Multi-level collaborative decision-making method and device

Abstract

The invention relates to a multi-level collaborative decision-making method and device, and relates to the field of unmanned aerial vehicles. The method comprises the following steps: aiming at each unmanned aerial vehicle in an unmanned aerial vehicle cluster, constructing a local decision function corresponding to each unmanned aerial vehicle; for a sub-cluster composed of a plurality of unmanned aerial vehicles, constructing a cluster decision function corresponding to each sub-cluster, and for an unmanned aerial vehicle cluster, constructing a global decision function corresponding to the unmanned aerial vehicle cluster; and optimizing the cooperative control decision of the unmanned aerial vehicle cluster according to the global decision function, the cluster decision function and the local decision function. Through a multi-level collaborative decision-making mechanism, it is ensured that the unmanned aerial vehicle cluster realizes high coordination and cooperation among local autonomous decision-making, cluster collaborative decision-making and global task planning. The multi-level decision model not only improves the autonomy of the unmanned aerial vehicle in the cluster, but also enhances the adaptability of the cluster to a complex and dynamic environment, can flexibly execute a multi-task scene in a complex combat environment, and achieves a more efficient and more reliable combat effect.

Description

Technical Field

[0001] The present application relates to the field of drone technology, and in particular to a multi-level collaborative decision-making method and device. Background Art

[0002] With the rapid development of drone technology, drone swarm formations are widely used in military, logistics, emergency rescue and other fields.

[0003] However, in a multi-task environment, the task allocation and collaborative decision-making mechanism of drone swarms is relatively basic and fails to fully consider the complexity of tasks and multi-level decision-making requirements. Most of the task allocation in existing systems is based on centralized decision-making, which cannot achieve adaptive task allocation and dynamic adjustment of large-scale drone swarms, and the flexibility and efficiency of task execution are limited. Summary of the invention

[0004] Based on this, it is necessary to provide a multi-level collaborative decision-making method and device that can improve the autonomy of each drone under the condition of high coordination and cooperation of the drone cluster, thereby realizing flexible and efficient control of the drone cluster.

[0005] In a first aspect, the present application provides a multi-level collaborative decision-making method. The method comprises:

[0006] For each drone in the drone cluster, the local decision function corresponding to each drone is used to obtain the drone's perception information and the status information of neighboring drones;

[0007] Based on the perception information and state information, for the sub-clusters composed of several drones, the cluster decision function corresponding to each sub-cluster is constructed, and for the drone cluster, the global decision function corresponding to the drone cluster is constructed;

[0008] According to the global decision function, cluster decision function and local decision function, the collaborative control decision of the UAV cluster is optimized.

[0009] In one embodiment, according to the global decision function, the cluster decision function and the local decision function, optimizing the coordinated control decision of the drone cluster includes:

[0010] The collaborative control decision-making adopts a top-down coordination mechanism and a bottom-up feedback mechanism; among them, the coordination order of the top-down coordination mechanism is UAV cluster, sub-cluster and UAV, and the feedback order of the bottom-up feedback mechanism is UAV, sub-cluster and UAV cluster.

[0011] In one embodiment, the method further comprises:

[0012] Dynamically allocate communication resources based on the decision-making needs of drones, sub-clusters, and drone clusters;

[0013] The computing resources of edge computing nodes are dynamically allocated according to the computing needs of drones, sub-clusters, and drone clusters.

[0014] In one embodiment, 6G communication modules are used for data exchange between drones and between drones and edge computing nodes.

[0015] In one embodiment, the method further comprises:

[0016] Use a combination of distributed computing and edge computing to obtain collaborative control strategies;

[0017] Among them, the distributed computing and edge computing collaborative control strategy allocates computing resources with the goal of minimizing the time cost of completing computing tasks.

[0018] In a second aspect, the present application also provides a multi-level collaborative decision-making device. The device includes:

[0019] The local decision module is used to obtain the perception information of each drone and the status information of neighboring drones by using the local decision function corresponding to each drone in the drone cluster;

[0020] The cluster and global decision module is used to construct the cluster decision function corresponding to each sub-cluster composed of several drones based on the perception information and state information, and to construct the global decision function corresponding to the drone cluster for the drone cluster;

[0021] The collaborative decision-making module is used to optimize the collaborative control decision of the UAV cluster based on the global decision function, cluster decision function and local decision function.

[0022] In a third aspect, the present application further provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps in the above-mentioned multi-level collaborative decision-making method when executing the computer program.

[0023] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps in the above-mentioned multi-level collaborative decision-making method when executed by a processor.

[0024] In a fifth aspect, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps in the multi-level collaborative decision-making method are implemented.

[0025] In a sixth aspect, the present application further provides a combined and divided attack method based on multi-level collaborative decision-making. Based on the above multi-level collaborative decision-making method, the combined and divided attack method includes:

[0026] Obtain mission requirements, space environment and drone state vector in real time; mission requirements include target relative position and target point position, and space environment includes obstacle position;

[0027] Obtain consistent control law based on the state vector and the target relative position;

[0028] Based on the state vector, a convergence attraction force is generated around the target point position, and a convergence repulsion force is generated around the obstacle position and the adjacent UAV. The convergence control law is obtained according to the convergence attraction and convergence repulsion forces.

[0029] Based on the state vector, a split repulsive force is generated around the center position of the drone cluster, and the split repulsive force and the combined repulsive force are used to obtain the split control law.

[0030] According to the consistency control law, the merging control law and the split-attack control law, an optimization model with the goal of optimizing the task execution efficiency is constructed. The formation and merging-split-attack strategies of the UAV cluster are obtained by solving the optimization model.

[0031] The above-mentioned multi-level collaborative decision-making method and device, for each drone in the drone cluster, uses the local decision function corresponding to each drone to obtain the perception information of the drone and the status information of the neighboring drones; based on the perception information and status information, for the sub-cluster composed of several drones, a cluster decision function corresponding to each sub-cluster is constructed, and for the drone cluster, a global decision function corresponding to the drone cluster is constructed; according to the global decision function, the cluster decision function and the local decision function, the collaborative control decision of the drone cluster is optimized. Through the multi-level collaborative decision-making mechanism, it is ensured that the drone cluster achieves a high degree of coordination and cooperation between local autonomous decision-making, cluster collaborative decision-making and global task planning. This multi-level decision-making model not only improves the autonomy of drones in the cluster, but also enhances the adaptability of the cluster in the face of complex and dynamic environments. It can flexibly execute multi-task scenarios in complex combat environments and achieve more efficient and reliable combat effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 A schematic diagram of a multi-level collaborative decision-making method in one embodiment;

[0033] Figure 2 It is a structural block diagram of a multi-level collaborative decision-making device in an embodiment. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0035] In order to effectively coordinate the decisions at different levels, a hierarchical optimization framework is introduced, which performs collaborative decision-making at three levels: local (i.e., individual), regional (i.e., cluster), and global.

[0036] The present application embodiment provides a multi-level collaborative decision-making method, such as Figure 1 As shown, the following steps are included:

[0037] Step 102, for each drone in the drone cluster, using the local decision function corresponding to each drone, obtain the perception information of the drone and the status information of the neighboring drones.

[0038] At the local level, each drone makes independent decisions based on its own perception information, mainly dealing with the drone’s path planning, obstacle avoidance, and local task execution. The local decision-making goal of drone i can be expressed as a local optimization problem

[0039]

[0040] in: is the local objective function of UAV i, which may include multiple factors such as path planning, obstacle avoidance, and energy consumption; u i (t) is the control input of UAV i at time t; x i (t) is the state vector of UAV i at time t (including position, speed, etc.). Local decision-making can obtain the perception information of the UAV and the state information of neighboring UAVs in real time to ensure the real-time performance of path planning and obstacle avoidance.

[0041] Step 104, based on the perception information and state information, for the sub-clusters composed of several drones, a cluster decision function corresponding to each sub-cluster is constructed, and for the drone cluster, a global decision function corresponding to the drone cluster is constructed.

[0042] At the cluster level, a sub-cluster consisting of several drones makes collaborative decisions based on regional mission objectives to ensure efficient collaboration between drones within the cluster. The goal of this level is to optimize the execution of local collaborative tasks of the entire cluster, such as collective advance, target siege, joint detection, etc.

[0043] The cluster collaborative decision-making problem can be expressed as a distributed optimization problem:

[0044]

[0045] Among them: J cluster is the objective function of the cluster, which may include tasks such as target capture, task allocation, and joint detection; X C (t) = [x 1 (t), x 2 (t), ..., xk (t)] is the state matrix of the UAVs in the cluster; U C (t)=[u 1 (t),u 2 (t), ..., u k (t)] is the control input of the drones in the cluster; k represents the number of drones in the cluster. The collaborative decision-making of the cluster can be achieved through the "consensus control" mechanism. This mechanism enables each drone in the cluster to gradually tend to the shared decision results through communication between adjacent drones, that is, the control input of each drone depends not only on its own state, but also on the state of neighboring drones. The collaborative control law of the cluster can be expressed as:

[0046]

[0047] Where: K is the control gain matrix; is the neighbor set of UAV i; gi(t) is the local task-driven control input of UAV i. This method ensures that the UAVs in the cluster maintain relative consistency when performing tasks, while being able to make individual adjustments based on local needs.

[0048] Step 106, optimizing the collaborative control decision of the UAV cluster according to the global decision function, the cluster decision function and the local decision function.

[0049] At the global level, the top-level command center or the global controller of the drone cluster is responsible for the strategic decision-making of the entire cluster, such as task allocation, target prioritization, resource allocation, etc. The goal of global decision-making is to coordinate all drones and sub-clusters to ensure the optimal completion of the entire cluster's tasks.

[0050] The global decision can be modeled as a multi-objective optimization problem:

[0051]

[0052] Among them: J global is the global objective function of the cluster, which may include task completion time, energy consumption, target capture rate, etc.; X(t) = [x 1 (t), x 2 (t), ..., x N (t)] is the state matrix of the entire cluster; U(t) = [u 1 (t),u 2 (t), ..., u N (t)] is the control input for the entire cluster.

[0053] Global decisions can be optimized through methods such as reinforcement learning or dynamic programming. At the global level, the edge computing nodes of the 6G network can provide powerful computing power to process the information flow from all drones in real time and make the best decision based on the global situation.

[0054] In one embodiment, according to the global decision function, the cluster decision function and the local decision function, the collaborative control decision of the UAV cluster is optimized, including: the collaborative control decision adopts a top-down coordination mechanism and a bottom-up feedback mechanism; wherein, the coordination order of the top-down coordination mechanism is UAV cluster, sub-cluster and UAV, and the feedback order of the bottom-up feedback mechanism is UAV, sub-cluster and UAV cluster.

[0055] To achieve high efficiency of multi-level collaborative decision-making, we adopt a hierarchical optimization framework to ensure that decisions at all levels can cooperate and optimize each other through a top-down coordination mechanism and a bottom-up feedback mechanism. The global level determines the task allocation and priority of each subcluster according to the overall requirements of the task, and generates a preliminary task allocation matrix T through the global task planner. assign :

[0056]

[0057] in: Represents the set of all subclusters, the task allocation matrix T assign It is generated by optimizing the global goal of the cluster and the synergy of the sub-clusters. The cluster level is based on the task allocation matrix T of the global decision assign , combined with the local goals of the cluster, optimize its internal task execution and collaborative operations. Individual drones at the local level make autonomous adjustments based on the collaborative decision-making of the cluster and the information they perceive locally. Each drone will perform real-time path optimization and local goal correction in its local environment based on the decisions of neighboring drones and its independently perceived tasks.

[0058] In one embodiment, the method further includes: dynamically allocating communication resources according to the decision-making requirements of drones, sub-clusters and drone clusters; dynamically allocating computing resources of edge computing nodes according to the computing requirements of drones, sub-clusters and drone clusters.

[0059] The effective implementation of multi-level collaborative decision-making mechanisms depends on efficient communication and computing resource allocation within the drone cluster. The ultra-high-speed communication and edge computing capabilities of 6G networks provide support for multi-level collaborative decision-making, especially at the cluster collaboration and global decision-making levels where data transmission and computing requirements are high.

[0060] For communication resource allocation, the network slicing technology of 6G networks can dynamically adjust the communication bandwidth and latency according to the decision-making requirements at different levels. The global decision-making level requires higher bandwidth and lower latency to obtain global situation information, while the cluster level needs to ensure low latency and high reliability of communication to achieve efficient collaborative control. The optimization goal of communication resource allocation can be expressed as:

[0061]

[0062] Where: B(t) is the communication bandwidth allocation matrix of each level of decision; It is the communication delay at different levels; are the delay weight coefficients of different levels.

[0063] For computing resource allocation, the edge computing nodes of the 6G network dynamically allocate computing resources according to the computing needs of each level. Decisions at the global level require strong computing power to handle complex task planning and optimization problems, while drones at the local level mainly rely on their own computing resources for real-time path planning and control. Therefore, the optimization of computing resource allocation can be expressed by the following formula:

[0064]

[0065] Where: fedge(t) is the allocation matrix of edge computing resources; is the computation time of each level; is the weight coefficient for computing resource allocation.

[0066] In one embodiment, 6G communication modules are used for data exchange between drones and between drones and edge computing nodes.

[0067] 6G communication technology has achieved higher communication bandwidth and lower latency. Compared with traditional 4G / 5G networks, 6G communication provides Tbps-level data transmission rates and millisecond-level ultra-low latency, which can effectively support the real-time communication needs of large-scale drone clusters. When drone clusters are working together or performing complex tasks, each drone can share status information, mission instructions, and perception data in real time, greatly improving the efficiency and reliability of information transmission, and ensuring that all drones in the cluster can complete task coordination and response within millisecond-level delays. This ultra-low latency communication ensures the real-time and accuracy of dynamic formation adjustment, path planning, and target tracking, and improves the efficiency of cluster collaboration and combat capabilities.

[0068] In one embodiment, the method further includes: using a combination of distributed computing and edge computing to obtain a collaborative control strategy; wherein the distributed computing and edge computing collaborative control strategy allocates computing resources with the goal of minimizing the time cost of completing the computing task.

[0069] In the coordinated control of large-scale drone swarms, each drone in the swarm not only needs to maintain communication with other drones, but also needs to calculate its flight trajectory, path planning, obstacle avoidance strategy, etc. in real time according to mission requirements. Traditional centralized computing architectures have problems such as data processing delays and excessive computing load when facing large-scale tasks, and cannot meet the real-time requirements of drone swarms.

[0070] The introduction of distributed computing and edge computing provides new ideas for solving these problems. Distributed computing distributes computing tasks to each drone, so that each drone not only acts as a task executor, but also as a computing node, sharing part of the computing tasks. Edge computing relies on edge devices close to drones, such as edge servers on 6G base stations, which can effectively balance the communication delay and computing burden of drones, improving the real-time and accuracy of collaborative control.

[0071] In the distributed computing model, considering that each drone in the drone cluster is both a computing node and a task executor, the total computing burden of each drone performing a task consists of two parts: local computing and network collaborative computing.

[0072] Let the N drones in the drone cluster be represented as U 1 , U 2 , ..., U N , each drone U i The computational requirements of the local computing task can be expressed as:

[0073] C local，i =f(x i (t),u i (t))

[0074] Where: C local,i is the local computing task of UAV i; x i (t) is the state vector of UAV i at time t, including its position, velocity, acceleration, etc.; u i (t) is the control input of UAV i at time t.

[0075] In order to reduce the local computing burden, distributed computing introduces a collaborative mechanism, namely, UAV i Part of the computing tasks can be distributed to other UAVs for computing, and the total burden of collaborative computing is defined as C coop，i In an ideal situation, the UAV i The total computational burden can be expressed by the following formula:

[0076]

[0077] Where: C total,iis the total computational burden of UAV i; It's a drone i The neighbor set of C (i.e., the set of drones that can perform collaborative computing through the 6G network); coop，ij It is the task that UAV i delegates to UAV j for calculation.

[0078] For each drone U i For example, the computational burden C total，i It is necessary to optimize between local computing and collaborative computing. The effect of collaborative computing depends on the communication bandwidth and delay between drones. Therefore, considering the communication delay τ ij (t) and bandwidth B ij The transmission time of the collaborative computing task can be expressed as:

[0079]

[0080] Where: D coop，ij is the amount of task data that UAV i assigns to UAV j for calculation; B ij is the communication bandwidth between UAVs i and j; τ ij (t) is the communication delay between UAVs i and j at time t. i The total task execution time T total,i It can be expressed as the sum of local computation time and collaborative computation time:

[0081]

[0082] Where: f local，i is the local computing capacity of UAV i; f coop，j is the collaborative computing capability of UAV j.

[0083] In order to minimize the task execution time of the entire cluster, it is necessary to allocate the computing tasks of each drone through an optimization algorithm. The optimization goal can be expressed as:

[0084]

[0085] This optimization problem can be solved by distributed optimization algorithms, such as gradient descent or Lagrange multiplier method, to dynamically adjust the task allocation between drones to achieve optimal collaborative computing performance.

[0086] Although distributed computing can effectively disperse computing tasks, when the task complexity is high, collaborative computing between drones alone may still not meet real-time requirements. For this reason, edge computing is introduced, which offloads part of the computing tasks to edge computing nodes close to drones, such as edge servers of 6G base stations. This architecture can significantly improve computing power while maintaining low communication latency.

[0087] For UAV i For example, the task offloading model of edge computing can be expressed as:

[0088]

[0089] Where: D offload，i It's a drone i The amount of task data that needs to be offloaded to the edge computing node; R edge，i It's a drone i Communication rate with edge nodes; C edge，i is the amount of computing tasks that the edge computing node needs to process; f edge It is the computing power of the edge computing node. i The total task execution time can be optimized as follows:

[0090]

[0091] The above shows that the UAV can choose the optimal strategy of local computing, collaborative computing or task offloading according to the current network status, task complexity and load of edge computing nodes to ensure that the task is completed in the shortest time.

[0092] Collaborative control based on distributed computing and edge computing is mainly reflected in the overall task execution and decision optimization of the UAV cluster. Each UAV in the cluster must make autonomous decisions based on its own local information, and optimize the action strategy of the entire cluster through communication and collaboration with other UAVs.

[0093] The cooperative control problem of UAV swarm can be modeled as the optimal control problem of a multi-agent system, where each UAV U i The control goal is to minimize its local cost function J i (x i (t),u i (t)), the cost function is related to factors such as the UAV’s energy consumption, mission completion, and path planning.

[0094] The global collaborative control objective can be expressed as the overall cost function J of the entire cluster: total :

[0095]

[0096] Where: ij is the coordination weight between neighboring UAVs, which controls the coordination degree between each UAV. The coordination control law can be derived from the overall cost function J by gradient descent method. total It is derived from the above to obtain the optimal control input u for each UAV. i (t):

[0097]

[0098] The control law shows that each drone must not only perform local optimization based on its own mission status, but also exchange information with neighboring drones to make collaborative decisions and adjust paths, thereby optimizing the overall action strategy of the cluster. By combining distributed computing with edge computing, the collaborative control of drone clusters has been greatly optimized. Distributed computing improves the autonomous decision-making ability of each drone and reduces the delay problem caused by centralized computing by distributing computing tasks to drone clusters; the introduction of edge computing provides a guarantee for real-time processing of complex tasks through powerful computing power and low-latency communication support.

[0099] The innovativeness of this collaborative control method is reflected in its combination of the high bandwidth and low latency characteristics of the 6G communication network, ensuring efficient communication and task allocation between drones. At the same time, the combination of distributed computing and edge computing has greatly improved the overall computing power of the cluster, reduced the delay in task execution, and improved the efficiency and accuracy of task completion. In the multi-task scenarios of future drone clusters, this method has broad application prospects, especially in complex mission environments such as military, logistics distribution, and disaster relief, and can significantly improve the combat and collaborative efficiency of drone clusters.

[0100] In one embodiment, a merge-and-divide method based on multi-level collaborative decision-making is provided. Based on the multi-level collaborative decision-making method, the merge-and-divide method includes: obtaining task requirements, spatial environment and state vectors of UAVs in real time; the task requirements include target relative position and target point position, and the spatial environment includes obstacle position; based on the state vector and target relative position, obtaining consistency control law; based on the state vector, generating merge attraction force around the target point position, generating merge repulsion force around the obstacle position and adjacent UAVs, and obtaining merge control law according to merge attraction and merge repulsion force; based on the state vector, generating divergence repulsion force around the center position of the UAV cluster, and obtaining divergence control law according to divergence repulsion force and merge repulsion force; based on the consistency control law, merge control law and divergence control law, constructing an optimization model with the goal of optimizing task execution efficiency, and obtaining the formation and merge-and-divide strategy of the UAV cluster by solving the optimization model.

[0101] When drone swarms are performing complex tasks, they usually need to flexibly adjust their formations according to changes in the mission scenario. This dynamic formation requires drone swarms to be able to reorganize their formations in real time based on information such as the environment, target location, and threat situation, in order to achieve efficient target approach and attack (co-attack) or rapid dispersion after attack (separate attack), thereby achieving efficient mission execution and self-defense.

[0102] Traditional formation control methods mostly use fixed formation modes, which are difficult to cope with changing environments and mission requirements. Dynamic formation and converge and diverge strategies use intelligent algorithms to achieve flexible adjustments of drone formations, allowing clusters to efficiently aggregate during the target approach phase and quickly disperse after completing attacks or other tasks to avoid enemy counterattacks and enhance overall flexibility and autonomy.

[0103] The formation problem of a drone swarm can be abstracted as a formation maintenance and adjustment problem of a multi-agent system. Let the number of drones in the swarm be N, and the state vector of each drone be x i (t) = [p i (t), v i (t), a i (t)] T , where p i (t), vi(t) and a i (t) represent the spatial position, velocity and acceleration of UAV i at time t, respectively.

[0104] The dynamic formation control goal of the UAV swarm can be expressed as: at any time t, the relative positions between the UAVs must satisfy the constraints of the formation. Define the expected relative position between UAVs i and j as d ij (t), then the control objective of the dynamic formation can be expressed as:

[0105]

[0106] where d ij (t) is the relative position of the dynamic target that changes with time, which is used to adapt to different mission scenarios. The dynamic formation goal of the UAV cluster is to enable each UAV to adjust its position and speed according to the state of the neighboring UAVs, so as to ensure that the formation of the entire cluster is maintained.

[0107] The consistency control law can be expressed as:

[0108]

[0109] Where: u i (t) is the control input (acceleration or velocity adjustment) of UAV i; K is the control gain matrix used to adjust the stability of the formation; is the set of neighbors of UAV i (i.e., the UAVs with which it has direct communication links).

[0110] Through the consistency control law, the UAVs can maintain the formation by adjusting the relative positions with neighboring UAVs, and at the same time, the relative position d ij(t) changes to achieve dynamic adjustment of the formation. For example, during the target approach phase, the drone cluster can gradually aggregate from a dispersed formation to an attack formation, and then quickly disperse after completing the attack.

[0111] The goal of the convergence strategy is to quickly gather the drone cluster near the target and form a siege situation during the attack phase to achieve concentrated firepower or mission execution. The convergence strategy generates an attractive potential field for the target and a repulsive potential field for threats or obstacles to guide the drone cluster to converge towards the target.

[0112] Define the target position as p target (t), the attraction force on drone i can be expressed as:

[0113]

[0114] Among them: U att,i (t) is the attractive potential of UAV i; k att is the gain coefficient of the attractive potential; p i (t) is the position of UAV i; p target (t) is the position of the target.

[0115] To avoid collisions between drones or with other obstacles, the repulsive potential field needs to be considered during the merging process. The position of obstacles or other drones is defined as p obs (t), then the repulsive force on UAV i is:

[0116]

[0117] Where: k rep is the gain coefficient of the repulsive potential; ||p i (t)-p j (t)|| is the distance between drones i and j.

[0118] The combined control law of UAV i can be expressed as the resultant force of attraction and repulsion:

[0119] u i (t) = F att,i (t)+F rep,i (t)

[0120] This control law ensures that the drones can gradually gather to form a siege situation when approaching the target, and maintain a certain distance while avoiding collision to achieve coordinated attack.

[0121] The split-attack strategy is to quickly disperse the drone cluster after the combined attack is completed to prevent the enemy's counterattack or interference. The split-attack strategy requires the drone cluster to be able to leave the battlefield as soon as possible after completing the attack mission and enter a dispersed defense posture. The split-attack strategy can be implemented through a potential field method similar to the combined attack strategy, but unlike the combined attack, the drone cluster is no longer attracted by the target during the split-attack, but is affected by the repulsive potential field to maintain the dispersion and mobility of the cluster.

[0122] Define the center position P center (t) is the center of mass of the drone cluster. The repulsive force on drone i during the attack process can be expressed as:

[0123]

[0124] Where: k split is the gain coefficient of the scattered potential field; It is the center of the drone cluster.

[0125] In this way, the drone cluster will quickly leave the central area after completing the mission and maintain a safe distance from neighboring drones to avoid being counterattacked by the enemy. The overall attack control law can be expressed as:

[0126]

[0127] This control law ensures that the drones remain dispersed during the strike process while avoiding collisions with each other.

[0128] The execution of the combined attack strategy needs to be dynamically optimized according to real-time mission requirements and environmental changes. With the support of 6G communication networks, drones can obtain information about neighboring drones and the environment through low-latency communication networks, and make dynamic formation adjustments and optimize the combined attack strategy based on this information.

[0129] The overall task execution efficiency E(t) of the drone cluster can be expressed as:

[0130]

[0131] Where: T complete，i (t) is the time it takes for UAV i to complete the task; E energy，i (t) is the energy consumption of UAV i; α i It is the weight coefficient of energy consumption, which is used to balance the relationship between task completion time and energy consumption.

[0132] The optimization goal is to minimize the total task completion time and reduce energy consumption. This problem can be solved by real-time dynamic programming or reinforcement learning algorithms to ensure that the drone cluster achieves optimal efficiency when executing the merge-and-divide strategy. The dynamic formation and merge-and-divide strategy of the present invention ensure that the drone cluster can flexibly adjust the formation and coordinate attacks according to mission requirements. Through the merge strategy, the drone cluster can efficiently gather near the target for attack; through the divergent strategy, the cluster can quickly disperse after the attack to avoid enemy counterattacks. The innovation of this method is reflected in the combination of the real-time communication capabilities of the 6G network, which improves the dynamic response capabilities of the drone cluster, ensures the efficient execution and autonomous defense of the cluster in complex tasks, and has significant practicality and application prospects.

[0133] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0134] Based on the same inventive concept, the embodiment of the present application also provides a multi-level collaborative decision-making device for implementing the multi-level collaborative decision-making method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more multi-level collaborative decision-making device embodiments provided below can refer to the limitations of the multi-level collaborative decision-making method above, and will not be repeated here.

[0135] In one embodiment, Figure 2 As shown, a multi-level collaborative decision-making device is provided, including: a local decision-making module 202, a cluster and global decision-making module 204 and a collaborative decision-making module 206, wherein:

[0136] The local decision module 202 is used to obtain the perception information of each drone and the status information of neighboring drones by using the local decision function corresponding to each drone in the drone cluster;

[0137] The cluster and global decision module 204 is used to construct a cluster decision function corresponding to each sub-cluster composed of a plurality of drones based on the perception information and the state information, and to construct a global decision function corresponding to the drone cluster for the drone cluster;

[0138] The collaborative decision module 206 is used to optimize the collaborative control decision of the UAV cluster according to the global decision function, the cluster decision function and the local decision function.

[0139] Each module in the above multi-level collaborative decision-making device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module above.

[0140] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in all the above method embodiments when executing the computer program.

[0141] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in all the above method embodiments are implemented.

[0142] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in all the above method embodiments when executed by a processor.

[0143] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0144] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0145] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0146] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A multi-level collaborative decision-making method, characterized in that: The method comprises: For each drone in the drone cluster, using the local decision function corresponding to each drone, the perception information of the drone and the status information of neighboring drones are obtained; Based on the perception information and the state information, for the subclusters composed of the plurality of drones, a cluster decision function corresponding to each of the subclusters is constructed, and for the drone cluster, a global decision function corresponding to the drone cluster is constructed; According to the global decision function, the cluster decision function and the local decision function, the collaborative control decision of the UAV cluster is optimized.

2. The method according to claim 1, characterized in that Optimizing the coordinated control decision of the drone cluster according to the global decision function, the cluster decision function and the local decision function includes: The collaborative control decision adopts a top-down coordination mechanism and a bottom-up feedback mechanism; wherein the coordination order of the top-down coordination mechanism is UAV cluster, sub-cluster and UAV, and the feedback order of the bottom-up feedback mechanism is UAV, sub-cluster and UAV cluster.

3. The method according to claim 1, characterized in that The method further comprises: Dynamically allocating communication resources according to the decision requirements of the drone, the sub-cluster, and the drone cluster; The computing resources of the edge computing nodes are dynamically allocated according to the computing requirements of the drone, the sub-cluster and the drone cluster.

4. The method according to claim 3, characterized in that: A 6G communication module is used for data exchange between the drones and between the drones and the edge computing nodes.

5. The method according to claim 1, characterized in that The method further comprises: The collaborative control strategy is obtained by combining distributed computing and edge computing; Among them, the distributed computing and edge computing collaborative control strategy allocates computing resources with the goal of minimizing the time cost of completing computing tasks.

6. A multi-level collaborative decision-making device, characterized in that: The device comprises: A local decision module is used to obtain the perception information of each drone in the drone cluster and the status information of neighboring drones by using the local decision function corresponding to each drone; A cluster and global decision module, for constructing a cluster decision function corresponding to each sub-cluster composed of a plurality of the drones based on the perception information and the state information, and for constructing a global decision function corresponding to the drone cluster; A collaborative decision-making module is used to optimize the collaborative control decision of the drone cluster based on the global decision function, the cluster decision function and the local decision function.

7. A combined and divided attack method based on multi-level collaborative decision-making, based on the multi-level collaborative decision-making method according to any one of claims 1 to 5, the combined and divided attack method comprising: Obtain mission requirements, space environment, and drone state vectors in real time; The task requirements include the relative position of the target and the position of the target point, and the spatial environment includes the position of obstacles; Acquire a consistency control law based on the state vector and the target relative position; Based on the state vector, a merging attraction force is generated around the target point position, a merging repulsion force is generated around the obstacle position and the adjacent UAV, and a merging control law is obtained according to the merging attraction force and the merging repulsion force; Based on the state vector, a split repulsive force is generated around the center position of the drone cluster, and a split repulsive force control law is obtained according to the split repulsive force and the combined repulsive force; According to the consistency control law, the merging control law and the split-attack control law, an optimization model with the goal of optimizing task execution efficiency is constructed, and the formation and merging-split-attack strategy of the drone cluster are obtained by solving the optimization model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images