Coal mine underground fire preventing and extinguishing system based on distributed internet of things

Abstract

The invention discloses a coal mine underground fire preventing and extinguishing system based on a distributed internet of things, and relates to the technical field of coal mine safety monitoring and intelligent control. Comprising a sensing sub-node module, a communication cooperation module, an edge decision control module, a linkage response module and a cloud management platform. Environmental data such as temperature, gas concentration and smoke are collected through a multi-mode sensor, and an edge AI chip is used for real-time analysis and generation of an early warning score. And by adopting a multi-scale scoring fusion and neighborhood reference mechanism, the accuracy and robustness of early fire recognition are improved. And a multi-node collaborative confirmation mechanism is introduced, so that distributed intelligent judgment is realized, and false alarm and missing alarm are effectively avoided. The system has a communication self-healing capability, and can be linked with a spraying device, a plugging device, a power-off device and the like on the basis of multi-node response, so that second-level response and local control are realized. And a cloud three-dimensional mine map and a fire behavior simulation platform are combined to construct an overground and underground cooperative fire prevention and extinguishing system, so that the early warning response capability and the emergency disposal efficiency of the coal mine fire are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine safety monitoring and intelligent control, and in particular to a coal mine underground fire prevention and extinguishing system based on a distributed Internet of Things. Background Art

[0002] Underground coal mine fires are one of the major hazards threatening coal mine production safety. They are characterized by sudden onset, rapid spread, the release of toxic and hazardous gases, and the difficulty of extinguishing them. If not promptly addressed, they can easily lead to casualties and significant property losses. Therefore, establishing an efficient and reliable underground fire prevention and extinguishing system is crucial to ensuring safe coal mine production.

[0003] Existing coal mine fire prevention and extinguishing systems mostly use a centralized IoT architecture, relying primarily on a large number of sensors to upload environmental data to a ground-based monitoring center, which then identifies fires and issues response commands. Although these systems have a certain degree of automation capabilities, they still have the following significant problems in practical applications:

[0004] Current coal mine fire warning systems mostly use a single sensor threshold trigger mechanism, which is unable to effectively identify the coordinated evolution of multi-parameter anomalies. This results in insufficient perception of early signs of fire, and often leads to false alarms, missed alarms, or delayed responses. In particular, in the presence of underground environmental fluctuations, dust interference, or localized gas anomalies, single-point alarm systems are susceptible to interference, reducing the reliability and practicality of early warnings. This problem seriously affects the safety management efficiency of mines and increases the risk of fire emergencies not being handled in a timely manner.

[0005] Existing systems generally rely on centralized processing architectures, requiring all sensor data to be uploaded to ground servers or dispatch centers before event analysis and assessment can be performed. This leads to lengthy processing times, high risk of communication interruptions, and significant single-point failure impacts. Faced with complex and ever-changing underground fire conditions, centralized mechanisms struggle to obtain comprehensive, reliable information in a timely manner for rapid assessment. This is especially true when communication links are damaged or transmission delays occur. Fire incidents can be delayed, missing the optimal time for intervention and increasing the likelihood of casualties and property damage. Summary of the Invention

[0006] In response to the above problems, a coal mine underground fire prevention and extinguishing system based on distributed Internet of Things is proposed.

[0007] The purpose of the present invention can be achieved through the following technical solutions: A coal mine underground fire prevention and extinguishing system based on a distributed Internet of Things, comprising: a perception sub-node module and an edge decision control module;

[0008] The perception sub-node module obtains the multi-modal sensor collection data deployed in different roadway areas, working faces and high-risk sections of the coal mine underground, and a local cache area is arranged in each node to record the sensor data in the last 30 seconds in real time, forming a multi-modal and time-series short-period monitoring data set; then the integrated micro control processing unit is used to analyze the short-period monitoring data set collected in real time to obtain a comprehensive score value, specifically: a plurality of time scale scores are designed to be superimposed, and the comprehensive score value under each scale is calculated; then time weight weighted fusion is adopted to comprehensively generate a final score value; the final score value is compared with a preset threshold value, and when the final score value exceeds the preset threshold value, the node determines the current state as a possible fire signal; when a node enters the possible fire signal state, a fire warning data packet containing key information is immediately constructed, and a distributed Mesh network architecture is adopted by the communication cooperation module; the node and any three or more adjacent nodes establish a communication relationship for response; when the adjacent nodes receive the fire warning data packet, the effectiveness thereof is verified, and the same is included in the local event pool; if subsequent multiple nodes continuously receive the same warning information, the subsequent joint confirmation mechanism will be triggered; the local final score value of each adjacent node is calculated in real time, and compared with the determination threshold value: if the final score value reaches more than 80% of the node determination threshold value, it is regarded as a potential co-perception response node; for the potential co-perception response node, a co-confirmation signal is constructed by combining the multi-modal data in the last 10 seconds, and returned.

[0009] The edge decision control module deploys an edge control node in each region; after receiving at least two co-confirmation signals, a multi-point verification is started to obtain a final confirmation value; when the final confirmation value S v reaches the determination threshold, a fire event confirmation signal is determined.

[0010] As a preferred embodiment of the present application, it further comprises a linkage response module and a cloud monitoring management platform.

[0011] After the linkage response module receives the fire event confirmation signal, it immediately starts the response mechanisms of spraying fire extinguishing, grouting sealing, ventilation adjustment, power cut-off and alarm evacuation;

[0012] The cloud monitoring management platform receives the underground fire warning data packet every 2 seconds to perform visual command.

[0013] As a preferred embodiment of the present application, the specific calculation process of the comprehensive score value is as follows:

[0014] Through outputting the comprehensive score value S, n is the number of selected key parameters; ω i is the risk weight of the i-th parameter; V i represents the standard deviation or variance of the parameter in a certain period of time. is the normalized rate of change of the i-th parameter in the last 10 seconds; α i is the neighborhood reference amplification factor.

[0015] As a preferred embodiment of the present invention, the calculation process of the normalized rate of change of the i-th parameter in the last 10 seconds is:

[0016] pass Output each item Among them, X i (t) is the sensor reading at the current time point t; σ is the change observation period; R i It is the safety reference variation range or the empirical normal fluctuation range of the corresponding parameters.

[0017] As a preferred embodiment of the present invention, the calculation process of the neighborhood reference amplification factor is:

[0018] Output neighborhood reference amplification factor α i ;in, is the average rate of change of the same parameter in the neighborhood, is the standard deviation of the rate of change of the parameter in the neighborhood.

[0019] As a preferred embodiment of the present invention, the specific process of using time weighted fusion to comprehensively generate the final score value is as follows:

[0020] Design the superposition of scores under multiple time scales, including short-term, medium-term and long-term scales; and calculate the comprehensive score value S under each scale, denoted as S 10s 、S 1min and S 5min ; Use time weighted fusion to comprehensively generate the final score value S z ;S z =λ1×S 10s +λ2×S 1min +λ3×S 5min ; λ1, λ2 and λ3 are weight coefficients that are dynamically adjusted according to the operating environment and historical false alarm rates.

[0021] As a preferred embodiment of the present invention, the specific process of starting multi-point verification to obtain the final confirmation value is as follows:

[0022] Extract the highest-scoring node and the earliest-reporting node records, and use the verification model to finalize the aggregated data: Output final confirmation value S v , S j The local score returned for the jth node; γ j is the time weighting factor; ρ j is the spatial similarity factor; δj is the node credibility coefficient; m is the number of nodes participating in collaborative confirmation.

[0023] As a preferred embodiment of the present invention, the time weighting factor and the spatial similarity factor;

[0024] Time weighting factor γ j Defined as: γ j =exp(-λ t ·(t j -t0)), t j is the response timestamp of the jth node; t0 is the timestamp of the earliest reporting node; λ t is the time attenuation coefficient;

[0025] Spatial similarity factor ρ j Used to measure the consistency of the change trend between the current node and the initial warning node to enhance the directional judgment ability of collaborative judgment; The multi-parameter change vector constructed for the j-th node; is the change vector of the initial warning node; κ j is the angle between the two; and ρ j ∈[-1,1].

[0026] As a preferred embodiment of the present invention, the node credibility factor δ j Used to express the long-term stability and historical reliability of each node, and dynamically analyze historical node operation data: ε j is the false alarm rate of the node; η is the penalty coefficient; expanded to the combined index: δ j =β1·(1-false alarm rate)+β2·response stability, where: stability is calculated by short-term score volatility; β1 and β2 are custom weights.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. This invention uses the edge AI chip in the perception subnode module to analyze short-term multimodal monitoring data in real time. Combining a standardized rate of change, a neighborhood reference amplification factor, and a multi-factor weighted fusion model, it can rapidly identify early signs of fire, such as temperature rise, changes in hazardous gas concentrations, and sudden changes in smoke. It also integrates scores at three timescales: 10 seconds, 1 minute, and 5 minutes. This improves the stability and robustness of early warning scores, significantly reducing the risk of false alarms and missed alerts. This allows for accurate identification and response to fires at their earliest stages, enhancing the sensitivity and timeliness of overall early warnings.

[0029] 2、The application introduces a cooperative confirmation mechanism, forms a defense ring through multi-hop self-organizing communication between nodes, and dynamically calculates parameters such as local score values, spatial variation trend similarity, response timeliness and historical reliability in multiple adjacent nodes to form a multi-dimensional weighted verification model. The final confirmation value not only combines the time sequence, but also refers to the consistency of the change direction between nodes and the historical stability, effectively improving the scientificity and robustness of event confirmation, avoiding misjudgment caused by single-point abnormality or signal island, and constructing an adaptive, multi-source cross-verification intelligent judgment system.

[0030] 3、The application can realize second-level starting after confirming the fire event through the edge decision control module linkage of multiple response devices such as spraying, grouting, ventilation, power cut-off and sound-light alarm. Rapid isolation and control of the fire area are completed. At the same time, the response strategy is dynamically adjusted according to parameters such as on-site temperature and smoke concentration, such as increasing the spraying density or closing the air flow channel, effectively containing the spread of the fire and assisting personnel evacuation. Combined with the three-dimensional mine map and dynamic fire simulation of the cloud visual management platform, the collaborative control of the surface and the underground is realized, and the active protection ability and emergency disposal efficiency of the whole coal mine fire extinguishing system are improved. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to facilitate the understanding of those skilled in the art, the application will be further described below in conjunction with the drawings.

[0032] Figure 1 The schematic diagram of the modules of the application is shown in the figure. DETAILED DESCRIPTION

[0033] The technical solutions of the application will be described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the application.

[0034] It should be understood that the terms "include" and "contain" used in the specification and claims of the present disclosure indicate the presence of the described features, whole, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, whole, steps, operations, elements, components and / or their sets.

[0035] It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used in this disclosure and the claims, "a," "an," and "the" are intended to include both singular and plural forms, unless the context clearly indicates otherwise. It is further to be understood that the terms "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items, as well as the lack of combinations when interpreted in the alternative ("and / or").

[0036] Referring to Figure 1 As shown in the figure, a coal mine underground fire prevention and extinguishing system based on a distributed Internet of Things includes a perception sub-node module, a communication coordination module, an edge decision control module, a linkage response module, and a cloud monitoring and management platform. Description: Each module is connected to each other through a distributed network mode and is deployed in different roadway areas, working faces, and high-risk sections of the coal mine underground.

[0037] The perception sub-node module acquires data collected by multi-modal sensors (including temperature sensors, carbon monoxide (CO) sensors, methane (CH4) sensors, smoke sensors, humidity sensors, etc.) deployed in different roadway areas, working faces, and high-risk sections of the coal mine underground, denoted as underground environmental parameters. Each perception node continuously acquires environmental parameters at a fixed frequency (for example, once per second), including temperature, carbon monoxide (CO), methane (CH4), smoke concentration, and humidity information. A local cache area is provided inside each node to record the latest 30 seconds of sensor data in real time, forming a multi-modal, time-series short-period monitoring data set.

[0038] The short-period monitoring data set collected in real time is analyzed by the micro-control processing unit (edge AI chip) integrated in the perception sub-node module.

[0039] Within a time window of 10 seconds, the node extracts the trend of the parameters in the short-period monitoring data set, calculates the comprehensive score value S using a standardized and multi-factor weighted fusion model, and outputs the value S through the model:

[0040] The comprehensive score value S is output, n is the number of selected key parameters (such as temperature, CO, smoke, etc., generally 3-5 items); ω i is the risk weight of the i-th parameter, indicating its importance in fire determination (such as CO being more important than humidity); V i represents the standard deviation or variance of the parameter within a certain time (such as 60 seconds) in history, which can dynamically reflect the stability of the channel; is the standardized change rate of the i-th parameter within the last 10 seconds, specifically: through Each item wherein, X i (t) is the sensor reading at the current time point t; σ is the change observation period; R i is the safety reference change range or the experience normal fluctuation range of the corresponding parameter, used for normalization processing so that data of different dimensions are comparable; α i is the neighborhood reference amplification factor, representing the amplification coefficient of the i-th parameter change of the current node relative to the average value of its neighborhood, and is calculated by: outputting the neighborhood reference amplification factor α i ; wherein, is the average change rate of the same parameter in the neighborhood, is the standard deviation of the change rate of the parameter in the neighborhood; if the change of the current node is higher than a times (such as a = 2) of the change of the neighborhood, it is considered as a local anomaly weighting, enhancing the sensitivity; if it is consistent with the neighborhood, the score amplification is inhibited to prevent false positives;

[0041] According to the obtained comprehensive score value S (in a time window of 10 seconds in length), score superposition at multiple time scales is designed, such as short-term (10 seconds), medium-term (1 minute) and long-term (5 minutes) three scales; and the comprehensive score values S at each scale are calculated, which are denoted as S 10s , S 1min and S 5min respectively; time weight is used for weighted fusion to generate the final score value S z ; S z = λ1×S 10s + λ2×S 1min + λ3×S 5min ; wherein, λ1, λ2 and λ3 are weight coefficients dynamically adjusted according to the running environment and the historical false positive rate; λ1+ λ2+ λ3=1. The specific size of λ1, λ2 and λ3 can be determined by objective weighting method, such as entropy weight method, principal component analysis (PCA), CRITIC method, etc.

[0042] The final score value is compared with a preset threshold value, and when the final score value exceeds the preset threshold value, the current state of the node is determined as a “possible fire” signal, that is, there is obvious temperature rise, gas concentration rise or smoke index mutation, which has early signs of fire.

[0043] When a node enters the “possible fire” signal state, a fire warning data packet containing key information is immediately constructed, and the communication coordination module uses a distributed Mesh network architecture to support multi-hop forwarding between nodes, self-organizing communication and path reconstruction functions. Function description: the node can establish a communication relationship with any three or more adjacent nodes; when a communication link is disconnected, it automatically finds other available paths for data forwarding, and has “communication self-healing” function; priority broadcast and multicast mode are supported, and the fire node can quickly notify the surrounding area to form a “defense ring”;

[0044] When the adjacent node receives the early warning package, it will verify its validity and include it in the local event pool; if subsequent multiple nodes continuously receive the same (obtained through similarity calculation) or similar early warning information, the subsequent joint confirmation mechanism will be triggered;

[0045] The current local final score value of each adjacent node is calculated in real time, denoted as S loca , and compared with the decision threshold value: if the value reaches more than 80% of the node decision threshold value (S thresh ), that is, S loca > 0.8·S thresh , it is considered as a potential co-sensing response node;

[0046] For the potential co-sensing response node, a "co-confirmation signal" will be constructed combining the multi-modal data within 10 seconds and returned;

[0047] The edge decision control module deploys an edge control node (such as an industrial control host or a strong computing power edge gateway) in each area (such as a district working face, return airway, etc.); after receiving at least two co-confirmation signals, multi-point verification is started:

[0048] All received "co-confirmation signals" are distributed cached and sorted for extraction:

[0049] The "highest score node" and "earliest reporting node" records are extracted, and abnormal nodes with inconsistent data, missing data or severe fluctuations are eliminated;

[0050] Next, the verification model is used to make a final confirmation on the summarized data: Output the final confirmation value S v , S j is the local score returned by the jth node; γ j is the time weighting factor (the earlier the report, the higher the weight); ρ j is the spatial similarity factor (the consistency of the sensing trend of the node with the center node); δ j is the node credibility coefficient (calculated based on the historical false alarm rate and stability of the node in the past); m is the number of nodes participating in co-confirmation.

[0051] Among them, the time weighting factor γ j is used to reflect the judgment principle of "the earlier the response, the more reliable", and is defined as:

[0052] γ j = exp(-λ t ·(t j -t0)), t j is the response timestamp of the jth node; t0 is the timestamp of the earliest reporting node (i.e. min(tj )) ; λ t is the time decay coefficient, controlling the priority of early warning signals, recommended value range [0.1, 1.0] ; γ j ∈(0, 1], the closer to 1, the greater the weight of the earlier reporter;

[0053] Spatial similarity factor p j Used to measure the consistency of the change trend of the current node and the initial warning node, to enhance the directional judgment ability of collaborative decision; The "multi-parameter change vector" constructed for the jth node, such as [AT, AC0, Asmoke]; is the change vector of the initial warning node; κ j is the included angle between the two; p j ∈[-1, 1], but in practice, negative correlation nodes can be eliminated, and only p j > 0 effective response;

[0054] Node credibility factor d j Used to express the long-term stability and historical reliability of each node, based on the dynamic evaluation of node operation history data: ε j is the false alarm rate / fault rate of the node, which can be obtained by historical event statistics, such as the false alarm proportion in the past N times of collaborative response; η is the penalty coefficient, controlling the influence degree of false alarm rate on weight (recommended value 2 to 5); can be extended to a combination index:

[0055] d j = β1·(1- false alarm rate) + β2· response stability, where: stability can be calculated by short-term score fluctuation rate (such as standard deviation / mean); β1 and β2 are self-defined weights (such as 0.6 and 0.4);

[0056] When the final confirmation value S v reaches the decision threshold (such as pass rate ≥ 85%), it is determined as "fire event confirmation" signal, and immediately enters the response stage;

[0057] The linkage response module receives the "fire event confirmation" signal; controls the fire extinguishing spray device to activate within 1 second after response determination; sprays high-pressure water mist or chemical fire extinguishing agent to the trigger node and adjacent area (radius 10-15 meters); if the environmental temperature > 80℃ or the smoke concentration > 70% FS, the spray density is increased; the spray cycle is 15s each time, and the interval is 10s cycle, until the center temperature drops below the set threshold; control the grouting sealing device to use high-viscosity fire-retardant materials (such as heat-resistant foam / mineral slurry) for rapid filling when the measured point temperature continues to rise > 100℃, and preferentially seal the space outlet consistent with the wind direction; control the ventilation adjustment device to close the air door near the fire area and adjust the main wind direction to cut off the oxygen supply or assist personnel evacuation; control the power cut-off device to preferentially cut off the main power supply trunk near the fire source; control the alarm light and sound evacuation device to start the high-brightness LED lamp, strong sound pressure buzzer and emergency indication arrow; the starting logic is: cover the whole mine area broadcast "fire alarm notice", node belongs to the partition light red flash + buzzer, green indicating light direction of emergency evacuation passage;

[0058] The cloud monitoring management platform receives the underground fire alarm data package every 2 seconds, including: fire alarm location (GIS coordinates + node ID), current environmental parameters (temperature, CO, smoke, etc.), alarm level and linkage device state and information automatically pushed to the dispatch center large screen + mobile terminal; visual command interface is performed, showing the three-dimensional mine structure diagram and the real-time state of each sensing node and execution device and the dynamic fire spread simulation diagram.

[0059] The preferred embodiments disclosed above are only used to help explain the present application. The preferred embodiments do not describe all the details, nor limit the present application to the specific implementation. Obviously, according to the content of the present application, many modifications and changes can be made. The present application selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present application, so that those skilled in the art can well understand and utilize the present application. The present application is limited by the claims and their entire scope and equivalents.

Claims

1. A coal mine underground fire prevention and extinguishing system based on a distributed Internet of Things, comprising: Perception sub-node module and edge decision control module; characterized by: The perception sub-node module acquires data collected by multi-modal sensors deployed in different tunnel areas, working faces and high-risk sections in the coal mine. A local cache area is set inside each node to record the sensor data of the last 30 seconds in real time to form a multi-modal, time-series short-period monitoring data set; the short-period monitoring data set collected in real time is then analyzed by the integrated micro-control processing unit to obtain a comprehensive score value, specifically: design a score superposition under multiple time scales, and calculate the comprehensive score value under each scale; then use time weighted fusion to comprehensively generate a final score value; compare the final score value with a preset threshold. When the final score value exceeds the preset threshold, the node determines the current state as a possible fire signal; when a node enters a possible fire signal After the signal state is detected, a fire warning data packet containing key information is immediately constructed, and the communication collaboration module adopts a distributed Mesh network architecture; the node establishes a communication relationship with any three or more neighboring nodes to respond; when the neighboring node receives the fire warning data packet, its validity is verified and included in the local event pool; if multiple nodes subsequently receive the same warning information continuously, the subsequent joint confirmation mechanism will be triggered; the current local final score value of each neighboring node is calculated in real time and compared with the judgment threshold: if the final score value reaches more than 80% of the judgment threshold of this node, it is regarded as a potential co-perception response node; for the potential co-perception response node, a collaborative confirmation signal will be constructed and returned based on its own multimodal data within 10 seconds; The edge decision control module deploys an edge control node in each area; after receiving at least two or more collaborative confirmation signals, it starts multi-point verification to obtain the final confirmation value; when the final confirmation value S v When the judgment threshold is reached, it is determined as a fire event confirmation signal.

2. The coal mine underground fire prevention and extinguishing system based on distributed Internet of Things according to claim 1 is characterized in that: It also includes a linkage response module and a cloud-based monitoring and management platform; After receiving the fire event confirmation signal, the linkage response module immediately activates multiple response mechanisms including spray fire extinguishing, grouting sealing, ventilation adjustment, power cut-off and alarm evacuation; The cloud-based monitoring and management platform receives underground fire warning data packets every 2 seconds and conducts visual command.

3. The coal mine underground fire prevention and extinguishing system based on distributed Internet of Things according to claim 1 is characterized in that: The specific calculation process of the comprehensive score is as follows: pass Output comprehensive score S, n is the number of key parameters selected; ω i is the risk weight of the i-th parameter; V i Indicates the standard deviation or variance of the parameter within a certain historical period; is the normalized rate of change of the i-th parameter in the last 10 seconds; α i is the neighborhood reference amplification factor.

4. The coal mine underground fire prevention and extinguishing system based on distributed Internet of Things according to claim 3 is characterized in that: The calculation process of the standardized rate of change of the i-th parameter in the last 10 seconds is: pass Output each item Among them, X i (t) is the sensor reading at the current time point t; σ is the change observation period; R i It is the safety reference variation range or the empirical normal fluctuation range of the corresponding parameters.

5. The coal mine underground fire prevention and extinguishing system based on distributed Internet of Things according to claim 4 is characterized in that: The calculation process of the neighborhood reference amplification factor is: Output neighborhood reference amplification factor α i ;in, is the average rate of change of the same parameter in the neighborhood, is the standard deviation of the rate of change of the parameter in the neighborhood.

6. The coal mine underground fire prevention and extinguishing system based on distributed Internet of Things according to claim 5 is characterized in that: The specific process of using time weighted fusion to comprehensively generate the final score value is as follows: Design the superposition of scores under multiple time scales, including short-term, medium-term and long-term scales; and calculate the comprehensive score value S under each scale, denoted as S 10s 、S 1min and S 5min ; Use time weighted fusion to comprehensively generate the final score value S z ;S z =λ1×S 10s +λ2×S 1min +λ3×S 5min ; λ1, λ2 and λ3 are weight coefficients that are dynamically adjusted according to the operating environment and historical false alarm rates.

7. The coal mine underground fire prevention and extinguishing system based on distributed Internet of Things according to claim 1 is characterized in that: The specific process of starting multi-point verification to obtain the final confirmation value is as follows: Extract the highest-scoring node and the earliest-reporting node records, and use the verification model to finalize the aggregated data: Output final confirmation value S v , S j The local score returned for the jth node; γ j is the time weighting factor; ρ j is the spatial similarity factor; δ j is the node credibility coefficient; m is the number of nodes participating in collaborative confirmation.

8. The coal mine underground fire prevention and extinguishing system based on distributed Internet of Things according to claim 1 is characterized in that: Temporal weighting factor and spatial similarity factor; Time weighting factor γ j Defined as: j =exp(-λ t ·(t j -t0)), t j is the response timestamp of the jth node; t0 is the timestamp of the earliest reporting node; t is the time attenuation coefficient; Spatial similarity factor ρ j Used to measure the consistency of the change trend between the current node and the initial warning node to enhance the directional judgment ability of collaborative judgment; The multi-parameter change vector constructed for the j-th node; is the change vector of the initial warning node; κ j is the angle between the two; and ρ j ∈[-1,1].

9. The coal mine underground fire prevention and extinguishing system based on distributed Internet of Things according to claim 7, characterized in that: Node credibility factor δ j Used to express the long-term stability and historical reliability of each node, and dynamically analyze historical node operation data: ε j is the false alarm rate of the node; η is the penalty coefficient; expanded to the combined index: δ j =β1·(1-false alarm rate)+β2·response stability, where: stability is calculated by short-term score volatility; β1 and β2 are custom weights.

Images