A High-Order Distributed Firefighting UAV Task Allocation Method and System

Abstract

The present invention provides a high-order distributed fire-fighting drone task allocation method and system, which relates to the technical field of drone task allocation, and includes the following steps: obtaining thermal infrared images and necessary parameters; calculating the length and width of the coverage area corresponding to a single pixel point of the thermal infrared image; converting the thermal infrared image into a grayscale image, performing binary processing on the fire area, and calculating the area of the fire area; extracting the boundary of the fire area, establishing a local coordinate system, and calculating the coordinates of the fire area boundary and the fire-fighting drones; distributed collaborative decision-making and task allocation. The beneficial effects of the present invention are as follows: avoiding the single-point failure problem in the centralized mode, having high system reliability and robustness, efficient information interaction and collaboration, simplifying the decision-making process, avoiding repeated task allocation, and being able to adjust the allocation strategy according to the dynamic changes of the fire area.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) mission allocation, and particularly to a high-order distributed fire-fighting UAV mission allocation method and system. Background Art

[0002] With the deep penetration of UAV technology in the fire-fighting field, the UAV mission allocation technology for fire-fighting has become a key research direction for improving the efficiency of fire fighting. In recent years, research in this field has focused on multi-UAV collaborative path planning and task scheduling algorithms, integrating geographic information systems, intelligent optimization algorithms, etc., aiming to optimize the allocation of UAV resources. Relevant technologies have made progress in single-UAV task execution and simple scenario scheduling, but in the face of distributed collaborative requirements in complex fire field environments, the existing technology system still needs to be further improved.

[0003] First, existing technologies mostly adopt a centralized mission allocation architecture, relying on a central node for information processing and decision-making. However, in a fire-fighting scenario, the central node is vulnerable to fire damage or communication interference. Once it fails, the system will collapse, and its robustness cannot be guaranteed, making it unable to adapt to the complex communication environment in the fire field. Second, there is a lack of an efficient distributed information interaction mechanism, and it is difficult for multiple UAVs to achieve the consistency convergence of global information under local communication conditions, resulting in the lack of accurate awareness of the global situation of the fire field in task allocation and limited decision-making scientificity. Finally, existing technologies mostly focus on task allocation based on static fire area information, ignoring the real-time relationship between the UAV and the fire boundary, as well as the competition and conflicts that may occur during the cooperation of multiple UAVs.

[0004] How to solve the above technical problems is the topic faced by the present invention. Summary of the Invention

[0005] To solve the deficiencies of the existing technology, the present invention provides a high-order distributed fire-fighting UAV system mission allocation method that avoids the single-point failure problem in the centralized mode, has high system reliability and robustness, efficient information interaction and collaboration, simplifies the decision-making process, avoids repeated task allocation, and can adjust the allocation strategy according to the dynamic changes of the fire area.

[0006] The technical solution adopted by the present invention to solve its technical problems is: The present invention provides a high-order distributed fire-fighting UAV mission allocation method, including the following steps:

[0007] Obtain thermal infrared images and necessary parameters.

[0008] Calculate the length and width of the coverage area of the thermal infrared image corresponding to a single pixel point.

[0009] Convert the thermal infrared image into a grayscale image, perform binary processing on the fire area, and calculate the area of the fire area.

[0010] Extract the boundary of the fire area, establish a local coordinate system, and calculate the coordinates of the fire area boundary and the fire-fighting UAV.

[0011] Distributed collaborative decision-making and task allocation.

[0012] Preferably, the obtaining of the thermal infrared image and necessary parameters is specifically as follows:

[0013] Vertically photograph the fire scene with a thermal infrared imaging UAV to obtain a high-precision thermal infrared image, the flight altitude H, the horizontal field of view angle , the vertical field of view angle , the number of pixels in the horizontal direction of the image , the number of pixels in the vertical direction of the image , the longitude of the infrared imaging UAV , the latitude of the infrared imaging UAV , the set of longitude and latitude of the fire-fighting UAV , where .

[0014] It should be noted that the thermal infrared image is obtained by vertically photographing with a thermal infrared imaging UAV, the field of view angles are evenly distributed, the horizontal field of view corresponds to the longitude direction in the geographical location, and the vertical field of view corresponds to the latitude direction in the geographical location. Therefore, from left to right in the image is the direction of increasing longitude, and from bottom to top is the direction of increasing latitude.

[0015] Preferably, the calculating of the length and width of the area covered by the thermal infrared image corresponding to a single pixel point is specifically as follows:

[0016] According to the collected flight altitude H, horizontal field of view angle and vertical field of view angle information, use the tangent function to calculate the length X and width Y of the area covered by the thermal infrared image;

[0017] Respectively use the ratio of the length X of the area covered by the thermal infrared image to the number of pixels in the horizontal direction of the image and the ratio of the width Y of the area covered by the thermal infrared image to the number of pixels in the vertical direction of the image to calculate the length and width of the area covered by the thermal infrared image corresponding to a single pixel point.

[0018] The relational expression of the actually covered area of the thermal infrared image is as follows:

[0019] ;

[0020] The expressions of the length and width of the area covered by the thermal infrared image corresponding to a single pixel point are as follows:

[0021] ;

[0022] It should be noted that when the thermal infrared imaging UAV shoots vertically, its field of view (horizontal field of view , vertical field of view ) forms a coverage area similar to "conic projection". The flight height H serves as the vertical axis and, together with the field of view, determines the size of the actual ground coverage area.

[0023] For the horizontal direction, taking half of the horizontal field of view as the included angle and the flight height H as the right-angled side, the half-length of the ground coverage area is calculated using the tangent function, that is , and then multiplied by 2 to obtain the full length.

[0024] Similarly, for the vertical direction, taking half of the vertical field of view as the included angle, the half-width is calculated through the tangent function and multiplied by 2 to obtain the width.

[0025] Preferably, load the thermal infrared image and convert it into a grayscale image, perform binary processing on the fire area, use a fixed threshold m, mark the area above the fixed threshold m as white, and the remaining areas as black to generate a binary mask of the fire scene, and count the number of pixels in the white area , combined with the length and width corresponding to the coverage area of the thermal infrared image for a single pixel point to calculate the fire area.

[0026] The calculation formula for the fire area is as follows:

[0027] ;

[0028] represents the actual geographical length of the coverage area of the thermal infrared image corresponding to a single pixel point, represents the actual geographical width of the coverage area of the thermal infrared image corresponding to a single pixel point, and the product of the two represents the actual geographical area covered by a single pixel point. Fire area = total number of pixel points in the fire area × actual geographical area covered by a single pixel point.

[0029] Preferably, the extraction of the fire area boundary, the establishment of a local coordinate system, and the calculation of the coordinates of the fire area boundary and the fire-fighting UAV are specifically as follows:

[0030] Use the contour search algorithm to extract the fire area boundary from the binary image to obtain the set of pixel point coordinates of the fire area boundary , where .

[0031] Take the shooting position of the thermal infrared imaging UAV as the origin A local coordinate system is established, where the increasing direction of the x-axis of the local coordinate system is consistent with the increasing direction of longitude, and the increasing direction of the y-axis is consistent with the increasing direction of latitude.

[0032] According to the pixel coordinates of the boundary of the fire area Relative to the image center The offset, combined with the length of the area covered by the thermal infrared image corresponding to a single pixel And width , calculate the coordinates of the boundary of the fire area in the local coordinate system, and the coordinate set of the pixel points of the boundary of the fire area in the local coordinate system , where the coordinates of the i-th pixel point of the boundary of the fire area in the local coordinate system Are calculated as follows:

[0033] ;

[0034] It should be noted that the image pixel coordinates usually take the upper left corner as the origin (x increases to the right, y increases downward), while the local coordinate system takes the position of the thermal infrared imaging drone as the origin (x corresponds to the increasing direction of longitude, y corresponds to the increasing direction of latitude). Therefore, when calculating the ordinate, a negative sign needs to be added to reverse the y-axis direction of the image pixel coordinate system to make it consistent with the local coordinate system.

[0035] According to the longitude and latitude coordinates of the fire-fighting drone Relative to the longitude and latitude coordinates of the thermal infrared imaging drone The offset, combined with the actual distance corresponding to a 1° difference in longitude and latitude, calculates the coordinates of the fire-fighting drone in the local coordinate system. The coordinates of the i-th fire-fighting drone in the local coordinate system The calculation formula is expressed as follows:

[0036] ;

[0037] It should be noted that 111320 is the approximate distance corresponding to a 1° difference in longitude and latitude (unit: meters), Is to consider that the actual distance corresponding to a 1° difference in longitude is different at different latitudes, and is corrected by taking the cosine value of the average of two latitudes.

[0038] Preferably, the distributed collaborative decision-making and task allocation is specifically as follows:

[0039] Taking several fire-fighting drones as nodes, combining the communication situation between the fire-fighting drones, introducing the drone auxiliary state variables and filter adjustment parameters to construct a distributed average consensus filter. Through the distributed average consensus filter, each drone only relies on local communication (neighbor node information) to gradually achieve the consistency convergence of global information and ensure that the system stably shares key data in a dynamic communication environment.

[0040] The expression of the distributed average consensus filter is as follows:

[0041] ;

[0042] In the formula, represents whether the i-th fire drone can receive the communication of the j-th fire drone. 0 means it cannot receive the communication, and 1 means it can receive the communication; is 's estimate; is the i-th drone-aided state variable, which is used to spread the global average information; is the adjustment parameter of the filter.

[0043] It should be noted that affects the convergence speed of the distributed average consensus filter. In practical applications, numerical simulation can be used to test the convergence performance under different to determine the optimal value.

[0044] In , determines the communication relationship, represents the comprehensive influence of the difference between the aided state variables of adjacent drones and itself and the local information . Subtracting is to adjust the estimated value in a more accurate direction. As time goes by, the of all drones will gradually converge to the global average information.

[0045] In , the change rate of the aided state variable of the i-th drone depends on the difference between its estimated value and that of adjacent drones. If the estimated values of adjacent drones are different from its own, then will change, thus promoting the spread of information among adjacent drones.

[0046] A winner-takes-all competition strategy model is established. The distance of the fire drone from the closest vertex of the fire field, the communication situation between drones, and the activation status of drones are used as the model inputs. Combining the size of the fire area, the winner-takes-all competition strategy model assigns tasks to fire drones. The winner-takes-all competition strategy model is as follows:

[0047] ;

[0048] In the formula, is the activation status of the i-th fire drone. If ​It means it is activated, and the i-th fire-fighting drone is dispatched to perform tasks. It means it is not activated. It is the adjustment parameter for the evolution speed of the winner-takes-all competition strategy model. It is the distance from the i-th fire-fighting drone to the nearest vertex of the fire scene. ; It is the adjustment parameter related to the fire area in the winner-takes-all competition strategy model. It is the offset parameter related to the fire area in the winner-takes-all competition strategy model to avoid the denominator being zero. It is the fire area; in the formula, It represents whether the p-th fire-fighting drone can receive the communication from the q-th fire-fighting drone. 0 means it cannot receive the communication, and 1 means it can receive the communication.

[0049] It should be noted that the parameters of the model are given based on numerical simulation tests combined with the experience of technical personnel.

[0050] Each fire-fighting drone participates in the competition according to the distance to the nearest vertex of the fire scene and the fire area. The smaller the distance to the nearest vertex of the fire scene, the easier it is to be activated. The larger the fire area, the more fire-fighting drones may be activated.

[0051] , considering the activation status of adjacent drones. If an adjacent drone has been activated, it will affect the activation status of the current drone, reflecting the cooperation and competition relationship between drones.

[0052] It is an inhibition term, and its function is to inhibit the over-activation of drones. When increases, the absolute value of will also increase, which will inhibit the growth of , ensuring that finally only a few dominant drones are activated, achieving the "winner-takes-all" effect.

[0053] The present invention also provides a high-order distributed fire-fighting drone task allocation system, including:

[0054] A data acquisition module, which acquires thermal infrared images and necessary parameters through a thermal infrared imaging drone;

[0055] A calculation module, which is used to calculate the length and width of the coverage area corresponding to a single pixel point of the thermal infrared image, the fire area, the fire area boundary coordinates, the fire-fighting drone coordinates, and the distance from the fire-fighting drone to the nearest vertex of the fire area;

[0056] An image processing module, which is used to convert the thermal infrared image into a grayscale image, perform binary processing on the fire area, and extract the fire area boundary.

[0057] The task allocation module inputs external data and allocates tasks to fire-fighting drones according to the distributed average consensus filter and the winner-takes-all competition strategy model.

[0058] The beneficial effects of the present invention are as follows: It avoids the single-point failure problem in the centralized mode, has high system reliability and robustness, efficient information interaction and collaboration, simplifies the decision-making process, avoids repeated task allocation, and can adjust the allocation strategy according to the dynamic changes of the fire area. By introducing a distributed task allocation mechanism, the single-point failure problem in the centralized mode is avoided. Each drone makes independent decisions based on local information and feedback from neighboring drones, thus improving the system's reliability and fault tolerance. By adopting a task allocation method based on the winner-takes-all competition strategy model, each drone competes according to its own position and the size of the fire area, and drones closer to the fire are preferentially dispatched to perform tasks. The larger the fire area, the more drones are dispatched. This method simplifies the decision-making process and avoids repeated task allocation. The present invention strengthens information sharing and collaboration among drones by constructing a high-order network, making task allocation more intelligent and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a flowchart of the method steps of the present invention.

[0060] Figure 2 It is a system module diagram of the present invention.

[0061] Figure 3 It is a schematic diagram of the change process when the fire area is 90 square meters in Embodiment 3 of the present invention

[0062] Figure 4 It is a schematic diagram of the change process when the fire area is 350 square meters in Embodiment 3 of the present invention DETAILED DESCRIPTION OF THE INVENTION

[0063] In order to clearly illustrate the technical features of the present solution, the present solution will be described below through specific embodiments.

[0064] Embodiment 1: Refer to Figure 1 As shown, this embodiment is a high-order distributed fire-fighting drone task allocation method, including the following steps:

[0065] S1. Obtain thermal infrared images and necessary parameters.

[0066] Vertically photograph the fire scene with a thermal infrared imaging drone to obtain high-precision thermal infrared images and the flight altitude H, horizontal field of view angle , vertical field of view angle , the number of pixels in the horizontal direction of the image​​ 、 The number of pixels in the vertical direction of the image 、 The longitude of the infrared imaging UAV 、 The latitude of the infrared imaging UAV 、 The set of longitude and latitude of the fire-fighting UAV , where .

[0067] It should be noted that the thermal infrared image is obtained by the vertical shooting of the thermal infrared imaging UAV, and the field of view angle is evenly distributed. The horizontal field of view corresponds to the longitude direction on the geographical location, and the vertical field of view corresponds to the latitude direction on the geographical location. Therefore, from left to right in the image is the direction of increasing longitude, and from bottom to top is the direction of increasing latitude.

[0068] S2. Calculate the length and width of the coverage area of the thermal infrared image corresponding to a single pixel point.

[0069] According to the collected flight altitude H, horizontal field of view angle and vertical field of view angle information, use the tangent function to calculate the length X and width Y of the coverage area of the thermal infrared image;

[0070] Respectively use the ratio of the length X of the coverage area of the thermal infrared image to the number of pixels in the horizontal direction of the image and the ratio of the width Y of the coverage area of the thermal infrared image to the number of pixels in the vertical direction of the image to calculate the length and width of the coverage area of the thermal infrared image corresponding to a single pixel point.

[0071] The relational expressions of the actual coverage area of the thermal infrared image are as follows:

[0072] ;

[0073] The expressions of the length and width of the coverage area of the thermal infrared image corresponding to a single pixel point are as follows:

[0074] ;

[0075] It should be noted that when the thermal infrared imaging UAV shoots vertically, its field of view angle (horizontal field of view angle , vertical field of view angle ) forms a coverage range similar to "conic projection". The flight altitude H is used as the vertical axis, and together with the field of view angle, it determines the size of the actual coverage area on the ground.

[0076] For the horizontal direction, take half of the horizontal field of view angle as the included angle, and the flight altitude H as the right-angled side, and use the tangent function to calculate the semi-length of the ground coverage area, that is , multiply by 2 to obtain the full length.

[0077] Similarly, in the vertical direction, using half of the vertical field of view angle as the included angle, calculate the half-width through the tangent function , multiply by 2 to obtain the width.

[0078] S3. Convert the thermal infrared image into a grayscale image, perform binary processing on the fire area, and calculate the area of the fire area.

[0079] Load the thermal infrared image and convert it into a grayscale image. Perform binary processing on the fire area. Use a fixed threshold m. Mark the area higher than the fixed threshold m as white, and the remaining areas as black to generate a binary mask of the fire scene. Count the number of pixels in the white area , combine with the length and width of the area covered by the thermal infrared image corresponding to a single pixel point to calculate the area of the fire area.

[0080] The formula for calculating the area of the fire area is as follows:

[0081] ;

[0082] represents the actual geographical length of the area covered by the thermal infrared image corresponding to a single pixel point, represents the actual geographical width of the area covered by the thermal infrared image corresponding to a single pixel point. The product of the two represents the actual geographical area covered by a single pixel point. Area of the fire area = Total number of pixel points in the fire area × Actual geographical area covered by a single pixel point.

[0083] S4. Extract the boundary of the fire area, establish a local coordinate system, and calculate the coordinates of the fire area boundary and the fire-fighting drone.

[0084] Use the contour search algorithm to extract the boundary of the fire area from the binary image to obtain the set of pixel point coordinates of the fire area boundary [[ID=, where .

[0085] Take the shooting position of the thermal infrared imaging drone as the origin , establish a local coordinate system. The increasing direction of the x-axis of the local coordinate system is the same as the increasing direction of longitude, and the increasing direction of the y-axis is the same as the increasing direction of latitude.

[0086] According to the pixel point coordinates of the fire area boundary relative to the image center , combine with the length and width , calculate the coordinates of the fire area boundary in the local coordinate system, and the coordinate set of the pixel points of the fire area boundary in the local coordinate system , where the coordinates of the i-th pixel point of the fire area boundary in the local coordinate system are calculated as follows:

[0087] ;

[0088] It should be noted that the image pixel coordinates usually take the upper left corner as the origin (x increases to the right, y increases downward), while the local coordinate system takes the position of the thermal infrared imaging UAV as the origin (x corresponds to the direction of increasing longitude, y corresponds to the direction of increasing latitude). Therefore, when calculating the ordinate, a negative sign needs to be added to reverse the y-axis direction of the image pixel coordinate system to make it consistent with the local coordinate system.

[0089] According to the longitude and latitude coordinates of the fire-fighting UAV relative to the longitude and latitude coordinates of the thermal infrared imaging UAV offset, combined with the actual distance corresponding to a 1° difference in longitude and latitude, calculate the coordinates of the fire-fighting UAV in the local coordinate system. The coordinates of the i-th fire-fighting UAV in the local coordinate system are calculated as follows:

[0090] ;

[0091] It should be noted that 111320 is the approximate distance corresponding to a 1° difference in longitude and latitude (unit: meter), is to correct for the fact that the actual distance corresponding to a 1° difference in longitude is different at different latitudes by taking the cosine value of the average of the two latitudes.

[0092] S5, Distributed collaborative decision-making and task allocation.

[0093] Take several fire-fighting UAVs as nodes, combine the communication situation between the fire-fighting UAVs, introduce UAV auxiliary state variables and filter adjustment parameters to construct a distributed average consensus filter. Through the distributed average consensus filter, each UAV only relies on local communication (neighbor node information) to gradually achieve the consistency convergence of global information and ensure that the system stably shares key data in a dynamic communication environment.

[0094] The expression of the distributed average consensus filter is as follows:

[0095] ;

[0096] In the formula, represents whether the i-th fire-fighting UAV can receive the communication of the j-th fire-fighting UAV. 0 represents that the communication cannot be received, and 1 represents that the communication can be received; is Valuation; is the i-th UAV-assisted state variable for propagating global average information; is the adjustment parameter of the filter.

[0097] It should be noted that affects the convergence speed of the distributed average consensus filter. In practical applications, numerical simulations can be used to test the convergence performance under different to determine the optimal value.

[0098] In , determines the communication relationship, represents the difference between the assisted state variables of adjacent UAVs and itself, as well as the comprehensive influence of local information . Subtracting is to adjust the estimated value in a more accurate direction. Over time, the of all UAVs will gradually converge to the global average information.

[0099] In , the change rate of the i-th UAV's assisted state variable depends on the difference between its estimated value and that of adjacent UAVs. If the estimated values of adjacent UAVs are different from its own, then will change, thus promoting the propagation of information between adjacent UAVs.

[0100] Establish a winner-takes-all competition strategy model. The distance of the fire-fighting UAV from the nearest vertex of the fire, the communication situation between UAVs, and the activation status of UAVs are used as model inputs. Combining the size of the fire area, the winner-takes-all competition strategy model assigns tasks to the fire-fighting UAVs. The winner-takes-all competition strategy model is as follows:

[0101] ;

[0102] In the formula, is the activation status of the i-th fire-fighting UAV. If it means it is activated, and the i-th fire-fighting UAV is dispatched to perform tasks. it means it is not activated; is the adjustment parameter for the evolution speed of the winner-takes-all competition strategy model; is the distance of the i-th fire-fighting UAV from the nearest vertex of the fire, ; is the adjustment parameter related to the fire area size in the winner-takes-all competition strategy model; is the offset parameter related to the fire area in the winner-takes-all competition strategy model to avoid the case of a zero denominator; is the fire area; where, represents whether the p-th fire drone can receive the communication of the q-th fire drone, 0 means it cannot receive the communication, and 1 means it can receive the communication.

[0103] It should be noted that the parameters of the model are given based on numerical simulation tests combined with the experience of technical personnel.

[0104] Each fire drone participates in the competition according to the distance to the nearest vertex of the fire scene and the fire area. The smaller the distance to the nearest vertex of the fire scene, the easier it is to be activated. The larger the fire area, the more fire drones may be activated.

[0105] , considering the activation status of adjacent drones. If an adjacent drone has been activated, it will affect the activation status of the current drone, reflecting the cooperation and competition relationship between drones.

[0106] is an inhibitory term, which functions to inhibit the over-activation of drones. When increases, the absolute value of will also increase, which will inhibit the growth of , ensuring that only a few dominant drones are finally activated to achieve the "winner-takes-all" effect.

[0107] Example 2: Refer to Figure 2 As shown, this embodiment is a high-order distributed fire drone task allocation system, including:

[0108] A data acquisition module that obtains thermal infrared images and necessary parameters through a thermal infrared imaging drone;

[0109] A calculation module for calculating the length and width of the area covered by the thermal infrared image corresponding to a single pixel point, the fire area, the fire area boundary coordinates, the fire drone coordinates, and the distance from the fire drone to the nearest vertex of the fire area;

[0110] An image processing module for converting the thermal infrared image into a grayscale image, performing binary processing on the fire area, and extracting the fire area boundary;

[0111] A task allocation module that inputs external data and allocates tasks to fire drones according to the distributed average consensus filter and the winner-takes-all competition strategy model.

[0112] Example 3: Refer to Figure 3 、 Figure 4As shown in the figure, in order to illustrate the effectiveness of a high-order distributed fire-fighting UAV task allocation method proposed by the present invention, in this embodiment, a high-order distributed network is constructed with 10 fire-fighting UAVs, and UAV task allocation simulation is carried out for the cases where the area of the fire area is 90 square meters and 350 square meters respectively.

[0113] In the two cases of the fire area area, the distributed average consensus filter and the winner-takes-all competition strategy model in this embodiment adopt the same parameters, and the specific values are as follows:

[0114] ;

[0115] ;

[0116] For the activation state of the i-th UAV , in this embodiment, the initial value is taken as 0.1, and the relevant values are brought into the distributed average consensus filter and the winner-takes-all competition strategy model for calculation. The changes of the activation state over time in the two cases of the fire area area are respectively as Figure 3 , Figure 4 shown. After calculation, when the fire area area is 90 square meters, , and are greater than 0, and the system dispatches the 1st, 4th, and 5th fire-fighting UAVs to perform tasks; when the fire area area is 350 square meters, , , , and are greater than 0, and the system dispatches the 1st, 4th, 5th, and 7th fire-fighting UAVs to perform tasks.

[0117] From the above simulation test results, it can be seen that a high-order distributed fire-fighting UAV task allocation method provided by the present invention simplifies the decision-making process by adopting a task allocation method based on a winner-takes-all competition strategy model. Each UAV competes according to its own position and the size of the fire field area, and preferentially dispatches the UAVs closer to the fire field to perform tasks. The larger the fire field area, the more UAVs are dispatched. This method avoids the problems of task duplication and single-point failure in the centralized mode, has high system reliability and robustness, efficient information interaction and coordination, and can adjust the allocation strategy according to the dynamic changes of the fire area.

[0118] The un-described technical features of the present invention can be achieved by or adopted from the prior art, and will not be elaborated herein. Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by those of ordinary skill in the art within the scope of the essence of the present invention shall also fall within the protection scope of the present invention.

Claims

1. A high-order distributed fire-fighting UAV task allocation method, characterized in that, Including the following steps: Obtain the thermal infrared image and necessary parameters; Calculate the length and width of the area covered by the thermal infrared image corresponding to a single pixel point; Convert the thermal infrared image into a grayscale image, perform binary processing on the fire area, and calculate the area of the fire area; Extract the boundary of the fire area, establish a local coordinate system, and calculate the coordinates of the fire area boundary and the fire-fighting UAV; Distributed collaborative decision-making and task allocation; The specific steps of extracting the boundary of the fire area, establishing a local coordinate system, and calculating the coordinates of the fire area boundary and the fire-fighting UAV are as follows: Using the contour search algorithm, the boundary of the fire area is extracted from the binary image, and the set S of pixel coordinates of the boundary of the fire area is obtained , where ; Take the position where the thermal infrared imaging drone is located as the origin , and establish a local coordinate system. The increasing direction of the x-axis of the local coordinate system is consistent with the increasing direction of longitude, and the increasing direction of the y-axis is consistent with the increasing direction of latitude; According to the pixel coordinates of the fire area boundary The offset relative to the image center ( , ), combined with the length and width of the area covered by the thermal infrared image corresponding to a single pixel, calculate the coordinates of the fire area boundary in the local coordinate system; According to the longitude and latitude coordinates of the fire-fighting UAV relative to the longitude and latitude coordinates of the thermal infrared imaging UAV ( , ), combined with the actual distance corresponding to a 1° difference in longitude and latitude, calculate the coordinates of the fire-fighting UAV in the local coordinate system; The specific steps of the distributed collaborative decision-making and task allocation are as follows: Regarding several fire-fighting UAVs as nodes, combining the communication situation among the fire-fighting UAVs, introducing UAV auxiliary state variables and filter adjustment parameters to construct a distributed average consensus filter. The formula of the distributed average consensus filter is expressed as follows: Wherein, is the number of fire-fighting UAVs, represents whether the i-th fire-fighting UAV can receive the communication of the j-th fire-fighting UAV, 0 means it cannot receive the communication, and 1 means it can receive the communication; and are the estimated values of the global information average; is the auxiliary state variable of the i-th fire-fighting UAV; is the auxiliary state variable of the j-th fire-fighting UAV; is the adjustment parameter of the filter; and are the activation states of the i-th fire-fighting UAV and the j-th fire-fighting UAV respectively. If it is greater than 0, it means it is activated. If it is equal to 0, it means it is not activated; Establish a winner-takes-all competition strategy model. The formula of the winner-takes-all competition strategy model is expressed as follows: Wherein, , , and are the activation states of the i-th, p-th, q-th, and k-th fire drones respectively. If it is greater than 0, it means being activated; if it is equal to 0, it means not being activated; is the adjustment parameter of the evolution speed of the winner-takes-all competition strategy model; is the distance from the i-th fire drone to the nearest vertex of the fire scene; is the adjustment parameter related to the fire area in the winner-takes-all competition strategy model; is the offset parameter related to the fire area in the winner-takes-all competition strategy model; is the fire area; represents whether the p-th fire drone can receive the communication from the q-th fire drone. 0 means it cannot receive the communication, and 1 means it can receive the communication; represents whether the i-th fire drone can receive the communication from the k-th fire drone. 0 means it cannot receive the communication, and 1 means it can receive the communication; Taking the distance between the fire-fighting UAV and the nearest vertex of the fire scene, the communication situation among the UAVs, and the activation state of the UAV as the model inputs, and combining the size of the fire area, the winner-takes-all competition strategy model allocates tasks to the fire-fighting UAVs.

2. The high-order distributed fire-fighting drone mission allocation method according to claim 1, characterized in that The specific steps of obtaining the thermal infrared image and necessary parameters are as follows: Vertically photograph the fire scene with a thermal infrared imaging UAV to obtain high-precision thermal infrared images and the flight altitude H, horizontal field of view , vertical field of view , the number of pixels in the horizontal direction of the image , the number of pixels in the vertical direction of the image , the longitude of the infrared imaging UAV , the latitude of the infrared imaging UAV , the set of longitude and latitude of the fire-fighting UAV , where .

3. The high-order distributed fire-fighting UAV mission allocation method according to claim 1, wherein The specific steps of calculating the length and width of the area covered by the thermal infrared image corresponding to a single pixel point are as follows: According to the collected flight altitude H, horizontal field of view and vertical field of view information, use the tangent function to calculate the length X and width Y of the thermal infrared image coverage area; The ratio of the length X of the area covered by the thermal infrared image to the number of pixels in the horizontal direction of the image and the ratio of the width Y of the area covered by the thermal infrared image to the number of pixels in the vertical direction of the image are used to calculate the length and width corresponding to a single pixel in the area covered by the thermal infrared image.

4. The high-order distributed fire-fighting drone task allocation method according to claim 1, wherein The specific steps of converting the thermal infrared image into a grayscale image, performing binary processing on the fire area, and calculating the area of the fire area are as follows: Load the thermal infrared image and convert it to a grayscale image. Perform binary processing on the fire area. Use a fixed threshold m. Mark the area above the fixed threshold m as white and the remaining area as black to generate a binary mask of the fire scene, and count the number of pixels in the white area , combined with the length of the area covered by the thermal infrared image corresponding to a single pixel point and width Calculate the area of the fire area.

5. A high-order distributed fire-fighting UAV mission allocation system, characterized in that, For implementing the high-order distributed fire-fighting UAV task allocation method according to any one of claims 1-4, it includes: A data acquisition module that obtains the thermal infrared image and necessary parameters through a thermal infrared imaging UAV; A calculation module for calculating the length and width of the area covered by the thermal infrared image corresponding to a single pixel point, the area of the fire area, the coordinates of the fire area boundary, the coordinates of the fire-fighting UAV, and the distance between the fire-fighting UAV and the nearest vertex of the fire area; An image processing module for converting the thermal infrared image into a grayscale image, performing binary processing on the fire area, and extracting the boundary of the fire area; A task allocation module that inputs external data and allocates tasks to the fire-fighting UAVs according to the distributed average consensus filter and the winner-takes-all competition strategy model.

Images