Forest fire prediction and response system based on AI

Abstract

The invention discloses an AI-based forest fire prediction and response system, and relates to the technical field of fire prediction and response. Comprising an establishment module which divides a forest into a plurality of areas to obtain a plurality of area items, each area item deploys at least one edge computing network and at least one monitoring system, the monitoring system is used for monitoring the forest in a target area, and a center computing network is deployed and used for receiving data of the edge computing networks. According to the invention, when the edge computing network processes the abnormal condition, the abnormal condition and the processing scheme are sent to the central computing network and the adjacent edge computing network at the same time, and the adjacent edge computing network receives the abnormal condition and the processing scheme. And meanwhile, the abnormal condition and the processing scheme are sent to the central computing network, so that the effects of data distribution and parallel transmission of prediction and response are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of fire prediction and response, and in particular to an AI-based forest fire prediction and response system. Background Art

[0002] A forest fire refers to a fire caused by natural causes such as lightning, drought, or human factors such as open fire and burning of garbage in forests or other vegetation-covered areas. Such fires not only cause serious damage to the ecological environment, but may also threaten human life and property safety. Forest fires not only destroy habitats and lead to a reduction in biodiversity, but also release large amounts of smoke and harmful gases, affecting air quality. The prediction and response to forest fires is an important part of protecting the ecological environment and human safety. Meteorological monitoring uses meteorological data such as temperature, humidity, wind speed, etc. to assess fire risks, and monitors drought index and precipitation to determine the dryness of vegetation.

[0003] Patent publication number CN106934451A is a BP neural network WSN forest fire prevention system based on ant colony optimization. Compared with the traditional forest fire prediction model, the prediction based on the BP neural network has higher accuracy and faster speed. The ant colony algorithm is adopted on the original basic BP neural network, which is beneficial to improve the shortcomings of the BP neural network such as easy to fall into local minima, slow convergence speed and causing oscillation, thereby improving the network convergence speed, with higher prediction accuracy, and integrated wireless sensor network applications, which can more effectively and quickly predict the occurrence trend of forest fires.

[0004] In the process of predicting and responding to forest fires, the above-mentioned and similar technical solutions are prone to delays in signal transmission due to network instability, as most forests are located in mountains and the places where deep forest fires occur. In the process of obtaining an abnormal situation somewhere and transmitting the abnormal situation, the monitoring system may easily experience delays in signal transmission due to network instability. From detecting the abnormality - transmitting signals - receiving signals - issuing instructions, the entire process will be extended indefinitely if the network is unstable, thereby affecting the prediction and response speed of forest fires. Summary of the invention

[0005] The purpose of the present invention is to provide an AI-based forest fire prediction and response system to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: an AI-based forest fire prediction and response system, comprising: Establishing a module: Dividing the forest into multiple regions to obtain multiple regional items, deploying at least one edge computing network and at least one monitoring system for each regional item, wherein the monitoring system is used to monitor the forest in the target region, and deploying a central computing network to receive data from the edge computing network; Acquisition module: sets a division threshold, and based on the division threshold, divides the abnormal situation into at least two levels, obtains abnormal level items, obtains real-time abnormal situations in the area, and classifies the abnormal situations into levels, obtains real-time abnormal items, and schedules resources in the area through scheduling methods; Detection module: Set the detection threshold. The edge computing networks in different areas send detection signals to the central computing network and to each other based on the detection threshold. The central computing network and the edge computing network rank the transmission speed according to the speed of receiving the detection signal, and obtain the central signal ranking item and the edge signal ranking item. Transmission module: While processing abnormal situations, the edge computing network in the target area sends the abnormal situations and processing solutions to the central computing network and the adjacent edge computing network respectively based on the central signal sorting items and the edge signal sorting items, thereby achieving data distribution and parallel transmission effects for prediction and response, and improving data transmission effects; The sending methods include: Step 1: Target determination, obtaining edge computing network information in the target area where abnormal conditions and processing solutions need to be sent, and obtaining target information items; Step 2: Sorting acquisition: based on the target information item and the edge signal sorting item, the sorting level of the target information item and the edge computing network is obtained to obtain the first sorting item; based on the first sorting item and the central signal sorting item, the signal speed sorting between the edge computing network connected to the target information item and the central computing network is obtained to obtain the second sorting item; Step 3: Sorting and combining. Based on the combination of the first sorting item and the second sorting item, a sorting combination item is obtained, and the edge computing network with the highest sorting combination item is selected as the parallel target.

[0007] Furthermore, the method for obtaining the region item includes: S1: area acquisition, obtaining forest area data based on the target device to obtain the forest area item; S2: Area division: set an area threshold, divide the forest area based on the forest area item and the area threshold, and obtain the area item.

[0008] Furthermore, the method for obtaining the region item further includes: S3: Specification limitation, setting the limited specifications, the limited specifications include the length and width of the area item, and obtaining the limited length item and the limited width item; S4: Area allocation: Based on the limited length item and the limited width item, the forest is divided into regions according to the set area threshold to obtain region items.

[0009] Furthermore, the monitoring system comprises: M1: First target device selection, select an infrared detection device as the first target device, and the number of infrared detection devices is at least two; M2: Interval setting: setting a detection interval, and based on the detection interval, obtaining hotspot information and hotspot distribution in the area through the first target device to obtain hotspot distribution items.

[0010] Furthermore, the method for obtaining the abnormal level item includes: setting a division range based on a division threshold, the division threshold includes a hotspot range, obtaining at least two division ranges, and then obtaining at least two division range items, assigning levels according to the division range items, and then obtaining the abnormal level item.

[0011] Furthermore, the monitoring system also includes: second target device selection, selecting a high-definition image acquisition device as the second target device, the number of high-definition image acquisition devices is at least two, and based on the detection interval, the image information in the area is acquired through the second target device to obtain image information items.

[0012] Furthermore, the method for obtaining the abnormal level item also includes: dividing the threshold value including the smoke range, obtaining at least two divided ranges, and then obtaining at least two divided range items, assigning levels according to the divided range items, and then obtaining the abnormal level item.

[0013] Furthermore, the scheduling method includes: Z1: Determine the level and set the matching range. Based on the abnormal level item, set different ranges of scheduling resources for abnormal situations of different levels, obtain at least two resource matching items, and determine the level of the real-time abnormal items based on the real-time abnormal items to obtain the real-time level items, and then obtain the target resource matching items; Z2: Resource acquisition, obtain the location of schedulable resources in the area and obtain the schedulable resource set; Z3: Resource correspondence, based on the target resource matching items and the scheduling resource set, obtain the resource scheduling items, generate the scheduling path, and schedule the resources based on the scheduling path and the resource scheduling items.

[0014] Furthermore, the scheduling method also includes: value-added judgment, based on the scheduling path, obtaining the resource scheduling duration, obtaining the scheduling duration item, predicting the real-time abnormal item based on the scheduling duration item, obtaining the predicted abnormal item, determining the level based on the predicted abnormal item, obtaining the predicted level item, and then obtaining the predicted resource matching item, based on the predicted resource matching item and the scheduling resource set, regenerating the scheduling path, obtaining the updated resource item and the updated path item, and scheduling the resources based on the updated path item and the updated resource item.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This AI-based forest fire prediction and response system divides the forest into multiple areas through edge computing, and deploys an edge computing network and at least one monitoring system in each area. When the monitoring system detects an abnormal situation in the area, the abnormal data is transmitted to the edge computing network, and the edge computing network analyzes the abnormal situation. The edge computing network classifies the abnormal situation based on the AI ​​system and initiates the corresponding response plan. While processing the abnormal situation, the edge computing network sends the abnormal situation and the processing plan to the central computing network and the adjacent edge computing network at the same time. After receiving the abnormal situation and the processing plan, the adjacent edge computing network sends the abnormal situation and the processing plan to the central computing network at the same time, thereby realizing the distribution and parallel transmission of data.

[0016] At the same time, by setting the detection threshold, the edge computing networks in different areas send detection signals to the central computing network at regular intervals based on the detection threshold, and the central computing network sorts the signal levels of the edge computing networks according to the speed of the received detection signals. When the edge computing network in a certain area needs to transmit abnormal situations and processing solutions, the target edge computing network is selected as the distribution target according to the signal sorting items. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of the relationship between the establishment module, the acquisition module, the detection module and the transmission module of the present invention; Figure 2 It is a schematic diagram of the sending method flow of the present invention; Figure 3 It is a schematic diagram of the relationship between the central signal sorting items of the present invention; Figure 4 It is a schematic diagram of the relationship between edge signal sorting items of the present invention; Figure 5 It is a schematic diagram of the sequencing and combination relationship of the present invention; Figure 6 A schematic diagram of the location of a scheduling resource set of the present invention; Figure 7It is a schematic diagram of resource scheduling of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] During the data transmission process, when passing through some areas, the transmission speed will be slower due to poor signals, and when passing through some areas, the data transmission speed will be faster due to strong signals. Therefore, data transmission in a single direction will be interfered with, affecting the timely reception and response of fire prediction information. Therefore, when the signal at the location of a forest fire is poor and the signal transmission is affected, the fire response speed will be affected to a certain extent. The technical solution provided by the present application divides the forest into multiple areas through edge computing, and deploys an edge computing network and at least one monitoring system in each area. When the monitoring system detects an abnormal situation in the area, the abnormal data is transmitted to the edge computing network, and the edge computing network analyzes the abnormal situation. The edge computing network divides the abnormal situation into at least two levels based on the AI ​​system, and each level corresponds to a different response plan. Based on the acquired abnormal situation, the abnormal situation is divided into levels, and the corresponding response plan is initiated. A response path is generated within the area according to the location information of the abnormal situation, and the response plan is responded to within the area. The resources are allocated. When the edge computing network handles the abnormal situation, it sends the abnormal situation and the processing plan to the central computing network and the adjacent edge computing network at the same time. When the adjacent edge computing network receives the abnormal situation and the processing plan, it sends the abnormal situation and the processing plan to the central computing network at the same time. When the edge computing network signal in the target area is unstable, resulting in a slow speed of sending to the central computing network, the adjacent edge computing network with a stable signal sends the abnormal situation and the processing plan to the target area, thereby achieving data distribution and parallel transmission effects, increasing the speed at which the central computing network accepts abnormal situations and processing plans, and by setting a detection threshold, the edge computing networks in different areas send detection signals to the central computing network at regular intervals based on the detection threshold, and the central computing network sorts the signal levels of the edge computing networks according to the speed of the received detection signals. When the edge computing network in a certain area needs to transmit abnormal situations and processing plans, the target edge computing network is selected as the distribution target according to the signal sorting item.

[0020] like Figure 1-Figure 5As shown, the present invention provides a technical solution: an AI-based forest fire prediction and response system, comprising: an establishment module: dividing the forest into multiple areas to obtain multiple area items, each area item deploys at least one edge computing network and at least one monitoring system, the monitoring system is used to monitor the forest in the target area, and deploys a central computing network for receiving data from the edge computing network; an acquisition module: setting a division threshold, and based on the division threshold, dividing the abnormal situation into at least two levels to obtain abnormal level items, obtaining real-time abnormal situations in the area, and classifying the abnormal situations to obtain real-time abnormal items, and scheduling resources in the area through a scheduling method; a detection module: setting a detection threshold, the edge computing networks in different areas send detection signals to the central computing network based on the detection threshold and send detection signals to each other, the central computing network and the edge computing network rank the transmission speed according to the speed of receiving the detection signal, and obtain a central signal ranking item and an edge signal ranking item; a transmission module :While processing abnormal situations, the edge computing network in the target area sends the abnormal situations and processing solutions to the central computing network and the adjacent edge computing network respectively based on the central signal sorting item and the edge signal sorting item, thereby realizing the data distribution and parallel transmission effect of prediction and response, and improving the data transmission effect; the sending method includes: step one: target determination, obtaining the edge computing network information in the target area to which the abnormal situations and processing solutions need to be sent, and obtaining the target information item; step two: sorting acquisition, based on the target information item and the edge signal sorting item, obtaining the sorting level of the target information item and the edge computing network, obtaining the first sorting item, based on the first sorting item and the central signal sorting item, obtaining the signal speed sorting between the edge computing network connected to the target information item and the central computing network, and obtaining the second sorting item; step three: sorting combination, based on the combination of the first sorting item and the second sorting item, obtaining the sorting combination item, and selecting the edge computing network with the highest sorting combination item as the parallel target.

[0021] It should be noted that the forests that need to be predicted and responded to forest fires are divided into regions, and at least one edge computing network and at least one monitoring system are deployed in each region. Due to the use of regionalized monitoring methods, the monitoring effect can be stronger, and the abnormal conditions in the forest are divided into three levels in advance according to the division threshold, namely, slight, moderate and severe. When the monitoring system detects abnormal conditions in the area, the abnormal conditions are graded according to the real-time abnormal conditions obtained, and the relevant resources in the area are scheduled according to the level of the abnormality. This process is completely scheduled by the edge computing network based on the AI ​​model. When the edge computing network obtains the abnormal conditions and generates a processing plan, the abnormal conditions and the processing plan will be sent to the central computing network and the adjacent edge computing network at the same time. The adjacent edge computing network that receives the abnormal conditions and the processing plan will assist in sending the data to the central computing network, realizing the data distribution and parallel transmission effect of the prediction and response, and improving the data transmission effect.

[0022] It should be noted that the detection threshold is set to 10 minutes. Every 10 minutes, all edge computing networks will send detection signals to the central computing network, and will also send detection signals to each other. The central computing network will sort the signals of different edge computing networks according to the time when the detection signal is received, and the edge computing network will sort the signals between the edge computing networks according to the time when the detection signal is received from the adjacent edge computing network. In this way, according to the combined results of the sorting, the best combination is selected to transmit the abnormal situation and processing plan.

[0023] In the specific implementation process, a forest is divided into three areas, namely area 1, area 2 and area 3. Edge computing networks are deployed in each area, namely network 1, network 2 and network 3. According to the set detection interval, network 1, network 2 and network 3 send detection signals to each other, and network 1, network 2 and network 3 send detection signals to the central computing network at the same time. The central computing network first receives the detection signal from network 3 and sets it to speed 1, then receives the detection signal from network 2 and sets it to speed 2, and finally receives the detection signal from network 1 and sets it to speed 3. Network 1 first receives the detection signal from network 3 and sets it to speed 1, then The detection signal from network 2 is received and set to speed 2. Network 2 first receives the detection signal from network 3 and sets it to speed 1. Then it receives the detection signal from network 1 and sets it to speed 2. Network 3 first receives the detection signal from network 1 and sets it to speed 1. Then it receives the detection signal from network 2 and sets it to speed 2. When an abnormality occurs in area 1 and the abnormality and the treatment plan are transmitted to the central computing network, the abnormality and the treatment plan are also transmitted to network 3 at the same time. Network 3 transmits the abnormality and the treatment plan to the central computing network at the same time, thereby realizing the distribution and parallel transmission of data and improving the data transmission effect.

[0024] The method for obtaining the regional item includes: S1: area acquisition, obtaining the area data of the forest based on the target device to obtain the forest area item; S2: area division, setting the area threshold, dividing the forest into regions based on the forest area item and the area threshold, and obtaining the regional item.

[0025] It should be noted that based on remote sensing technology, the overall forest area data is first obtained, and the area threshold is set to 1km². Under the area threshold of 1km², the obtained forest area data is divided. In the specific implementation process, the area of ​​a certain forest is obtained to be 2.6km². Under the area threshold of 1km², the target forest is divided into three areas.

[0026] The method for obtaining the regional item also includes: S3: specification limitation, setting the limited specifications, the limited specifications include the length and width of the regional item, and obtaining the limited length item and the limited width item; S4: area allocation, based on the limited length item and the limited width item, according to the set area threshold, the forest is divided into areas to obtain the regional items.

[0027] It should be noted that the set length item and width item are 1 km respectively, and the forest area is divided according to the set length item and width item.

[0028] The monitoring system includes: M1: first target device selection, selecting an infrared detection device as the first target device, and the number of infrared detection devices is at least two; M2: interval setting, setting the detection interval, based on the detection interval, obtaining the hotspot information and hotspot distribution in the area through the first target device, and obtaining the hotspot distribution item.

[0029] It should be noted that an infrared detection device is used as the first detection device, and the infrared detection device is mounted on a drone. At a set interval of 5 minutes, the drone equipped with the infrared detection device acquires information about the area every 5 minutes to obtain hotspot data in the area.

[0030] The acquisition method of the abnormal level item includes: setting a division range based on a division threshold, the division threshold includes a hotspot range, obtaining at least two division ranges, and then obtaining at least two division range items, assigning levels according to the division range items, and then obtaining the abnormal level item.

[0031] It should be noted that the division threshold is the range of the hotspot. Since the abnormal conditions in the forest are divided into three levels, namely, slight, moderate and severe, there are also three division range items. The hotspot range corresponding to the slight level is 10㎡, the hotspot range corresponding to the moderate level is 10-50㎡, and the hotspot range corresponding to the severe level is above 50㎡.

[0032] In the specific implementation process, a drone equipped with infrared detection equipment obtained a 24-square-meter hotspot at a certain location when acquiring information about the area. According to the classification results of the abnormal level items, it was determined that the abnormal situation in the area was at a medium level.

[0033] The monitoring system also includes: second target device selection, selecting a high-definition image acquisition device as the second target device, the number of high-definition image acquisition devices is at least two, based on the detection interval, the image information in the area is acquired through the second target device to obtain image information items.

[0034] It should be noted that a high-definition image acquisition device is used as the second target device, and the high-definition image acquisition device is mounted on the drone to collect and acquire images in the area at a set interval of 5 minutes.

[0035] The method for obtaining the abnormal level item also includes: dividing the threshold value including the smoke range, obtaining at least two divided ranges, and then obtaining at least two divided range items, assigning levels according to the divided range items, and then obtaining the abnormal level item.

[0036] It should be noted that the threshold for division is the range of smoke. Since the abnormal conditions in the forest are divided into three levels, namely, slight, moderate and severe, there are also three division range items. The smoke range corresponding to the slight level is 20 square meters, the hotspot range corresponding to the moderate level is 20-100 square meters, and the hotspot range corresponding to the severe level is more than 100 square meters.

[0037] In the specific implementation process, a drone equipped with a high-definition camera obtained information about the area and found a 56-square-meter smoke group at a certain location. According to the classification results of the abnormal level items, the abnormal situation in the area was determined to be medium level.

[0038] like Figure 6-Figure 7 As shown, the scheduling method includes: Z1: level determination, setting the matching range, based on the abnormal level item, setting different ranges of scheduling resources for different levels of abnormal situations, obtaining at least two resource matching items, and based on the real-time abnormal item, determining the level of the real-time abnormal item, obtaining the real-time level item, and then obtaining the target resource matching item; Z2: resource acquisition, obtaining the location of the schedulable resources in the area, and obtaining the scheduling resource set; Z3: resource correspondence, based on the target resource matching item and the scheduling resource set, obtaining the resource scheduling item, generating the scheduling path, and scheduling the resources based on the scheduling path and the resource scheduling item.

[0039] It should be noted that the set matching range is the number of alarm stations. Since the abnormal conditions in the forest are divided into three levels, namely, minor, medium and severe, there are also three matching ranges, of which the matching range of the minor level is 2, the matching range of the medium level is 4, and the matching range of the severe level is 6. Three resource matching items are obtained. According to the obtained level of the abnormal condition, the corresponding resource matching items are obtained, and the location of the schedulable resources in the area is obtained at the same time. According to the location of the schedulable resources and the resource matching items, a scheduling path is generated to schedule the resources.

[0040] During the specific implementation process, a medium-level abnormal situation was discovered in a certain area. At this time, alarm stations were set up at different locations in the area. According to the level of the abnormal situation, it was necessary to dispatch the resources of four alarm stations. According to the specific locations of different alarm stations and the principle of proximity, the four nearest alarm stations were selected for resource dispatch.

[0041] The scheduling method also includes: value-added judgment, based on the scheduling path, obtaining the resource scheduling duration, obtaining the scheduling duration item, predicting the real-time abnormal item based on the scheduling duration item, obtaining the predicted abnormal item, judging the level based on the predicted abnormal item, obtaining the predicted level item, and then obtaining the predicted resource matching item, based on the predicted resource matching item and the scheduling resource set, regenerating the scheduling path, obtaining the updated resource item and the updated path item, and scheduling the resources based on the updated path item and the updated resource item.

[0042] It should be noted that the resource scheduling process will take a certain amount of time. At this time, the growth of abnormal situations is predicted based on the resource scheduling time, and the scheduling path is generated again to schedule resources based on the abnormal situation level after the prediction.

[0043] During the specific implementation process, a minor level of abnormality was found in a certain area, and the resources of two alarm stations needed to be dispatched. During the resource scheduling process, the resource scheduling time was calculated. Based on this time, it was predicted that the level of the abnormality would increase to a medium level. Therefore, based on the original scheduling path, a scheduling path was generated again, and the resources of the two alarm stations were dispatched.

[0044] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is limited by the attached embodiments and their equivalents.

Claims

1. An AI-based forest fire prediction and response system, characterized in that: include: Establishing a module: Dividing the forest into multiple regions to obtain multiple regional items, deploying at least one edge computing network and at least one monitoring system for each regional item, wherein the monitoring system is used to monitor the forest in the target region, and deploying a central computing network to receive data from the edge computing network, and deploying the central computing network; Acquisition module: sets a division threshold, and based on the division threshold, divides the abnormal situation into at least two levels, obtains abnormal level items, obtains real-time abnormal situations in the area, and classifies the abnormal situations into levels, obtains real-time abnormal items, and schedules resources in the area through scheduling methods; Detection module: Set the detection threshold. The edge computing networks in different areas send detection signals to the central computing network and to each other based on the detection threshold. The central computing network and the edge computing network rank the transmission speed according to the speed of receiving the detection signal, and obtain the central signal ranking item and the edge signal ranking item. Transmission module: While processing abnormal situations, the edge computing network in the target area sends the abnormal situations and processing solutions to the central computing network and the adjacent edge computing network respectively based on the central signal sorting items and the edge signal sorting items, thereby achieving data distribution and parallel transmission effects for prediction and response, and improving data transmission effects; The sending methods include: Step 1: Target determination, obtaining edge computing network information in the target area where abnormal conditions and processing solutions need to be sent, and obtaining target information items; Step 2: Sorting acquisition: based on the target information item and the edge signal sorting item, the sorting level of the target information item and the edge computing network is obtained to obtain the first sorting item; based on the first sorting item and the central signal sorting item, the signal speed sorting between the edge computing network connected to the target information item and the central computing network is obtained to obtain the second sorting item; Step 3: Sorting and combining. Based on the combination of the first sorting item and the second sorting item, a sorting combination item is obtained, and the edge computing network with the highest sorting combination item is selected as the parallel target.

2. The AI-based forest fire prediction and response system according to claim 1, characterized in that: The method for obtaining the region item includes: S1: area acquisition, obtaining forest area data based on the target device to obtain the forest area item; S2: Area division: set an area threshold, divide the forest area based on the forest area item and the area threshold, and obtain the area item.

3. The AI-based forest fire prediction and response system according to claim 2, characterized in that: The method for obtaining the region item further includes: S3: Specification limitation, setting the limited specifications, the limited specifications include the length and width of the area item, and obtaining the limited length item and the limited width item; S4: Area allocation: Based on the limited length item and the limited width item, the forest is divided into regions according to the set area threshold to obtain region items.

4. The AI-based forest fire prediction and response system according to claim 1, characterized in that: The monitoring system comprises: M1: First target device selection, select an infrared detection device as the first target device, and the number of infrared detection devices is at least two; M2: Interval setting: setting a detection interval, and based on the detection interval, obtaining hotspot information and hotspot distribution in the area through the first target device to obtain hotspot distribution items.

5. The AI-based forest fire prediction and response system according to claim 4, characterized in that: The method for obtaining the abnormal level item includes: setting a division range based on a division threshold, the division threshold includes a hotspot range, obtaining at least two division ranges, and then obtaining at least two division range items, assigning levels according to the division range items, and then obtaining the abnormal level item.

6. The AI-based forest fire prediction and response system according to claim 4, characterized in that: The monitoring system also includes: second target device selection, selecting a high-definition image acquisition device as the second target device, the number of high-definition image acquisition devices is at least two, and based on the detection interval, the image information in the area is acquired by the second target device to obtain image information items.

7. The AI-based forest fire prediction and response system according to claim 1, characterized in that: The method for obtaining the abnormal level item also includes: dividing the threshold value including the smoke range, obtaining at least two divided ranges, and then obtaining at least two divided range items, assigning levels according to the divided range items, and then obtaining the abnormal level item.

8. The AI-based forest fire prediction and response system according to claim 1, characterized in that: The scheduling methods include: Z1: Determine the level and set the matching range. Based on the abnormal level item, set different ranges of scheduling resources for abnormal situations of different levels, obtain at least two resource matching items, and determine the level of the real-time abnormal items based on the real-time abnormal items to obtain the real-time level items, and then obtain the target resource matching items; Z2: Resource acquisition, obtain the location of schedulable resources in the area and obtain the schedulable resource set; Z3: Resource correspondence, based on the target resource matching items and the scheduling resource set, obtain the resource scheduling items, generate the scheduling path, and schedule the resources based on the scheduling path and the resource scheduling items.

9. The AI-based forest fire prediction and response system according to claim 8, characterized in that: The scheduling method also includes: value-added judgment, based on the scheduling path, obtaining the resource scheduling duration, obtaining the scheduling duration item, predicting the real-time abnormal item based on the scheduling duration item, obtaining the predicted abnormal item, judging the level based on the predicted abnormal item, obtaining the predicted level item, and then obtaining the predicted resource matching item, based on the predicted resource matching item and the scheduling resource set, regenerating the scheduling path, obtaining the updated resource item and the updated path item, and scheduling the resources based on the updated path item and the updated resource item.

Images