtain the radiation temperature data of that monitoring point.

[0043] Next, using the location of the monitoring point as an index, the gas composition, point temperature data, radiation temperature data, goaf pressure difference and wind speed and direction data of the same monitoring point are aligned according to the collection time to generate a time series dataset.

[0044] Then, based on the time series dataset, for point temperature data, the point temperature difference between the current time and the previous adjacent time is calculated, and then divided by the time interval to obtain the point temperature change rate; similarly, the same processing is performed on the radiation temperature data and pressure difference data to obtain the radiation temperature change rate and pressure difference change rate.

[0045] Meanwhile, for wind speed and direction data, the wind speed is decomposed into east-west and north-south components. Based on the ratio of the east-west component to the north-south component, the wind direction angle relative to due north is calculated using the arctangent function, with a value range of 0 to 360 degrees.

[0046] Subsequently, the range of fire source temperature, range of smoke spread rate, concentration values ​​of characteristic fire gases, and fire type were extracted from historical fire cases as fire characteristic parameters. Historical fire cases were selected by collecting mine fire accident reports, experimental data, and literature, and cases containing monitoring data such as gas composition, point temperature, thermal imaging video, pressure difference, wind speed and direction, and the fire development process were selected.

[0047] The ignition source temperature range refers to the interval between the lowest and highest flame area temperatures in historical fire cases. The smoke spread velocity range refers to the interval between the lowest and highest smoke movement velocities calculated based on the spatial distance between the monitoring point and the moment when the gas concentration at the monitoring point first exceeded or equaled the concentration of the fire characteristic gas in historical fire cases.

[0048] The concentration values ​​of characteristic gases in a fire refer to the critical concentration values ​​at which fire-marking gases such as carbon monoxide and ethylene reach the alarm state in the early stages of a fire. Fire types include spontaneous combustion of coal, electrical fires, and conveyor belt fires, among others.

[0049] At the same time, gas composition, point temperature change rate, radiation temperature change rate, pressure difference change rate, and wind direction angle were extracted from historical fire cases as monitoring parameters.

[0050] Finally, using monitoring parameters as input and fire feature parameters as output, an association rule mining algorithm (such as the Apriori algorithm) is employed to train the monitoring parameters and corresponding fire feature parameters from historical fire cases, generating association rules between each parameter and fire feature parameter at each monitoring point, thereby providing a quantitative basis for subsequent fire confirmation.

[0051] Please see Figure 4 The verification and positioning module is used to compare thermal imaging video and point temperature data according to association rules to confirm the fire, and to infer the fire source area after confirming the fire. By cross-checking thermal imaging video and point temperature data, misjudgments of the fire caused by single-point sensor failure or environmental thermal radiation interference can be ruled out.

[0052] The process of confirming a fire involves inputting the monitoring parameters of the current monitoring point into the association rule and outputting the corresponding fire characteristic parameters.

[0053] If, at the current moment and at least two adjacent moments prior, both the radiant temperature and the point temperature at the current monitoring point reach the lower limit of the ignition source temperature range, it indicates that the location of the current monitoring point has been in a potentially ignitable high-temperature state for a certain period of time, and a fire is initially suspected. If, at the current moment and at least two adjacent moments prior, both the radiant temperature and the point temperature at the current monitoring point do not reach the lower limit of the ignition source temperature range, monitoring continues.

[0054] It should be noted that in this invention, the time conditions for the above judgment can be adjusted based on the wind speed and direction data at the current monitoring point. When the wind speed is greater than 2 m / s, the airflow will accelerate heat dissipation. To avoid missed alarms, the time condition is changed to the current time and at least one previous adjacent time. When the wind speed is less than 0.5 m / s, heat accumulates more slowly. To avoid false alarms, the time condition is changed to the current time and at least three previous adjacent time. At other wind speeds, the current time condition is maintained.

[0055] Upon initial suspicion of a fire, all pixels with temperatures greater than or equal to the lower limit of the fire source temperature range are immediately extracted from the thermal imaging video of the current monitoring point and marked as anomalous pixels. Connectivity analysis is performed on all anomalous pixels, merging adjacent anomalous pixels within four or eight neighboring regions into connected regions. Each connected region represents a temperature anomaly area.

[0056] Then, for temperature anomaly regions spanning at least N consecutive time periods, the number of pixels contained within the temperature anomaly region at each time period is counted. The number of pixels is then multiplied by the actual spatial area corresponding to each pixel to obtain the area of ​​the temperature anomaly region at that time period. Here, N is the number of adjacent time periods plus 1 in the above time condition.

[0057] The difference between the area of ​​the temperature anomaly region at the next moment and the area at the previous moment is divided by the time interval to obtain the rate of change of area at adjacent moments; if the area of ​​the temperature anomaly region at the previous moment is zero, then the rate of change of area at adjacent moments is recorded as zero.

[0058] The arithmetic mean of the area change rates at all adjacent time points is taken as the average area change rate over at least N consecutive time points. The average area change rate is then divided by the cross-sectional width of the tunnel at the current monitoring point to obtain the linear velocity of the flame front spread.

[0059] If the flame front spread velocity falls within the range of smoke spread velocity, it indicates that the expansion rate of the abnormal temperature area matches the fire smoke spread velocity, and a fire is confirmed at the current monitoring point. Otherwise, continue monitoring and issue an abnormal temperature rise alert, but do not confirm it as a fire at this time.

[0060] After confirming the fire, considering that the smoke and heat generated by the fire usually spread with the airflow, the method of inverting the air leakage path to infer the fire source area was adopted to locate the fire source.

[0061] Specifically, the monitoring points where a fire has been confirmed are first designated as alarm monitoring points, and the node numbers of the branches where the alarm monitoring points are located are extracted from the three-dimensional ventilation network model.

[0062] Based on the relationship between the start and end node numbers of each branch, find the branch whose end node is the start node of the branch where the alarm monitoring point is located, and take it as the upstream adjacent branch; take the branch whose start node is the end node of the branch where the alarm monitoring point is located, and take it as the downstream adjacent branch; and take the monitoring points on the upstream and downstream adjacent branches as the upstream and downstream monitoring points.

[0063] Then, the wind direction angle is obtained based on the wind speed and direction data from the upstream and downstream monitoring points. If the wind direction angle points towards the end node of the branch where the upstream and downstream monitoring points are located, then the measured airflow direction at the upstream and downstream monitoring points is from the starting node to the ending node; conversely, the measured airflow direction at the upstream and downstream monitoring points is from the ending node to the starting node. Furthermore, the measured pressure difference between the alarm monitoring point and the upstream and downstream monitoring points is calculated based on the pressure difference data, and the gas composition at the upstream and downstream monitoring points is determined based on the gas concentration data.

[0064] Subsequently, the direction of airflow from the starting node to the ending node in each branch of the three-dimensional ventilation network model is taken as the theoretical airflow direction. Monitoring points from upstream and downstream that meet any of the following conditions are selected as anomalies: the measured airflow direction is opposite to the corresponding theoretical airflow direction; or the gas composition contains gases exceeding the concentration values ​​characteristic of fire. Reverse airflow indicates potential air leakage causing turbulent airflow, while exceeding gas concentration limits indicates that the monitoring point's location has been affected by fire smoke.

[0065] Furthermore, starting from the alarm monitoring point, the upstream branch connected to it is searched along the opposite direction of the measured airflow. If there is an abnormal point on the upstream branch, all branches passed from the alarm monitoring point to the abnormal point and the monitoring points on them are recorded. If there is no abnormal point on the upstream branch, the tracing continues along the opposite direction of the upstream branch until the abnormal point is found, or the tracing stops when there are no other branches connected at the starting node of the upstream branch.

[0066] Subsequently, based on the branch connection relationship in the three-dimensional ventilation network model, all branches passed through during the tracing process are connected in series according to the connection order to form multiple paths from the alarm monitoring point to each abnormal point. The union of all paths is taken as the leakage path.

[0067] After determining the air leakage path, retrieve the equivalent diameter, air leakage volume, air leakage velocity, and corresponding node coordinates of each branch along the air leakage path from the three-dimensional ventilation network model.

[0068] The equivalent radius is obtained from the equivalent diameter, and the cross-sectional area of ​​the leakage channel is obtained by multiplying the square of the equivalent radius by pi. The leakage air volume is divided by the cross-sectional area of ​​the leakage channel to obtain the measured value of the leakage air velocity of each branch.

[0069] Next, branches whose measured leakage velocity values ​​are not equal to the corresponding retrieved leakage velocity are filtered. If only one branch is selected, its location is designated as a candidate area; if multiple branches are selected, a second filtering process is performed. "Not equal" means that the absolute value of the difference between the measured leakage velocity and the corresponding retrieved leakage velocity is greater than 2.5 times the anemometer's accuracy value; the anemometer's accuracy value can be obtained from the manufacturer's product manual.

[0070] The secondary screening process is as follows: from the thermal imaging video of the alarm monitoring point at the time of fire confirmation, extract all pixels whose temperature reaches the lower limit of the fire source temperature range to form the temperature anomaly area at the time of fire confirmation. From multiple branches, select branches whose node coordinates are located in the temperature anomaly area and use the location of the secondary-screened branches as candidate areas.

[0071] After determining the candidate area, the Euclidean distance between adjacent branch nodes in the candidate area is calculated based on the node coordinates. If the Euclidean distance is less than the cross-sectional width of the shaft represented by the corresponding branch, and the adjacent branch nodes do not belong to the same branch, it indicates that the adjacent branch nodes jointly cover the same fire source area, and the adjacent branch nodes are merged. If the Euclidean distance is greater than or equal to the cross-sectional width of the shaft represented by the corresponding branch, the nodes are not merged.

[0072] Finally, using the coordinates of each node in the merged node set as the center and the cross-sectional width of the corresponding branch as the radius, the circular region corresponding to each center is determined, and the union of all circular regions is taken as the fire source area.

[0073] The analysis and early warning module is used to generate a fire spread prediction trajectory based on the temporal changes of gas composition on the downwind side of the fire source area, and to determine the early warning level by combining the point temperature data on the downwind side.

[0074] Specifically, the node coordinates of the fire source area in the three-dimensional ventilation network model are first identified. Then, starting from the fire source area node, a breadth-first traversal is performed along the theoretical airflow direction of each branch. All traversed branches are taken as downstream connected branches of the fire source area, and the monitoring points on the downstream connected branches of the fire source area are taken as downwind monitoring points.

[0075] Next, the gas composition time series data of the downwind monitoring points after the fire confirmation time are retrieved, and the moment when the concentration of the fire characteristic gas first exceeds or equals the corresponding fire characteristic gas concentration value (retrieved from the association rules) is extracted. The difference between this moment and the fire confirmation time is calculated as the propagation time of the fire characteristic gas from the fire source area to each downwind monitoring point.

[0076] Subsequently, in the three-dimensional ventilation network model, using the branch length as the edge weight, the Dijkstra algorithm can be employed to calculate the shortest path from the fire source node to the node where the monitoring point is located on the leeward side. The spatial distance is obtained by summing the lengths of all branches on the shortest path. The specific shortest path calculation process is existing technology and will not be elaborated here.

[0077] Divide the spatial distance by the propagation time to obtain the smoke spread velocity in the direction of each downwind monitoring point, and store the smoke spread velocity as an additional attribute of the corresponding branch.

[0078] Next, starting from the fire source node, the node coordinates of the monitoring points on the downwind side are connected sequentially along the downstream branch direction to generate a fire spread prediction trajectory. This fire spread prediction trajectory allows for the prediction of the diffusion path and arrival time of fire smoke on the downwind side, providing decision support for personnel evacuation and firefighting rescue.

[0079] After obtaining the predicted fire spread trajectory, the downwind monitoring points on the predicted fire spread trajectory are extracted as target monitoring points, and the point temperature of the target monitoring points at the current moment is retrieved.

[0080] Then, target monitoring points whose point temperatures fall within the range of the fire source temperature are designated as risk monitoring points. The risk monitoring points on the predicted fire spread trajectory are grouped according to their respective branches, and all risk monitoring points on each branch constitute an initial segment.

[0081] Furthermore, all branches directly connected to the branches of each initial segment are extracted. If the extracted branch contains other initial segments, these other initial segments are merged with the current initial segment to obtain a merged combination; if the extracted branch does not contain other initial segments, the current initial segment remains unchanged. Identical risk monitoring points are removed from all merged combinations. The union of all merged combinations after deduplication represents the distribution range of risk monitoring points on the fire spread prediction trajectory. The boundary of the distribution range is the smallest circumscribed region of all risk monitoring points in the merged combination.

[0082] Then, the warning level is determined based on the number and distribution of risk monitoring points. The specific determination rules are as follows: if the number of risk monitoring points is less than or equal to 3 and the distribution range is limited to a single branch, it is designated as a Level 3 warning; if the number of risk monitoring points is greater than 3 but less than or equal to 8, or the distribution range involves two branches, it is designated as a Level 2 warning; if the number of risk monitoring points is greater than 8, or the distribution range involves three or more branches, it is designated as a Level 1 warning.

[0083] Different warning levels can be distinguished by different colors and frequencies of audible and visual alarms.

[0084] The linkage control module is used to match the extinguishing medium and control method according to the warning level and the fire protection level of the fire source area. This ensures the fire extinguishing effect while reducing the impact on production and the risk of secondary disasters.

[0085] The fire prevention level of a fire source area refers to the fire prevention importance level of different areas underground, as classified according to the mine safety regulations. This level can be obtained by consulting the mine safety assessment report or by on-site markings. Generally, it is divided into three levels, with the fire prevention importance gradually decreasing: Level 1, Level 2, and Level 3.

[0086] The specific matching rules are as follows: when the fire source area has a fire protection level of Level 1, regardless of the warning level, water or water-based fire extinguishing medium will be used, and the sprinkler system will be started using a remote automatic control method.

[0087] When the fire source area has a fire protection level of Level II, if the warning level is Level I, then inert gas extinguishing medium is used and remote automatic control is adopted; if the warning level is Level II or Level III, then dry powder extinguishing medium is used and a control method that starts after remote manual confirmation is adopted.

[0088] When the fire source area is classified as Level 3, if the warning level is Level 1, foam extinguishing medium will be used and remote automatic control will be employed; if the warning level is Level 2, water medium will be used and a control method that requires remote manual confirmation before activation will be adopted; if the warning level is Level 3, only an alarm will be issued, and on-site personnel will handle the situation as appropriate.

[0089] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0090] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0091] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0092] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0093] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A mine fire early warning management system based on multi-source data fusion analysis, characterized in that, include: The model building module is used to build upon ventilation facility parameters, tunnel engineering parameters including start and end node numbers, and goaf leakage channel detection data including start and end coordinates, equivalent diameter, leakage air volume, and leakage air velocity. Using the start and end node numbers as nodes, each node is assigned a unique coordinate. Each set of start and end node numbers is used as a branch connecting the corresponding nodes to form a ventilation network topology. The equivalent diameter, leakage air volume, and leakage air velocity are assigned to the target branch connecting the goaf nodes and roadway nodes to generate a three-dimensional ventilation network model and mark the location of monitoring points. The rule building module is used to time-align the gas composition, point temperature data, thermal imaging video, goaf pressure difference and wind speed and direction data of the monitoring points, and establish association rules between each parameter and fire characteristic parameters in combination with the location of the monitoring points. The verification and positioning module is used to compare thermal imaging videos with point temperature data according to association rules to confirm the fire. After confirming the fire, it retrieves the wind speed, wind direction, pressure difference, and gas concentration data of the alarm monitoring point and its upstream and downstream. Based on the numbering relationship of the start and end nodes of each branch in the three-dimensional ventilation network model, it determines the upstream and downstream adjacent branches of the branch where the alarm monitoring point is located, and uses the monitoring points on the upstream and downstream adjacent branches as upstream and downstream monitoring points. From the upstream and downstream monitoring points, it selects the monitoring points where the measured airflow direction is opposite to the corresponding theoretical airflow direction, as well as the monitoring points where the gas composition contains gas exceeding the concentration value of fire characteristic gas, as abnormal points. Starting from the alarm monitoring point and using the abnormal point as the middle point, it traces the connection path along the opposite direction of the measured airflow direction according to the branch connection relationship in the three-dimensional ventilation network model to generate the air leakage path to infer the fire source area. The analysis and early warning module is used to determine all connected branches downstream of the fire source area and their monitoring points based on the node coordinates of the fire source area in the three-dimensional ventilation network model and the theoretical airflow direction, which are then used as downwind monitoring points. Based on the gas composition time-series data of the downwind monitoring points after the fire is confirmed, and the moment when the concentration of the fire characteristic gas first reaches the corresponding fire characteristic gas concentration value, the propagation duration is calculated. The cumulative length of the fire source area node and the downwind monitoring point node along the branches in the three-dimensional ventilation network model is calculated as the spatial distance, and the smoke spread speed is calculated in combination with the propagation duration. Starting from the fire source area node, the nodes of the downwind monitoring points are connected sequentially along the downstream branches to generate a fire spread prediction trajectory, which, combined with the downwind point temperature data, determines the early warning level. The linkage control module is used to match the extinguishing medium and control method according to the warning level and the fire protection level of the fire source area.

2. The mine fire early warning management system based on multi-source data fusion analysis according to claim 1, characterized in that, The parameters for shaft and tunnel engineering also include the cross-sectional dimensions, length, and frictional resistance of the shaft and tunnel; the parameters for ventilation facilities include the location coordinates of the facilities, frictional resistance, local resistance, and dynamic air pressure.

3. The mine fire early warning management system based on multi-source data fusion analysis according to claim 2, characterized in that, The process of generating a three-dimensional ventilation network model also includes: Based on the cross-sectional dimensions, tunnel length, and tunnel friction resistance associated with the node number, calculate and assign values ​​to the tunnel ventilation resistance values ​​of each branch; Based on the facility's frictional resistance, local resistance, and dynamic wind pressure associated with the facility's location coordinates, find nodes with the same coordinates and assign values ​​to them; Based on the equivalent diameter, leakage air volume, and leakage air velocity associated with the start and end position coordinates, nodes with the same coordinates are identified as nodes of the goaf and roadway, respectively. If there is already a direct connection between the nodes of the goaf and the roadway, then this branch is taken as the target branch; otherwise, the node connecting the goaf and the roadway forms the target branch.

4. The mine fire early warning management system based on multi-source data fusion analysis according to claim 1, characterized in that, The process of establishing association rules is as follows: Decode the thermal imaging video and extract the radiation temperature data of the corresponding pixels of the monitoring points in chronological order; Using the location of the monitoring point as an index, the gas composition, point temperature data collected by the sensor, radiation temperature data, pressure difference and wind speed and direction data of the goaf area at the same monitoring point are aligned according to the collection time to generate a time series dataset; Based on the time series dataset, differential calculations were performed on point temperature data, goaf pressure difference and radiation temperature data to obtain the point temperature change rate, radiation temperature change rate and pressure difference change rate. Vector decomposition was performed on wind speed and wind direction data to obtain wind direction angle. The range of fire source temperature, range of smoke spread rate, concentration of characteristic fire gases and fire type were extracted from historical fire cases as fire characteristic parameters; the gas composition, point temperature change rate, radiation temperature change rate, pressure difference change rate and wind direction angle of the monitoring points were extracted as monitoring parameters. Using monitoring parameters as input and fire characteristic parameters as output, establish association rules between each parameter at each monitoring point and the fire characteristic parameters.

5. The mine fire early warning management system based on multi-source data fusion analysis according to claim 4, characterized in that, The process of confirming the fire is as follows: Input the monitoring parameters of the current monitoring point into the association rule, and output the corresponding fire characteristic parameters; If, at the current moment and at least two adjacent moments prior, the radiation temperature and point temperature of the current monitoring point both reach the lower limit of the fire source temperature range, then all pixels whose temperature reaches the lower limit of the fire source temperature range are extracted from its thermal imaging video to form a temperature anomaly area. Calculate the rate of change of the area of ​​the temperature anomaly region at multiple consecutive time points; divide the rate of change of the area by the cross-sectional width of the tunnel at the current monitoring point to obtain the linear velocity of the flame front spread; If the linear velocity of the flame front falls within the range of the smoke spread velocity, then a fire is confirmed to have occurred at the current monitoring point.

6. The mine fire early warning management system based on multi-source data fusion analysis according to claim 1, characterized in that, The process of generating the air leakage path also includes: The monitoring point where the fire is confirmed is used as the alarm monitoring point, and the node number of the branch where it is located is extracted from the three-dimensional ventilation network model. Based on wind speed and direction data, determine the measured airflow direction of upstream and downstream monitoring points; based on pressure difference data, calculate the measured pressure difference between the alarm monitoring point and the upstream and downstream monitoring points; based on gas concentration data, determine the gas composition of upstream and downstream monitoring points. The direction of airflow from the starting node to the ending node in each branch of the three-dimensional ventilation network model is taken as the theoretical airflow direction.

7. The mine fire early warning management system based on multi-source data fusion analysis according to claim 3, characterized in that, The process of inferring the fire source area is as follows: From the three-dimensional ventilation network model, retrieve the equivalent diameter, leakage air volume, leakage air velocity, and corresponding node coordinates of each branch along the leakage path; The cross-sectional area of ​​the leakage channel of each branch is calculated based on the equivalent diameter. The leakage air volume is divided by the cross-sectional area of ​​the leakage channel to obtain the measured value of the leakage air velocity of each branch. Filter out branches where the measured leakage velocity does not match the corresponding retrieved leakage velocity, and use their locations as candidate areas; Based on the node coordinates, adjacent branch nodes in the candidate area whose distance between nodes is less than the corresponding branch cross-section size are merged, and the area covered by the merged nodes is taken as the fire source area.

8. The mine fire early warning management system based on multi-source data fusion analysis according to claim 1, characterized in that, The process of generating a fire spread prediction trajectory also includes: The difference between the moment when the concentration of the fire characteristic gas first reaches the corresponding fire characteristic gas concentration value and the moment when the fire is confirmed is calculated as the propagation time. Divide the spatial distance by the propagation time to obtain the smoke spread velocity in the direction of each downwind monitoring point and assign it to the corresponding branch.

9. The mine fire early warning management system based on multi-source data fusion analysis according to claim 1, characterized in that, The process for determining the warning level is as follows: Extract the downwind monitoring points on the predicted fire spread trajectory as target monitoring points, and retrieve their current point-based temperature. Target monitoring points whose point temperatures fall within the range of the fire source temperature are designated as risk monitoring points; The warning level is determined based on the number and distribution of risk monitoring points; the warning levels include Level 1, Level 2, and Level 3.

10. The mine fire early warning management system based on multi-source data fusion analysis according to claim 1, characterized in that, The process of determining the distribution range is as follows: The risk monitoring points on the predicted fire spread trajectory are grouped according to their respective branches, and all the risk monitoring points on each branch constitute an initial segment; Extract all branches directly connected to the branch containing each initial segment. If the extracted branch contains other initial segments, merge the other initial segments with the current initial segment to obtain a merged combination. After merging and deduplicating all combinations, the distribution range of risk monitoring points on the predicted fire spread trajectory is obtained.

Images