Unmanned aerial vehicle automatic bomb dropping fire extinguishing method, device and equipment based on Beidou positioning

Abstract

The invention provides an unmanned aerial vehicle automatic bomb dropping fire extinguishing method, device and equipment based on Beidou positioning, and the method comprises the steps: collecting region feature data of a plurality of sub-regions of a to-be-monitored region according to a preset time interval in a first preset time period, and representing the region feature data of each time point through a first feature sequence; preprocessing the first feature sequence to obtain a second feature sequence; constructing a region graph structure according to the second feature sequence; based on the regional graph structure, constructing a fire behavior prediction model and predicting the size of a fire behavior, and outputting fire behavior scores of a plurality of sub-regions within a second preset time period; according to the invention, based on the fire behavior score and Beidou positioning, fire extinguishing is carried out through automatic bomb dropping of the unmanned aerial vehicle, fire behavior monitoring is carried out in real time, and when the fire behavior is monitored, fire extinguishing is carried out in time through the unmanned aerial vehicle, so that the timeliness of fire behavior detection is ensured, and the fire extinguishing efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fire fighting and fire extinguishing, and in particular to a method, device and equipment for automatic bomb dropping and fire extinguishing by unmanned aerial vehicles based on Beidou positioning. Background Art

[0002] At present, the main means of firefighting in high-rise buildings are high-pressure water guns and ladders equipped by fire trucks. However, even the most advanced fire ladder trucks have a maximum extension height of no more than 100 meters, and the ladders cannot reach higher-rise buildings. In addition, the deployment and use of the ladders require a certain amount of time and space. On narrow urban roads, the operation of the ladders will be hindered. For high-rise buildings, there may also be problems with insufficient water pressure, which limits the effectiveness of traditional firefighting methods.

[0003] Therefore, how to detect fire in time and improve the efficiency of firefighting has become a technical problem that technical personnel in this field urgently need to solve. Summary of the invention

[0004] The present invention provides a method, device and equipment for automatic bomb dropping and fire extinguishing by unmanned aerial vehicle based on Beidou positioning, which are used to solve the defects of untimely fire detection and low fire fighting efficiency in the prior art.

[0005] In a first aspect, the present invention provides a method for automatically dropping bombs and extinguishing fires by a drone based on Beidou positioning, comprising: Within a first preset time period, regional characteristic data of a plurality of sub-regions of the area to be monitored are collected at preset time intervals, and the regional characteristic data at each time point is represented by a first characteristic sequence; Preprocessing the first feature sequence to obtain a second feature sequence; constructing a regional graph structure according to the second feature sequence; Based on the regional graph structure, a fire prediction model is constructed to predict the fire size, and the fire scores of the multiple sub-regions within the second preset time period are output; Based on the size of the fire score and Beidou positioning, firefighting is carried out by automatically dropping bombs through drones.

[0006] According to a method for automatic bombing and fire extinguishing by a drone based on Beidou positioning provided by the present invention, the regional feature data includes: image data and sensor data, and the sensor data includes: temperature, humidity, wind speed, wind direction and smoke concentration; The preprocessing of the first feature sequence to obtain a second feature sequence includes: The image data, the temperature, the humidity, the wind speed, the wind direction and the smoke density in the first feature sequence are screened respectively, and the missing values ​​are filled by taking the average value of the value of the previous sequence unit and the value of the next sequence unit; For the data after the filling process, a sliding time window method is used to determine the local average value, create a local average value feature, and splice it after the regional feature data; The spliced ​​regional feature data are normalized to obtain a second feature sequence.

[0007] According to a method for automatic bombing and fire extinguishing by a drone based on Beidou positioning provided by the present invention, the step of constructing a regional graph structure according to the second feature sequence includes: Using the regional feature data of the sub-region in the second feature sequence as the initial feature of the node, establishing a mapping relationship between the sub-region and the node; A region graph structure is constructed based on the mapping relationship, wherein the region graph structure includes nodes, initial features of the nodes, and edges.

[0008] According to a method for automatic bombing and fire extinguishing by a drone based on Beidou positioning provided by the present invention, the edge in the construction area graph structure includes: Build bidirectional edges between adjacent sub-regions and set basic weights; When the angle between the line connecting the center points of the adjacent sub-areas and the wind direction is less than a first threshold, constructing a unidirectional edge between the sub-areas along the wind direction; Calculating the Manhattan distance of the initial features of the adjacent sub-regions, and when the Manhattan distance is less than a second threshold, constructing a bidirectional edge between the two adjacent sub-regions; The Manhattan distance is calculated as follows: ; in, represents the Manhattan distance between two sub-regions, k represents the dimension of the regional feature data of the sub-region, and are the values ​​of the i-th dimension of two adjacent sub-regions respectively; The edge weight is set according to the Manhattan distance, which is determined as follows: ; in, represents the weight of edge ab, Indicates the second threshold.

[0009] According to a method for automatically dropping bombs to extinguish fire by a drone based on Beidou positioning provided by the present invention, the fire prediction model includes a first hidden layer, a splicing layer, a second hidden layer, a combination layer, a third hidden layer, a segmentation layer and a first classifier; The first hidden layer includes M*N first units, the m*nth first unit is used to input the image data of the mth sub-region at the nth time point, and output the first updated feature; The splicing layer includes M*N splicers, the m*nth splicer is used to expand the first updated feature into a vector representation and splice it with the sensor data of the mth sequence unit of the second feature sequence to obtain a splicing vector, and use the splicing vector as the initial feature of the node; The second hidden layer includes N second units, and the nth second unit is used to input the initial features of the nodes and output the second updated features of the M nodes; The combination layer includes N combination units, and the nth combination unit is used to combine the second updated features into a combination vector; The third hidden layer includes N third units, and the nth third unit is used to input the combination vector and output a third updated feature; The segmentation layer is used to segment the Nth third update feature into M fourth update features, each of which corresponds to a node update feature of a node; Inputting each of the node update features into the first classifier, the classification space of the first classifier is used to represent the fire severity score of each sub-area; Among them, m, n, M, and N are natural numbers.

[0010] According to a method for automatic bombing and fire extinguishing by a drone based on Beidou positioning provided by the present invention, the second hidden layer is: ; ; ; in, represents the second updated feature of node i, and They all represent the initial features of the nodes. represents the normalized attention weight between the node i of the lth attention head and its adjacent node j, represents the set of nodes adjacent to node i, ReLU represents the ReLU activation function, represents the attention coefficient between node i of the lth attention head and its adjacent node j, represents the attention coefficient between node i of the lth attention head and its adjacent node k, LeakyReLU represents the LeakyReLU activation function, exp represents the exponential function, T represents the transposition operation, and concat represents the concatenation operation. , , and They respectively represent the first weight parameter, the second weight parameter, the third weight parameter and the fourth weight parameter of the lth attention head.

[0011] According to the present invention, a method for automatically dropping bombs to extinguish fire using a drone based on Beidou positioning also includes: Generate label data through the generative adversarial network model and train the fire prediction model; The generative adversarial network model includes a generator and a discriminator, and generating label data includes: first fixing the generator and training the discriminator, then fixing the discriminator and training the generator, using the cross entropy function as the loss function to update the generator and the discriminator, and using the generated data finally obtained as the label data of the fire prediction model; The generator takes as input a random noise vector containing a timestamp and outputs a new regional graph structure and a fire severity score of the corresponding node; The discriminator inputs the collected real data or the data generated by the generator, and outputs the probability value that the input data is the real data.

[0012] According to a method for automatically dropping bombs to extinguish a fire by a drone based on Beidou positioning provided by the present invention, firefighting is extinguished by automatically dropping bombs by a drone based on the fire severity score and Beidou positioning, including: When the fire severity score indicates level one, continuously monitoring the corresponding sub-area; When the fire severity score is level 2, a firefighting instruction is sent to the drone, and the drone is controlled to carry a first number of fire extinguishing bombs and to be delivered to the corresponding sub-area through Beidou positioning navigation; When the fire severity score is level three, a firefighting instruction is sent to the drone, and the drone is controlled to carry a second number of fire extinguishing bombs and to be deployed at multiple points in the corresponding sub-area through Beidou positioning navigation; When the fire severity score is level 4, a firefighting instruction is sent to the drone, and the drone is controlled to carry a third number of fire extinguishing bombs and navigate to the corresponding sub-area through Beidou positioning for coverage and delivery; Among them, the first level means no fire, the second level fire is smaller than the third level fire, the third level fire is smaller than the fourth level fire, the first number is smaller than the second number, and the second number is smaller than the third number.

[0013] In a second aspect, the present invention provides a drone automatic bombing and fire extinguishing device based on Beidou positioning, comprising: A monitoring module, configured to collect regional characteristic data of a plurality of sub-regions of a to-be-monitored region at preset time intervals within a first preset time period, and represent the regional characteristic data at each time point by a first characteristic sequence; A construction module, used for preprocessing the first feature sequence to obtain a second feature sequence; and constructing a region map structure according to the second feature sequence; A prediction module, configured to construct a fire intensity prediction model and predict the fire intensity based on the regional map structure, and output fire intensity scores of the plurality of sub-regions within a second preset time period; The fire fighting module is used to automatically drop bombs by drones to extinguish fires based on the size of the fire score and Beidou positioning.

[0014] In a third aspect, the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements any of the above-mentioned methods for automatic bombing and fire extinguishing by unmanned aerial vehicles based on Beidou positioning.

[0015] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for automatic bombing and fire extinguishing by a UAV based on Beidou positioning.

[0016] In a fifth aspect, the present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned Beidou positioning-based unmanned aerial vehicle automatic bombing and fire extinguishing methods.

[0017] The present invention provides a method, device and equipment for automatic bomb dropping and fire extinguishing by unmanned aerial vehicles based on Beidou positioning. Within a first preset time period, regional feature data of multiple sub-areas of a to-be-monitored area are collected at preset time intervals, and the regional feature data of each time point are represented by a first feature sequence; the first feature sequence is preprocessed to obtain a second feature sequence; a regional map structure is constructed according to the second feature sequence; based on the regional map structure, a fire prediction model is constructed and the size of the fire is predicted, and the fire scores of the multiple sub-areas within the second preset time period are output; based on the size of the fire score and Beidou positioning, firefighting is extinguished by automatic bomb dropping by unmanned aerial vehicles, and fire monitoring is performed in real time. When a fire is detected, firefighting is extinguished in a timely manner by means of unmanned aerial vehicles, which not only ensures the timeliness of fire detection, but also improves the efficiency of firefighting. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 : is a flowchart of a method for automatically dropping bombs to extinguish a fire using a drone based on Beidou positioning provided in this embodiment; Figure 2 : is a schematic diagram of the structure of the automatic bombing fire extinguishing device of the unmanned aerial vehicle based on Beidou positioning provided in this embodiment; Figure 3 It is a schematic diagram of the structure of the electronic device provided in this embodiment. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.

[0021] Figure 1 It is a flow chart of the method for automatic fire extinguishing by drone dropping bombs based on Beidou positioning provided in this embodiment.

[0022] like Figure 1 As shown, the automatic bombing and fire extinguishing method of a drone based on Beidou positioning provided by an embodiment of the present invention mainly includes the following steps: 101. Within a first preset time period, collect regional characteristic data of multiple sub-regions of a to-be-monitored region at preset time intervals, and represent the regional characteristic data at each time point by a first characteristic sequence.

[0023] In a specific implementation process, the data acquisition module can collect regional feature data of M sub-areas of the monitored area within a first preset time period T and at a preset time interval t, and represent the regional feature data at each time point by a first feature sequence.

[0024] Among them, the mth sequence unit of the first feature sequence at the nth time point represents the regional feature data of the mth sub-region at the nth time point, where 1≤m≤M, 1≤n≤N, N=T / t, m, n, M and N are all natural numbers, T can be set to 5 minutes, and t can be set to 10 seconds.

[0025] The regional characteristic data includes image data and sensor data. The sensor data includes temperature, humidity, wind speed, wind direction and smoke concentration, etc. The image data of the monitored area can be collected by ground fire monitoring cameras, and the sensor data of the monitored area can be collected by arranging temperature sensors, humidity sensors, wind direction sensors, wind speed sensors and smoke concentration sensors.

[0026] 102. Preprocess the first feature sequence to obtain a second feature sequence.

[0027] After obtaining the first feature sequence, the image data, temperature, humidity, wind speed, wind direction and smoke concentration in the first feature sequence are screened respectively, and the missing values ​​are filled by taking the average value of the previous sequence unit and the next sequence unit; for the data after the filling process, the local average value is determined by the sliding time window method, and the local average value feature is created and spliced ​​after the regional feature data; the spliced ​​regional feature data is normalized to the maximum and minimum to obtain the second feature sequence.

[0028] Among them, the sliding time window method can be to determine the time window size as 30 seconds, that is, a time window composed of 3 time points; place the time window at the starting position of the sensor data of the regional feature data of M sub-areas at N time points; calculate the average value of the data in each dimension in the window, which is used as the local average feature of the third time point; move the window forward by one time point, and continue to calculate the average value of the data in each dimension in the window; repeat the previous step until the data ends; the data of the first two time points are empty, and the data of the third time point is used to fill the data, so as to determine the local average value.

[0029] 103. Construct a regional graph structure according to the second feature sequence.

[0030] The regional graph structure includes: nodes, initial features of nodes and edges. Specifically, the regional feature data of the sub-region in the second feature sequence is used as the initial feature of the node, and a mapping relationship between the sub-region and the node is established; and the regional graph structure is constructed based on the mapping relationship.

[0031] Among them, constructing the edge in the regional graph structure includes: constructing a bidirectional edge between adjacent sub-regions and setting a basic weight, which can be set to 1. When the angle between the line connecting the center points of adjacent sub-regions and the wind direction is less than a first threshold, construct a unidirectional edge between the sub-regions along the wind direction. That is, if the wind direction is from sub-region A to sub-region B, then construct a unidirectional edge from A to B. Preferably, the first threshold is set to 30 degrees and the edge weight is set to 2.

[0032] The Manhattan distance of the initial features of adjacent sub-regions is calculated, and when the Manhattan distance is less than the second threshold, a bidirectional edge is constructed between the two adjacent sub-regions.

[0033] The calculation method of Manhattan distance is (1): (1) in, represents the Manhattan distance between two sub-regions, k represents the dimension of the regional feature data of the sub-region, and are the values ​​of the i-th dimension of two adjacent sub-regions respectively; express and The absolute value of the difference between the two, the second threshold can be set to 5.

[0034] Then set the edge weight according to the Manhattan distance, and the determination method is (2): (2) in, represents the weight of edge ab, Represents the second threshold. The comprehensive weight of an edge is the sum of the weights obtained by the edge construction method.

[0035] 104. Based on the regional graph structure, a fire prediction model is constructed to predict the fire size, and the fire scores of multiple sub-regions within a second preset time period are output.

[0036] The input of the fire prediction model is a regional graph structure, and the output is the fire scores of M sub-regions within a second preset time period G. The fire prediction model includes a first hidden layer, a concatenation layer, a second hidden layer, a combination layer, a third hidden layer, a segmentation layer and a first classifier.

[0037] Among them, the first hidden layer includes M*N first units, the m*nth first unit is used to input the image data of the mth sub-region at the nth time point, and output the first updated feature; the splicing layer includes M*N splicers, the m*nth splicer is used to expand the first updated feature into a vector representation and splice it with the sensor data of the mth sequence unit of the second feature sequence to obtain a splicing vector, and the splicing vector is used as the initial feature of the node; the second hidden layer includes N second units, the nth second unit is used to input the initial feature of the node, and output the second updated feature of M nodes; the combination layer includes N combination units, and the nth combination unit is used to combine the second updated features into a combination vector; the third hidden layer includes N third units, and the nth third unit is used to input the combination vector and output the third updated feature; the segmentation layer is used to segment the Nth third update feature into M fourth update features, each fourth update feature corresponds to the node update feature of a node; each node update feature is input into the first classifier, and the classification space of the first classifier is used to represent the fire score of each sub-region. The fire severity score can be an integer ranging from 0 to 10.

[0038] The fire prediction model first updates the node features of the regional graph structure at each time point, and then makes predictions based on the feature changes at different time points. The first unit is built based on a convolutional neural network, and the third unit is built based on a long short-term memory network unit. The third unit can also be built based on a gated recurrent unit or a recurrent neural unit.

[0039] The second hidden layer is: (3) (4) (5) in, represents the second updated feature of node i, and They all represent the initial features of the nodes. represents the normalized attention weight between the node i of the lth attention head and its adjacent node j, represents the set of nodes adjacent to node i, ReLU represents the ReLU activation function, represents the attention coefficient between node i of the lth attention head and its adjacent node j, represents the attention coefficient between node i of the lth attention head and its adjacent node k, LeakyReLU represents the LeakyReLU activation function, exp represents the exponential function, T represents the transposition operation, and concat represents the concatenation operation. , , and They respectively represent the first weight parameter, the second weight parameter, the third weight parameter and the fourth weight parameter of the lth attention head.

[0040] Before training the fire prediction model, the first unit is pre-trained. During the pre-training process, the first updated feature output by the first unit is expanded into a vector representation and then input into the second classifier. The classification space of the second classifier represents whether there are signs of burning in the objects in the area, and the classification labels of the second classifier are manually marked.

[0041] Then, the label data is generated by the generative adversarial network model, and the fire prediction model is trained. The generative adversarial network model includes a generator and a discriminator. The generation of label data includes: first fixing the generator and training the discriminator, then fixing the discriminator and training the generator, using the cross entropy function as the loss function to update the generator and the discriminator, and using the generated data finally obtained as the label data of the fire prediction model; the input of the generator is a random noise vector containing a timestamp, and the output is a new regional graph structure and the fire score of the corresponding node; the discriminator inputs the collected real data or the data generated by the generator, and outputs the probability value of the input data being the real data.

[0042] Finally, the fire prediction model is used to timely and accurately predict the fire situation in each sub-area and output the fire score. Different fire scores represent the presence and size of the fire, thus ensuring that when a fire occurs, it can be discovered in time and corresponding fire-fighting measures can be implemented in time.

[0043] 105. Based on the fire severity score and Beidou positioning, firefighting is carried out by automatically dropping bombs through drones.

[0044] Specifically, firefighting measures are generated according to the fire intensity scores of the M sub-areas within the second preset time period G. Different fire intensity scores correspond to different firefighting measures, including different numbers of fire-extinguishing bombs and different placement locations. Use drones to carry fire-extinguishing bombs and drop them into the fire area to achieve a fire-extinguishing effect. The drone includes a Beidou positioning module, a visual image recognition module, and a data transmission module. After a fire occurs, the visual image recognition module of the drone is used to monitor the fire in the fire area in real time, and the image is transmitted back to the command center, and the command center is controlled to carry out firefighting. T, t, M, m, N, n, and G are custom parameters.

[0045] For example, the fire severity score can be divided into four levels, 0 to 1 for level 1, indicating no fire, 2 to 3 for level 2, indicating a small fire, 4 to 6 for level 3, indicating a medium fire, and 7 to 10 for level 4, indicating a large fire. Level 2 fire severity is smaller than level 3 fire severity, level 3 fire severity is smaller than level 4 fire severity, the first number is smaller than the second number, and the second number is smaller than the third number.

[0046] Therefore, when the fire severity score indicates level one, the firefighting measure is to take no action and continue to monitor the corresponding sub-area.

[0047] When the fire severity score is level 2, a firefighting command is sent to the drone, which controls the drone to carry the first number of fire-extinguishing bombs, such as 1-2 fire-extinguishing bombs, and uses Beidou positioning navigation to drop them in the corresponding sub-area. This ensures that the fire-extinguishing bombs can hit the corresponding area of ​​the small fire and extinguish the fire in time.

[0048] When the fire severity score is level three, a firefighting command is sent to the drone, which controls the drone to carry the second number of fire-extinguishing bombs, such as 3-5 fire-extinguishing bombs, and uses Beidou positioning navigation to carry out multi-point delivery to the corresponding sub-area. In addition to extinguishing the fire at the starting point, multi-point delivery can also cool down and isolate the surrounding area to prevent the fire from further spreading, thereby ensuring the efficiency of fire extinguishing.

[0049] When the fire severity score is level 4, a firefighting command is sent to the drone, which controls the drone to carry the third number of fire-extinguishing bombs, such as 6-10 fire-extinguishing bombs, and uses Beidou positioning navigation to cover and deploy them in the corresponding sub-area. This not only ensures that the fire point is extinguished, but also covers the preset range of the fire point, ensuring the speed of fire extinguishing and improving the efficiency of fire extinguishing.

[0050] Carrying different numbers of fire extinguishing bombs by drones and delivering them in different ways can not only ensure the efficiency of fire fighting, but also effectively solve the problem of fire trucks being unable to reach high-rise buildings and narrow urban road areas quickly.

[0051] Based on the same general inventive concept, the present invention also protects an automatic bomb dropping fire extinguishing device for unmanned aerial vehicles based on Beidou positioning. The automatic bomb dropping fire extinguishing device for unmanned aerial vehicles based on Beidou positioning provided by the present invention is described below. The automatic bomb dropping fire extinguishing device for unmanned aerial vehicles based on Beidou positioning described below and the automatic bomb dropping fire extinguishing method for unmanned aerial vehicles based on Beidou positioning described above can be referenced to each other.

[0052] Figure 2 It is a structural schematic diagram of the automatic bomb dropping fire extinguishing device of the unmanned aerial vehicle based on Beidou positioning provided in this embodiment.

[0053] like Figure 2 As shown, this embodiment provides a drone automatic bombing fire extinguishing device based on Beidou positioning, including: Monitoring module 201, used for collecting regional characteristic data of multiple sub-regions of the monitored region at preset time intervals within a first preset time period, and representing the regional characteristic data at each time point by a first characteristic sequence; A construction module 202 is used to preprocess the first feature sequence to obtain a second feature sequence; and to construct a regional graph structure according to the second feature sequence; Prediction module 203, used to construct a fire intensity prediction model and predict the fire intensity based on the regional graph structure, and output the fire intensity scores of multiple sub-regions within the second preset time period; The fire fighting module 204 is used to extinguish fires by automatically dropping bombs through drones based on the fire severity score and Beidou positioning.

[0054] Figure 3 It is a schematic diagram of the structure of the electronic device provided in this embodiment.

[0055] like Figure 3As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute the automatic bombing fire extinguishing method of the unmanned aerial vehicle based on Beidou positioning, the method comprising: within a first preset time period, collecting regional feature data of multiple sub-areas of the monitored area according to a preset time interval, and representing the regional feature data at each time point by a first feature sequence; preprocessing the first feature sequence to obtain a second feature sequence; constructing a regional map structure according to the second feature sequence; based on the regional map structure, constructing a fire prediction model and predicting the size of the fire, and outputting the fire scores of the multiple sub-areas within the second preset time period; based on the size of the fire score and Beidou positioning, automatically dropping bombs by the unmanned aerial vehicle to extinguish the fire.

[0056] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0057] On the other hand, the present invention also provides a computer program product, which includes a computer program, and the computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the Beidou positioning-based drone automatic bombing fire extinguishing method provided by the above methods, and the method includes: within a first preset time period, collecting regional feature data of multiple sub-areas of the monitored area at preset time intervals, and representing the regional feature data at each time point through a first feature sequence; preprocessing the first feature sequence to obtain a second feature sequence; constructing a regional map structure according to the second feature sequence; based on the regional map structure, constructing a fire prediction model and predicting the size of the fire, and outputting the fire scores of the multiple sub-areas within the second preset time period; based on the size of the fire score and Beidou positioning, fire fighting is carried out by automatically dropping bombs by drones.

[0058] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it is implemented to execute the Beidou positioning-based drone automatic bombing fire extinguishing method provided by the above-mentioned methods. The method includes: within a first preset time period, collecting regional feature data of multiple sub-areas of the monitored area at preset time intervals, and representing the regional feature data at each time point by a first feature sequence; preprocessing the first feature sequence to obtain a second feature sequence; constructing a regional map structure according to the second feature sequence; based on the regional map structure, constructing a fire prediction model and predicting the size of the fire, and outputting the fire scores of the multiple sub-areas within the second preset time period; based on the size of the fire score and Beidou positioning, firefighting is carried out by automatically dropping bombs by drones.

[0059] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0060] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for automatic bombing and fire extinguishing by drone based on Beidou positioning, characterized in that: include: Within a first preset time period, regional characteristic data of a plurality of sub-regions of the area to be monitored are collected at preset time intervals, and the regional characteristic data at each time point is represented by a first characteristic sequence; Preprocessing the first feature sequence to obtain a second feature sequence; constructing a regional graph structure according to the second feature sequence; Based on the regional graph structure, a fire prediction model is constructed to predict the fire size, and the fire scores of the multiple sub-regions within the second preset time period are output; Based on the size of the fire score and Beidou positioning, firefighting is carried out by automatically dropping bombs through drones.

2. The method for automatic bombing and fire extinguishing by unmanned aerial vehicle based on Beidou positioning according to claim 1 is characterized in that: The regional characteristic data includes: image data and sensor data, and the sensor data includes: temperature, humidity, wind speed, wind direction and smoke concentration; The preprocessing of the first feature sequence to obtain a second feature sequence includes: The image data, the temperature, the humidity, the wind speed, the wind direction and the smoke density in the first feature sequence are screened respectively, and the missing values ​​are filled by taking the average value of the value of the previous sequence unit and the value of the next sequence unit; For the data after the filling process, a sliding time window method is used to determine the local average value, create a local average value feature, and splice it after the regional feature data; The spliced ​​regional feature data are normalized to obtain a second feature sequence.

3. The method for automatic bombing and fire extinguishing by unmanned aerial vehicle based on Beidou positioning according to claim 1 is characterized in that: The step of constructing a regional map structure according to the second feature sequence comprises: Using the regional feature data of the sub-region in the second feature sequence as the initial feature of the node, establishing a mapping relationship between the sub-region and the node; A region graph structure is constructed based on the mapping relationship, wherein the region graph structure includes nodes, initial features of the nodes, and edges.

4. The method for automatic bombing and fire extinguishing by unmanned aerial vehicle based on Beidou positioning according to claim 3 is characterized in that: The constructing of the edges in the regional graph structure includes: Build bidirectional edges between adjacent sub-regions and set basic weights; When the angle between the line connecting the center points of the adjacent sub-areas and the wind direction is less than a first threshold, constructing a unidirectional edge between the sub-areas along the wind direction; Calculating the Manhattan distance of the initial features of the adjacent sub-regions, and when the Manhattan distance is less than a second threshold, constructing a bidirectional edge between the two adjacent sub-regions; The Manhattan distance is calculated as follows: ； in, represents the Manhattan distance between two sub-regions, k represents the dimension of the regional feature data of the sub-region, and are the values ​​of the i-th dimension of two adjacent sub-regions respectively; The edge weight is set according to the Manhattan distance, which is determined as follows: ； in, represents the weight of edge ab, Indicates the second threshold.

5. The method for automatic fire extinguishing by drone dropping bombs based on Beidou positioning according to claim 1 is characterized in that: The fire prediction model includes a first hidden layer, a splicing layer, a second hidden layer, a combination layer, a third hidden layer, a segmentation layer and a first classifier; The first hidden layer includes M*N first units, the m*nth first unit is used to input the image data of the mth sub-region at the nth time point, and output the first updated feature; The splicing layer includes M*N splicers, the m*nth splicer is used to expand the first updated feature into a vector representation and splice it with the sensor data of the mth sequence unit of the second feature sequence to obtain a splicing vector, and use the splicing vector as the initial feature of the node; The second hidden layer includes N second units, and the nth second unit is used to input the initial features of the nodes and output the second updated features of the M nodes; The combination layer includes N combination units, and the nth combination unit is used to combine the second updated features into a combination vector; The third hidden layer includes N third units, and the nth third unit is used to input the combination vector and output a third updated feature; The segmentation layer is used to segment the Nth third update feature into M fourth update features, each of which corresponds to a node update feature of a node; Inputting each of the node update features into the first classifier, the classification space of the first classifier is used to represent the fire severity score of each sub-area; Among them, m, n, M, and N are natural numbers.

6. The method for automatic bombing and fire extinguishing by unmanned aerial vehicle based on Beidou positioning according to claim 5 is characterized in that: The second hidden layer is: ； ； ； in, represents the second updated feature of node i, and They all represent the initial features of the nodes. represents the normalized attention weight between the node i of the lth attention head and its adjacent node j, represents the set of nodes adjacent to node i, ReLU represents the ReLU activation function, represents the attention coefficient between node i of the lth attention head and its adjacent node j, represents the attention coefficient between node i of the lth attention head and its adjacent node k, LeakyReLU represents the LeakyReLU activation function, exp represents the exponential function, T represents the transposition operation, and concat represents the concatenation operation. , , and They respectively represent the first weight parameter, the second weight parameter, the third weight parameter and the fourth weight parameter of the lth attention head.

7. The method for automatic fire extinguishing by drone dropping bombs based on Beidou positioning according to claim 1 is characterized in that: Also includes: Generate label data through the generative adversarial network model and train the fire prediction model; The generative adversarial network model includes a generator and a discriminator, and generating label data includes: first fixing the generator and training the discriminator, then fixing the discriminator and training the generator, using the cross entropy function as the loss function to update the generator and the discriminator, and using the generated data finally obtained as the label data of the fire prediction model; The generator takes as input a random noise vector containing a timestamp and outputs a new regional graph structure and a fire severity score of the corresponding node; The discriminator inputs the collected real data or the data generated by the generator, and outputs the probability value that the input data is the real data.

8. The method for automatic fire extinguishing by drone dropping bombs based on Beidou positioning according to any one of claims 1 to 7, characterized in that: Based on the fire severity score and Beidou positioning, firefighting is carried out by automatically dropping bombs through drones, including: When the fire severity score indicates level one, continuously monitoring the corresponding sub-area; When the fire severity score is level 2, a firefighting instruction is sent to the drone, and the drone is controlled to carry a first number of fire extinguishing bombs and to be delivered to the corresponding sub-area through Beidou positioning navigation; When the fire severity score is level three, a firefighting instruction is sent to the drone, and the drone is controlled to carry a second number of fire extinguishing bombs and to be deployed at multiple points in the corresponding sub-area through Beidou positioning navigation; When the fire severity score is level 4, a firefighting instruction is sent to the drone, and the drone is controlled to carry a third number of fire extinguishing bombs and navigate to the corresponding sub-area through Beidou positioning for coverage and delivery; Among them, the first level means no fire, the second level fire is smaller than the third level fire, the third level fire is smaller than the fourth level fire, the first number is smaller than the second number, and the second number is smaller than the third number.

9. A drone automatic bombing fire extinguishing device based on Beidou positioning, characterized in that: include: A monitoring module, configured to collect regional characteristic data of a plurality of sub-regions of a to-be-monitored region at preset time intervals within a first preset time period, and represent the regional characteristic data at each time point by a first characteristic sequence; A construction module, used for preprocessing the first feature sequence to obtain a second feature sequence; and constructing a region map structure according to the second feature sequence; A prediction module, configured to construct a fire intensity prediction model and predict the fire intensity based on the regional map structure, and output fire intensity scores of the plurality of sub-regions within a second preset time period; The fire fighting module is used to automatically drop bombs by drones to extinguish fires based on the size of the fire score and Beidou positioning.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the automatic bombing and fire extinguishing method of a drone based on Beidou positioning as described in any one of claims 1 to 8 is implemented.

Images