A Dynamic Grouping Method for Heterogeneous UAV Swarms Based on Coalition Game Theory

Abstract

A dynamic grouping method for heterogeneous UAV swarms based on coalition game theory is presented, belonging to the field of large-scale heterogeneous UAV collaborative task execution. This method executes sequentially: receiving dynamic grouping instructions, collecting heterogeneous UAV node states, modeling complex sequential tasks, generating initial coalition structures, calculating coalition comprehensive utility, constructing joint switching operations, iteratively updating the coalition structure, and triggering dynamic regrouping. This method significantly improves the response speed and collaborative reliability of large-scale heterogeneous UAV swarms in continuous tasks such as search-attack missions without altering the UAVs' preset flight paths. Furthermore, by eliminating inefficient communication links and overlapping nodes, it effectively enhances the communication security and anti-interference capabilities of the swarm in sensitive mission environments.

Description

Technical Field

[0001] This invention belongs to the field of large-scale heterogeneous UAV collaborative task execution, and proposes a dynamic grouping method for heterogeneous UAV swarms based on alliance game theory. Background Technology

[0002] In the field of heterogeneous UAV swarms collaboratively executing complex continuous tasks, existing research mainly revolves around alliance formation game theory for task allocation and resource optimization to improve overall swarm efficiency. Traditional alliance formation game theory models treat UAVs as independent players, optimizing individual or collective gains through single-node switching operations between alliances. These models have achieved significant results in static or quasi-static resource optimization problems related to spectrum allocation, bandwidth management, and relay selection. However, when applied to dynamic grouping to optimize communication topology, their limitations become increasingly apparent: First, most traditional models are based on single-node switching mechanisms, meaning a single UAV can only move between two alliances at a time. In dynamic scenarios where heterogeneous UAVs must meet the minimum number of models required for complex tasks, single-point switching easily disrupts the task feasibility of the original alliance, leading to frequent grouping structure failures, slow and inefficient convergence, and difficulty adapting to the real-time response requirements of large-scale swarms. Second, existing typical preference orders, such as Pareto order and selfish order, tend towards extremes. Pareto ordering requires that handovers must not harm the interests of any related nodes, which is too stringent and often makes it difficult to find feasible handover operations under complex constraints. Selfish ordering, on the other hand, focuses only on the interests of the switcher, easily leading to decreased overall utility and getting trapped in local optima, thus failing to guarantee the global stability of the grouping structure. Although some studies have proposed compromise solutions such as bilateral reciprocal transfer ordering, these are still limited to single-node and dual-alliance interaction frameworks and cannot address the complex decision-making needs of multi-node and multi-alliance collaborative handovers. Furthermore, while existing research focuses on communication performance indicators such as energy consumption and signal-to-noise ratio, it often considers grouping structure separately from communication topology optimization and task security constraints. It lacks a unified utility framework that can comprehensively quantify intra-group cohesion, link durability, scale benefits, and cross-group penalties, resulting in a singular optimization objective that fails to balance communication efficiency and security while ensuring basic task completion. Finally, in the grouping initialization phase, most methods employ random or simple enumeration strategies. When the number of UAVs and the complexity of task constraints are high, the computational cost is enormous, failing to meet the requirement of rapid response after task commands are issued, thus limiting the practicality of dynamic grouping mechanisms in real-world tasks. Summary of the Invention

[0003] The main objective of this invention is to overcome the shortcomings of existing alliance-based game theory methods in solving the dynamic grouping problem of heterogeneous UAV swarms, such as low switching efficiency, single optimization objective, weak convergence guarantee, and slow initialization speed. This invention proposes a dynamic grouping method for heterogeneous UAV swarms based on alliance game theory. This method aims to construct an autonomous grouping system that supports multi-node collaborative switching, integrates multiple performance indicators, has stable convergence guarantees, and can be rapidly initialized. This allows for dynamic, secure, and efficient optimization of the communication topology of heterogeneous UAV swarms while meeting the basic requirements of complex continuous tasks. Specifically, the key technical problems to be solved include: game theory modeling and efficiency improvement for multi-node joint switching under multiple constraints; how to overcome the bottleneck of single-node switching; designing a game theory model that allows multiple UAV participants to perform synchronous and collaborative switching between multiple alliances; and defining corresponding joint switching operations and gain calculation rules to significantly accelerate the optimization process of the grouping structure while ensuring that each alliance still meets the minimum number of mission-specific UAVs after switching. The problems include: 1) Designing a novel utility function and preference order driven by comprehensive performance: How to design a multi-dimensional utility function that integrates dynamic stability, static scale benefits, intra-group aggregation degree, and cross-group penalties to comprehensively and balancedly evaluate grouping quality. Based on this, a joint switching preference order that simultaneously considers the interests of switching participants and affected nodes should be designed to guide the grouping process towards convergence in the direction of improving overall utility and avoid getting trapped in local optima. 2) Rapid generation of initial grouping structures under highly complex constraints: How to avoid the huge computational overhead of exhaustive search after receiving grouping instructions? Design a hierarchical greedy selection strategy based on task execution sequences to quickly construct an initial alliance structure that meets all basic task requirements and has a relatively balanced distribution, laying a good foundation for subsequent iterative optimization and significantly reducing initial computational complexity. 3) Theoretical guarantees of convergence and stability of dynamic grouping processes: How to prove that under the proposed game model and preference order, the dynamic grouping process can converge to a Nash-stable final alliance structure after a finite number of iterations, and provide a rigorous theoretical convergence proof from the perspective of exact potential game theory to ensure the reliability and robustness of the proposed method.

[0004] The technical solution of this invention to solve the key technical problem is as follows: To achieve the above objective, this invention provides a dynamic grouping method for heterogeneous UAV swarms based on coalition game theory. This method is executed in the following time sequence: receiving dynamic grouping instructions, collecting heterogeneous UAV node status, modeling complex sequential tasks, generating initial coalition structure, calculating coalition comprehensive utility, constructing joint switching operations, iteratively updating coalition structure, and triggering dynamic regrouping. Specifically, it includes the following steps: S1: Receive dynamic grouping instructions and identify heterogeneous drone cluster objects; Within the mission area, the heterogeneous UAV swarm receives dynamic grouping instructions. These instructions include the basic mission requirements for the complex sequential task, the types of heterogeneous UAV nodes participating in the task, the number of nodes of each type, and communication constraints within the mission area. Based on these dynamic grouping instructions, the heterogeneous UAV swarm participating in the dynamic grouping is represented as follows: in, This indicates a heterogeneous drone swarm participating in dynamic grouping; This represents the total number of types of heterogeneous drone nodes; Indicates the first The first of the types One heterogeneous drone node; Indicates the heterogeneous drone node type number, and ; This indicates the sequence number of heterogeneous drone nodes of the same type, and ; Indicates the first The number of heterogeneous drone nodes of each type; Represents the total number of heterogeneous drone nodes, and satisfies: Based on the heterogeneous drone cluster obtained in step S1 The subsequent step S2 further collects the status information of each heterogeneous drone node, and uses this as the data basis for the initial alliance structure generation and alliance comprehensive utility calculation.

[0005] S2: Collect the status information of heterogeneous drone nodes and construct the initial communication topology model of the heterogeneous drone cluster; Based on the heterogeneous drone swarm determined in step S1 At the current moment Collect the status information of each heterogeneous drone node, and then... Type 1 Heterogeneous drone nodes The status information is represented as: in, Representing heterogeneous drone nodes At the present moment Status information; Representing heterogeneous drone nodes Position and heading information in a fixed coordinate system; Representing heterogeneous drone nodes Speed ​​information within the machine system; Representing heterogeneous drone nodes The inherent task performance information.

[0006] in: in, Representing heterogeneous drone nodes The horizontal coordinate in a fixed coordinate system; Representing heterogeneous drone nodes The vertical coordinate in a fixed coordinate system; Representing heterogeneous drone nodes The heading angle; Representing heterogeneous drone nodes Longitudinal velocity within the machine system; Representing heterogeneous drone nodes Lateral velocity within the machine system; Representing heterogeneous drone nodes angular velocity; superscript Represents the transpose of a matrix or vector.

[0007] Based on the state information of the heterogeneous drone nodes, an initial communication topology model for the heterogeneous drone cluster is constructed: in, This represents the initial communication topology model of a heterogeneous drone swarm. Represents a heterogeneous set of drone nodes; Let represent the set of communication edges between heterogeneous UAV nodes. Let the maximum communication radius of the heterogeneous UAV nodes be . When the distance between two heterogeneous drone nodes does not exceed At that time, a communication edge can be established between the two.

[0008] Based on the state information obtained in step S2 and initial communication topology model The next step, S3, is to construct a complex sequential task model and determine the number of task execution alliances and the basic task requirements of each task execution alliance.

[0009] S3: Construct a complex sequential task model and determine the basic task requirements; Based on the dynamic grouping command received in step S1 and the state information of the heterogeneous UAV nodes obtained in step S2, the complex sequential task model is represented as follows: in, Represents a complex sequential task model; Indicates the first Sub-tasks; Indicates the subtask sequence number, and ; This indicates the total number of subtasks.

[0010] The execution order of subtasks in a complex sequential task model is represented as follows: in, This indicates the sequential execution relationship between subtasks, meaning that the later subtask is executed after the previous subtask is completed or meets the conditions for execution.

[0011] A basic task requirement that can independently satisfy the execution requirements of subtasks in a complex sequential task model can be represented as: in, This represents the basic task requirements that a single task execution alliance must satisfy to complete a complex sequential task model. Indicates the first task in a single task execution alliance Minimum required number of heterogeneous drone nodes for each type; Indicates the first A collection of heterogeneous drone nodes of various types.

[0012] Assuming that the first type of heterogeneous UAV nodes are used to execute the initial subtask, the number of task execution alliances is determined by the number of the first type of heterogeneous UAV nodes: in, Indicates the number of task execution alliances; This indicates the number of heterogeneous drone nodes of type 1.

[0013] Based on the basic task requirements obtained in step S3 and the number of mission execution alliances The subsequent step S4 uses a hierarchical greedy selection strategy to generate an initial alliance structure that meets the basic task requirements.

[0014] S4: Generate the initial alliance structure based on a hierarchical greedy selection strategy; Based on the number of task execution alliances obtained in step S3 First establish Initial task execution alliance: in, Indicates the initial alliance structure; Indicates the first An initial task execution alliance; Indicates the task execution alliance number, and .

[0015] The first The three types of heterogeneous UAV nodes serve as the leader nodes of each initial task execution alliance, resulting in: in, Indicates the first The first of the types One heterogeneous drone node; Further indicate that The first leader node An initial task execution alliance.

[0016] Then, following the execution order of the subtasks, perform the following... Hierarchical classification of heterogeneous UAV nodes of different types: in, Indicates the first The supply and demand gap of different types of heterogeneous drone nodes.

[0017] when When, it indicates the first The number of heterogeneous drone nodes of this type is insufficient, necessitating the generation of cross-federation heterogeneous drone nodes; when When, it indicates the first The number of nodes for each type of heterogeneous UAV is equal to the number required for the basic mission; when When, it indicates the first There is a sufficient number of heterogeneous drone nodes of this type, and there are redundant heterogeneous drone nodes.

[0018] In the In the process of allocating heterogeneous UAV nodes of various types, based on the results obtained in step S2 Calculate the first The virtual centroid position of the previous type of heterogeneous drone node in the task execution consortium: in, Indicates the first The first task execution alliance The virtual centroid location of heterogeneous UAV nodes of various types; Indicates the first The first task execution alliance The number of heterogeneous drone nodes of each type; Indicates the first The first of the types One heterogeneous drone node; Representing heterogeneous drone nodes Position and heading information; Indicates the first A task execution alliance.

[0019] Further calculation of candidate heterogeneous drone nodes With the Distance between virtual centroids of heterogeneous drone nodes of the same type in a task execution consortium: in, Indicates candidate heterogeneous drone nodes With the The distance between the virtual centroids of a heterogeneous UAV node of a certain type in a task execution consortium; This represents the L2 norm.

[0020] Based on distance The size, arranged in ascending order of the number of... Different types of heterogeneous UAV nodes are sequentially added to each task execution alliance until all task execution alliances meet the basic task requirements. When the first When the number of heterogeneous drone nodes of a certain type is insufficient, the already allocated nodes will be used to... One type of heterogeneous drone node joins other task execution alliances that do not meet basic task requirements as a cross-alliance heterogeneous drone node; when the first When there are enough heterogeneous drone nodes of each type, redundant heterogeneous drone nodes will be allocated to the corresponding task execution alliance according to the principle of proximity.

[0021] Based on the above processing, an initial alliance structure that meets the basic task requirements is obtained. This initial alliance structure serves as input for subsequent steps S5 to calculate the alliance's overall utility and S6 to construct a candidate set of joint switching operations.

[0022] S5: Calculate the overall utility of the alliance based on the initial alliance structure; Based on the initial alliance structure obtained in step S4 Execute the alliance for any of these tasks. Calculate the overall utility of the consortium. First, calculate the task execution consortium. Virtual centroid: in, Indicates the Task Execution Alliance The virtual centroid position; Indicates the Task Execution Alliance Total number of heterogeneous unmanned aerial vehicle (UAV) nodes in China; Representing heterogeneous drone nodes Belongs to the Task Execution Alliance .

[0023] According to the Mission Execution Alliance The distance from each heterogeneous UAV node to the virtual centroid is used to calculate the coalition cohesion index: in, Indicates the Task Execution Alliance The alliance cohesion index; the larger the value, the more concentrated the spatial distribution of heterogeneous UAV nodes within the mission execution alliance.

[0024] Calculate the first The first task execution alliance Scale and return metrics for different types of heterogeneous drone nodes: in, Indicates the Task Execution Alliance The Middle Scale and revenue metrics for different types of heterogeneous drone nodes; The slope adjustment parameter represents the scale return function; This represents the adjustment parameter for the maximum effective return of the scale-return function; This represents a reference value for returns to scale. Indicates the Task Execution Alliance The Middle The number of heterogeneous drone nodes of each type; Represents the natural constant.

[0025] Mission Execution Alliance The total scale return metric is: in, Indicates the Task Execution Alliance Total size return metric.

[0026] Computational Task Execution Alliance Cross-league penalty indicators: in, Indicates the Task Execution Alliance Cross-league penalty indicators; Indicates the cross-league penalty coefficient, and ; This indicates the number of task execution alliances that the same heterogeneous drone node participates in simultaneously; Indicates the Task Execution Alliance China participated simultaneously The number of cross-alliance heterogeneous drone nodes in each task execution alliance.

[0027] Further Computation Task Execution Consortium stability index If two heterogeneous drone nodes execute adjacent subtasks, the link uptime is expressed as: Indicates the first The first of the types Heterogeneous drone nodes: in, Indicates execution of the first Heterogeneous UAV nodes for each sub-task and execution of the first Link maintenance time between heterogeneous drone nodes in each sub-task; Indicates the first The first of the types One heterogeneous drone node; Indicates the first The serial number of the heterogeneous UAV node of each type; Representing heterogeneous drone nodes Speed ​​information; Representing heterogeneous drone nodes Position and heading information; Indicates the Task Execution Alliance The Middle The number of heterogeneous drone nodes of each type.

[0028] If two heterogeneous drone nodes perform the same subtask, the link uptime is expressed as: in, This indicates the link maintenance time between two heterogeneous drone nodes performing the same subtask; Indicates the first The first of the types One heterogeneous drone node; Indicates the difference between the same type and Another heterogeneous UAV node serial number; Representing heterogeneous drone nodes Position and heading information; Representing heterogeneous drone nodes Speed ​​information.

[0029] Mission Execution Alliance The stability index is: in, Indicates the Task Execution Alliance The shortest duration of the communication link; the larger the value, the stronger the task execution alliance. The better the communication stability.

[0030] Finally, the Mission Execution Alliance The overall utility of the alliance is: in, Indicates the Task Execution Alliance The overall effectiveness of the alliance; Indicates the weight of the stability index; Indicates the weight of the total scale return indicator; Indicates the weight of the alliance cohesion index; Represents the weight of the cross-alliance penalty indicator, and satisfies: Alliance Structure The total utility is: in, Indicates alliance structure Total utility.

[0031] Based on the overall utility of the alliance obtained in step S5 Total utility of alliance structure The subsequent step S6 constructs a joint switching alliance to form a game model, and selects executable joint switching operations based on the joint switching gain.

[0032] S6: Construct a joint handover alliance to form a game model and generate a candidate set of joint handover operations; Based on the comprehensive utility of the alliance obtained in step S5, the dynamic grouping problem is modeled as a game model of joint switching alliance formation with transferable utility: in, This indicates a game theory model for the formation of a joint switching alliance; This represents a set of non-cross-alliance heterogeneous drone nodes; This represents a collection of heterogeneous drone nodes across alliances. This represents the group decision vector for all heterogeneous drone nodes.

[0033] For any heterogeneous UAV node Calculate individual utility according to the principle of proportionality: in, Representing heterogeneous drone nodes Individual utility; This indicates the presence of heterogeneous drone nodes. A collection of task execution alliances; Represents a set The members of the group.

[0034] Define the joint handover operation as follows: in, This represents the set of heterogeneous drone nodes to be jointly switched. In the relevant task execution alliance The joint handover operation performed on the above; This represents the set of heterogeneous UAV nodes that perform the joint handover operation. This represents the set of task execution federations whose membership has changed due to a federation switching operation; Indicates the first before the joint handover operation A related task execution alliance; Indicates the first handover operation after the joint handover. A related task execution alliance; The relevant task execution alliance number indicates the membership change. Indicates from the mission execution alliance A collection of heterogeneous drone nodes that have left the region; Indicates joining the mission execution alliance A heterogeneous set of drone nodes; symbol Represents the set difference operation; symbol This indicates the set union operation.

[0035] The joint handover gain for the joint handover operation is: in, Indicates joint handover operation Joint switching gain; This indicates the number of task execution federations whose membership has changed due to a federation handover operation; Indicates the first handover operation after the joint handover. The overall effectiveness of a coalition of related task execution alliances; Indicates the first time before the joint handover operation The overall effectiveness of the alliance for executing related tasks.

[0036] A joint handover operation is added to the candidate set of joint handover operations when it simultaneously meets the following two conditions: First, each task execution federation after the joint handover operation still meets the basic task requirements determined in step S3; Second, the joint handover gain... .

[0037] Based on the joint handover operation candidate set obtained in step S6, the subsequent step S7 performs iterative updates according to the joint handover common improvement preference order.

[0038] S7: Iterative update of the alliance structure based on joint switching to improve preference order; Based on the candidate set of joint handover operations obtained in step S6, the optimal joint handover operation is selected by jointly improving the preference order through joint handover. This applies to a heterogeneous set of UAV nodes that share the same joint handover operation. The two alliance structures formed and If the following conditions are met: Then determine: in, This represents the first candidate alliance structure; This indicates the second candidate alliance structure; Indicates the set of heterogeneous drone nodes to be jointly switched. Preferences regarding alliance structures; This represents a set of heterogeneous UAV nodes that are affected by the joint handover operation but do not directly perform the joint handover operation. Represents a set Heterogeneous drone nodes in the process; Represents a set Heterogeneous drone nodes in the process; Representing heterogeneous drone nodes The second candidate alliance structure Individual utility under; Representing heterogeneous drone nodes The second candidate alliance structure Individual utility under; Representing heterogeneous drone nodes First candidate alliance structure Individual utility under; Representing heterogeneous drone nodes First candidate alliance structure Individual utility.

[0039] In each iteration, calculate the joint switching gain of all joint switching operations that satisfy the joint switching common improvement preference order, and select the joint switching operation with the largest joint switching gain: in, This represents the optimal joint switching operation executed in the current iteration. Indicates the candidate set for joint handover operations; This represents the joint handover operation that maximizes the joint handover gain; This indicates the joint switching operations that can be executed in the current iteration; This represents the difference between the total utility after performing the joint handover operation and the total utility before performing the joint handover operation.

[0040] implement Then, update the current alliance structure: in, Indicates the first Alliance structure before round iteration; Indicates the first The alliance structure after rounds of iteration; Indicates the iteration round.

[0041] Repeat steps S6 and S7 until no joint handover operation exists that satisfies the basic task requirements and has a joint handover gain greater than zero. At this point, the final alliance structure is obtained. in, Indicates the final alliance structure; Indicates the first in the final alliance structure A task execution alliance.

[0042] The final alliance structure satisfies: in, In the final alliance structure The next constructible joint switching operation; this formula indicates that in the final coalition structure, there is no joint switching operation that can further improve the overall utility of the coalition for related task execution, therefore the final coalition structure is a Nash stable coalition structure.

[0043] S8: Output the final alliance structure and execute dynamic regrouping trigger; Based on the final alliance structure obtained in step S7 The alliance affiliation result of each heterogeneous drone node is output to the heterogeneous drone cluster communication management module, enabling heterogeneous drone nodes within the same task execution alliance to establish communication links according to the task execution order, and enabling different task execution alliances to exchange information through their respective leader nodes.

[0044] To adapt to the communication topology changes caused by the movement of heterogeneous UAV nodes, the total utility of the alliance structure in step S5 is considered. Configure a dynamic regrouping trigger mechanism. When the total utility of the current alliance structure falls below a preset utility threshold, trigger threshold-based regrouping: in, This indicates the preset utility threshold.

[0045] When the total utility of the current alliance structure is not lower than a preset utility threshold, time-based regrouping is triggered according to a preset time interval: in, Indicates the current time; Indicates the time when dynamic regrouping was last performed; This indicates the preset regrouping time interval.

[0046] When the threshold-based or time-based regrouping conditions are met, the current alliance structure is used as the new initial alliance structure. The process returns to step S5 to recalculate the overall alliance utility, and steps S6 to S8 are then executed. This achieves continuous dynamic optimization of the communication topology of heterogeneous UAV swarms.

[0047] The beneficial effects of this invention are: The above-described method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory offers the following advantages compared to existing methods for task allocation and grouping of heterogeneous UAV swarms based on coalition game theory: 1) Addressing the shortcomings of existing coalition formation game theory methods, such as low efficiency and easy destruction of grouping feasibility due to single-node switching mechanisms under complex task constraints, this invention innovatively proposes a joint switching coalition formation game model supporting multi-node collaboration and designs corresponding joint switching operations and gain calculation rules. This method allows a group of heterogeneous UAVs to switch synchronously between multiple communication groups, and can adjust the structure of multiple groups at once while meeting the minimum task UAV type constraint. This significantly reduces the number of iterations required for the grouping structure to reach a stable state, greatly improving the grouping response speed and optimization efficiency of large-scale heterogeneous swarms in dynamic environments. 2) Addressing the problems of excessively strict Pareto constraints and selfish orders being prone to local optima in existing preference orders, this invention proposes joint switching to jointly improve preference orders and designs a comprehensive utility function that integrates dynamic stability, static scale benefits, intra-group cohesion, and cross-group penalties. JSCI preference order guides the search process towards improving global returns by simultaneously considering the overall utility changes of switching participants and affected nodes, effectively avoiding the optimization process from getting trapped in local optima. The multi-dimensional utility function ensures that the optimization objective not only focuses on communication performance but also strengthens the grouping's ability to guarantee basic task requirements and suppress security risks, thereby improving task execution reliability while enhancing the security and anti-interference capabilities of the cluster communication topology. 3) Addressing the problem of high computational overhead in initial group generation under high constraints and the lack of fast initialization strategies in existing methods, this invention designs a greedy selection algorithm for forming basic alliance structures based on task hierarchy. This algorithm, according to the subtask execution order, uses a "nearest distance" greedy strategy to aggregate heterogeneous UAVs layer by layer. After receiving the grouping instruction for the first time, it can quickly generate an initial group that meets all basic task requirements and has a relatively reasonable distribution with extremely low computational complexity, providing a high-quality starting point for subsequent iterative optimization. This effectively solves the real-time bottleneck in the dynamic grouping initialization stage and improves the overall practicality and scalability of the method. 4) Addressing the limitation of existing dynamic grouping methods lacking strict convergence theoretical guarantees, this invention conducts in-depth stability analysis of the proposed game model and algorithm. By constructing an exact potential function, we rigorously prove that the proposed joint switching alliance formation game under the JSCI preference order is an exact potential game, and therefore at least one pure policy Nash equilibrium exists. Furthermore, combining the algorithm's iterative process, we prove that regardless of the feasible initial structure, the proposed algorithm converges to a Nash-stable final alliance structure within a finite number of iterations. This theoretical guarantee lays a solid foundation for the reliability and robustness of the method, ensuring its deterministic performance in complex dynamic environments. Attached Figure Description

[0048] Figure 1 This is a typical scenario for collaborative tasks in complex sequences; Figure 2 It is the basic alliance structure in the dynamic grouping process of a heterogeneous drone swarm consisting of four U1 drones, seven U2 drones, and three U3 drones in scenarios where the number of drones is insufficient. Figure 3 It is the final alliance structure after joint switching operations during the dynamic grouping process of a heterogeneous drone cluster consisting of four U1 drones, seven U2 drones, and three U3 drones in scenarios where the number of drones is insufficient. Figure 4 This refers to the convergence behavior of the algorithm during the dynamic grouping process of a heterogeneous drone swarm consisting of four U1 drones, seven U2 drones, and three U3 drones in scenarios where the number of drones is insufficient. Figure 5 It is the basic alliance structure in the dynamic grouping process of a heterogeneous drone swarm consisting of four U1 drones, eight U2 drones, and four U3 drones in a balanced quantity scenario. Figure 6 It is the final formation structure after joint switching operations during the dynamic grouping process of a heterogeneous drone swarm consisting of four U1 drones, eight U2 drones, and four U3 drones in a balanced quantity scenario. Figure 7 It is the convergence behavior of the algorithm during the dynamic grouping process of a heterogeneous drone swarm consisting of four U1 drones, eight U2 drones, and four U3 drones in a balanced quantity scenario. Figure 8 It is the basic alliance structure in the dynamic grouping process of a heterogeneous drone swarm consisting of four U1 drones, ten U2 drones, and four U3 drones in a scenario with surplus numbers. Figure 9 It is the final formation structure after joint switching operations during the dynamic grouping process of a heterogeneous drone cluster consisting of four U1 drones, ten U2 drones, and four U3 drones in a scenario with surplus numbers. Figure 10 It refers to the convergence behavior of the algorithm during the dynamic grouping process of a heterogeneous drone swarm consisting of four U1 drones, ten U2 drones, and four U3 drones in a scenario with surplus drones. Detailed Implementation

[0049] The technical solution of the present invention will be further described below through specific embodiments.

[0050] A dynamic grouping method for heterogeneous UAV swarms based on coalition game theory is proposed. This method executes in the following temporal order: receiving dynamic grouping instructions, collecting heterogeneous UAV node states, modeling complex sequential tasks, generating the initial coalition structure, calculating the coalition's overall utility, constructing joint switching operations, iteratively updating the coalition structure, and triggering dynamic regrouping. Specifically, it includes the following steps: S1: Receive dynamic grouping instructions and identify heterogeneous drone cluster objects; Within the mission area, the heterogeneous UAV swarm receives dynamic grouping instructions. These instructions include the basic mission requirements for the complex sequential task, the types of heterogeneous UAV nodes participating in the task, the number of nodes of each type, and communication constraints within the mission area. Based on these dynamic grouping instructions, the heterogeneous UAV swarm participating in the dynamic grouping is represented as follows: in, This indicates a heterogeneous drone swarm participating in dynamic grouping; This represents the total number of types of heterogeneous drone nodes; Indicates the first The first of the types One heterogeneous drone node; Indicates the heterogeneous drone node type number, and ; This indicates the sequence number of heterogeneous drone nodes of the same type, and ; Indicates the first The number of heterogeneous drone nodes of each type; Represents the total number of heterogeneous drone nodes, and satisfies: Based on the heterogeneous drone cluster obtained in step S1 The subsequent step S2 further collects the status information of each heterogeneous drone node, and uses this as the data basis for the initial alliance structure generation and alliance comprehensive utility calculation.

[0051] S2: Collect the status information of heterogeneous drone nodes and construct the initial communication topology model of the heterogeneous drone cluster; Based on the heterogeneous drone swarm determined in step S1 At the current moment Collect the status information of each heterogeneous drone node, and then... Type 1 Heterogeneous drone nodes The status information is represented as: in, Representing heterogeneous drone nodes At the present moment Status information; Representing heterogeneous drone nodes Position and heading information in a fixed coordinate system; Representing heterogeneous drone nodes Speed ​​information within the machine system; Representing heterogeneous drone nodes The inherent task performance information.

[0052] in: in, Representing heterogeneous drone nodes The horizontal coordinate in a fixed coordinate system; Representing heterogeneous drone nodes The vertical coordinate in a fixed coordinate system; Representing heterogeneous drone nodes The heading angle; Representing heterogeneous drone nodes Longitudinal velocity within the machine system; Representing heterogeneous drone nodes Lateral velocity within the machine system; Representing heterogeneous drone nodes angular velocity; superscript Represents the transpose of a matrix or vector.

[0053] Based on the state information of the heterogeneous drone nodes, an initial communication topology model for the heterogeneous drone cluster is constructed: in, This represents the initial communication topology model of a heterogeneous drone swarm. Represents a heterogeneous set of drone nodes; Let represent the set of communication edges between heterogeneous UAV nodes. Let the maximum communication radius of the heterogeneous UAV nodes be . When the distance between two heterogeneous drone nodes does not exceed At that time, a communication edge can be established between the two.

[0054] Based on the state information obtained in step S2 and initial communication topology model The next step, S3, is to construct a complex sequential task model and determine the number of task execution alliances and the basic task requirements of each task execution alliance.

[0055] S3: Construct a complex sequential task model and determine the basic task requirements; Based on the dynamic grouping command received in step S1 and the state information of the heterogeneous UAV nodes obtained in step S2, the complex sequential task model is represented as follows: in, Represents a complex sequential task model; Indicates the first Sub-tasks; Indicates the subtask sequence number, and ; This indicates the total number of subtasks.

[0056] The execution order of subtasks in a complex sequential task model is represented as follows: in, This indicates the sequential execution relationship between subtasks, meaning that the later subtask is executed after the previous subtask is completed or meets the conditions for execution.

[0057] A basic task requirement that can independently satisfy the execution requirements of subtasks in a complex sequential task model can be represented as: in, This represents the basic task requirements that a single task execution alliance must satisfy to complete a complex sequential task model. Indicates the first task in a single task execution alliance Minimum required number of heterogeneous drone nodes for each type; Indicates the first A collection of heterogeneous drone nodes of various types.

[0058] Assuming that the first type of heterogeneous UAV nodes are used to execute the initial subtask, the number of task execution alliances is determined by the number of the first type of heterogeneous UAV nodes: in, Indicates the number of task execution alliances; This indicates the number of heterogeneous drone nodes of type 1.

[0059] Based on the basic task requirements obtained in step S3 and the number of mission execution alliances The subsequent step S4 uses a hierarchical greedy selection strategy to generate an initial alliance structure that meets the basic task requirements.

[0060] S4: Generate the initial alliance structure based on a hierarchical greedy selection strategy; Based on the number of task execution alliances obtained in step S3 First establish Initial task execution alliance: in, Indicates the initial alliance structure; Indicates the first An initial task execution alliance; Indicates the task execution alliance number, and .

[0061] The first The three types of heterogeneous UAV nodes serve as the leader nodes of each initial task execution alliance, resulting in: in, Indicates the first The first of the types One heterogeneous drone node; Further indicate that The first leader node An initial task execution alliance.

[0062] Then, following the execution order of the subtasks, perform the following... Hierarchical classification of heterogeneous UAV nodes of different types: in, Indicates the first The supply and demand gap of different types of heterogeneous drone nodes.

[0063] when When, it indicates the first The number of heterogeneous drone nodes of this type is insufficient, necessitating the generation of cross-federation heterogeneous drone nodes; when When, it indicates the first The number of nodes for each type of heterogeneous UAV is equal to the number required for the basic mission; when When, it indicates the first There is a sufficient number of heterogeneous drone nodes of this type, and there are redundant heterogeneous drone nodes.

[0064] In the In the process of allocating heterogeneous UAV nodes of various types, based on the results obtained in step S2 Calculate the first The virtual centroid position of the previous type of heterogeneous drone node in the task execution consortium: in, Indicates the first The first task execution alliance The virtual centroid location of heterogeneous UAV nodes of various types; Indicates the first The first task execution alliance The number of heterogeneous drone nodes of each type; Indicates the first The first of the types One heterogeneous drone node; Representing heterogeneous drone nodes Position and heading information; Indicates the first A task execution alliance.

[0065] Further calculation of candidate heterogeneous drone nodes With the Distance between virtual centroids of heterogeneous drone nodes of the same type in a task execution consortium: in, Indicates candidate heterogeneous drone nodes With the The distance between the virtual centroids of a heterogeneous UAV node of a certain type in a task execution consortium; This represents the L2 norm.

[0066] Based on distance The size, arranged in ascending order of the number of... Different types of heterogeneous UAV nodes are sequentially added to each task execution alliance until all task execution alliances meet the basic task requirements. When the first When the number of heterogeneous drone nodes of a certain type is insufficient, the already allocated nodes will be used to... One type of heterogeneous drone node joins other task execution alliances that do not meet basic task requirements as a cross-alliance heterogeneous drone node; when the first When there are enough heterogeneous drone nodes of each type, redundant heterogeneous drone nodes will be allocated to the corresponding task execution alliance according to the principle of proximity.

[0067] Based on the above processing, an initial alliance structure that meets the basic task requirements is obtained. This initial alliance structure serves as input for subsequent steps S5 to calculate the alliance's overall utility and S6 to construct a candidate set of joint switching operations.

[0068] S5: Calculate the overall utility of the alliance based on the initial alliance structure; Based on the initial alliance structure obtained in step S4 Execute the alliance for any of these tasks. Calculate the overall utility of the consortium. First, calculate the task execution consortium. Virtual centroid: in, Indicates the Task Execution Alliance The virtual centroid position; Indicates the Task Execution Alliance Total number of heterogeneous unmanned aerial vehicle (UAV) nodes in China; Representing heterogeneous drone nodes Belongs to the Task Execution Alliance .

[0069] According to the Mission Execution Alliance The distance from each heterogeneous UAV node to the virtual centroid is used to calculate the coalition cohesion index: in, Indicates the Task Execution Alliance The alliance cohesion index; the larger the value, the more concentrated the spatial distribution of heterogeneous UAV nodes within the mission execution alliance.

[0070] Calculate the first The first task execution alliance Scale and return metrics for different types of heterogeneous drone nodes: in, Indicates the Task Execution Alliance The Middle Scale and revenue metrics for different types of heterogeneous drone nodes; The slope adjustment parameter represents the scale return function; This represents the adjustment parameter for the maximum effective return of the scale-return function; This represents a reference value for returns to scale. Indicates the Task Execution Alliance The Middle The number of heterogeneous drone nodes of each type; Represents the natural constant.

[0071] Mission Execution Alliance The total scale return metric is: in, Indicates the Task Execution Alliance Total size return metric.

[0072] Computational Task Execution Alliance Cross-league penalty indicators: in, Indicates the Task Execution Alliance Cross-league penalty indicators; Indicates the cross-league penalty coefficient, and ; This indicates the number of task execution alliances that the same heterogeneous drone node participates in simultaneously; Indicates the Task Execution Alliance China participated simultaneously The number of cross-alliance heterogeneous drone nodes in each task execution alliance.

[0073] Further Computation Task Execution Consortium stability index If two heterogeneous drone nodes execute adjacent subtasks, the link uptime is expressed as: Indicates the first The first of the types Heterogeneous drone nodes: in, Indicates execution of the first Heterogeneous UAV nodes for each sub-task and execution of the first Link maintenance time between heterogeneous drone nodes in each sub-task; Indicates the first The first of the types One heterogeneous drone node; Indicates the first The serial number of the heterogeneous UAV node of each type; Representing heterogeneous drone nodes Speed ​​information; Representing heterogeneous drone nodes Position and heading information; Indicates the Task Execution Alliance The Middle The number of heterogeneous drone nodes of each type.

[0074] If two heterogeneous drone nodes perform the same subtask, the link uptime is expressed as: in, This indicates the link maintenance time between two heterogeneous drone nodes performing the same subtask; Indicates the first The first of the types One heterogeneous drone node; Indicates the difference between the same type and Another heterogeneous UAV node serial number; Representing heterogeneous drone nodes Position and heading information; Representing heterogeneous drone nodes Speed ​​information.

[0075] Mission Execution Alliance The stability index is: in, Indicates the Task Execution Alliance The shortest duration of the communication link; the larger the value, the stronger the task execution alliance. The better the communication stability.

[0076] Finally, the Mission Execution Alliance The overall utility of the alliance is: in, Indicates the Task Execution Alliance The overall effectiveness of the alliance; Indicates the weight of the stability index; Indicates the weight of the total scale return indicator; Indicates the weight of the alliance cohesion index; Represents the weight of the cross-alliance penalty indicator, and satisfies: Alliance Structure The total utility is: in, Indicates alliance structure Total utility.

[0077] Based on the overall utility of the alliance obtained in step S5 Total utility of alliance structure The subsequent step S6 constructs a joint switching alliance to form a game model, and selects executable joint switching operations based on the joint switching gain.

[0078] S6: Construct a joint handover alliance to form a game model and generate a candidate set of joint handover operations; Based on the comprehensive utility of the alliance obtained in step S5, the dynamic grouping problem is modeled as a game model of joint switching alliance formation with transferable utility: in, This indicates a game theory model for the formation of a joint switching alliance; This represents a set of non-cross-alliance heterogeneous drone nodes; This represents a collection of heterogeneous drone nodes across alliances. This represents the group decision vector for all heterogeneous drone nodes.

[0079] For any heterogeneous UAV node Calculate individual utility according to the principle of proportionality: in, Representing heterogeneous drone nodes Individual utility; This indicates the presence of heterogeneous drone nodes. A collection of task execution alliances; Represents a set The members of the group.

[0080] Define the joint handover operation as follows: in, This represents the set of heterogeneous drone nodes to be jointly switched. In the relevant task execution alliance The joint handover operation performed on the above; This represents the set of heterogeneous UAV nodes that perform the joint handover operation. This represents the set of task execution federations whose membership has changed due to a federation switching operation; Indicates the first before the joint handover operation A related task execution alliance; Indicates the first handover operation after the joint handover. A related task execution alliance; The relevant task execution alliance number indicates the membership change. Indicates from the mission execution alliance A collection of heterogeneous drone nodes that have left the region; Indicates joining the mission execution alliance A heterogeneous set of drone nodes; symbol Represents the set difference operation; symbol This indicates the set union operation.

[0081] The joint handover gain for the joint handover operation is: in, Indicates joint handover operation Joint switching gain; This indicates the number of task execution federations whose membership has changed due to a federation handover operation; Indicates the first handover operation after the joint handover. The overall effectiveness of a coalition of related task execution alliances; Indicates the first time before the joint handover operation The overall effectiveness of the alliance for executing related tasks.

[0082] A joint handover operation is added to the candidate set of joint handover operations when it simultaneously meets the following two conditions: First, each task execution federation after the joint handover operation still meets the basic task requirements determined in step S3; Second, the joint handover gain... .

[0083] Based on the joint handover operation candidate set obtained in step S6, the subsequent step S7 performs iterative updates according to the joint handover common improvement preference order.

[0084] S7: Iterative update of the alliance structure based on joint switching to improve preference order; Based on the candidate set of joint handover operations obtained in step S6, the optimal joint handover operation is selected by jointly improving the preference order through joint handover. This applies to a heterogeneous set of UAV nodes that share the same joint handover operation. The two alliance structures formed and If the following conditions are met: Then determine: in, This represents the first candidate alliance structure; This indicates the second candidate alliance structure; Indicates the set of heterogeneous drone nodes to be jointly switched. Preferences regarding alliance structures; This represents a set of heterogeneous UAV nodes that are affected by the joint handover operation but do not directly perform the joint handover operation. Represents a set Heterogeneous drone nodes in the process; Represents a set Heterogeneous drone nodes in the process; Representing heterogeneous drone nodes The second candidate alliance structure Individual utility under; Representing heterogeneous drone nodes The second candidate alliance structure Individual utility under; Representing heterogeneous drone nodes First candidate alliance structure Individual utility under; Representing heterogeneous drone nodes First candidate alliance structure Individual utility.

[0085] In each iteration, calculate the joint switching gain of all joint switching operations that satisfy the joint switching common improvement preference order, and select the joint switching operation with the largest joint switching gain: in, This represents the optimal joint switching operation executed in the current iteration. Indicates the candidate set for joint handover operations; This represents the joint handover operation that maximizes the joint handover gain; This indicates the joint switching operations that can be executed in the current iteration; This represents the difference between the total utility after performing the joint handover operation and the total utility before performing the joint handover operation.

[0086] implement Then, update the current alliance structure: in, Indicates the first Alliance structure before round iteration; Indicates the first The alliance structure after rounds of iteration; Indicates the iteration round.

[0087] Repeat steps S6 and S7 until no joint handover operation exists that satisfies the basic task requirements and has a joint handover gain greater than zero. At this point, the final alliance structure is obtained. in, Indicates the final alliance structure; Indicates the first in the final alliance structure A task execution alliance.

[0088] The final alliance structure satisfies: in, In the final alliance structure The next constructible joint switching operation; this formula indicates that in the final coalition structure, there is no joint switching operation that can further improve the overall utility of the coalition for related task execution, therefore the final coalition structure is a Nash stable coalition structure.

[0089] S8: Output the final alliance structure and execute dynamic regrouping trigger; Based on the final alliance structure obtained in step S7 The alliance affiliation result of each heterogeneous drone node is output to the heterogeneous drone cluster communication management module, enabling heterogeneous drone nodes within the same task execution alliance to establish communication links according to the task execution order, and enabling different task execution alliances to exchange information through their respective leader nodes.

[0090] To adapt to the communication topology changes caused by the movement of heterogeneous UAV nodes, the total utility of the alliance structure in step S5 is considered. Configure a dynamic regrouping trigger mechanism. When the total utility of the current alliance structure falls below a preset utility threshold, trigger threshold-based regrouping: in, This indicates the preset utility threshold.

[0091] When the total utility of the current alliance structure is not lower than a preset utility threshold, time-based regrouping is triggered according to a preset time interval: in, Indicates the current time; Indicates the time when dynamic regrouping was last performed; This indicates the preset regrouping time interval.

[0092] When the threshold-based or time-based regrouping conditions are met, the current alliance structure is used as the new initial alliance structure. The process returns to step S5 to recalculate the overall alliance utility, and steps S6 to S8 are then executed. This achieves continuous dynamic optimization of the communication topology of heterogeneous UAV swarms.

[0093] Simulation experiments verify: To further demonstrate that the heterogeneous UAV swarm dynamic grouping method based on alliance game theory described in this invention can be repeatedly implemented by those skilled in the art, and to verify its effectiveness under different heterogeneous UAV node resource conditions and complex sequential task constraints, this simulation experiment takes the execution of complex sequential tasks by a heterogeneous UAV swarm as the object, and proceeds in the order of "simulation parameter setting - initial alliance structure generation - joint switching iterative optimization - result comparison and analysis" to prove that this invention can achieve rapid dynamic grouping and stable optimization of the communication topology of heterogeneous UAV swarms while meeting basic task requirements.

[0094] 1. Simulation experiment scenario and task settings: The simulation experiment sets up three types of heterogeneous UAV nodes, denoted as Type 1 heterogeneous UAV nodes. Type 2 heterogeneous drone nodes And the third type of heterogeneous drone node A complex sequential task consists of three subtasks, executed in the following order: in, Indicates the first subtask. This indicates the second subtask. Indicates the third subtask. This indicates the sequential execution relationship between subtasks. To maintain consistency with the basic task requirements in the technical solution of this invention, the basic task requirements for a single task execution alliance to complete a full, complex sequential task are defined as follows: in, Represents the basic task requirement vector; This represents the minimum required number of heterogeneous drone nodes of type 1 in a single task execution alliance; This represents the minimum required number of heterogeneous drone nodes of type 2 in a single task execution alliance; This represents the minimum required number of heterogeneous drone nodes of type 3 in a single task execution alliance. For example... Figure 1 As shown, this setup corresponds to a typical complex sequential task scenario in which the first type of heterogeneous drone node performs the initial subtask, two second type heterogeneous drone nodes collaboratively perform the intermediate subtask, and a third type heterogeneous drone node performs the subsequent subtask.

[0095] The number of mission execution alliances is determined by the number of heterogeneous UAV nodes of type 1, that is: in, Indicates the number of task execution alliances. This represents the number of heterogeneous UAV nodes of type 1. To verify the feasibility of this invention under different resource conditions, the following three typical simulation scenarios are set up:

[0096] The three scenarios described above are used to verify whether the present invention can maintain mission feasibility by using heterogeneous drone nodes across alliances when there are insufficient heterogeneous drone nodes; whether the present invention can improve the overall utility of the alliance structure by using joint switching operations when there are balanced heterogeneous drone nodes; and whether the present invention can evenly distribute redundant heterogeneous drone nodes to different task execution alliances when there are sufficient heterogeneous drone nodes, thereby avoiding excessive aggregation or excessive communication burden in some task execution alliances.

[0097] 2. Assigning values ​​to simulation experiment parameters: In the simulation experiment, heterogeneous UAV nodes are randomly distributed within the mission area and execute dynamic grouping commands without changing the preset flight paths. The flight speeds of each heterogeneous UAV node are randomly generated within a given range, and the communication capability between the heterogeneous UAV nodes is limited by the maximum communication radius. The simulation experiment parameters are set as follows:

[0098] in, Indicates the size of the task area; A unified representation of the velocity of heterogeneous UAV nodes; Indicates the maximum communication radius of heterogeneous drone nodes; Indicates the number of task execution alliances; , , , Let represent the weights of the stability index, total size return index, alliance cohesion index, and cross-alliance penalty index in the alliance's overall utility function, respectively, and satisfy the following: In each simulation experiment, an initial alliance structure is first generated based on the number of heterogeneous UAV nodes and basic mission requirements. Then, iterative optimization is performed based on the joint handover alliance formation algorithm until there are no joint handover operations with a joint handover gain greater than zero that satisfy the basic mission requirements. To reduce the impact of random initial distribution on simulation results, 50 independent simulations are performed for each set of comparative experiments, and the average value is used for evaluation.

[0099] 1. Evaluation Indicator Setting: To quantitatively evaluate the effectiveness of the method of this invention, average total utility, average number of convergence iterations, proportion of cross-alliance heterogeneous UAV nodes, relative communication overhead, and average running time are set as evaluation indicators for simulation experiments.

[0100] Average total utility represents the average total utility of the final alliance structure after 50 independent simulations, and is calculated as follows: in, This represents average total utility; Indicates the number of independent simulations; Indicates the first Sub-independent simulation; Indicates the first The final alliance structure obtained from each independent simulation; Indicates the first The final total utility of the alliance structure obtained from each independent simulation.

[0101] The average number of convergence iterations represents the average number of iterations required for the algorithm to converge from the initial coalition structure to the final coalition structure, and is calculated as follows: in, Indicates the average number of convergence iterations; Indicates the first The number of convergence iterations in each independent simulation.

[0102] The cross-alliance heterogeneous drone node ratio represents the proportion of cross-alliance heterogeneous drone nodes to the total number of heterogeneous drone nodes in the final alliance structure. The calculation formula is: in, Indicates the proportion of heterogeneous drone nodes across alliances; Indicates the number of heterogeneous drone nodes across alliances; This represents the total number of heterogeneous drone nodes participating in dynamic grouping.

[0103] Average running time represents the average computation time of the algorithm across multiple independent simulations, and is calculated as follows: in, Indicates the average running time; Indicates the first Runtime of each independent simulation.

[0104] The relative communication overhead is normalized based on the joint handover alliance formation algorithm described in this invention, and the calculation formula is as follows: in, Indicates relative communication overhead; Indicates the communication overhead of the algorithms to be compared; This represents the communication overhead of the joint handover alliance formation algorithm described in this invention. When When, it indicates that the communication overhead of the algorithm to be compared is the same as that of the method of this invention; when When the communication overhead of the algorithm to be compared is higher than that of the method of the present invention, it indicates that the communication overhead of the algorithm to be compared is higher than that of the method of the present invention.

[0105] 2. Simulation Experiment Process: The simulation experiment was conducted according to the following procedure: 1) In Three types of heterogeneous drone nodes are generated within the task area, and each heterogeneous drone node is randomly assigned a... Flight speed; 2) Based on the basic task requirements The number of task execution alliances is determined by the number of heterogeneous drone nodes of type 1. ; 3) Based on a hierarchical greedy selection strategy, heterogeneous UAV nodes are aggregated layer by layer according to the execution order of subtasks to generate an initial alliance structure that meets the basic task requirements; 4) Calculate the overall alliance utility of each task execution alliance based on the initial alliance structure, and generate a candidate set of joint switching operations according to the joint switching operation rules; 5) For each joint handover operation candidate set, calculate the joint handover gain and determine whether it meets the basic task requirements; 6) Select the joint handover operation that satisfies the joint handover common improvement preference order and maximizes the joint handover gain, and update the current alliance structure; 7) Repeat steps 4 through 6 until no further joint switching operations are available that can improve the overall utility of the task execution alliance, and output the final group structure. and alliance structure; 8) Statistical analysis of average total utility, average number of convergence iterations, proportion of heterogeneous drone nodes across alliances, relative communication overhead, and average runtime.

[0106] 3. Simulation results and discussion under different resource conditions: In scenarios with insufficient heterogeneous drone nodes, the system needs to minimize the number of heterogeneous drone nodes across alliances while ensuring that each task execution alliance meets its basic task requirements. For example... Figure 2 , Figure 3 , Figure 4 As shown, in Under certain conditions, a hierarchical greedy selection strategy can first generate an initial alliance structure that meets basic task requirements. Subsequently, through joint switching operations, the members of relevant task execution alliances are further adjusted, so that the final alliance structure improves the overall utility of the alliance structure while satisfying task feasibility. This result shows that, under conditions of insufficient resources, this invention can avoid task execution alliance failure by setting up controlled cross-alliance heterogeneous UAV nodes.

[0107] In heterogeneous drone node balancing scenarios, such as Figure 5 , Figure 6 , Figure 7 As shown, the number of heterogeneous drone nodes just meets the basic mission requirements, that is... At this point, the hierarchical greedy selection strategy enables heterogeneous drone nodes of various types to be evenly distributed to different task execution alliances according to basic task requirements, and avoids the generation of unnecessary independent nodes or heterogeneous drone nodes across alliances. Furthermore, the joint switching operation can improve the overall utility of the alliance structure without destroying the basic task requirements, and eventually converge to a Nash-stable alliance structure.

[0108] In scenarios with an abundance of heterogeneous drone nodes, such as Figure 8 , Figure 9 , Figure 10 As shown, the heterogeneous UAV node composition is set as follows: Under these conditions, the second type of heterogeneous UAV nodes exhibits redundancy. Simulation results show that the hierarchical greedy selection strategy can first ensure that each task execution alliance meets the basic task requirements, and then evenly distribute redundant heterogeneous UAV nodes to different task execution alliances. The joint switching alliance formation algorithm further optimizes the alliance structure through joint switching operations, preventing redundant heterogeneous UAV nodes from being overly concentrated in a single task execution alliance, thereby reducing communication congestion and security risks.

[0109] The convergence curves of the three scenarios described above show that after a finite number of joint switching operations, the total utility of the alliance structure converges to a stable value. Even if the local utility of individual task execution alliances decreases in some iterations, the total utility of the alliance structure can still continuously improve or remain stable, indicating that joint switching to improve the preference order can avoid the local optimum problem caused by only considering the interests of a single heterogeneous UAV node.

Claims

1. A method for dynamic grouping of heterogeneous drone swarms based on coalition game theory, characterized in that, Specifically, the following steps are included: S1: Receive dynamic grouping instructions and identify heterogeneous drone cluster objects; S2: Collect the status information of heterogeneous drone nodes and construct the initial communication topology model of the heterogeneous drone cluster; S3: Construct a complex sequential task model and determine the basic task requirements; Based on the dynamic grouping command received in step S1 and the state information of the heterogeneous UAV nodes obtained in step S2, the complex sequential task model is represented as follows: in, Represents a complex sequential task model; Indicates the first Sub-tasks; Indicates the subtask sequence number, and ; Indicates the total number of subtasks; A basic task requirement that can independently satisfy the execution requirements of subtasks in a complex sequential task model can be represented as: in, This represents the basic task requirements that a single task execution alliance must satisfy to complete a complex sequential task model. Indicates the first task in a single task execution alliance Minimum required number of heterogeneous drone nodes for each type; Indicates the first A collection of heterogeneous drone nodes of various types; S4: Generate the initial alliance structure based on a hierarchical greedy selection strategy; S5: Calculate the overall utility of the alliance based on the initial alliance structure; S6: Construct a joint handover alliance to form a game model and generate a candidate set of joint handover operations; Based on the comprehensive utility of the alliance obtained in step S5, the dynamic grouping problem is modeled as a game model of joint switching alliance formation with transferable utility: in, This indicates a game theory model for the formation of a joint switching alliance; This represents a set of non-cross-alliance heterogeneous drone nodes; This represents a collection of heterogeneous drone nodes across alliances. This represents the group decision vector for all heterogeneous drone nodes; Indicates an alliance structure; Indicates the Task Execution Alliance The overall effectiveness of the alliance; A joint handover operation is added to the candidate set of joint handover operations when it simultaneously meets the following two conditions: First, each task execution federation after the joint handover operation still meets the basic task requirements determined in step S3; Second, the joint handover gain... ; in, Indicates joint handover operation Joint switching gain; S7: Iterative update of the alliance structure based on joint switching to improve preference order; Based on the candidate set of joint handover operations obtained in step S6, the optimal joint handover operation is selected by jointly improving the preference order of joint handover; for a heterogeneous set of UAV nodes with the same joint handover operation to be executed. The two alliance structures formed and If the following conditions are met: Then determine: in, This represents the first candidate alliance structure; This indicates the second candidate alliance structure; Indicates the set of heterogeneous drone nodes to be jointly switched. Preference relationship regarding alliance structure; This represents a set of heterogeneous UAV nodes that are affected by the joint handover operation but do not directly perform the joint handover operation. Represents a set Heterogeneous drone nodes in the process; Represents a set Heterogeneous drone nodes in the process; Representing heterogeneous drone nodes Second candidate alliance structure Individual utility under; Representing heterogeneous drone nodes The second candidate alliance structure Individual utility under; Representing heterogeneous drone nodes First candidate alliance structure Individual utility under; Representing heterogeneous drone nodes First candidate alliance structure Individual utility under; S8: Output the final alliance structure and execute dynamic regrouping trigger.

2. The method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory according to claim 1, characterized in that, The specific steps of S1 are as follows: Within the mission area, the heterogeneous UAV swarm receives dynamic grouping instructions. These instructions include the basic mission requirements for the complex sequential task, the types of heterogeneous UAV nodes participating in the task, the number of nodes of each type, and communication constraints within the mission area. Based on these dynamic grouping instructions, the heterogeneous UAV swarm participating in the dynamic grouping is represented as follows: in, This indicates a heterogeneous drone swarm participating in dynamic grouping; This represents the total number of types of heterogeneous drone nodes; Indicates the first The first of the types One heterogeneous drone node; Indicates the heterogeneous drone node type number, and ; This indicates the sequence number of heterogeneous drone nodes of the same type, and ; Indicates the first The number of heterogeneous drone nodes of each type; Represents the total number of heterogeneous drone nodes, and satisfies: 。 3. The method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory according to claim 2, characterized in that, The specific steps of step S2 are as follows: Based on the heterogeneous drone swarm determined in step S1 At the current moment Collect the status information of each heterogeneous drone node, and then... Type 1 Heterogeneous drone nodes The status information is represented as: in, Representing heterogeneous drone nodes At the present moment Status information; Representing heterogeneous drone nodes Position and heading information in a fixed coordinate system; Representing heterogeneous drone nodes Speed ​​information within the machine system; Representing heterogeneous drone nodes The inherent task performance information; in: in, Representing heterogeneous drone nodes The horizontal coordinate in a fixed coordinate system; Representing heterogeneous drone nodes The vertical coordinate in a fixed coordinate system; Representing heterogeneous drone nodes The heading angle; Representing heterogeneous drone nodes Longitudinal velocity within the machine system; Representing heterogeneous drone nodes Lateral velocity within the machine system; Representing heterogeneous drone nodes angular velocity; superscript Represents the transpose of a matrix or vector; Based on the state information of the heterogeneous drone nodes, an initial communication topology model for the heterogeneous drone cluster is constructed: in, This represents the initial communication topology model of a heterogeneous drone swarm. Represents a heterogeneous set of drone nodes; Let represent the set of communication edges between heterogeneous UAV nodes; let the maximum communication radius of the heterogeneous UAV nodes be . When the distance between two heterogeneous drone nodes does not exceed At that time, a communication edge can be established between the two.

4. The method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory according to claim 3, characterized in that, The specific steps of step S3 are as follows: The execution order of subtasks in a complex sequential task model is represented as follows: in, This indicates the sequential execution relationship between subtasks, meaning that the later subtask is executed after the previous subtask is completed or meets the conditions for execution. Assuming that the first type of heterogeneous UAV nodes are used to execute the initial subtask, the number of task execution alliances is determined by the number of the first type of heterogeneous UAV nodes: in, Indicates the number of task execution alliances; This indicates the number of heterogeneous drone nodes of type 1.

5. The method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory according to claim 4, characterized in that, The specific steps of S4 are as follows: Based on the number of task execution alliances obtained in step S3 First establish Initial task execution alliance: in, Indicates the initial alliance structure; Indicates the first An initial task execution alliance; Indicates the task execution alliance number, and ; The first The three types of heterogeneous UAV nodes serve as the leader nodes of each initial task execution alliance, resulting in: in, Indicates the first The first of the types One heterogeneous drone node; Further indicate that The first leader node An initial task execution alliance; Then, following the execution order of the subtasks, perform the following... Hierarchical classification of heterogeneous UAV nodes of various types: in, Indicates the first The supply and demand gap of different types of heterogeneous drone nodes; when When, it indicates the first The number of heterogeneous drone nodes of this type is insufficient, necessitating the generation of cross-federation heterogeneous drone nodes; when When, it indicates the first The number of nodes for each type of heterogeneous UAV is equal to the number required for the basic mission; when When, it indicates the first There is a sufficient number of heterogeneous drone nodes of this type, and there are redundant heterogeneous drone nodes. In the In the process of allocating heterogeneous UAV nodes of various types, based on the results obtained in step S2 Calculate the first The virtual centroid position of the previous type of heterogeneous drone node in the task execution consortium: in, Indicates the first The first task execution alliance The virtual centroid location of heterogeneous UAV nodes of various types; Indicates the first The first task execution alliance The number of heterogeneous drone nodes of each type; Indicates the first The first of the types One heterogeneous drone node; Representing heterogeneous drone nodes Position and heading information; Indicates the first A task execution alliance; Further calculation of candidate heterogeneous drone nodes With the Distance between virtual centroids of heterogeneous drone nodes of the same type in a task execution consortium: in, Indicates candidate heterogeneous drone nodes With the The distance between the virtual centroids of heterogeneous drone nodes of the same type in a task execution consortium; Represents the L2 norm; Based on distance The size, arranged in ascending order of the number of... Different types of heterogeneous UAV nodes are sequentially added to each task execution alliance until all task execution alliances meet the basic task requirements. When the first When the number of heterogeneous drone nodes of a certain type is insufficient, the already allocated nodes will be used to... One type of heterogeneous drone node joins other task execution alliances that do not meet basic task requirements as a cross-alliance heterogeneous drone node; when the first When there are enough heterogeneous drone nodes of each type, redundant heterogeneous drone nodes will be allocated to the corresponding task execution alliance according to the principle of nearest distance.

6. The method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory according to claim 5, characterized in that, The specific steps of step S5 are as follows: Based on the initial alliance structure obtained in step S4 Execute the alliance for any of these tasks. Calculate the overall utility of the consortium; first, calculate the task execution consortium. Virtual centroid: in, Indicates the Task Execution Alliance The virtual centroid position; Indicates the Task Execution Alliance Total number of heterogeneous drone nodes in China; Representing heterogeneous drone nodes Belongs to the Task Execution Alliance ; According to the Mission Execution Alliance The distance from each heterogeneous UAV node to the virtual centroid is used to calculate the coalition cohesion index: in, Indicates the Task Execution Alliance The alliance cohesion index; the larger the value, the more concentrated the spatial distribution of heterogeneous UAV nodes within the mission execution alliance. Calculate the first The first task execution alliance Scale and return metrics for different types of heterogeneous drone nodes: in, Indicates the Task Execution Alliance The Middle Scale and revenue metrics for different types of heterogeneous drone nodes; The slope adjustment parameter represents the scale return function; This represents the adjustment parameter for the maximum effective return of the scale-return function; This represents a reference value for returns to scale. Indicates the Task Execution Alliance The Middle The number of heterogeneous drone nodes of each type; Represents the natural constant; Mission Execution Alliance The total scale return metric is: in, Indicates the Task Execution Alliance Total scale return metric; Computational Task Execution Alliance Cross-league penalty indicators: in, Indicates the Task Execution Alliance Cross-league penalty indicators; Indicates the cross-league penalty coefficient, and ; This indicates the number of task execution alliances that the same heterogeneous drone node can participate in simultaneously; Indicates the Task Execution Alliance China participated simultaneously The number of cross-alliance heterogeneous drone nodes in each task execution alliance; Further Computation Task Execution Consortium stability index If two heterogeneous drone nodes execute adjacent subtasks, the link uptime is expressed as: Indicates the first The first of the types Heterogeneous drone nodes: in, Indicates execution of the first Heterogeneous UAV nodes for each sub-task and execution of the first Link maintenance time between heterogeneous drone nodes in each sub-task; Indicates the first The first of the types One heterogeneous drone node; Indicates the first The serial number of the heterogeneous UAV node of each type; Representing heterogeneous drone nodes Speed ​​information; Representing heterogeneous drone nodes Position and heading information; Indicates the Task Execution Alliance The Middle The number of heterogeneous drone nodes of each type; If two heterogeneous drone nodes perform the same subtask, the link uptime is expressed as: in, This indicates the link maintenance time between two heterogeneous drone nodes performing the same subtask; Indicates the first The first of the types One heterogeneous drone node; Indicates the difference between the same type and Another heterogeneous UAV node serial number; Representing heterogeneous drone nodes Position and heading information; Representing heterogeneous drone nodes Speed ​​information; Mission Execution Alliance The stability index is: in, Indicates the Task Execution Alliance The shortest duration of the communication link; the larger the value, the stronger the task execution alliance. The better the communication stability; Finally, the Mission Execution Alliance The overall utility of the alliance is: in, Indicates the Task Execution Alliance The overall effectiveness of the alliance; Indicates the weight of the stability index; Indicates the weight of the total scale return indicator; Indicates the weight of the alliance cohesion index; Represents the weight of the cross-alliance penalty indicator, and satisfies: Alliance Structure The total utility is: in, Indicates alliance structure Total utility.

7. A method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory as described in claim 6, characterized in that, The specific steps of step S6 are as follows: For any heterogeneous UAV node Calculate individual utility according to the principle of proportionality: in, Representing heterogeneous drone nodes Individual utility; This indicates the presence of heterogeneous drone nodes. A collection of task execution alliances; Represents a set Members; Define the joint handover operation as follows: in, This represents the set of heterogeneous drone nodes to be jointly switched. In the relevant task execution alliance The joint handover operation performed on the above; This represents the set of heterogeneous UAV nodes that perform the joint handover operation. This represents the set of task execution federations whose membership has changed due to a federation switching operation; Indicates the first before the joint handover operation A related task execution alliance; Indicates the first handover operation after the joint handover. A related task execution alliance; The relevant task execution alliance number indicates the membership change. Indicates from the mission execution alliance A collection of heterogeneous drone nodes that have left the region; Indicates joining the mission execution alliance A heterogeneous set of drone nodes; symbol Represents the set difference operation; symbol This represents the set union operation; The joint handover gain for the joint handover operation is: in, This indicates the number of task execution federations whose membership has changed due to a federation handover operation; Indicates the first handover operation after the joint handover. The overall effectiveness of a coalition of related task execution alliances; Indicates the first time before the joint handover operation The overall effectiveness of the alliance for executing related tasks.

8. The method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory according to claim 7, characterized in that, The specific steps of step S7 are as follows: In each iteration, calculate the joint switching gain of all joint switching operations that satisfy the joint switching common improvement preference order, and select the joint switching operation with the largest joint switching gain: in, This represents the optimal joint switching operation executed in the current iteration. Indicates the candidate set for joint handover operations; This represents the joint handover operation that maximizes the joint handover gain; This indicates the joint switching operations that can be executed in the current iteration; This represents the difference between the total utility after performing the joint handover operation and the total utility before performing the joint handover operation; implement Then, update the current alliance structure: in, Indicates the first Alliance structure before round iteration; Indicates the first The alliance structure after rounds of iteration; Indicates the iteration round; Repeat steps S6 and S7 until no joint handover operation exists that satisfies the basic task requirements and has a joint handover gain greater than zero. At this point, the final alliance structure is obtained. in, Indicates the final alliance structure; Indicates the first in the final alliance structure A task execution alliance; The final alliance structure satisfies: in, In the final alliance structure The next constructible joint switching operation; this formula indicates that in the final coalition structure, there is no joint switching operation that can further improve the overall utility of the coalition for related task execution, therefore the final coalition structure is a Nash stable coalition structure.

9. A method for dynamic grouping of heterogeneous UAV swarms based on coalition game theory as described in claim 8, characterized in that, The specific steps of S8 are as follows: Based on the final alliance structure obtained in step S7 The alliance affiliation result of each heterogeneous drone node is output to the heterogeneous drone cluster communication management module, so that heterogeneous drone nodes in the same task execution alliance can establish communication links according to the task execution order, and different task execution alliances can exchange information through their respective leader nodes. To adapt to the communication topology changes caused by the movement of heterogeneous UAV nodes, the total utility of the alliance structure in step S5 is considered. Configure a dynamic regrouping trigger mechanism; when the total utility of the current alliance structure is lower than a preset utility threshold, trigger threshold-based regrouping: in, Indicates the preset utility threshold; When the total utility of the current alliance structure is not lower than a preset utility threshold, time-based regrouping is triggered according to a preset time interval: in, Indicates the current moment; Indicates the time when dynamic regrouping was last performed; Indicates the preset regrouping time interval; When the threshold-based regrouping condition or the time-based regrouping condition is met, the current alliance structure is used as the new initial alliance structure, and the process returns to step S5 to recalculate the alliance's overall utility, and continues to execute steps S6 to S8; thereby achieving continuous dynamic optimization of the heterogeneous UAV swarm communication topology.

Images