Forest fire extinguishing cooperative scheduling system based on multi-unmanned aerial vehicle ad hoc network

Abstract

The invention belongs to the field of unmanned aerial vehicle cluster scheduling and wireless self-organizing networks, and relates to a multi-unmanned aerial vehicle forest fire extinguishing cooperative scheduling system based on a Mesh network, which comprises a ground control station responsible for strategy generation, task distribution coordination and remote scheduling control; a plurality of unmanned aerial vehicles equipped with communication and sensing modules form a honeycomb Mesh network to realize detection, relaying and fire extinguishing; the air Mesh communication network is dynamically constructed, comprises a backbone layer and an access layer, and supports distributed control and collaborative operation; the data sensing and sensing system is used for fusing information such as images, temperature and infrared rays to carry out fire behavior identification; and the task scheduling and energy management module supports task division, path planning and endurance guarantee. According to the system, an integrated closed-loop process of sensing, communication, scheduling and fire extinguishing is realized, and the response speed of forest fire extinguishing, the resource scheduling efficiency and the fire extinguishing accuracy are improved.

Description

Technical Field

[0001] The present invention belongs to the field of unmanned aerial vehicle (UAV) cluster scheduling and wireless self-organizing networks, and specifically relates to a forest fire fighting collaborative scheduling system based on a multi-UAV self-organizing network. Background Art

[0002] Drones (UAVs) are widely used in forest firefighting due to their high maneuverability and real-time monitoring capabilities. Currently, multi-UAV forest firefighting technology relies on a simple star-topology network. Specifically, a ground control station (GCS) establishes a communication connection with each UAV, sends real-time tasks to each UAV based on a generated scheduling policy, and receives real-time feedback from each UAV. However, this star-topology-based multi-UAV forest firefighting technology has several drawbacks. First, forest fires often occur in valleys and canyons, where conventional signal coverage is difficult. This can lead to poor communication between UAVs and the GCS, resulting in task allocation conflicts and untimely task adjustments, making it difficult to achieve multi-UAV collaborative operations. Second, the GCS needs to process real-time feedback from each UAV and generate real-time tasks for each UAV. This places high demands on the scientific and complex scheduling policy and the GCS's computing power. The high computational load will affect the GCS's responsiveness, thereby affecting firefighting efficiency. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention is implemented through the following technical solutions: A forest fire fighting collaborative dispatching system based on multi-UAV ad hoc network is proposed, including: a Mesh network composed of multiple UAVs; each UAV includes: a real-time positioning module, used to obtain the real-time position of the UAV and broadcast the real-time position to other UAVs; a network partitioning module, used to divide the Mesh network into multiple cellular subnets according to each real-time position, and identify the cellular subnet where the UAV is located; a status monitoring module, used to monitor the real-time status of the UAV and send the real-time status to other UAVs in the cellular subnet; a network topology module, used to establish a hierarchical network topology of the cellular subnet according to each real-time status; the hierarchical network topology includes: a backbone layer node and multiple access layer nodes; a selection trigger module, used to trigger the fire awareness module to work when the UAV is an access layer node, and to trigger the fire awareness module to work when the UAV is a backbone layer node. At this time, the fire awareness module, task allocation module, task parsing module, scheduling request module and resource scheduling module are triggered; the fire awareness module is used to perceive the fire situation in real time and send the real-time fire situation to the other nodes of this cellular subnet; the task allocation module is used to generate the fire extinguishing task of each node in this cellular subnet according to each real-time fire situation and each real-time status, and send each fire extinguishing task to each node of this cellular subnet accordingly; the scheduling request module is used to predict whether this cellular subnet can complete the fire extinguishing task according to each real-time fire situation and each real-time status during the fire extinguishing process, and if so, predict the resource surplus; if not, predict the resource demand and send the resource demand to each backbone layer node; the resource scheduling module is used to match the resource surplus of this cellular subnet with each resource demand, and generate the corresponding resource scheduling strategy according to the matching result.

[0004] Compared with the existing technology, the present invention has the following advantages and beneficial effects: based on the Mesh self-organizing network technology, a Mesh network with multiple drones as nodes is established to realize mutual communication between multiple drones and solve the problem of poor communication between drones and ground control stations; on this basis, a "backbone layer + access layer" two-layer network structure is established for each cellular subnet, and tasks are allocated to each node of the cellular subnet through the backbone layer nodes, and resources are coordinated and scheduled between the cellular subnets, thereby avoiding ground centralized control, reducing the computing load of the ground control station, and realizing efficient forest fire fighting. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings: Figure 1 This is a diagram of the Mesh network architecture consisting of multiple drones provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0006] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with the examples. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention. The embodiments described below 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 making creative work are within the scope of protection of the present invention.

[0007] In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that these specific details are not necessarily required to practice the present invention. In other examples, well-known structures, materials, or methods are not specifically described to avoid obscuring the present invention. The materials, instruments, and reagents used in the following examples, unless otherwise specified, are commercially available. The techniques used in the examples, unless otherwise specified, are conventional techniques well known to those skilled in the art.

[0008] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. Throughout the description of this application, "plurality" means two or more, unless otherwise specifically defined.

[0009] Example 1: A forest fire fighting coordination and dispatching system based on a multi-UAV ad hoc network is provided, which is composed of multiple UAVs. Figure 1 The mesh network shown in the figure. Each drone is equipped with: a wireless mesh communication module, a real-time positioning module, a network partitioning module, a status monitoring module, a network topology module, a selection trigger module, a fire awareness module, a fire extinguishing device, a task allocation module, a task analysis module, a scheduling request module, a resource scheduling module, a support request module, a support return module, a cellular subnet adjustment module, a blind spot monitoring module, a relay node screening module, a relay link construction module, a relay node adjustment module, a 3D model creation module, a flight trajectory adjustment module, a directional antenna control module, and an omnidirectional antenna control module. The following details the purpose of each of these modules, their internal operation, and their communication relationships with other modules.

[0010] 1. Wireless Mesh Communication Module A wireless mesh communication module is a communication device that supports dynamic routing protocols (such as AODV and OLSR) and enables self-organizing networks between devices. It wirelessly connects multiple nodes to form a mesh structure, supporting multi-hop transmission and dynamic routing. In this embodiment, each drone is equipped with this wireless mesh communication module, enabling multiple drones to autonomously establish a mesh network. In this mesh network, data is relayed through multiple intermediate drone nodes, expanding coverage and improving network reliability and resilience. Furthermore, this mesh network can automatically discover and configure nodes. If a node fails, other nodes can automatically reorganize the network structure to ensure continuous data transmission. Therefore, by equipping each drone with a wireless mesh communication module, a self-organizing mesh network with multiple drones as nodes is established. Multipath data transmission and redundant design reduce the impact of single points of failure. Compared to a star-topology network composed of multiple drones, this achieves efficient data transmission and task coordination, addressing communication issues in valleys and canyons.

[0011] 2. Real-time positioning module Used to obtain the real-time location of this drone. Suitable real-time positioning modules for drones include: the CUAV C-RTK 9Ps (supports RTK and PPK, provides centimeter-level accuracy, and is suitable for applications requiring real-time, high-precision navigation), the Unicore UM220-III (supports multiple systems such as Beidou and GPS, achieving centimeter-level positioning accuracy, and is suitable for high-precision missions), and the E108-GN03BD (a single Beidou positioning module that supports Beidou-2 and Beidou-3 signals and features high integration, low power consumption, and a small size). It should be understood that in addition to the real-time positioning modules listed above in this embodiment, other devices that can achieve high-precision real-time positioning may also be selected.

[0012] By configuring a real-time positioning module on a drone, the drone's real-time location can be collected. The real-time positioning module connects to the Mesh communication module, broadcasting the drone's real-time location through the Mesh communication module. This allows each node drone in the Mesh network to obtain the real-time location of every other drone, allowing each drone to understand its real-time position relationship with every other drone.

[0013] 3. Network partition module Based on the real-time positioning module and the mesh communication module, this drone can collect the real-time locations of all drones in the mesh network. Based on all collected real-time locations, the network partitioning module divides the mesh network into multiple cellular subnets through distributed clustering, identifying the cellular subnet where the drone is located. The forest firefighting collaborative dispatch system provided in this embodiment, equipped with a network partitioning module on each drone, aims to build a hierarchical, distributed collaborative architecture to address the high communication complexity, inefficient task allocation, and extensive resource management issues in large-scale drone firefighting scenarios.

[0014] The network partitioning module consists of a distributed clustering unit and a cell identification unit.

[0015] (1) Distributed clustering unit Used to perform distributed clustering of real-time locations to form multiple clusters. Each cluster corresponds to a cellular subnet. Through distributed clustering, each drone gradually forms a cluster using local rules based solely on its own real-time location and the real-time locations of other drones. No global topology information is required. The clustering process is driven by a distributed algorithm between nodes to avoid overload or failure of a single node. In addition, distributed clustering allows drones to recalculate adjacency relationships through real-time interaction when they move, join, or exit a cluster, triggering an update to the cluster structure. If individual nodes lose connection, it does not affect the overall cluster, and the remaining nodes can automatically reorganize.

[0016] (2) Cellular identification unit This function matches the real-time location of the drone with the real-time location of each drone in each cellular subnet, and determines the cellular subnet corresponding to the matched real-time location as the cellular subnet where the drone is located. For example, if the real-time location of the drone is (x1, y1, z1), and the matching finds that the drone corresponding to the real-time location (x1, y1, z1) is located in the first cellular subnet, then the first cellular subnet can be determined as the cellular subnet where the drone is located.

[0017] 4. Condition monitoring module This module is used to monitor the real-time status of the drone. This real-time status includes the drone's current remaining battery power, fire extinguishing agent level, communication radius, received signal strength, and computing power. The status monitoring module is connected to the Mesh communication module, which transmits the drone's real-time status to all other drones in the same cellular subnet. This allows every drone in the subnet to understand the real-time status of every other drone.

[0018] In the status monitoring module: (1) The battery management system (BMS) can be used to monitor the current remaining power of the drone in real time. The battery management system (BMS) is integrated into the drone battery, collects battery voltage, current, and temperature data in real time, and calculates the remaining power and health status. (2) A weight sensor can be used to monitor the current remaining amount of fire extinguishing agent in the drone in real time. The weight sensor directly weighs the container containing the fire extinguishing agent and obtains the remaining amount of fire extinguishing agent after deducting the weight of the empty container. (3) The NRF24L01 wireless module can be used to monitor the communication radius and received signal strength indicators in real time. (4) The performance counter (such as the PMU unit of ARM Cortex) can be used to monitor the computing power of the drone in real time. The performance counter is built into the drone's control chip and monitors the node computing power consumption by calling the top command or nodelet component in real time.

[0019] 5. Network topology module Based on the network partitioning module and the status monitoring module, the network topology module establishes a hierarchical network topology for the cellular subnet based on the real-time status of the drone and the real-time status of other drones in the cellular subnet received. The hierarchical network topology consists of backbone layer nodes and multiple access layer nodes.

[0020] This embodiment uses the network partition module to build a hierarchical and distributed collaborative architecture. It further uses the network topology module to establish a two-layer network topology of "backbone layer nodes + access layer nodes" for each cellular subnet. The purpose is to: Reduce communication complexity and improve collaboration efficiency. On the one hand, localized communication is achieved—access layer nodes within each cellular subnet communicate directly only with backbone layer nodes within that cellular subnet. Backbone layer nodes, acting as cellular subnet "managers," aggregate information and issue commands, avoiding the exponential complexity of "two-to-two communication" in star topology networks. On the other hand, global communication overhead is reduced—cross-subnet collaboration is accomplished solely through communication between backbone layer nodes. Non-essential information is not transmitted across subnets, improving communication bandwidth utilization.

[0021] This system implements task management by domain, adapting to the local characteristics of a fire scene. On one hand, it enables regionalized task allocation—each cellular subnet corresponds to a local area of ​​the fire scene. Backbone nodes can dynamically adjust the task division of drones within the cellular subnet based on the local fire situation, avoiding the delays in responding to local demands that would be encountered with the centralized dispatching of ground control stations based on a star-topology network. On the other hand, it achieves strategic flexibility—different subnets can adopt differentiated control strategies to meet the multi-dimensional firefighting needs of a fire scene.

[0022] Enhance system robustness and resist localized failures. On the one hand, a decentralized architecture is established—a two-tier network topology without a global central controller. Each cellular subnet operates autonomously through backbone nodes. When communication in a cellular subnet is interrupted or a backbone node fails, other subnets can still independently execute tasks, preventing a "single point of failure" from causing a system crash. On the other hand, fault isolation is achieved—failures within a cellular subnet only affect tasks in that area, and backbone nodes can quickly redistribute tasks, minimizing the scope of the failure.

[0023] Improve system scalability and adapt to dynamic changes in scale and flexible networking capabilities - new drones can be quickly added to adjacent subnets through a distributed clustering algorithm without reconstructing the entire network topology, supporting the dynamic expansion of drone clusters during firefighting missions.

[0024] Furthermore, the backbone layer nodes and access layer nodes in each cellular subnet are screened by the weight calculation unit, the first screening unit, and the second screening unit to obtain: (1) Weight calculation unit It is used to obtain the comprehensive weight of the UAV in the cellular subnet according to the real-time status, and send the comprehensive weight to the other UAVs in the cellular subnet.

[0025] Combined with the explanation of the status monitoring module, the real-time status of the drone includes: current remaining power, current fire extinguishing agent remaining, communication radius, current received signal strength index and current computing power. The comprehensive weight of the drone in this cellular subnet can be obtained through the weight calculation model. The weight calculation model is expressed as: comprehensive weight = a 1×Remaining power+ a 2× Fire Extinguishing Agent Remaining + a 3×communication radius+ a 4×Received Signal Strength Index+ a 5× computing power. Among them, a 1 is the weight coefficient of the remaining power, a 2 is the weight coefficient of the remaining amount of fire extinguishing agent, a 3 is the weight coefficient of the communication radius, a 4 is the weight coefficient of the received signal strength index, a 5 is the weight coefficient of computing power.

[0026] It should be noted that: (1) Before calculating the comprehensive weight, the remaining, extinguishing agent residue, communication radius, received signal strength index and computing power need to be normalized. The purpose is to convert indicators of different dimensions and value ranges into a unified dimensionless value, usually mapped to the interval [0,1] or [-1,1]. (2) Each weight coefficient is set according to the importance of each type of real-time status. If the real-time status of each type of task is of equal importance, the weight coefficients can be set to the same, for example a1 to a 5 are all 0.2.

[0027] (2) First screening unit Used to filter out drones corresponding to multiple comprehensive weights greater than a threshold.

[0028] (3) Second screening unit It is used to select the drone closest to the geometric center or geometric center of gravity of the current cellular subnet as the backbone layer drone based on the real-time positions of the multiple drones selected, and use the remaining drones in the current cellular subnet as access layer nodes.

[0029] It should be noted that: in combination with the role of the above-mentioned backbone layer nodes, the drone closest to the geometric center or geometric center of gravity of this cellular subnet is used as the backbone layer drone. This is based on the requirements of the core goals of optimizing the communication efficiency of the backbone layer nodes, balancing coverage, maximizing network stability and resource utilization.

[0030] 6. Select trigger module When this drone is an access layer node, it triggers the fire awareness module to work; when this drone is a backbone layer node, it triggers the fire awareness module, task allocation module, task analysis module, scheduling request module and resource scheduling module to work.

[0031] 7. Fire awareness module Used to perceive fire conditions in real time.

[0032] The acquisition and perception module is connected to the mesh communication module, which transmits real-time fire information to the remaining nodes in the cellular subnet through the mesh communication module. The fire perception module described in this embodiment is composed of a region division unit, a visible light camera, a thermal imaging infrared sensor, a temperature sensor, a feature extraction unit, and an information processing unit.

[0033] (1) Regional division unit It is used to divide the fire area into multiple management units according to the hierarchical structure of each cellular subnet, so that one cellular subnet corresponds to one management unit. Based on the above introduction to the functions of the backbone layer nodes, it can be seen that the backbone layer nodes, as the "managers" of the cellular subnet, are responsible for aggregating the information of this cellular subnet and issuing task instructions to each node of this cellular subnet, as well as realizing cross-subnet communication with other cellular subnets. Among them, the task instructions issued by the backbone layer nodes include real-time perception of the fire situation, so each drone needs to clearly define the perception area it is responsible for. The entire fire area is divided into cellular subnets through regional division units, so that cellular subnets and management units correspond one to one, and the tasks are clarified.

[0034] (2) Visible light camera Used to collect visible light images of this management unit in real time. Visible light cameras can perform conventional flame and smoke recognition, and combined with convolutional neural networks (such as YOLOv5), they can quickly detect fire targets.

[0035] (3) Thermal imaging infrared sensor Used to collect real-time thermal images of this management unit. The thermal imaging infrared sensor can capture hot spots, fire points, and smoke areas in real time, and can detect fire sources even at night or in thick smoke.

[0036] (4) Temperature sensor Used to collect real-time temperature distribution maps of this management unit. The temperature sensor can monitor the local ambient temperature rise and, by integrating it with image data, assist in determining the severity of the fire.

[0037] (5) Feature extraction unit This method uses an image fusion algorithm to fuse visible light images, thermal images, and temperature distribution maps. It then uses a convolutional neural network to extract features from the fused image to obtain characteristic information. Specifically, before image fusion, the visible light image, thermal image, and temperature distribution map are preprocessed to unify the image size, eliminate noise, and correct for color and grayscale differences. The SIFT or SURF algorithm is used to extract feature points from the visible light image, thermal image, and temperature distribution map, respectively, and then feature matching is performed. The matched features are then fused, and the fused image is reconstructed using interpolation, transformation, and other methods based on the fused feature information.

[0038] (6) Information processing unit This feature is used to analyze feature information using deep learning models (such as the YOLOv5 model) to obtain real-time fire information, including the fire's affected area and flame spread rate.

[0039] For the calculation of the fire affected area: the YOLOv5 model calculates the bounding box area based on the bounding box coordinates (upper left corner coordinates, upper right corner coordinates), optimizes the area within the bounding box through morphological operations (such as erosion and dilation), and then converts the pixel area into the actual physical area based on the image resolution. Specifically, first, the YOLOv5 model infers the input image and outputs the bounding box coordinates of the flame target in the form of upper left corner coordinates ( x 1, y 1) and the lower right corner coordinates ( x 2, y 2), the minimum enclosing area of ​​the corresponding rectangular frame. For the pixel area of ​​a single flame target S px The calculation formula is: S px =( x 2- x 1)×(y 2- y 1) If there are multiple flame targets, the area of ​​each target needs to be calculated separately and then accumulated to obtain the total pixel area. Then, morphological operations are used to remove noise in the detection results (such as small areas of false detection) and fill the holes in the flame area to make the boundary smoother and improve the accuracy of area calculation, including: corrosion operation - use a structural element of a specified size (such as a 3×3 or 5×5 rectangular or elliptical kernel) to traverse the binary flame mask within the bounding box, delete edge pixels, and eliminate small particle noise; expansion operation - the shape and size of the structural element need to be adjusted according to the resolution and noise characteristics of the flame image. The optimal value is usually determined through experiments (such as a kernel size of 5×5 and 1-2 iterations). Finally, the pixel area is converted to the physical area. The mapping relationship between the image resolution and the actual physical size needs to be calibrated in advance, that is, the actual area corresponding to the unit pixel (such as meters). 2 There are two cases: If the camera position is fixed, the pixel density can be calculated using a reference object of known size (such as a ground marker). , ,in, d 1 indicates the actual physical width (in meters); d 2 represents the image width in pixels, h 1 represents the actual physical altitude (in meters), h 2 represents the image height (in pixels). If perspective transformation is involved (such as tilted shooting), the image needs to be corrected through the homography matrix (Homography Matrix), converted into a bird's-eye view, and then the physical area is calculated. Finally, the total physical area .

[0040] For the calculation of flame spread speed: After the YOLOv5 model obtains multiple consecutive frames of images, it performs flame detection on each frame of the image to obtain the bounding box coordinates of multiple flame targets, determines the correspondence between flame targets in different frames, and indirectly calculates the flame spread speed by calculating the displacement distance of the flame center of mass between different frames. Specifically, first, continuous frame flame detection and target tracking are performed - YOLOv5 detection is performed on N consecutive frames of images (such as 30 frames per second, taking the most recent 5 frames) in turn to obtain the bounding box coordinates of the flame target in each frame; a target tracking algorithm (such as IOU-based greedy tracking or DeepSORT) is used to associate the same flame target in different frames to ensure the accuracy of cross-frame correspondences; the Hungarian Algorithm is used to match the intersection-over-union (IOU) of the current frame detection frame and the previous frame tracking frame to avoid target ID jumps. Then, the center of mass is calculated - for the bounding box of a single flame target, the center of mass ( c x , cy ) Take the bounding box center: c x =( x 1+ x 2) / 2, c y =( y 1+ y 2) / 2. Next, calculate the frame displacement - let the target be t The centroid of the frame is ( c x t , c y t ), No. t The centroid of the +1 frame is ( c x t+1 , c y t+1 ), then the pixel displacement distance is d px , , converting pixel displacement distance into actual physical displacement d phys , , is the proportional factor between the pixel displacement distance and the actual physical displacement, which can be obtained through experiments or calibration process. Finally, calculate the flame spread speed - assuming the video frame rate is f (such as 30FPS), the time interval between two adjacent frames is △ t =1 / f seconds, then the average speed of the flame v for v = d phys / △ t (m / s).

[0041] 8. Task Assignment Module It is used to generate fire extinguishing tasks for each node in this cellular subnet according to each real-time fire situation and each real-time status, and send each fire extinguishing task to each node in this cellular subnet accordingly.

[0042] In the drone cellular subnet architecture, after the backbone nodes receive real-time fire information and node status information from the cellular subnet, they use multi-dimensional data processing and intelligent algorithms to dynamically and accurately allocate firefighting tasks to access-layer nodes, thereby improving overall firefighting efficiency and resource utilization. Specifically, the task allocation module includes: (1) Fire analysis unit Used to divide the management unit into multiple zones based on real-time fire conditions. These zones include the flame front, flame spread, and cooling zone. The flame front is the area where the fire is currently most intense, with direct and advancing flames. The flame spread zone, located behind the flame front, is the area where the flames may spread. The cooling zone is the area where the fire has been extinguished or effectively controlled, with relatively low temperatures. For the flame front, edge detection (such as the Canny operator) can be performed on the fused image to extract the flame edge outline, and the boundary area at the forefront of the flame is identified as the flame front. For the flame spread, machine learning algorithms (such as LSTM and GRU) can be used to learn from the historical movement data of the flame front to predict the future direction and range of the flame spread. For the cooling zone, areas with temperatures below a set threshold and no visible flames can be marked as cooling zones in the temperature distribution map.

[0043] (2) Priority quantization unit It is used to quantify the priority of each flame front zone, each flame spread zone, and each cooling zone through the fire priority quantification model. The purpose of priority quantification is to: 1) Achieve rational allocation of resources: In firefighting missions, resources such as the number of drones, the amount of fire extinguishing agent carried, and flight time are limited. By quantifying the priorities of different areas, we can clearly judge the urgency and resource requirements of each area, thereby prioritizing the allocation of limited resources to high-priority areas.

[0044] 2) Promote multi-drone collaborative operations: After the priorities of different areas are quantified, each drone can clearly understand the importance and goals of its own mission, which helps achieve better coordination and cooperation.

[0045] The expression of the fire priority quantitative model is: (1); In formula (1), w 1 represents the weight of flame spread speed, w 2 represents the weight of the fire affected area, V Indicates the flame spread rate, V mas Indicates the maximum flame spread speed, S Indicates the area affected by the fire. S max Indicates the maximum fire area; w 1 and w 2 is calculated by the hierarchical analysis method. For example, the fire priority is divided into three levels by the fire priority quantitative model: the first priority ( P ≥0.7), second priority (0.3< P <0.7)、Third priority( P≤0.3). It should be noted that before calculating the comprehensive weight, the flame spread speed, fire affected area and maximum flame spread speed need to be normalized.

[0046] In addition, the quantization results are sent to the remaining backbone layer nodes through the mesh communication module to provide data support for subsequent priority matching.

[0047] (3) Priority matching unit It is used to match the comprehensive weights from each weight calculation unit with the priority levels from each priority quantization unit through a support vector machine to establish a priority matching table.

[0048] (4) Task allocation unit It is used to assign firefighting tasks to each node in this cellular subnet based on the priority matching table using a multi-objective optimization algorithm. Specifically: First, define the core objective function of multi-objective optimization, including: Fire extinguishing efficiency function: ,in, F eff It is a measure of the area of ​​fire extinguished per unit time. j is the node number in the cellular subnet, k Indicates the number of nodes in this subnet. S shed,j Indicates the j The fire source area extinguished by each node, t total Indicates the total fire extinguishing time.

[0049] Resource utilization function: ,in, F res Indicates the comprehensive resource utilization rate, m use,j Indicates the j The amount of fire extinguishing agent used by nodes, m total,j Indicates the j The total amount of fire extinguishing agent per node, E left,j Indicates the j The remaining power of each node, E total,j Indicates the j The total power of the node.

[0050] Task completion time function: F time =max{ t finish,1 , t finish,2 ,…, tfinish,k},in, F time Indicates the maximum value of all tasks completed, t finish,j Indicates the j The task completion time of each node, j=1 , 2 , … , k .

[0051] The multi-objective optimization algorithm described in this embodiment is the NSGA-II algorithm. Based on the NSGA-II algorithm, firefighting tasks are assigned to each node in the cellular subnet. The specific operations include: initializing the population, fitness calculation, non-dominated sorting, congestion calculation, genetic operation, population update iteration and solution selection. Initialize the population - randomly generate a certain number (e.g., 100) of task allocation solutions as the initial population, ensuring that each task is covered by at least one node (to avoid invalid solutions) and that node energy and communication constraints are met.

[0052] Fitness calculation - Calculate the fitness value of each individual based on the above three objective functions.

[0053] Non-dominated sorting—stratifies individuals in a population according to non-dominance relationships. Individuals in the same stratum are mutually non-dominant, and the smaller the stratum number, the more dominant the individual. For example, for any two individuals A and B in a population, if A is not inferior to B in all objectives and is superior to B in at least one objective, then A is said to dominate B. By sorting each stratum, the population is divided into different non-dominated levels (level 1 is the most optimal).

[0054] Crowding calculation reflects the distribution density of individuals in the solution space and is used to maintain population diversity. Crowding calculation calculates the sum of the distances between adjacent individuals on each target dimension for individuals in the same non-dominated layer. A higher crowding indicates a sparser distribution of individuals around the individual and better diversity.

[0055] Genetic operations include: (1) Selection operation: Using the tournament selection method, k individuals are randomly selected from the population, and individuals with high non-dominated levels and high crowding are selected to enter the mating pool; (2) Crossover operation: Using two-point crossover or uniform crossover, the task assignment segments of the selected chromosomes are randomly exchanged, while ensuring the legitimacy of the solution after crossover (such as task coverage constraints); (3) Mutation operation: Flip the genes in the chromosomes (0→1 or 1→0) with a low probability, and repair the solutions that violate the constraints (such as reallocating nodes to uncovered tasks).

[0056] Population update iteration - merge the parent and child populations, retain the top 100 excellent individuals (based on non-dominated sorting and crowding) to form a new generation of population; repeat the selection, crossover, and mutation operations until the maximum number of iterations is reached or the Pareto front converges.

[0057] Solution selection - Select the task allocation solution that best suits the current fire situation and node status from the Pareto front solution set. Specifically, selecting the task allocation solution that best suits the current fire situation and node status from the Pareto front solution set requires combining dynamic evaluation indicators, real-time data feedback, and decision-making strategies to transform the multi-objective optimization results into specific executable solutions: (1) Dynamic evaluation indicators Including: fire urgency (flame spread speed, population / asset value in the threatened area, and type of burning material), mission timeliness (remaining time to the optimal fire extinguishing window), spatial distribution (fire source location, spread direction), drone resources (remaining battery power, range, fire extinguishing agent capacity, distance between current location and fire source), historical mission load (number of missions executed, cumulative flight time), etc.

[0058] (2) Real-time data feedback Weights are assigned to each objective based on real-time data, transforming the multi-objective problem into a single-objective optimization problem. For example, objective weights are adjusted in real time based on pre-set expert domain knowledge. This expert domain knowledge includes: when the fire severity is ≥ Level 3, the weight for "firefighting speed" is > 0.6, and the weight for "drone safety" is > 0.3. When people are trapped in a certain area, the weight for "prioritizing coverage of this area" is increased to 0.5. Objective weights can be adjusted in real time, and weights can be dynamically modified using sensor data (such as thermal imaging and wind speed). For example, as wind speed increases, the weight for "rapidly blocking the direction of fire spread" increases.

[0059] (3) Decision-making strategy A fuzzy inference system was constructed with fire urgency (low / medium / high) and node health (excellent / good / poor) as input variables, and the output weight coefficients for each objective were calculated. For example, when fire urgency = high and node health = excellent, task timeliness was weighted 0.7 and energy consumption was weighted 0.3. When fire urgency = medium and node health = poor, load balancing was weighted 0.6 and completion rate was weighted 0.4.

[0060] (4) Solution screening Use evaluation metrics to quantitatively score each solution in the Pareto solution set and select the solution that best suits the current situation. This can be done by using the approximate ideal solution ranking method. The method steps are as follows: First, standardize the indicator data for each solution (e.g., normalized power consumption, distance, etc.); then, determine the "positive ideal solution" (optimal value for each indicator) and the "negative ideal solution" (worst value for each indicator); finally, calculate the Euclidean distance between each solution and the positive / negative ideal solution to obtain a comprehensive score: C i = d i - / ( d i - + d i + ),in, C i represents the comprehensive score, C i The closer to 1, the better the solution. d i - is a negative ideal solution, d i + is a positive ideal solution.

[0061] 9. Scheduling Request Module During the fire extinguishing process, it is used to predict whether the cellular subnet can complete the fire extinguishing task based on the real-time fire situation and real-time status. If so, it predicts the resource surplus. If not, it predicts the resource demand and sends the resource demand to each backbone layer node.

[0062] Among them, the resource surplus indicates the remaining resources after the flame in the management unit is completely extinguished, including: surplus electricity and surplus fire extinguishing agent; the resource demand indicates the amount of resources required to completely extinguish the flame in the management unit, including: electricity demand and fire extinguishing agent demand.

[0063] The scheduling request module includes: (1) Fire situation and status classification unit It is used to input the real-time fire situation and status of each node in the cellular subnet into the random forest model and output the classification results. The classification results include: can complete the fire extinguishing task or cannot complete the fire extinguishing task.

[0064] (2) Surplus power prediction unit This algorithm inputs the elapsed flight time, power consumption rate, current flight speed, current flight altitude, and current fire extinguishing agent remaining of each node in the cellular subnet into a time series prediction model (such as an LSTM / GRU neural network) to output the remaining power for each node. Specifically, remaining power prediction essentially captures the dynamic relationship between multiple features and power consumption through time series modeling, providing endurance prediction for task scheduling. Before prediction, data on elapsed flight time, power consumption rate, current flight speed, current flight altitude, and current fire extinguishing agent remaining is cleaned, normalized, and features extracted. This processed data is then fed into a trained time series prediction model (using a long short-term memory network as an example). In the LSTM, the input layer receives sequence data within a time window with a dimension of K×5 (K is the time window length, 5 is the number of features). In the LSTM hidden layer, the number of layers is selected based on data complexity (e.g., 1-3), with each layer containing N hidden units (e.g., N=64). In the output layer, a fully connected (Dense) layer maps the LSTM output to a single-value prediction, outputting the remaining power.

[0065] (3) Fire extinguishing agent surplus prediction unit This algorithm inputs the fire extinguishing area executed by each node in the cellular subnet, the average temperature of the management unit, the average extinguishing time of each node, and the flame spread rate into a regression model (such as a gradient boosting tree (XGBoost) or LightGBM) to output the remaining fire extinguishing agent for each node. Specifically, the gradient boosting tree algorithm can capture complex nonlinear relationships between features (such as the synergistic effect between temperature and spread rate), is insensitive to outliers, and is suitable for processing potential noise in sensor data. It also outputs the contribution of each feature to the prediction result, helping to understand the key drivers of fire extinguishing agent consumption. In actual tasks, the appropriate gradient boosting tree is selected based on the data volume, accuracy requirements, scenario complexity, and real-time requirements. This embodiment uses the LightGBM model, which is suitable for scenarios with large data volumes and high real-time requirements (such as dynamic updates of cellular subnets). Similarly, before prediction, the fire extinguishing area, the average temperature of the management unit, and the average extinguishing time of each node and the flame spread rate are cleaned, normalized, and features extracted. The processed data is then input into the trained LightGBM model. In the LightGBM model, the model outputs the surplus amount of fire extinguishing agent through weighted summation of the prediction results of multiple trees.

[0066] (4) Electricity demand forecasting unit It is used to establish a power demand prediction model, input the mission flight distance, average flight speed, total firefighting operation duration, average power and endurance time of each node into the power demand prediction model, output the power demand of this cellular subnet, and send the power demand to the remaining backbone layer nodes in the form of a request signal.

[0067] The expression of the power demand forecast model is: (3); In formula (3), j is the node number in the cellular subnet, k Indicates the number of nodes in this subnet. T req Indicates the power demand, P j Indicates the j The average power of a node, L j Indicates the j The mission flight distance of each node, V j Indicates the j The average flight speed of each node, T total,j Indicates the j The total duration of firefighting operations for each node, T already,j Indicates the j The elapsed time of each node.

[0068] The fire extinguishing agent demand prediction unit is used to establish a fire extinguishing agent demand prediction model, input the area of ​​each area in this management unit, the fire extinguishing agent dosage per unit area in each area, the priority quantization value of each area in this management unit and the current fire extinguishing agent balance of each node into the fire extinguishing agent demand prediction model, output the fire extinguishing agent demand of this cellular subnet, and send the fire extinguishing agent demand to the remaining backbone layer nodes in the form of a request signal.

[0069] The expression of the fire extinguishing agent demand prediction model is: (2); In formula (2), Q Indicates the demand for fire extinguishing agent, i is the area number in the management unit, m Indicates the number of areas divided in this management unit. S i Indicates the i The area of ​​the region, q i Indicates the i The amount of fire extinguishing agent per unit area in the area, Indicates the i The priority quantization value of each region, C j Indicates the j The current fire extinguishing agent remaining at each node.

[0070] 10. Resource Scheduling Module It is used to match the resource surplus of the cellular subnet with the resource demand, and generate the corresponding resource scheduling strategy based on the matching results.

[0071] Specifically, the resource scheduling module includes: (1) Demand Analysis Unit It is used to judge whether the resource surplus of this cellular subnet satisfies the conditions that surplus power > power demand and fire extinguishing agent surplus > fire extinguishing agent demand. If so, the demand response unit and timing analysis unit are triggered to work; otherwise, no demand response is performed.

[0072] (2) Demand response unit It is used to generate a demand response signal and send the demand response signal and the generation time of the demand response signal to the remaining backbone layer nodes.

[0073] (3) Timing analysis unit It is used to sort the generation time of each demand response signal in order, generate a time queue, and determine whether the generation time of the demand response signal of this node is at the head of the queue. If so, trigger the resource scheduling unit to work, otherwise, wait for receiving the next request signal.

[0074] (4) Resource scheduling unit Used to generate corresponding resource scheduling strategies based on resource demand; resource scheduling strategies include: the number of drones.

[0075] 11. Support Request Module Used to send a support request signal to the ground control station based on resource demand when condition A is met. Condition A: The classification result is that the firefighting mission cannot be completed and no demand response signal is received.

[0076] 12. Support return module This function is used to send a control signal back to the ground control station to each node in the local cellular subnet after the mission is completed when either condition B or condition C is met. Condition B: The classification result indicates that the firefighting mission was completed and no request signal was received during the mission. Condition C: The classification result indicates that the firefighting mission was completed and no request response was made during the mission.

[0077] 13. Cellular subnet adjustment module It is used to make the newly added drone the access layer node of this cellular subnet and update the hierarchical structure of this cellular subnet. When a drone exits this cellular subnet, the hierarchical structure of this cellular subnet is updated.

[0078] Furthermore, the selection trigger module is also used to trigger the cellular structure adjustment module to work when the drone is a backbone layer node and a drone joins / exits the cellular subnet.

[0079] 14. Blind spot monitoring module It monitors the received signal strength (RSS) of each communication link in the cellular subnet in real time, screens out communication links with an RSS below a threshold, and marks the area within the management unit where each selected communication link is projected as a communication blind spot edge. This information is then sent to the remaining backbone layers. The Received Signal Strength Indicator (RSSI) is a parameter that measures wireless signal strength. The receiver of the wireless mesh communication module determines the RSSI value by measuring the signal power. Furthermore, the communication blind spot edge refers to the boundary of an area in a communication system where, due to signal propagation limitations, no effective signal is received or the signal strength is extremely low, resulting in communication failure. Natural and man-made obstacles such as mountains, buildings, and tunnels can block the signal, creating shadowed areas or blind spots.

[0080] 15. Relay node screening module It is used to select relay nodes in this cellular subnet according to the hierarchical structure of this cellular subnet and the remaining power of each node. (1) Constructing the node model and weight graph of the cellular subnet Constructing a weighted undirected graph G ( V , E ).in, V Represents the node set of this cellular subnet, E Represents the communication link between nodes, and the weight on the communication link is ,in, represents the comprehensive weight of the communication link, Represents the weight coefficient of distance, distance represents the distance between the nodes at both ends of the communication link, represents the weight coefficient of the remaining circuit, energy Indicates the remaining power of the node. The closer the distance, the lower the communication energy consumption. Nodes with higher remaining power are more suitable for relaying. In this embodiment, the weight of the remaining power is set to the inverse of the remaining power.

[0081] (2) Determine the set of candidate relay nodes Initially select nodes that meet the basic conditions from the cellular subnet as candidate sets C ∈ V ,The conditions include: the remaining power is higher than the threshold (e.g., the remaining power ≥ 50% of the initial power); the communication coverage can connect to at least two other nodes (avoiding isolated nodes); the geographical locations are evenly distributed, covering potential communication blind spots.

[0082] (3) Apply the minimum spanning tree algorithm to generate relay links Using Kruskal's algorithm on weighted graphs GGenerate a minimum spanning tree on . First, the edge set E By weight Sort the nodes from smallest to largest; then, create a Union-Find structure, initializing each node as an independent set. Finally, traverse each edge in sorted order. If the two nodes corresponding to the currently queried edge belong to different sets, merge the sets and add the edge for the single-signature query to the spanning tree until all nodes are connected.

[0083] (4) Extract relay nodes from relay links After the minimum spanning tree is generated, the non-leaf nodes in the minimum spanning tree are marked as relay nodes.

[0084] 16. Relay link construction module Used to build relay links containing relay nodes through topology-aware algorithms.

[0085] Based on the minimum spanning tree generated in the relay node screening module, a path formed from the root node to the node is a relay link containing the relay node.

[0086] 17. 3D model creation module It is used to create a three-dimensional model of the communication blind area of ​​this cellular subnet based on the edge of the communication blind area and the fused image.

[0087] (1) Establishing a spatial coordinate system Use a UTM projection or local coordinate system to ensure that the model accurately corresponds to the actual geographic location, and define the Z axis based on mean sea level (MSL) or ground elevation.

[0088] (2) Three-dimensional blind spot boundary First, the two-dimensional blind spot edge points are combined with the terrain elevation data to assign a corresponding Z coordinate to each boundary point; then, the discrete boundary points are smoothed using a B-spline curve or polynomial fitting algorithm to generate a continuous three-dimensional boundary curve.

[0089] (3) Generate blind area First, Delaunay triangulation is applied to the area enclosed by the 3D boundary curve to generate a triangular mesh covering the blind area; then, a height value is assigned to each triangular facet according to the terrain relief and obstacle distribution.

[0090] (4) Build a 3D model The triangular mesh model is converted into voxel representation, each voxel is marked as "blind area" or "non-blind area", and a three-dimensional reconstruction algorithm (point cloud-based reconstruction algorithm) is used to establish a three-dimensional model of the communication blind area.

[0091] 18. Flight trajectory adjustment module This algorithm dynamically adjusts the flight paths of each drone in this cellular subnet using the NSGA-II algorithm based on terrain elevation and a 3D model of communication blind spots. This algorithm also references the principles and process of assigning firefighting tasks to nodes in this cellular subnet using the NSGA-II algorithm.

[0092] The selection trigger module is also used to trigger the three-dimensional model creation module and the flight trajectory adjustment module when the UAV is a backbone layer node and receives the edge of the communication blind area.

[0093] 19. Directional antenna control module Used to start the directional antenna and adjust the direction of the directional antenna.

[0094] 20. Omnidirectional antenna control module Used to start the omnidirectional antenna and adjust the direction of the directional antenna.

[0095] The selection trigger module is also used to trigger the directional antenna control module to work when the drone is a backbone layer node, and to trigger the omnidirectional antenna control module to work when the drone is an access layer node.

[0096] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0097] It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for understanding and reading by those familiar with this technology, and are not used to limit the conditions for implementation of the present invention. Therefore, they have no substantial technical significance. Any modification of the structure, change in the proportional relationship, or adjustment of the size should still fall within the scope of the technical content disclosed by the present invention without affecting the efficacy and purpose of the present invention. At the same time, the terms such as "upper", "lower", "left", "right", "middle", etc. quoted in this specification are only for the convenience of description and are not used to limit the scope of implementation of the present invention. Changes or adjustments in their relative relationships should also be regarded as the scope of implementation of the present invention without substantially changing the technical content.

Claims

1. A forest fire fighting collaborative dispatching system based on multi-UAV ad hoc network, characterized by: include: A Mesh network consisting of multiple drones; Each drone includes: Real-time positioning module, used to obtain the real-time location of the drone and broadcast the real-time location to other drones; The network division module is used to divide the Mesh network into multiple cellular subnets according to the real-time location and identify the cellular subnet where the drone is located; The status monitoring module is used to monitor the real-time status of the drone and send the real-time status to other drones in the cellular subnet; The network topology module is used to establish a hierarchical network topology of the cellular subnet according to each real-time status; the hierarchical network topology includes: a backbone layer node and multiple access layer nodes; Select a trigger module to trigger the fire awareness module when the drone is an access layer node, and to trigger the fire awareness module, task allocation module, task parsing module, scheduling request module, and resource scheduling module when the drone is a backbone layer node; Fire awareness module, used to sense fire conditions in real time and send the real-time fire conditions to other nodes in the cellular subnet; The task allocation module is used to generate fire extinguishing tasks for each node in the cellular subnet according to each real-time fire situation and each real-time status, and send each fire extinguishing task to each node in the cellular subnet accordingly; The scheduling request module is used to predict whether the cellular subnet can complete the fire extinguishing task according to the real-time fire situation and status during the fire extinguishing process. If so, it predicts the resource surplus. If not, it predicts the resource demand and sends the resource demand to each backbone node. The resource scheduling module is used to match the resource surplus of the cellular subnet with the resource demand, and generate a corresponding resource scheduling strategy based on the matching result.

2. The forest fire fighting coordinated dispatching system based on multi-UAV ad hoc network according to claim 1 is characterized in that: The network partitioning module includes: A distributed clustering unit is used to perform distributed clustering on each real-time location to obtain multiple clusters; one cluster corresponds to one cellular subnet; The cellular identification unit is used to match the real-time position of the drone with the real-time position of each drone in each cellular subnet, and determine the cellular subnet corresponding to the successfully matched real-time position as the cellular subnet where the drone is located.

3. A forest fire fighting collaborative dispatching system based on a multi-UAV ad hoc network according to claim 1 or 2, characterized in that: Real-time status includes: remaining battery power, fire extinguishing agent level, communication radius, received signal strength index, and computing power; The network topology module includes: A weight calculation unit is used to obtain the comprehensive weight of the UAV in the cellular subnet according to the real-time status and send the comprehensive weight to the other UAVs in the cellular subnet; The first screening unit is used to screen out drones corresponding to multiple comprehensive weights greater than a threshold; The second screening unit is used to screen out the drone closest to the geometric center or geometric center of gravity of the current cellular subnet as the backbone layer drone based on the real-time positions of the multiple drones screened out, and use the remaining drones of the current cellular subnet as access layer nodes.

4. The forest fire fighting coordinated dispatching system based on multi-UAV ad hoc network according to claim 3 is characterized in that: The fire awareness module includes: The area division unit is used to divide the fire area into multiple management units according to the hierarchical structure of each cellular subnet; one cellular subnet corresponds to one management unit; A visible light camera, used to collect visible light images of the management unit in real time; Thermal imaging infrared sensor, used to collect thermal images of this management unit in real time; Temperature sensor, used to collect the temperature distribution map of this management unit in real time; A feature extraction unit is used to fuse the visible light image, thermal image and temperature distribution map using an image fusion algorithm, and extract features from the fused image using a convolutional neural network to obtain feature information; The information processing unit is used to analyze feature information using a deep learning model to obtain real-time fire conditions; the real-time fire conditions include: the fire-affected area and the flame spread speed.

5. The forest fire fighting coordinated dispatching system based on multi-UAV ad hoc network according to claim 4 is characterized in that: The task allocation module includes: The fire situation analysis unit is used to divide the management unit into multiple areas according to the real-time fire situation; the multiple areas include: flame front area, flame spread area and cooling area; The priority quantification unit is used to quantify the priority of each flame front zone, each flame spread zone, and each cooling zone using the fire priority quantification model, and send the quantification results to the remaining backbone layer nodes; A priority matching unit is used to match the comprehensive weight of each node with each priority level through a support vector machine to establish a priority matching table; The task allocation unit is used to allocate fire-fighting tasks to each node of the cellular subnet according to the priority matching table and adopt a multi-objective optimization algorithm.

6. The forest fire fighting coordinated dispatching system based on multi-UAV ad hoc network according to claim 5 is characterized in that: Resource surplus includes: surplus electricity and surplus fire extinguishing agent; resource demand includes: electricity demand and fire extinguishing agent demand; The scheduling request module includes: The fire condition and status classification unit is used to input the real-time fire condition and status of each node in the cellular subnet into the random forest model and output the classification results; the classification results include: can complete the fire extinguishing task or cannot complete the fire extinguishing task; A surplus power prediction unit is used to input the flight time, power consumption rate, current flight speed, current flight altitude and current fire extinguishing agent remaining of each node in the cellular subnet into a time series prediction model and output the surplus power of each node; The fire extinguishing agent surplus prediction unit is used to input the area of ​​the fire extinguishing area executed by each node in the cellular subnet, the average temperature of the management unit, the average fire extinguishing time of each node, and the flame spread speed into the regression model, and output the fire extinguishing agent surplus of each node; The power demand prediction unit is used to establish a power demand prediction model, input the mission flight distance, average flight speed, total firefighting operation duration, average power, and elapsed flight time of each node into the power demand prediction model, output the power demand of the cellular subnet, and send the power demand to the remaining backbone nodes in the form of a request signal; The fire extinguishing agent demand prediction unit is used to establish a fire extinguishing agent demand prediction model, input the area of ​​each area in this management unit, the fire extinguishing agent dosage per unit area in each area, the priority quantization value of each area in this management unit and the current fire extinguishing agent balance of each node into the fire extinguishing agent demand prediction model, output the fire extinguishing agent demand of this cellular subnet, and send the fire extinguishing agent demand to the remaining backbone layer nodes in the form of a request signal.

7. The forest fire fighting coordinated dispatching system based on multi-UAV ad hoc network according to claim 6 is characterized in that: The expression of the fire priority quantitative model is: (1); In formula (1), w 1 represents the weight of flame spread speed, w 2 represents the weight of the fire affected area, V Indicates the flame spread rate, V mas Indicates the maximum flame spread speed, S Indicates the area affected by the fire. S max Indicates the maximum fire area; w 1 and w 2 calculated by analytic hierarchy process; The expression of the power demand forecast model is: (3); In formula (3), j is the node number in the cellular subnet, k Indicates the number of nodes in this subnet. T req Indicates the power demand, P j Indicates the j The average power of a node, L j Indicates the j The mission flight distance of each node, V j Indicates the j The average flight speed of each node, T total,j Indicates the j The total duration of firefighting operations for each node, T already,j Indicates the j The lifetime of each node; The expression of the fire extinguishing agent demand prediction model is: (2); In formula (2), Q Indicates the demand for fire extinguishing agent, i is the area number in the management unit, m Indicates the number of areas divided in this management unit. S i Indicates the i The area of ​​the region, q i Indicates the i The amount of fire extinguishing agent per unit area in the area, Indicates the i The priority quantization value of each region, C j Indicates the j The current fire extinguishing agent remaining of each node.

8. A forest fire fighting coordinated dispatching system based on a multi-UAV ad hoc network according to claim 6 or 7, characterized in that: The resource scheduling module includes: The demand analysis unit is used to determine whether the resource surplus of the cellular subnet satisfies the following conditions: surplus power > power demand and surplus fire extinguishing agent > fire extinguishing agent demand. If so, the demand response unit and the timing analysis unit are triggered to work; otherwise, no demand response is performed. A demand response unit is used to generate a demand response signal and send the demand response signal and the generation time of the demand response signal to the remaining backbone layer nodes; The timing analysis unit is used to sort the generation time of each demand response signal in order, generate a time queue, and determine whether the generation time of the demand response signal of this node is at the head of the queue. If so, the resource scheduling unit is triggered to work; otherwise, it waits for receiving the next request signal; The resource scheduling unit is used to generate corresponding resource scheduling strategies according to resource demand; the resource scheduling strategies include: the number of drones.

9. The forest fire fighting coordinated dispatching system based on multi-UAV ad hoc network according to claim 8 is characterized in that: Also includes: A support request module is used to send a support request signal to the ground control station according to the resource demand when condition A is met; Support return module, used to send a control signal back to the ground control station to each node of the cellular subnet after the mission is completed when condition B or condition C is met; Condition A: The classification result is that the fire extinguishing task cannot be completed and no demand response signal is received; Condition B: The classification result is that the fire extinguishing task can be completed and no request signal is received during the execution of the task; Condition C: The classification result is that the fire-fighting task can be completed and no demand response is made during the execution of the task.

10. A forest fire fighting coordinated dispatching system based on a multi-UAV ad hoc network according to claim 1 or 2, characterized in that: Also includes: The cellular subnet adjustment module is used to treat the newly added drone as the access layer node of the cellular subnet and update the hierarchical structure of the cellular subnet. When a drone leaves the cellular subnet, the hierarchical structure of the cellular subnet is updated; The selection trigger module is also used to trigger the cellular structure adjustment module to work when the current drone is a backbone layer node and a drone joins / exits the current cellular subnet.

11. The forest fire fighting coordinated dispatching system based on multi-UAV ad hoc network according to claim 4 is characterized in that: Also includes: The blind spot monitoring module is used to monitor the received signal strength index of each communication link in the cellular subnet in real time, screen out the communication links with received signal strength index less than the threshold, mark the area projected by each screened communication link within the management unit as the communication blind spot edge, and send the communication blind spot edge to the remaining backbone layers; The relay node screening module is used to screen relay nodes in the cellular subnet according to the hierarchical structure of the cellular subnet and the remaining power of each node; A relay link construction module is used to construct a relay link containing relay nodes through a topology-aware algorithm; The selection trigger module is also used to trigger the blind spot detection module, relay node screening module and relay link construction module when the drone is a backbone layer node and receives the edge of the communication blind spot.

12. The forest fire fighting coordinated dispatching system based on multi-UAV ad hoc network according to claim 11, characterized in that: Also includes: A three-dimensional model creation module is used to create a three-dimensional model of the communication blind area of ​​the cellular subnet based on the edge of the communication blind area and the fused image; The flight trajectory adjustment module is used to dynamically adjust the flight trajectory of each drone in the cellular subnet using the NSGA-II algorithm based on the terrain elevation and the three-dimensional model of the communication blind area; The selection trigger module is also used to trigger the three-dimensional model creation module and the flight trajectory adjustment module when the UAV is a backbone layer node and receives the edge of the communication blind area.

13. A forest fire fighting coordinated dispatching system based on a multi-UAV ad hoc network according to claim 1 or 2, characterized in that: Also includes: Directional antenna control module, used to start the directional antenna and adjust the direction of the directional antenna; Omnidirectional antenna control module, used to start the omnidirectional antenna and adjust the direction of the directional antenna; The selection trigger module is also used to trigger the directional antenna control module to work when the drone is a backbone layer node, and to trigger the omnidirectional antenna control module to work when the drone is an access layer node.

Images