Disaster relief-oriented multi-unmanned aerial vehicle cooperative control method and system

Abstract

The application discloses a kind of multi-unmanned aerial vehicle cooperative control methods and systems for disaster relief, it is related to unmanned aerial vehicle control technical field, including the following steps: acquisition disaster data, pre-processing generates disaster relief cooperative data;Unmanned aerial vehicle cooperation relationship graph is constructed, and unmanned aerial vehicle cooperation relationship graph data is generated;Determine the task association relationship of unmanned aerial vehicle, and generate task association attention feature data;Update connection relationship, and generate communication dynamic relationship graph data;Task association attention feature data and communication dynamic relationship graph data are input into improved Graph Attention Network graph attention network model, and generate multi-unmanned aerial vehicle cooperative feature data;Calculate the cooperative relationship of parent machine and child unmanned aerial vehicle, generate control strategy and execute rescue task, update cooperative relationship graph data.The application realizes multi-unmanned aerial vehicle cooperative control of disaster relief, improves task scheduling efficiency and complex environment adaptability.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a multi-UAV collaborative control method and system for disaster relief. Background Technology

[0002] During disaster relief efforts, sudden events such as earthquakes, floods, and fires can cause road blockages, communication disruptions, and complex environments in the rescue area, making it difficult for traditional manual rescue methods to obtain timely information about the disaster area. With the development of drone technology, drones are being used for disaster area patrols, target searches, environmental surveys, and material delivery. By carrying cameras, sensors, and communication equipment, they acquire information from the disaster site, improving the efficiency of rescue operations. Current drone rescue solutions primarily use a single drone to perform the mission, while some solutions employ multi-drone collaboration to achieve area coverage and task allocation.

[0003] Existing multi-UAV cooperative control methods primarily rely on UAV positional relationships, pre-defined task plans, or fixed communication topologies for control, utilizing information exchange between UAVs to complete path planning and task execution. However, in disaster relief scenarios, task locations, environmental conditions, communication conditions, and UAV operational states are constantly changing. Fixed relationship models struggle to reflect the dynamic cooperative relationships between UAVs in real time, resulting in insufficient matching between task allocation results and on-site requirements. Existing methods lack sufficient integration of task associations, communication states, and spatial relationships between UAVs, making it difficult to achieve multi-dimensional state information-driven cooperative control.

[0004] Therefore, how to provide a multi-UAV collaborative control method and system for disaster relief is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a multi-UAV collaborative control method and system for disaster relief. This invention utilizes multi-UAV collaborative control, graph attention networks, and dynamic task scheduling technology to construct an improved GraphAttention Network model, which realizes disaster area information fusion, UAV task collaborative allocation, and dynamic control strategy generation. It has the advantages of high task matching accuracy, strong collaborative control capability, and good disaster adaptability.

[0006] A multi-UAV cooperative control method for disaster relief according to an embodiment of the present invention includes the following steps: Collect environmental data, rescue mission data, and drone status data from disaster areas, perform preprocessing, and generate disaster relief collaborative data; Based on the location, communication status, and mission status of drones in disaster relief collaboration data, a drone collaboration relationship diagram is constructed, and drone collaboration relationship diagram data is generated. Based on the drone collaboration relationship graph data and rescue mission data, the drone mission relationships are determined, and mission association attention feature data is generated. Update the connection relationships in the UAV collaboration relationship graph data according to the UAV communication status to generate communication dynamic relationship graph data; Task-related attention feature data and communication dynamic relationship graph data are input into the improved Graph Attention Network model to extract UAV collaborative features and generate multi-UAV collaborative feature data. The collaborative relationship between the mother drone and the daughter drones is calculated based on the collaborative feature data of the multiple drones, and the collaborative control strategy data of the multiple drones is generated. Based on the multi-UAV collaborative control strategy, control UAVs to perform disaster relief missions and collect UAV mission execution status data; The UAV collaborative relationship graph data and communication dynamic relationship graph data are updated based on the UAV mission execution status data. The updated data is then input into the improved Graph Attention Network model to generate updated multi-UAV collaborative control strategy data.

[0007] Optionally, the preprocessing process includes: Collect environmental data of disaster areas, rescue mission data, and drone status data; Coordinate transformation and obstacle information extraction are performed on environmental data in disaster areas to generate environmental status data; The rescue mission data is parsed for mission location, encoded for mission type, and labeled for mission priority to generate mission status data; Time synchronization and state feature encoding are performed on UAV state data to generate UAV state feature data; By integrating environmental status data, mission status data, and UAV status characteristic data, collaborative disaster relief data can be generated.

[0008] Optionally, the process of constructing the UAV collaborative relationship graph includes: The mother drone and the child drones in the disaster relief collaborative data are taken as drone nodes, and the location status features, communication status features and mission status features of the corresponding drone nodes are extracted. Spatial relationships are determined based on the location distance between drone nodes, communication relationships are determined based on the communication quality between drone nodes, and task relationships are determined based on the task status corresponding to the drone nodes. By integrating spatial, communication, and task-related relationships, the connection relationships between UAV nodes are constructed, generating UAV collaborative relationship graph data.

[0009] Optionally, the process of generating task-related attention feature data includes: Obtain the characteristics of drone nodes from the drone collaboration graph data and the task location, task type, and task priority from the rescue task data; The location matching relationship is determined based on the spatial distance between the UAV node location and the mission location, and the mission execution capability matching relationship is determined based on the remaining battery power and payload status in the UAV status data. Based on location matching, task execution capability matching, and task priority, generate task association weights between drones and rescue missions; Attention is calculated on the features of UAV nodes based on task association weights to generate task association attention feature data.

[0010] Optionally, the process of generating the communication dynamic relationship graph data includes: Obtain the communication status of UAV nodes in the UAV collaborative relationship graph data, and extract the communication signal strength, data transmission delay, and communication link stability between UAV nodes; The communication association weights between UAV nodes are calculated based on communication signal strength, data transmission delay, and communication link stability. Adjust the connection relationships in the UAV collaborative relationship graph data according to the communication association weight, and update the communication association edges between UAV nodes; Generate dynamic communication relationship graph data based on the updated communication association edges.

[0011] Optionally, the step of inputting task-related attention feature data and communication dynamic relationship graph data into the improved Graph Attention Network model includes: Map task-related attention feature data to the UAV node feature space to generate task-enhanced node feature data; Input the task-enhanced node feature data and the communication dynamic relationship graph data into the Graph Attention Network model, and calculate the attention weights between nodes based on the UAV node connection relationships in the communication dynamic relationship graph data. Spatial collaborative feature data is generated by aggregating features of adjacent UAV nodes based on attention weights between nodes. The task-related attention features in the task-enhanced node feature data are fused with the spatial collaboration feature data to generate multi-dimensional collaboration feature data. Graph attention propagation processing is applied to multi-dimensional collaborative feature data to generate multi-UAV collaborative feature data.

[0012] Optionally, the process of establishing a collaborative relationship between the host computer and the sub-UAVs includes: Acquire drone node features from multi-drone collaborative feature data, and identify the parent and child drone nodes corresponding to the drone nodes; Extract the regional scheduling features corresponding to the mother node and the task execution features corresponding to the child UAV nodes; Calculate the scheduling association weight of the mother machine based on the characteristic association relationship between the mother machine node and the child UAV node, and determine the task allocation relationship between the mother machine node and the child UAV node; The collaborative association weights of sub-drone nodes are calculated based on the task execution characteristics of sub-drone nodes to determine the collaborative execution relationships between sub-drone nodes; By integrating the scheduling association weights of the mother machine and the collaborative association weights of the sub-UAVs, multi-UAV collaborative control strategy data is generated.

[0013] Optionally, the process of updating the UAV cooperative relationship graph data includes: Acquire the mission completion status, remaining battery status, communication status, and flight status from the drone mission execution status data; Adjust the task status characteristics of the drone nodes according to the task completion status, adjust the energy status characteristics of the drone nodes according to the remaining power status, and update the communication association between drone nodes according to the communication status. Based on the flight status detection of spatial position changes between drone nodes, update the spatial association between drone nodes; By integrating updated mission status characteristics, energy status characteristics, communication relationships, and spatial relationships, the UAV collaborative relationship graph data is updated; The updated UAV cooperative relationship graph data is input into the improved Graph Attention Network model to generate new multi-UAV cooperative control strategy data.

[0014] A multi-UAV collaborative control system for disaster relief according to an embodiment of the present invention includes the following modules: The data acquisition and processing module is used to collect environmental data, rescue mission data, and drone status data in disaster areas, perform preprocessing, and generate disaster relief collaborative data. The collaborative relationship graph construction module is used to construct a drone collaborative relationship graph based on the drone's location, communication status, and mission status in disaster relief collaborative data, and generate drone collaborative relationship graph data. The task association attention generation module is used to determine the drone task association relationship based on the drone cooperation relationship graph data and rescue mission data, and generate task association attention feature data. The communication dynamic relationship update module is used to update the connection relationships in the UAV collaborative relationship graph data according to the UAV communication status and generate communication dynamic relationship graph data. The graph attention feature extraction module is used to input task-related attention feature data and communication dynamic relationship graph data into the improved Graph Attention Network model to generate multi-UAV collaborative feature data. The collaborative control strategy generation module is used to calculate the collaborative relationship between the mother drone and the daughter drones based on the collaborative feature data of multiple drones, and generate collaborative control strategy data of multiple drones. The status feedback optimization module is used to update the UAV cooperative relationship graph data based on the UAV mission execution status data, and generate updated multi-UAV cooperative control strategy data.

[0015] The beneficial effects of this invention are: (1) This invention collects environmental data of disaster areas, rescue mission data and UAV status data, constructs a UAV collaborative relationship diagram that includes UAV location, communication status and mission status, and performs fusion analysis by combining mission-related attention feature data and communication dynamic relationship diagram data, thereby improving the comprehensive perception capability of multi-UAV status information in disaster relief scenarios and realizing collaborative control driven by multi-dimensional status information.

[0016] (2) This invention introduces an improved Graph Attention Network model to perform attention calculation on the relationship between UAV nodes, and integrates the features of task relationship, communication state relationship and spatial collaboration relationship, thereby improving the accuracy of multi-UAV task allocation and collaborative decision-making ability, enabling UAVs to dynamically adjust rescue missions according to changes in the disaster site.

[0017] (3) By constructing a hierarchical collaborative control relationship between the mother machine and the sub-UAVs and updating the UAV collaborative relationship diagram data according to the UAV mission execution status, the present invention realizes the dynamic control strategy adjustment during the rescue process, and improves the operational stability and rescue mission execution efficiency of the multi-UAV system in complex disaster environments. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a multi-UAV cooperative control method for disaster relief proposed in this invention; Figure 2 This is a block diagram of a multi-UAV collaborative control system for disaster relief proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-2 A multi-UAV collaborative control method for disaster relief includes the following steps: Collect environmental data, rescue mission data, and drone status data from disaster areas, perform preprocessing, and generate disaster relief collaborative data; Based on the location, communication status, and mission status of drones in disaster relief collaboration data, a drone collaboration relationship diagram is constructed, and drone collaboration relationship diagram data is generated. Based on the drone collaboration relationship graph data and rescue mission data, the drone mission relationships are determined, and mission association attention feature data is generated. Update the connection relationships in the UAV collaboration relationship graph data according to the UAV communication status to generate communication dynamic relationship graph data; Task-related attention feature data and communication dynamic relationship graph data are input into the improved Graph Attention Network model to extract UAV collaborative features and generate multi-UAV collaborative feature data. The collaborative relationship between the mother drone and the daughter drones is calculated based on the collaborative feature data of the multiple drones, and the collaborative control strategy data of the multiple drones is generated. Based on the multi-UAV collaborative control strategy, control UAVs to perform disaster relief missions and collect UAV mission execution status data; The UAV collaborative relationship graph data and communication dynamic relationship graph data are updated based on the UAV mission execution status data. The updated data is then input into the improved Graph Attention Network model to generate updated multi-UAV collaborative control strategy data.

[0021] In this embodiment, the preprocessing process includes: Collect environmental data of disaster areas, rescue mission data, and drone status data; Coordinate transformation and obstacle information extraction are performed on environmental data in disaster areas to generate environmental status data; The rescue mission data is parsed for mission location, encoded for mission type, and labeled for mission priority to generate mission status data; Time synchronization and state feature encoding are performed on UAV state data to generate UAV state feature data; By integrating environmental status data, mission status data, and UAV status characteristic data, collaborative disaster relief data can be generated.

[0022] In this embodiment, the process of constructing the UAV collaborative relationship graph includes: The mother drone and the child drones in the disaster relief collaborative data are taken as drone nodes, and the location status features, communication status features and mission status features of the corresponding drone nodes are extracted. Spatial relationships are determined based on the location distance between drone nodes, communication relationships are determined based on the communication quality between drone nodes, and task relationships are determined based on the task status corresponding to the drone nodes. By integrating spatial, communication, and task-related relationships, the connection relationships between UAV nodes are constructed, generating UAV collaborative relationship graph data.

[0023] In this embodiment, the process of generating task-related attention feature data includes: Obtain the characteristics of drone nodes from the drone collaboration graph data and the task location, task type, and task priority from the rescue task data; The location matching relationship is determined based on the spatial distance between the UAV node location and the mission location, and the mission execution capability matching relationship is determined based on the remaining battery power and payload status in the UAV status data. Based on location matching, task execution capability matching, and task priority, generate task association weights between drones and rescue missions; Attention is calculated on the features of UAV nodes based on task association weights to generate task association attention feature data.

[0024] In this embodiment, the process of generating the communication dynamic relationship graph data includes: Obtain the communication status of UAV nodes in the UAV collaborative relationship graph data, and extract the communication signal strength, data transmission delay, and communication link stability between UAV nodes; The communication association weights between UAV nodes are calculated based on communication signal strength, data transmission delay, and communication link stability. Adjust the connection relationships in the UAV collaborative relationship graph data according to the communication association weight, and update the communication association edges between UAV nodes; Generate dynamic communication relationship graph data based on the updated communication association edges.

[0025] In this embodiment, the step of inputting task-related attention feature data and communication dynamic relationship graph data into the improved Graph Attention Network model includes: Map task-related attention feature data to the UAV node feature space to generate task-enhanced node feature data; Input the task-enhanced node feature data and the communication dynamic relationship graph data into the Graph Attention Network model, and calculate the attention weights between nodes based on the UAV node connection relationships in the communication dynamic relationship graph data. Spatial collaborative feature data is generated by aggregating features of adjacent UAV nodes based on attention weights between nodes. The task-related attention features in the task-enhanced node feature data are fused with the spatial collaboration feature data to generate multi-dimensional collaboration feature data. Graph attention propagation processing is applied to multi-dimensional collaborative feature data to generate multi-UAV collaborative feature data.

[0026] In this embodiment, the process of establishing a collaborative relationship between the host computer and the sub-UAV includes: Acquire drone node features from multi-drone collaborative feature data, and identify the parent and child drone nodes corresponding to the drone nodes; Extract the regional scheduling features corresponding to the mother node and the task execution features corresponding to the child UAV nodes; Calculate the scheduling association weight of the mother machine based on the characteristic association relationship between the mother machine node and the child UAV node, and determine the task allocation relationship between the mother machine node and the child UAV node; The collaborative association weights of sub-drone nodes are calculated based on the task execution characteristics of sub-drone nodes to determine the collaborative execution relationships between sub-drone nodes; By integrating the scheduling association weights of the mother machine and the collaborative association weights of the sub-UAVs, multi-UAV collaborative control strategy data is generated.

[0027] In this embodiment, the process of updating the UAV collaborative relationship graph data includes: Acquire the mission completion status, remaining battery status, communication status, and flight status from the drone mission execution status data; Adjust the task status characteristics of the drone nodes according to the task completion status, adjust the energy status characteristics of the drone nodes according to the remaining power status, and update the communication association between drone nodes according to the communication status. Based on the flight status detection of spatial position changes between drone nodes, update the spatial association between drone nodes; By integrating updated mission status characteristics, energy status characteristics, communication relationships, and spatial relationships, the UAV collaborative relationship graph data is updated; The updated UAV cooperative relationship graph data is input into the improved Graph Attention Network model to generate new multi-UAV cooperative control strategy data.

[0028] A multi-UAV collaborative control system for disaster relief according to an embodiment of the present invention includes the following modules: The data acquisition and processing module is used to collect environmental data, rescue mission data, and drone status data in disaster areas, perform preprocessing, and generate disaster relief collaborative data. The collaborative relationship graph construction module is used to construct a drone collaborative relationship graph based on the drone's location, communication status, and mission status in disaster relief collaborative data, and generate drone collaborative relationship graph data. The task association attention generation module is used to determine the drone task association relationship based on the drone cooperation relationship graph data and rescue mission data, and generate task association attention feature data. The communication dynamic relationship update module is used to update the connection relationships in the UAV collaborative relationship graph data according to the UAV communication status and generate communication dynamic relationship graph data. The graph attention feature extraction module is used to input task-related attention feature data and communication dynamic relationship graph data into the improved Graph Attention Network model to generate multi-UAV collaborative feature data. The collaborative control strategy generation module is used to calculate the collaborative relationship between the mother drone and the daughter drones based on the collaborative feature data of multiple drones, and generate collaborative control strategy data of multiple drones. The status feedback optimization module is used to update the UAV cooperative relationship graph data based on the UAV mission execution status data, and generate updated multi-UAV cooperative control strategy data.

[0029] Example 1: To verify the feasibility of this invention in practice, it was applied to a disaster relief simulation scenario. The scenario was set up so that after a disaster, the rescue area experienced building collapses, road blockages, and unstable communication conditions, requiring the use of multiple UAVs in a collaborative manner to complete disaster area search, target detection, and material delivery tasks. One eVTOL mother drone and multiple sub-UAVs were deployed within the rescue area. The mother drone was responsible for rapidly arriving at the mission area, maintaining airborne communication connections, and scheduling the sub-UAVs. The sub-UAVs then entered specific areas to perform specific rescue tasks.

[0030] During the application process, environmental data of the disaster area, rescue mission data, and UAV status data are first collected. The environmental data includes regional image information, obstacle location information, and environmental hazard level information. The rescue mission data includes mission location, mission type, and mission priority information. The UAV status data includes UAV location coordinates, remaining battery power, payload status, and communication status. The collected data undergoes coordinate transformation, status encoding, and time synchronization processing to generate disaster relief collaborative data, providing a data foundation for subsequent UAV collaborative control.

[0031] Based on the location, communication status, and mission status of UAVs in disaster relief collaborative data, a UAV collaborative relationship diagram is constructed, with the mother UAV and sub-UAVs as UAV nodes. Spatial relationships are determined based on the spatial distance between UAVs, communication relationships are determined based on communication quality, and mission relationships are determined based on mission status, forming a UAV collaborative relationship diagram data containing multi-dimensional relational information. This UAV collaborative relationship diagram describes the information interaction relationships between the mother UAV and sub-UAVs, and among the sub-UAVs themselves, enabling the UAV collaborative control process to adapt to the dynamic changes at the disaster site.

[0032] During task allocation, drone task relationships are determined based on drone collaboration graph data and rescue mission data. Combining the distance between the drone's current location and the mission location, the drone's remaining battery power, drone payload status, and mission priority, the matching relationship between drones and rescue missions is calculated, generating task association attention feature data. Based on this data, drone task execution relationships are adjusted to prioritize drones that are closer, have sufficient energy, and possess the corresponding execution capabilities, thereby improving the efficiency of rescue resource utilization.

[0033] To address the changing communication environment in disaster areas, the system acquires real-time data on communication signal strength, data transmission latency, and communication link stability between drones. Based on this communication status, it updates the connections in the drone collaboration graph data, generating a dynamic communication graph. When some drones enter obstructed areas, causing a decline in communication quality, the system adjusts the information transmission relationships between drones according to the dynamic communication graph data, re-establishing stable communication connections and ensuring the transmission of control commands from the mother drone to the drones.

[0034] Task-related attention feature data and communication dynamic relationship graph data are input into an improved Graph Attention Network model. The association features between UAV nodes are extracted using graph attention calculation methods. Task association relationships, communication state relationships, and spatial cooperation relationships are fused to generate multi-UAV cooperative feature data. Based on this multi-UAV cooperative feature data, the cooperative relationship between the parent UAV and the child UAVs is calculated, generating multi-UAV cooperative control strategy data.

[0035] The system controls drones to perform search, detection, and supply delivery tasks in disaster areas based on multi-drone collaborative control strategy data. During execution, drone mission status data is continuously collected. When the system detects a decrease in drone battery power, reduced communication quality, or a change in mission status, it updates the drone collaborative relationship graph data and communication dynamic relationship graph data based on the drone mission status data. This data is then re-input into an improved Graph Attention Network model to generate new multi-drone collaborative control strategy data, enabling dynamic adjustments during the rescue process.

[0036] To verify the technical effectiveness of this invention, its method was compared with traditional fixed task allocation methods and traditional graph network cooperative control methods. During the tests, multiple UAVs were set up to collaboratively execute search, detection, and material delivery tasks, and indicators such as task completion rate, task response time, communication maintenance rate, and path adjustment frequency were recorded. Test results show that this invention achieves multi-UAV state fusion and dynamic cooperative control through task-related attention features, a dynamic communication relationship graph, and an improved Graph Attention Network model, thereby improving task execution efficiency and control stability in complex disaster environments.

[0037] Table 1: Comparison of the effects of multi-UAV collaborative control As can be seen from the test results in Table 1, the performance of multi-UAV cooperative control is significantly improved after adopting the method of this invention. The task completion rate of the fixed task allocation method is 82.6%, the average task response time is 46.8s, the communication maintenance rate is 78.4%, and the average number of path adjustments is 12.5. The task completion rate of the traditional graph network cooperative control method is increased to 88.3%, the average task response time is reduced to 39.6s, the communication maintenance rate is increased to 86.7%, and the average number of path adjustments is reduced to 9.2.

[0038] After adopting the method of this invention, the task completion rate reached 96.1%, which is 13.5 percentage points higher than the fixed task allocation method and 7.8 percentage points higher than the traditional graph network collaborative control method. This shows that the present invention, through task-related attention feature data and the improved Graph Attention Network model, can more accurately match the UAV status with the rescue mission requirements and improve the collaborative execution capability of multiple UAV missions.

[0039] In terms of mission response efficiency, the method of this invention reduces the average mission response time to 27.4 seconds, which is 19.4 seconds less than the fixed mission allocation method and 12.2 seconds less than the traditional graph network cooperative control method. By constructing a UAV cooperative relationship graph and combining it with communication dynamic relationship graph data, the information interaction relationship between UAVs can be quickly adjusted according to changes in the disaster environment, thereby improving the response speed of rescue missions.

[0040] Regarding communication stability and path adjustment, the method of this invention achieves a communication retention rate of 95.3% and reduces the average number of path adjustments to 5.6. Compared to traditional methods, this invention integrates communication status changes into the multi-UAV collaborative control process by updating UAV cooperative relationship graph data in real time, reducing control fluctuations caused by communication anomalies and path changes, and improving the operational stability of multi-UAV systems in complex disaster environments. The above test data verifies the effectiveness and feasibility of this invention in disaster relief scenarios.

[0041] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A multi-UAV cooperative control method for disaster relief, characterized in that, Includes the following steps: Collect environmental data, rescue mission data, and drone status data from disaster areas, perform preprocessing, and generate disaster relief collaborative data; Based on the location, communication status, and mission status of drones in disaster relief collaboration data, a drone collaboration relationship diagram is constructed, and drone collaboration relationship diagram data is generated. Based on the drone collaboration relationship graph data and rescue mission data, the drone mission relationships are determined, and mission association attention feature data is generated. Update the connection relationships in the UAV collaboration relationship graph data according to the UAV communication status to generate communication dynamic relationship graph data; Task-related attention feature data and communication dynamic relationship graph data are input into the improved Graph Attention Network model to extract UAV collaborative features and generate multi-UAV collaborative feature data. The collaborative relationship between the mother drone and the daughter drones is calculated based on the collaborative feature data of the multiple drones, and the collaborative control strategy data of the multiple drones is generated. Based on the multi-UAV collaborative control strategy, control UAVs to perform disaster relief missions and collect UAV mission execution status data; The UAV collaborative relationship graph data and communication dynamic relationship graph data are updated based on the UAV mission execution status data. The updated data is then input into the improved Graph Attention Network model to generate updated multi-UAV collaborative control strategy data.

2. The multi-UAV collaborative control method for disaster relief according to claim 1, characterized in that, The preprocessing process includes: Collect environmental data of disaster areas, rescue mission data, and drone status data; Coordinate transformation and obstacle information extraction are performed on environmental data in disaster areas to generate environmental status data; The rescue mission data is parsed for mission location, encoded for mission type, and labeled for mission priority to generate mission status data; Time synchronization and state feature encoding are performed on UAV state data to generate UAV state feature data; By integrating environmental status data, mission status data, and UAV status characteristic data, collaborative disaster relief data can be generated.

3. The multi-UAV collaborative control method for disaster relief according to claim 1, characterized in that, The process of constructing the drone collaborative relationship graph includes: The mother drone and the child drones in the disaster relief collaborative data are taken as drone nodes, and the location status features, communication status features and mission status features of the corresponding drone nodes are extracted. Spatial relationships are determined based on the location distance between drone nodes, communication relationships are determined based on the communication quality between drone nodes, and task relationships are determined based on the task status corresponding to the drone nodes. By integrating spatial, communication, and task-related relationships, the connection relationships between UAV nodes are constructed, generating UAV collaborative relationship graph data.

4. The multi-UAV cooperative control method for disaster relief according to claim 1, characterized in that, The process of generating task-related attention feature data includes: Obtain the characteristics of drone nodes from the drone collaboration graph data and the task location, task type, and task priority from the rescue task data; The location matching relationship is determined based on the spatial distance between the UAV node location and the mission location, and the mission execution capability matching relationship is determined based on the remaining battery power and payload status in the UAV status data. Based on location matching, task execution capability matching, and task priority, generate task association weights between drones and rescue missions; Attention is calculated on the features of UAV nodes based on task association weights to generate task association attention feature data.

5. A multi-UAV cooperative control method for disaster relief according to claim 1, characterized in that, The process of generating the communication dynamic relationship graph data includes: Obtain the communication status of UAV nodes in the UAV collaborative relationship graph data, and extract the communication signal strength, data transmission delay, and communication link stability between UAV nodes; The communication association weights between UAV nodes are calculated based on communication signal strength, data transmission delay, and communication link stability. Adjust the connection relationships in the UAV collaborative relationship graph data according to the communication association weight, and update the communication association edges between UAV nodes; Generate dynamic communication relationship graph data based on the updated communication association edges.

6. A multi-UAV cooperative control method for disaster relief according to claim 1, characterized in that, The step of inputting task-related attention feature data and communication dynamic relationship graph data into the improved GraphAttentionNetwork model includes: Map task-related attention feature data to the UAV node feature space to generate task-enhanced node feature data; Input the task-enhanced node feature data and the communication dynamic relationship graph data into the Graph Attention Network model, and calculate the attention weights between nodes based on the UAV node connection relationships in the communication dynamic relationship graph data. Spatial collaborative feature data is generated by aggregating features of adjacent UAV nodes based on attention weights between nodes. The task-related attention features in the task-enhanced node feature data are fused with the spatial collaboration feature data to generate multi-dimensional collaboration feature data. Graph attention propagation processing is applied to multi-dimensional collaborative feature data to generate multi-UAV collaborative feature data.

7. A multi-UAV cooperative control method for disaster relief according to claim 1, characterized in that, The process of calculating the collaborative relationship between the mother machine and the daughter drones includes: Acquire drone node features from multi-drone collaborative feature data, and identify the parent and child drone nodes corresponding to the drone nodes; Extract the regional scheduling features corresponding to the mother node and the task execution features corresponding to the child UAV nodes; Calculate the scheduling association weight of the mother machine based on the characteristic association relationship between the mother machine node and the child UAV node, and determine the task allocation relationship between the mother machine node and the child UAV node; The collaborative association weights of sub-drone nodes are calculated based on the task execution characteristics of sub-drone nodes to determine the collaborative execution relationships between sub-drone nodes; By integrating the scheduling association weights of the mother machine and the collaborative association weights of the sub-UAVs, multi-UAV collaborative control strategy data is generated.

8. A multi-UAV cooperative control method for disaster relief according to claim 1, characterized in that, The process of updating the drone collaboration graph data includes: Acquire the mission completion status, remaining battery status, communication status, and flight status from the drone mission execution status data; Adjust the task status characteristics of the drone nodes according to the task completion status, adjust the energy status characteristics of the drone nodes according to the remaining power status, and update the communication association between drone nodes according to the communication status. Based on the flight status detection of spatial position changes between drone nodes, update the spatial association between drone nodes; By integrating updated mission status characteristics, energy status characteristics, communication relationships, and spatial relationships, the UAV collaborative relationship graph data is updated; The updated UAV cooperative relationship graph data is input into the improved Graph Attention Network model to generate new multi-UAV cooperative control strategy data.

9. A multi-UAV collaborative control system for disaster relief, applied to the multi-UAV collaborative control method for disaster relief as described in any one of claims 1 to 8, characterized in that, Includes the following modules: The data acquisition and processing module is used to collect environmental data, rescue mission data, and drone status data in disaster areas, perform preprocessing, and generate disaster relief collaborative data. The collaborative relationship graph construction module is used to construct a drone collaborative relationship graph based on the drone's location, communication status, and mission status in disaster relief collaborative data, and generate drone collaborative relationship graph data. The task association attention generation module is used to determine the drone task association relationship based on the drone cooperation relationship graph data and rescue mission data, and generate task association attention feature data. The communication dynamic relationship update module is used to update the connection relationships in the UAV collaborative relationship graph data according to the UAV communication status and generate communication dynamic relationship graph data. The graph attention feature extraction module is used to input task-related attention feature data and communication dynamic relationship graph data into the improved Graph Attention Network model to generate multi-UAV collaborative feature data. The collaborative control strategy generation module is used to calculate the collaborative relationship between the mother drone and the daughter drones based on the collaborative feature data of multiple drones, and generate collaborative control strategy data of multiple drones. The status feedback optimization module is used to update the UAV cooperative relationship graph data based on the UAV mission execution status data, and generate updated multi-UAV cooperative control strategy data.

Images