Fire scene situation information processing and distributing method and system based on Bayesian network
Through the Bayesian network-based fire situation information processing and distribution method, the problem of difficulty in transmitting situation information at large fire scenes is solved, and fast and accurate situation information processing and decision-making support is achieved, providing a new solution for fire emergency rescue.
Patent Information
- Application Number
- CN202510003328.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-02
- Publication Date
- 2025-04-15
AI Technical Summary
At large fire scenes, it is difficult to transmit real-time, accurate and comprehensively, the fire situation information is difficult, resulting in low command efficiency, difficult decision-making, and inconvenient post-war summary and commentary.
The fire situation information processing and distribution method based on Bayesian network is adopted to collect, summarize, process and distribute fire situation information, use Bayesian network model to deduce the fire situation development trend, and provide situation playback function after the fire extinguishing and rescue is completed.
It has achieved rapid and accurate processing and distribution of fire situation information, improved the decision-making efficiency of fire scene commanders and the effectiveness of emergency response, and provided a comprehensive and authentic basis for the post-war summary and commentary.
Smart Images
Figure CN119830136A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for processing and distributing fire situation information based on a Bayesian network. Background Art
[0002] With the expansion of urban scale, the increase of population density and the improvement of industrialization, large-scale fires in high-rise buildings, urban complexes, oil storage facilities, etc. occur frequently, which can easily cause a large number of losses of life and property. At present, in the process of fire fighting and rescue, after obtaining information such as fire conditions, trapped persons, and fire environment, fire commanders generally use handheld walkie-talkies to convey the fire situation and various instructions to each position, resulting in low efficiency in the transmission of command instructions and situation information. The real-time situation information of the fire scene is often difficult to be timely, accurately and comprehensively transmitted to other combatants and superior commanders on the scene; when a large-scale fire occurs, the fire situation information with large amounts of multi-source and heterogeneous characteristics brings great difficulties to the fire commander to make quick and accurate decisions; because walkie-talkies are generally used to convey situation information and instructions during the battle, there is a lack of recording function for a large amount of multi-source heterogeneous situation information, so in the post-war summary and review stage, it is not conducive to effective review and summary of the command and dispatch during the battle. Therefore, there is an urgent need for a method and system that can collect, summarize, process, distribute and display fire situation information in real time, quickly and stably, so as to improve the efficiency of fire situation perception of commanders and fighters at various positions in the fire scene, provide fast and accurate auxiliary decision-making support for fire commanders, thereby improving the command efficiency of large-scale fire scenes, and providing a comprehensive and true basis for post-war summary and evaluation. Summary of the invention
[0003] The purpose of the present invention is to provide a method and system for processing and distributing fire situation information based on a Bayesian network, which can solve the technical problems mentioned in the background technology.
[0004] A method for processing and distributing fire situation information based on a Bayesian network, the method comprising the following steps:
[0005] Collect situation information: obtain map images of the fire area;
[0006] Situation information summary: by dragging icons, hand-drawing, inputting options and texts on the acquired fire area map image, multi-source situation information can be quickly generated. The multi-source situation information includes the location of the fire point, the duration of occurrence, combustibles, trapped personnel, water supply, fire environment and roads, etc.;
[0007] Situation information processing: Based on the Bayesian network model, the fire development trend is deduced according to multi-source situation information;
[0008] Situation information distribution: Fire situation information, images, videos, etc. are distributed and reported between the fire scene commander and various combat positions and superiors.
[0009] As a further technical solution of the present invention, the method also includes that after the fire fighting and rescue is completed, if it is necessary to summarize and comment on the entire battle situation, a certain period of time or the entire battle process can be selected in the situation playback module corresponding to the fire scene commander terminal or the superior commander terminal to play back the images, video and audio information sent and received during the period of time.
[0010] As a further technical solution of the present invention, the system involved in the method also includes a sound pickup device (pickup), which is configured near the handheld walkie-talkies of the fire scene commander and each combat position personnel, and sends the collected audio information to their terminals; thus, multi-source situation information, images and videos, and walkie-talkie audio information can be synchronously distributed to the corresponding combat positions or superior commander terminals.
[0011] As a further technical solution of the present invention, the Bayesian network model is a Bayesian network model suitable for deducing the development trend of fire in commercial complexes. The process of constructing the Bayesian network model and reasoning and calculating the probability of each scenario state node is as follows:
[0012] (1) Define the problem and variables:
[0013] The evolution of commercial complex fires is a continuous process. Depending on the effectiveness of the response measures, it will have a positive or negative drive on the development of the fire, making the evolution path develop in the optimistic and pessimistic directions respectively;
[0014] The evolution of commercial complex fires is affected by three main factors: fire scenario state S, emergency measures M, and disposal target T. The interaction between these factors makes the fire accidents in commercial complexes evolve continuously. According to the pessimistic decision-making criterion, the evolution scenarios of such fire accidents are selected, and finally 13 scenario states (S0, S1, S2, ..., S 13 ), 12 disaster response (R0, R1, R2, ..., R 12 ), 12 emergency response targets (T0, T1, T2, ..., T 12 ), and obtain the scenario table of commercial complex fire accidents; this type of fire will appear S0, S1, S2, ..., S3 over time from the start of the fire to the rapid development of the fire and then to the complete extinguishing of the fire. n There are multiple scenario states, where S0 represents the initial scenario before the fire occurs, S n Indicates the fire end scenario; the initial scenario S0, under the influence of emergency measures R0, regardless of whether the disposal target T0 can be achieved, will develop to the next scenario state until the fire accident disappears S nIf every disposal goal can be achieved after the intervention of emergency measures, the fire can be stopped as soon as possible; otherwise, the fire fighting and rescue will be prolonged, causing more serious losses;
[0015] (2) Building a network structure:
[0016] By dividing the fire accident scenario states and summarizing the occurrence rules of the same type of accidents based on historical accident cases, the causal relationship of network nodes is obtained; the various network nodes in the commercial complex fire accident scenario table are linked together according to the logical causal relationship to form a commercial complex fire accident scenario evolution path diagram, in which the horizontal direction is the most optimistic evolution path and the vertical direction is the most pessimistic evolution path;
[0017] (3) Determine the prior probability and conditional probability table:
[0018] According to the evolution path diagram of commercial complex fire accident scenarios and historical cases of commercial complex fire accidents, the prior probability of the initial event is determined. For network node variables with preceding variables, the conditional probability is obtained through historical case analysis and expert scoring. Based on the pessimistic decision criterion, the Bayesian network node probability is corrected to obtain the prior probability and conditional probability table of network node variables.
[0019] (4) Model reasoning and probability calculation:
[0020] First, the fire commander inputs multi-source situation information through the system interface; then, based on the evolution path diagram of the commercial complex fire accident scenario and combined with Formula 1 and the prior probability and conditional probability table of the network node variables, the probability of each scenario is deduced respectively; finally, the fire situation is predicted based on the calculated probability;
[0021] Formula 1: ;
[0022] Among them, P(B) is the total probability, P(A) is the prior probability, and P(B|A) is the conditional probability.
[0023] Another object of the present invention is to provide a fire situation information processing and distribution system based on a Bayesian network, the system comprising terminal software for multiple different roles in fire fighting and rescue operations, a Bayesian network model for deducing the fire development situation, and a database and background software running on a cloud server; the system is used to execute the above-mentioned fire situation information processing and distribution method based on a Bayesian network.
[0024] Beneficial effects achieved by the present invention:
[0025] (1) The developed software system can quickly process the collected multi-source heterogeneous situation information based on the actual needs of the fire commander (such as quickly generating a fire situation map by dragging preset common fire elements and firefighting facility icons or hand-drawing); and quickly draw conclusions after situation simulation to provide auxiliary decision support for the fire commander. The system fills the existing market gap in terms of information integration, real-time simulation and collaborative operation, and provides a brand-new solution for fire emergency rescue.
[0026] (2) A Bayesian network model suitable for the deduction of fire situation development trends in commercial complexes was constructed and integrated into the commander terminal software. In the situation information processing stage, the fire commander can process the input situation information based on the model and thus derive the fire situation development trend over a period of time. The deduction results are not only conducive to improving the decision-making efficiency of fire commanders, but also to improving the effectiveness of emergency response. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Figure 1 The figure is a flowchart of a fire situation information processing and distribution method based on Bayesian network.
[0028] Figure 2 This is a composition diagram of a fire situation information processing and distribution system based on Bayesian network.
[0029] Figure 3 This is the commercial complex fire scenario evolution path diagram provided by the present invention. DETAILED DESCRIPTION
[0030] The technical solution of the present invention is described in detail below in conjunction with the specific drawings.
[0031] See also Figure 1 , an embodiment of the present invention provides a method for processing and distributing fire situation information based on a Bayesian network, the method comprising the following steps;
[0032] Collect situation information: obtain map images of the fire area;
[0033] Specifically, it refers to: During the stage of collecting situation information, when the fire commander is on his way to or after arriving at the fire scene, he can use the commander terminal of this system to obtain fire scene map information in the following three ways: 1. Directly retrieve from the terminal. The commander needs to pre-set the map information of key fire prevention units in the jurisdiction to the local terminal and the cloud in advance; 2. Call the electronic map through the terminal, locate the fire area and then take a screenshot; 3. Receive the original image of the fire scene collected by the firefighting drone and pre-process the image;
[0034] Situation information summary: by dragging icons, hand-drawing, inputting options and texts on the acquired fire area map image, multi-source situation information can be quickly generated. The multi-source situation information includes the location of the fire point, the duration of occurrence, combustibles, trapped personnel, water supply, fire environment and roads, etc.; it also includes temperature, humidity, smoke concentration, air pressure, oxygen concentration, toxic gas concentration, etc.;
[0035] Specifically, it means: in the situation information summary stage, the fire commander uses the commander terminal to quickly generate a situation map of the fire scene by dragging preset common fire elements and firefighting facility icons or hand-drawing on the fire area map image obtained above; in the preset situation information input interface, quickly input the multi-source situation information that has been mastered, such as the location of the fire point, duration of occurrence, combustibles, trapped personnel, water supply, fire environment and roads, etc. The situation information input interface is divided into high-rise buildings, commercial complexes, etc. according to the type of fire. Correspondingly, the situation information that needs to be input is also different;
[0036] Situation information processing: Based on the Bayesian network model, the fire development trend is deduced according to multi-source situation information;
[0037] Situation information distribution: Multi-source situation information, images, videos, etc. are distributed and reported between the fire scene commander and various combat positions and superiors.
[0038] Specifically, it can refer to: in the situation information distribution stage, the commander forwards multi-source situation information, images, and videos according to actual needs (point-to-point, point-to-multipoint, and broadcast modes can be selected) under the situation information distribution module of his terminal software; the system involved in the method also includes a sound pickup device (pickup), which is configured near the handheld walkie-talkies of the fire scene commander and the personnel at each battle position, and sends the collected audio information to their terminal; thus, the multi-source situation information, images, and videos, and the walkie-talkie audio information can be synchronously distributed to the corresponding battle position or the superior commander's terminal. This system is compatible with the existing wireless voice communication mode of fire fighting and rescue operations, and can also be used as a backup means for storing voice communications. During the entire fire fighting and rescue process, the above-mentioned situation information aggregation, processing, and distribution will be repeated many times. In addition to the commander constantly summarizing, updating, and distributing various situation information and instructions, the status and location of the combatants or unmanned equipment at each battle position will also be reported to the fire scene commander. Instructions and situation information from the superior commander will also be sent to the fire scene commander. In addition to being stored on their respective terminals, all of this situation information and instructions are also stored on the cloud server.
[0039] In this embodiment, the method also includes that after the fire fighting and rescue is completed, if it is necessary to summarize and comment on the entire battle situation, a certain period of time or the entire battle process can be selected in the situation playback module corresponding to the fire scene commander terminal or the superior commander terminal to play back the images, videos and audio information sent and received during the period of time.
[0040] Another object of the present invention is to provide a software system, specifically a fire situation information processing and distribution system based on a Bayesian network, the system includes terminal software (running on the Hongmeng operating system) for multiple different roles in firefighting and rescue operations (including superior commander terminals, fire commander terminals, and terminals at various stations), a Bayesian network model for deducing the development of the fire situation, and a database and background software running on a cloud server; the system is used to execute the above-mentioned fire situation information processing and distribution method based on a Bayesian network. The system composed of this software system and related terminals, communication facilities and cloud servers is as follows Figure 2 shown.
[0041] In this embodiment, the Bayesian network model is a Bayesian network model suitable for deducing the development trend of fire in a commercial complex. The process of constructing the Bayesian network model and reasoning and calculating the probability of each scenario state node is as follows:
[0042] (1) Define the problem and variables:
[0043] The evolution of fire in commercial complex is a continuous process. In addition to being affected by the laws of fire development and evolution, it is also affected by the emergency measures taken by various departments. According to the effectiveness of the response measures, it will have a positive or negative drive on the development of the fire, making the evolution path develop in the optimistic and pessimistic directions respectively. Positive drive means that the event is transferred to the safe side. Whether the expected disposal goals can be achieved and the fire hazards can be mitigated is used as a reference to judge the effectiveness of emergency measures and the direction of fire development. When the emergency measures achieve the expected emergency measures goals, that is, the fire develops in the direction expected by people, the fire spread is controlled, the spread speed is slowed down, and the fire hazards gradually stop, that is, the fire evolves in the optimistic direction. When the emergency measures have little effect on the spread of the fire, that is, the disposal goals cannot be fully achieved, the fire scenario deviates from the expected direction, and the fire develops in the pessimistic direction. When the emergency measures are completely ineffective, all disposal goals cannot be achieved, and the fire presents a state of barbaric development until all the energy is released, the fire impact will stop. The fire develops in the most pessimistic direction, which often brings serious damage and losses.
[0044] The evolution of commercial complex fires is affected by three main factors: fire scenario state S, emergency measures M, and disposal target T. The interaction between these factors makes the fire accidents in commercial complexes evolve continuously. According to the pessimistic decision-making criterion, the evolution scenarios of such fire accidents are selected, and finally 13 scenario states (S0, S1, S2, ..., S 13 ), 12 disaster response (R0, R1, R2, ..., R 12 ), 12 emergency response targets (T0, T1, T2, ..., T 12 ), and obtain the commercial complex fire accident scenario table, as shown in Table 1; this type of fire from the fire to the rapid development of the fire and then to the complete extinguishing of the fire, over time there will be S0, S1, S2, ..., S n There are multiple scenario states, where S0 represents the initial scenario before the fire occurs, S n Indicates the fire end scenario, in this embodiment, n=13; the initial scenario S0, under the influence of emergency measures R0, regardless of whether the disposal target T0 can be achieved, will develop to the next scenario state until the fire accident disappears S n If every disposal goal can be achieved after the intervention of emergency measures, the fire can be stopped as soon as possible; otherwise, the fire fighting and rescue will be prolonged, causing more serious losses;
[0045] Table 1- Commercial complex fire accident scenarios
[0046]
[0047] (2) Building a network structure:
[0048] By dividing the fire accident scenario states and summarizing the occurrence patterns of the same type of accidents based on historical accident cases, the causal relationship of network nodes is obtained; the various network nodes in the commercial complex fire accident scenario table are linked together according to the logical causal relationship to form a commercial complex fire accident scenario evolution path diagram, in which the horizontal direction is the most optimistic evolution path, and the vertical direction is the most pessimistic evolution path, such as Figure 3 As shown;
[0049] Through the analysis of the fire evolution path of this case, it can be seen that there are 9 evolution paths in the fire development process;
[0050] a. Fire initial evolution path:
[0051] 1) S0→S 13 ; For some reason, a fire source appears (S0); measures are taken, the fire is extinguished / the fire is under control, and the destruction is stopped (S 13 ).
[0052] 2) S0→S1→S2→S 13 ; For some reason, a fire source appears (S0); a fire accident occurs (S1); on-site personnel discover the fire and use a fire extinguisher to put it out, the fire is under control, and residual fire remains (S2); the fire is under control and destruction stops (S 13 ).
[0053] 3) S0→S1→S3→S4→S 13 ; For some reason, a fire source appears (S0); a fire accident occurs (S1); the fire begins to spread (S3); the fire shutter is intact and the fire situation is stable (S4); the fire is under control and destruction stops (S 13 ).
[0054] b. Evolution path of fire development stage:
[0055] 4) S0→S1→S3→S4→S5→S 13 ; For some reason, a fire source appears (S0); a fire accident occurs (S1); the fire begins to spread (S3); the fire shutter is intact and the fire situation is stable (S4); the fire spreads horizontally and vertically (S5); the fire is controlled and destruction stops (S 13 ).
[0056] 5) S0→S1→S3→S6→S7→S 13 ; For some reason, a fire source appears (S0); a fire accident occurs (S1); the fire begins to spread (S3); there are a lot of flammable materials; the fire shutter is opened; the fire spreads on the floor (S6); the spread of the fire is stopped, the fire is controlled, and the burning is stable (S7); the fire is controlled and the destruction stops (S 13 ).
[0057] 6) S0→S1→S3→S6→S8→S9→S 13 ; For some reason, a fire source appears (S0); a fire accident occurs (S1); the fire begins to spread (S3); there are a lot of flammable materials; the fire shutter is opened; the fire spreads on the floor (S6); the fire spreads to the adjacent floors; the building components are damaged and collapsed; open flames appear on multiple floors (S8); the fire is under control and the fire tends to weaken (S9); the fire is under control and the destruction is stopped (S 13 ).
[0058] c. Path of the fierce burning stage of fire:
[0059] 7) S0→S1→S3→S6→S8→S9→S 10 →S 13; For some reason, a fire source appears (S0); a fire accident occurs (S1); the fire begins to spread (S3); there are a lot of flammable materials; the fire shutter is opened; the fire spreads on the floor (S6); the fire spreads to the adjacent floors; the building components are damaged and collapsed; open flames appear on multiple floors (S8); the fire is under control and the fire tends to weaken (S9); the open flames continue to burn on each floor and the fire spreads to the entire complex (S 10 ); The fire is under control and destruction is stopped (S 13 ).
[0060] 8) S0→S1→S3→S6→S8→S 11 →S 13 ; For some reason, a fire source appears (S0); a fire accident occurs (S1); the fire begins to spread (S3); there are a lot of flammable materials; the fire shutter is opened; the fire spreads on the floor (S6); the fire spreads to the adjacent floors; building components are damaged and collapsed; open flames appear on multiple floors (S8); fire separation facilities are damaged and the fire spreads to the entire complex (S 11 ); The fire is under control and destruction is stopped (S 13 ).
[0061] 9) S0→S1→S3→S6→S8→S 11 →S 12 →S 13 ; For some reason, a fire source appears (S0); a fire accident occurs (S1); the fire begins to spread (S3); there are a lot of flammable materials; the fire shutter is opened; the fire spreads on the floor (S6); the fire spreads to the adjacent floors; building components are damaged and collapsed; open flames appear on multiple floors (S8); fire separation facilities are damaged and the fire spreads to the entire complex (S 11 ); the entire complex continued to burn, and the building partially collapsed; the fire spread to adjacent buildings (S 12 ); The fire is under control and destruction is stopped (S 13 ).
[0062] (3) Determine the prior probability and conditional probability table (CPT):
[0063] According to the evolution path diagram of the commercial complex fire accident scenario, based on the historical cases of commercial complex fire accidents, the prior probability of the initial event is determined, and the conditional probability of the network node variables with preceding variables is obtained through historical case analysis and expert scoring. Based on the pessimistic decision criterion, the Bayesian network node probability is corrected, so as to obtain the prior probability and conditional probability tables of the network node variables, as shown in Tables 2 and 3;
[0064] Table 2 - Scenario network node prior probability table
[0065]
[0066] Table 3 - Conditional probability table of scenario network nodes
[0067]
[0068]
[0069] (4) Model reasoning and probability calculation:
[0070] The following example shows how to use the dynamic Bayesian network to deduce the fire scenario of a commercial complex based on known information, and use the deduced conclusions for situation prediction in this situation distribution system. First, the fire commander inputs the multi-source situation information through the system interface. Then, based on the evolution path diagram of the commercial complex fire accident scenario and combined with formula 1 and the prior probability and conditional probability table of network node variables, the probability of each scenario is deduced (the following scenarios S2 S5 and S6 are used as examples). 10 Take this as an example); finally, the fire situation is predicted based on the calculated probability;
[0071] Formula 1: ;
[0072] Among them, P(B) is the total probability, P(A) is the prior probability, and P(B|A) is the conditional probability.
[0073] Scenario 1:
[0074] In a commercial complex, a maintenance worker violated regulations by welding without fire separation at the site. High-temperature molten material fell into the stairwell on the first floor below ground, generating sparks that ignited the decoration materials and oil stains on components in the elevator shaft, causing the fire to spread (S1 occurred). The worker used a fire extinguisher to put out the fire (R1 occurred) and successfully prevented the fire from spreading further (T1 was achieved).
[0075] Depend on Figure 3 It can be seen that if the disposal target T1 is achieved, the scenario S2 will occur, otherwise S3 will occur. Therefore, the probability of fire scenario S2 occurring is consistent with the probability of disposal target T1 being achieved. The probability of S1 occurring is calculated to be 0.21, and the probability calculation process of S2 is as follows:
[0076] P(S2=T)=P(T1=T)
[0077] =P(S1=T)P(R1=T)P(T1=T|S1=T,R1=T)
[0078] +P(S1=T)P(R1=F)P(T1=T|S1=T,R1=F)
[0079] +P(S1=F)P(R1=T)P(T1=T|S1=F,R1=T)
[0080] +P(S1=F)P(R1=F)P(T1=T|S1=F,R1=F)
[0081] =0.21×0.72×0.1+0.21×0.28×0.2+0.79×0.72×0.3+0.79×0.28×0.67=0.35;
[0082] That is, the probability of S2 occurring (the fire is under control, with only residual fire remaining) is 0.35. Since the probability of S1 occurring is relatively low (0.21) as calculated previously, the probability of S2 occurring is also relatively low.
[0083] Scenario 2:
[0084] In the above scenario 1, if T1 is not achieved, that is, the fire is not successfully prevented from further expansion, the residual fire will reignite and the fire will begin to spread (S3 occurs). Firefighters arrive at the scene and activate the emergency plan and start firefighting operations (R3 occurs). If the fire is controlled (T3 is achieved), the fire shutter is intact, the fire burns steadily, and other businesses on the floor are not affected (S4 occurs). At this time, the water sprinkler system is activated, and the firefighters use fire extinguishers to put out the fire (R4 occurs). If the fire is not successfully prevented from expanding (T4 is not achieved), the fire will spread horizontally and vertically (S5 occurs).
[0085] Depend on Figure 3 It can be seen that when S4 occurs, if the treatment target T4 is achieved, S 13 This scenario; otherwise S5 occurs. Therefore, the probability of fire scenario S5 occurring is consistent with the probability of disposal target T4 not being achieved.
[0086] The probability calculation process of S4 is as follows:
[0087] P(S4=T)=P(T3=T)
[0088] =P(S3=T)P(R3=T)P(T3=T|S3=T,R3=T)
[0089] +P(S3=T)P(R3=F)P(T3=T|S3=T,R3=F)
[0090] +P(S3=F)P(R3=T)P(T3=T|S3=F,R3=T)
[0091] +P(S3=F)P(R3=F)P(T3=T|S3=F,R3=F)
[0092] =0.65×0.78×0.11+0.65×0.22×0.7+0.35×0.78×0.81+0.35×0.22×0.91=0.056+0.1+0.22113+0.07=0.45
[0093] The probability calculation process of S5 is as follows:
[0094] P(S5=T)=P(T4=F)=1-P(T4=T)
[0095] =1-(P(S4=T)P(R4=T)P(T4=T|S4=T,R4=T)
[0096] +P(S4=T)P(R4=F)P(T4=T|S4=T,R4=F)
[0097] +P(S4=F)P(R4=T)P(T4=T|S4=F,R4=T)
[0098] +P(S4=F)P(R4=F)P(T4=T|S4=F,R4=F))
[0099] =1-(0.45×0.5×0.1+0.45×0.5×0.2+0.55×0.5×0.3+0.55×0.5×0.6=0.0225+0.045+0.0825+0.165)=1-0.32=0.68;
[0100] That is, the probability of S5 occurring (fire spreading horizontally or vertically) is 0.68, which is relatively high, indicating that the fire at this stage can easily spread horizontally or vertically.
[0101] Scenario 3:
[0102] In the above scenario 2, if T3 is not achieved, that is, the fire is not successfully controlled, then in the presence of a large amount of flammable materials and the fire shutters are open, the fire spreads on the floor (S6 occurs). Firefighters continue to extinguish the fire with water guns while clearing the flammable materials; (R6 occurs). If the fire is not successfully prevented from spreading (T6 is not achieved), the fire spreads to adjacent floors; building components are damaged and collapsed; open flames appear on multiple floors (S8 occurs). At this time, firefighters attack from inside to extinguish the fire and establish partitions. (R8 occurs). If the fire is successfully prevented from spreading (T8 is achieved), the fire is under control and the fire tends to weaken (S9 occurs). At this time, firefighters continue to extinguish the fire and cool down the surrounding objects. If the fire is not successfully controlled, open flames continue to burn on each floor and the fire spreads to the entire complex (S 10 occur).
[0103] Depend on Figure 3It can be seen that when S9 occurs, if the treatment target T9 is achieved, S 13 This scenario; otherwise S 10 Therefore, the probability of fire scenario S5 occurring is consistent with the probability of disposal target T9 not being achieved. 10 The probability calculation process is as follows:
[0104] P(S 10 =T)=P(T9=F)=1-P(T9=T)
[0105] =1-(P(S9=T)P(R9=T)P(T9=T|S9=T,R9=T)
[0106] +P(S9=T)P(R9=F)P(T9=T|S9=T,R9=F)
[0107] +P(S9=F)P(R9=T)P(T9=T|S9=F,R9=T)
[0108] +P(S9=F)P(R9=F)P(T9=T|S9=F,R9=F))
[0109] =1-(0.51×0.69×0.4+0.51×0.31×0.3+0.49×0.69×0.64+0.49×0.31×0.6=0.14+0.0474+0.216384+0.09114)=1-0.50=0.50;
[0110] That is, S 10 The probability of occurrence (open flames continue to burn on each floor and the fire spreads to the entire complex) is 0.5, which means that at this stage the probability of the fire spreading to the entire complex is equivalent to the probability of not spreading.
[0111] (5) State probability of each scenario network node:
[0112] The node probabilities of each scenario state are calculated by calling the API of Netica software in the system, which is consistent with the results of direct calculation, as shown in Table 4;
[0113] Table 4 - Probability distribution of node states in scenario network
[0114]
[0115] The software system developed by the present invention can quickly process the collected multi-source heterogeneous situation information (such as quickly generating a fire situation map by dragging preset common fire elements and firefighting facility icons or hand-drawing) based on the actual needs of fire scene commanders, and quickly draw conclusions after situation deduction to provide auxiliary decision-making support for fire scene commanders. The system fills the existing market gap in terms of information integration, real-time deduction and collaborative operation, and provides a new solution for fire emergency rescue.
[0116] The present invention constructs a Bayesian network model suitable for deducing the development trend of fire situations in commercial complexes, and integrates it into the commander terminal software, so that in the situation information processing stage, the fire commander can process the input situation information based on the model, thereby deriving the fire development trend within a period of time. This deduction result is not only conducive to improving the decision-making efficiency of fire commanders, but also to improving the effectiveness of emergency response. Through preliminary literature research, it was found that the existing related results failed to quickly draw conclusions after situation deduction based on the actual needs of fire commanders, and provide auxiliary decision-making support for fire commanders.
[0117] The following further illustrates the technical advancement effect of the present invention by describing a scenario in which a fire commander uses the software system according to the present invention to handle the fire situation at a commercial complex fire scene.
[0118] In a commercial complex, a maintenance worker performed welding work in violation of regulations. No fire separation was performed on site, and the high-temperature molten material fell into the stairwell on the first floor below ground, generating sparks that ignited the decoration materials and oil stains on components in the elevator shaft, causing the fire to spread (S1 occurred). The worker used a nearby fire extinguisher to put out the fire (R1 occurred), but failed to prevent the fire from spreading further (T1 was not achieved), and the fire spread to the adjacent floors (S3 occurred).
[0119] After receiving the alarm, the first firefighting force in the jurisdiction of the commercial complex launched an initial response on the way to the scene. The fire commander kept in touch with the caller through the on-board communication equipment to obtain preliminary fire information, such as the location of the fire, the size of the fire, and whether there were people trapped.
[0120] After the fire truck arrived at the scene, the fire fighters quickly got off the car, and the commander led the reconnaissance team to conduct a preliminary assessment. Including the location of the fire source, the spread of the fire, the smoke concentration, the evacuation of personnel, etc. At the same time, a command center was set up, and the fire situation information obtained at this stage (the duration of the fire, combustibles, trapped personnel, water supply, etc.) was distributed to each combat team using this situation distribution system. Immediately afterwards, a fire investigation was quickly carried out, and a reconnaissance team was sent into the commercial complex to use thermal imagers, gas detectors and other equipment to find out in detail the location of the fire source, the path of the fire spread, the distribution of smoke, and whether there were people trapped; an external reconnaissance team was sent to operate a fire scene drone to take pictures of the exterior of the complex from a bird's-eye view, and at the same time, people were sent to check the perimeter of the building, evaluate the stability of the building structure, and find possible escape exits and fire water sources; each reconnaissance team will report the collected information to the fire scene commander in a timely manner through this situation distribution system. The fire scene commander will summarize the multi-source heterogeneous situation information received (by dragging and dropping preset common fire elements and firefighting facility icons or by hand-drawing to quickly generate a fire scene situation map; in the preset situation information input interface, quickly input the acquired fire scene situation information), save it and report it to the superior commander, and distribute it to each combat team; formulate a preliminary rescue plan based on the acquired situation information and report it to the superior commander at the same time.
[0121] The fire commander issued orders to each combat team (firefighting team, search and rescue team, evacuation team, water supply team and safety monitoring team); each team quickly arrived at each combat position according to the instructions and joined the battle (R3 occurred). At this time, the fire curtain was not closed in time, and there were a large number of flammable commodities on the scene. The fire quickly spread on this floor (S6). The firefighting team continued to use water guns to extinguish the fire while cleaning the combustibles (R6 occurred), and reported the situation at the fire scene to the fire commander. The commander input the received real-time situation information into this situation distribution system, and used the Bayesian network model embedded in the system to deduce the development of the fire, and concluded that if the fire was not successfully prevented from expanding (T6 was not achieved), the fire would spread to the adjacent floors, causing damage and collapse of building components; the probability of open fire on multiple floors (S8 occurred) was 77%, which was very likely; if the firefighters went inside to extinguish the fire, set up partitions (R8 occurred), and successfully prevented the fire from expanding (T8 was achieved), the fire would be controlled and the fire would tend to weaken (S9 occurred, probability 51%). Otherwise, if the building fire separation facilities are destroyed, the fire will spread to the entire complex (S 11 Occurs, probability 51%).
[0122] According to the above simulation results, the fire commander immediately adjusted the firefighting and rescue plan, promptly dispatched additional firefighting forces to the firefighting team, added a water gun position in the fire area, and used water cannons to extinguish the fire and cool down from the outside, while increasing the water supply guarantee. During this process, each team and fire commander used this system to update and distribute various relevant situation information. The commander reported the adjusted firefighting and rescue plan to the superior commander through this system and communication equipment. At the same time, according to the external road conditions of the complex building in the situation information (including text, images, audio and video), the external road conditions were distributed to the water cannon truck and water supply group through this system and communication equipment, and instructions to adjust the positions of the water cannon truck and water supply vehicles were issued to provide strong water supply guarantee for the newly added water guns and water cannon positions.
[0123] After the above deployment and adjustment, the firefighting force inside the building was strengthened, successfully preventing the fire from expanding (T6 was achieved); the firefighting team blocked the spread of the flames, reducing the trend of the fire, and the fire was under control and burning steadily (S7 occurred); at the same time, water cannons wrapped around the fire from the outside of the building and cooled the burning parts (R7 occurred). After a period of time, the fire gradually decreased, the smoke was effectively suppressed, the fire was under control, and the damage stopped (S 13 occur).
[0124] After all firefighting and rescue activities are completed, the fire scene commander will combine the situation information of the different stages of the entire battle recorded in this system to brief the superior commander, summarize the experience and lessons, and propose improvement measures; the fire scene commander and each combat team will review the performance of each team and summarize the experience and lessons of this battle based on the above situation information and voice communication records.
[0125] It should be noted that, in this article, the term "comprises" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of more restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.
[0126] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A method for processing and distributing fire situation information based on Bayesian network, characterized in that: The method comprises the following steps: Collect situation information: obtain map images of the fire area; Situation information summary: quickly generate multi-source situation information by dragging icons, hand drawing, inputting options and text on the acquired fire area map image; Situation information processing: Based on the Bayesian network model, the fire development trend is deduced according to multi-source situation information; Situation information distribution: Multi-source situation information, images, and videos are distributed and reported between the fire scene commander and various combat positions and superiors.
2. A method for processing and distributing fire situation information based on Bayesian network according to claim 1, characterized in that: The method also includes summarizing and commenting on the entire battle situation after the fire fighting and rescue is completed, and selecting a certain period of time or the entire battle process in the situation playback module corresponding to the fire scene commander terminal or the superior commander terminal to play back the images, videos and audio information sent and received during the period of time.
3. A method for processing and distributing fire situation information based on Bayesian network according to claim 1, characterized in that: The system involved in the method further comprises a sound pickup device, which is arranged near the handheld walkie-talkies of the fire scene commander and the personnel at each combat position, and sends the collected audio information to their terminals; As a result, multi-source situation information, images and videos, and walkie-talkie audio information can be synchronously distributed to the corresponding combat positions or superior commander terminals.
4. A method for processing and distributing fire situation information based on Bayesian network according to claim 1, characterized in that: The Bayesian network model is a Bayesian network model suitable for the deduction of the development trend of fire in commercial complexes. The process of constructing the Bayesian network model and reasoning and calculating the probability of each scenario state node is as follows: (1) Define the problem and variables: The evolution of commercial complex fires is a continuous process. Depending on the effectiveness of the response measures, it will have a positive or negative drive on the development of the fire, making the evolution path develop in the optimistic and pessimistic directions respectively; The evolution of commercial complex fires is affected by three main factors: fire scenario state S, emergency measures M, and disposal target T. The interaction between these factors makes commercial complex fire accidents evolve continuously. According to the pessimistic decision-making criterion, the evolution scenarios of such fire accidents are selected, and finally 13 scenario states (S0, S1, S2, ..., S 13 ), 12 disaster response (R0, R1, R2, ..., R 12 ), 12 emergency response targets (T0, T1, T2, ..., T 12 ), and obtain the scenario table of commercial complex fire accidents. This type of fire will appear S0, S1, S2, ..., S3 over time, from the start of the fire to the rapid development of the fire and then to the complete extinguishing of the fire. n Multiple scenario states, where S0 represents the initial scenario before the fire occurs, S n Indicates the fire end scenario; the initial scenario S0, under the influence of emergency measures R0, regardless of whether the disposal target T0 can be achieved, will develop to the next scenario state until the fire accident disappears S n If every disposal goal can be achieved after the intervention of emergency measures, the fire can be stopped as soon as possible; otherwise, the fire fighting and rescue will be prolonged, causing more serious losses; (2) Building a network structure: By dividing the fire accident scenario states and summarizing the occurrence rules of the same type of accidents based on historical accident cases, the causal relationship of network nodes is obtained; the various network nodes in the commercial complex fire accident scenario table are linked together according to the logical causal relationship to form a commercial complex fire accident scenario evolution path diagram, in which the horizontal direction is the most optimistic evolution path and the vertical direction is the most pessimistic evolution path; (3) Determine the prior probability and conditional probability table: According to the evolution path diagram of commercial complex fire accident scenarios and historical cases of commercial complex fire accidents, the prior probability of the initial event is determined. For network node variables with preceding variables, the conditional probability is obtained through historical case analysis and expert scoring. Based on the pessimistic decision criterion, the Bayesian network node probability is corrected to obtain the prior probability and conditional probability table of network node variables. (4) Model reasoning and probability calculation: First, the fire commander inputs multi-source situation information through the system interface; then, based on the evolution path diagram of the commercial complex fire accident scenario and combined with Formula 1 and the prior probability and conditional probability table of the network node variables, the probability of each scenario is deduced respectively; finally, the fire situation is predicted based on the calculated probability; Formula 1: ; Among them, P(B) is the total probability, P(A) is the prior probability, and P(B|A) is the conditional probability.
5. A fire situation information processing and distribution system based on Bayesian network, characterized in that: The system includes terminal software for multiple different roles in fire fighting and rescue operations, a Bayesian network model for deducing the development situation of a fire scene, and a database and background software running on a cloud server; the system is used to execute a Bayesian network-based fire scene situation information processing and distribution method as described in any one of claims 1 to 4.