Unmanned cluster grey wolf simulation task dynamic scheduling method under severe communication

Abstract

The invention discloses an unmanned cluster grey wolf simulation task dynamic scheduling method under severe communication, which comprises the following steps: firstly, constructing an unmanned cluster system model, a task demand and constraint model and a complex communication limited environment model, and providing a theoretical modeling basis for task dynamic scheduling; and finally, in a task redistribution stage, redistribution is started based on a time-event double-trigger mechanism, a load balancing strategy is adopted to construct a joint optimization target, and task dynamic adjustment is realized under the condition that constraint conditions are met. Based on the grey wolf society level, a three-level dynamic leader election mechanism is adopted, and an event-time double-trigger strategy is combined, so that the decision continuity in a high-interference environment is ensured, the task success rate is effectively improved, and the problem that an existing cluster task allocation method is insufficient in adaptability in extreme communication conditions and complex dynamic environments is solved.

Description

Technical Field

[0001] This invention relates to the field of unmanned swarm collaborative control technology, and in particular to a dynamic scheduling method for unmanned swarm tasks mimicking the gray wolf under poor communication conditions. Background Technology

[0002] Unmanned swarm systems have broad application prospects, covering many important aspects such as environmental monitoring, disaster relief, logistics and distribution, and public security. However, with increasingly complex application scenarios, more interference factors, and more dynamic changes, problems such as building obstruction, severe weather, spectrum congestion, and equipment failure may lead to a decline in communication quality, failure of critical nodes, and interruption of system functions, posing significant challenges to the working efficiency, operational reliability, and continuous survivability of unmanned swarm systems.

[0003] Task allocation, as a core component of collaborative decision-making in unmanned swarms, directly impacts system resource utilization and task completion quality. Based on the execution phase, it can be divided into pre-allocation and reallocation: the former involves offline planning based on prior information such as environmental maps, task objectives, and resource constraints before task execution, with typical methods including contract net protocols and optimization models; the latter involves online adjustments for dynamic emergencies such as equipment failures and task changes. Based on system architecture, it can be divided into centralized, distributed, and hybrid architectures: Centralized architecture relies on a central node to gather global information and can achieve theoretically optimal decision-making under a star communication topology. It has the advantages of high coordination efficiency and strong decision consistency, and is suitable for deterministic scenarios such as aerospace control. However, its single-point failure risk increases significantly with the expansion of network scale, and the mismatch between the common static update mechanism under its architecture and the dynamic network environment can lead to decision delays, limiting its application in dynamic tasks such as reconnaissance and rescue. Distributed methods, such as task allocation mechanisms based on biomimetic algorithms such as ant colony and wolf pack, and local negotiation strategies based on game theory, rely on local interaction to achieve autonomous decision-making. They have strong robustness and scalability, but are prone to getting trapped in local optima. Hybrid architecture attempts to combine the advantages of both, balancing global coordination and local autonomy through hierarchical control. However, the complexity of the architecture and the difficulty in designing the coordination mechanism limit its practical promotion.

[0004] In recent years, data-driven deep reinforcement learning and large-scale models have demonstrated environmental adaptability advantages in task allocation, but they remain limited by data quality, training costs, and interpretability issues. More importantly, existing research is largely based on ideal communication assumptions. In real-world, constrained communication environments, information asymmetry weakens global situational awareness; co-channel interference caused by spectrum resource competition creates a negative feedback loop of "communication degradation - decreased decision quality"; and traditional algorithms' reliance on stable links makes them unsuitable for rapid adjustments in emergency scenarios such as disaster relief. These challenges urgently necessitate more robust and adaptive task allocation strategies. Summary of the Invention

[0005] The purpose of this invention is to provide a dynamic scheduling method for unmanned swarm tasks mimicking the gray wolf under harsh communication conditions. Based on the gray wolf social hierarchy, it adopts a three-level dynamic leader election mechanism and combines an event-time dual-trigger strategy to ensure the continuity of decision-making in high-interference environments, effectively improve the task success rate, and solve the problem of insufficient adaptability of existing swarm task allocation methods in extreme communication conditions and complex dynamic environments.

[0006] To achieve the above objectives, the present invention provides the following solution: A dynamic scheduling method for unmanned swarm tasks mimicking the Grey Wolf troop under poor communication conditions, the method includes the following steps: S1, System and Environment Modeling: Construct an unmanned swarm system model, a task requirement and constraint model, and a complex communication-constrained environment model to provide a theoretical modeling basis for dynamic task scheduling. Among them, the task requirement and constraint model pre-sets multiple constraints. S2, Task Pre-allocation: Based on the models constructed in step S1, dynamic leader election, task assignment, and task scheduling are performed, specifically including: S201, Dynamic Leadership Election: Based on the gray wolf-like dynamic cluster hierarchical architecture, a three-level dynamic leadership decision-making architecture is established through candidate node screening, leadership hierarchy initialization and dynamic switching control, and a dynamic election and assignment center is dynamically elected. S202, Task Assignment: The assignment center, which is elected, uses the Hungarian algorithm to solve the cluster task assignment scheme and performs task assignment. S203, Task Scheduling: For each unmanned platform's task set, the satisfaction of preset constraints is verified sequentially. Through an iterative task filtering mechanism, tasks that meet the constraints are included in the execution sequence, while tasks that do not meet the preset constraints are returned to the waiting queue. S3, Task Reassignment: Reassignment is initiated based on a time-event dual-trigger mechanism, and a joint optimization objective is constructed using a load balancing strategy to achieve dynamic adjustment of tasks under the condition of satisfying constraints.

[0007] Furthermore, the method for constructing the unmanned swarm system model in S1 specifically includes: A single unmanned platform Use the quintuple This indicates that the state set Describe the platform's intrinsic state at time t, including the platform's current two-dimensional coordinates x. i ∈R 2 Speed ​​of motion v i Type identifier Resource reserves Cost matrix Task queue ; It is a set of executable actions, including state refresh, task processing, and cost estimation functions; Store shared information, including global task sets. The status summary of neighboring platforms and the reconnaissance area are synchronized globally through a central server in centralized control mode. Representative decision-making mechanism, based on and Choose the optimal operation; It is the state transition function, defined as ; The entire cluster consists of The platform consists of several components, denoted as... .

[0008] Furthermore, in S1, multiple constraints include synergy constraints, goal dependency constraints, resource constraints, and task timing constraints. The method for constructing the task requirements and constraint model specifically includes: Each combat target Characterized as containing spatial coordinates Type identifier Resource requirements during the reconnaissance / strike / assessment phase The set of prerequisite targets for task execution and completion status The tuple exists; and the following constraints apply: Coordination constraints: in, This indicates whether the i-th unmanned platform participated in the m-stage of the objective j. The value is 1 if the allocated resources are greater than 0, and 0 otherwise, ensuring that at least one unmanned platform participated. ≥ The unmanned platforms coordinate to execute, where R is the minimum threshold number of cooperating platforms; Target dependency constraints: in, This indicates the start time of objective j in task phase m. Table 1 The completion time of each objective is such that objective j can only begin after all its dependent objective tasks have been completed. Resource constraints: in, This represents the amount of resources allocated to target j by the i-th unmanned platform during task phase m, ensuring that the task's resource requirements are met. Task timing constraints: Equation (4) represents the timing requirements of the mission phases, with reconnaissance, strike, and assessment missions for the same target being carried out sequentially.

[0009] Furthermore, the method for constructing the complex communication-constrained environment model in S1 specifically includes: The set of obstacles is represented as: Each obstacle is defined as: ,in The coordinates of the obstacle's center are... The length and width dimensions of the obstacle. The angle of rotation of the obstacle. These are the physical characteristic parameters of the obstacle; Randomly distributed communication interference areas are used to simulate the interference characteristics in real wireless communication by parameterizing the regional interference. Equation (5) is used to simulate the phenomenon of linear attenuation of electromagnetic wave intensity with distance: in, This indicates the actual interference intensity at the location of the unmanned platform. Let be the distance between the unmanned platform i and the interference source w with an effective interference radius of r; The maximum value obtained by superimposing multiple regions reflects the "dominant effect" of electromagnetic interference: In the formula, K represents the number of interference zones where the unmanned platform i is located.

[0010] Furthermore, in S201, the constraints for candidate node selection in the dynamic leader election are as follows: in, Represents a node Communication quality, Represents a node The number of times you interact with the network. This is the minimum communication quality threshold; The comprehensive scoring function for initializing leadership levels is defined as follows: in, This is a normalized value for communication quality. For window internal nodes Compared to the average packet loss rate of its neighbors, This represents the weight used to balance communication quality and packet loss rate.

[0011] Furthermore, in S201, the three criteria for dynamic switching control in dynamic leadership election include: The main leader is out of contact: heartbeat signal lost or interrupted; Local communication quality advantage: There exists a node whose communication quality is significantly better than that of the current node α, defined by a threshold. : Potential interference zone identification: If the communication quality of node α shows a downward trend within the sliding window, and neighboring nodes show an abnormally high packet loss rate, then it is considered to be in an interference environment.

[0012] Furthermore, S202, task assignment, specifically includes: For N available unmanned platforms and Q target missions, a cost matrix is ​​constructed, considering the three phases of reconnaissance, strike, and assessment. ,in The cost function is defined as: in, For the normalization path cost, For the cost of communication links, At the cost of environmental threats, For ability compatibility, For collaborative gain; all components are normalized to the interval [0,1], weighted... Supports dynamic adjustments to enhance adaptability; The optimization objective of task assignment is to minimize the total cost and satisfy the synergy constraints, objective dependency constraints, and resource constraints in equations (1), (2), and (3): Where N represents the number of available unmanned platforms; Q represents the number of target tasks to be executed; For unmanned platforms Execute target task The overall cost under task phase m is defined by formula (14); To assign variables, if there is no platform The assigned task phase m executes the task. ,but If the value is 1, then it is 0 otherwise; J is the total cost objective function, which needs to be minimized and solved using the Hungarian function.

[0013] Furthermore, S203, task scheduling, specifically includes: After task assignment is completed, the task timing constraints in set equation (4) and the aforementioned target dependency constraints are used to sort and allocate time slots for the assigned tasks; the goal of task scheduling is to minimize task completion time and ensure that dependency logic is not violated. in, For the target task The completion time; the task scheduling result ensures the temporal consistency of the dependencies between tasks and the platform's execution capabilities, thereby forming an executable task sequence.

[0014] Furthermore, S3, task reassignment, specifically includes: Task reassignment trigger set Defined as: Among them, time triggering is every Periodically check task and platform status; event trigger is when a node goes offline. Task failed or node interference strength Exceeding the threshold Immediately initiate reallocation; Task redistribution considers task completion, platform load balancing, and communication costs simultaneously, constructing a joint optimization objective and solving it using a Hungarian function: in, Penalty based on the importance of the uncompleted task; This incurs additional communication costs; The cost of load balancing is represented by the mean squared error of resource utilization: , , They represent , , The corresponding weighting coefficients; Let be the resource utilization rate of unmanned platform i, and ... Task redistribution must meet the pre-defined constraints of collaboration, goal dependency, resources, and task timing in the task requirements and constraint model.

[0015] The present invention also provides an electronic device comprising one or more processors; a memory; and one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, and the one or more application programs are configured to perform a dynamic scheduling method for unmanned swarm mimicking gray wolf missions under adverse communication conditions as described above.

[0016] According to specific embodiments provided by the present invention, the unmanned swarm task dynamic scheduling method for simulated gray wolf missions under adverse communication conditions disclosed by the present invention has the following technical effects: (1) Strong robustness: A three-level dynamic leadership decision-making architecture is constructed based on the gray wolf-like hierarchical architecture. Through redundancy fault tolerance mechanism and dynamic switching control, single point of failure is avoided and the continuity of the decision chain is guaranteed in harsh communication environments. (2) Good adaptability: The task redistribution mechanism is triggered by both time and event, which can respond quickly when the environment changes, avoid excessive scheduling, and adapt to complex dynamic environments; (3) Excellent performance: The unbalanced assignment strategy is adopted in the task pre-allocation stage to ensure the rationality of engineering practice, and the load balancing strategy is adopted in the re-allocation stage to maximize resource utilization. Experimental verification shows that the task success rate is improved and the task time is reduced compared with the existing methods. (4) High practicality: It fully covers the entire process of system modeling, leader election, task pre-allocation and redistribution. The model and algorithm have good engineering feasibility and can be widely applied to unmanned swarm collaborative task scenarios in multiple fields such as military and disaster relief. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the unmanned swarm task dynamic scheduling method for the Grey Wolf simulation under harsh communication conditions according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the comparative simulation environment setup for an embodiment of the present invention; Figure 3 This diagram illustrates a comprehensive performance comparison between the task scheduling method based on activity and the Hungarian algorithm based on multiple indicators in this embodiment of the invention and the method described in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] This invention aims to address the problem of insufficient adaptability of existing cluster task allocation methods under extreme communication conditions and complex dynamic environments. Based on the gray wolf social hierarchy, a three-level dynamic leader election mechanism is proposed, combined with an event-time dual-trigger strategy, to ensure the continuity of decision-making under high interference environments and effectively improve the task success rate.

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] like Figure 1 As shown in the embodiment of the present invention, the unmanned swarm task dynamic scheduling method for the Grey Wolf simulation under poor communication conditions includes two major stages: task pre-allocation and re-allocation.

[0023] Task pre-allocation comprises three stages: dynamic leader election, task assignment, and task scheduling. First, based on a simulated Grey Wolf dynamic cluster hierarchical architecture, and considering the communication environment and platform status, a dynamic assignment center is elected. Then, the Hungarian algorithm is used to quickly solve the cluster task assignment scheme. This stage primarily focuses on resource and coordination constraints, employing an unbalanced assignment strategy. Although it may not obtain a globally optimal solution, in a dynamic and uncertain environment, given that the unmanned platform only possesses local situational awareness capabilities, this allocation strategy based on the current moment's optimum has practical engineering rationale. In the task scheduling stage, for each unmanned platform's task set, the satisfaction of constraints such as timing and dependencies is verified sequentially. Through an iterative task filtering mechanism, tasks that satisfy the constraints are included in the execution sequence, while tasks that do not satisfy the constraints are returned to the waiting queue.

[0024] After task scheduling is completed, each platform re-confirms the tasks based on its own status. Accepted tasks are added to the execution queue, while rejected tasks are re-entered into the allocation sequence. To cope with dynamic environmental changes, a time- and event-triggered mechanism is used for task redistribution. Time-triggered tasks are executed based on a fixed period, while event-triggered tasks respond to abnormal situations such as node failures. Unlike the initial assignment phase, the redistribution algorithm employs a load balancing strategy to maximize system resource utilization as its optimization objective. This hierarchical and progressive allocation mechanism effectively balances the requirements of computational efficiency and allocation quality.

[0025] The specific steps of the unmanned swarm-based gray wolf-like task dynamic scheduling method under harsh communication conditions described in this invention are analyzed as follows: 1. System and Environment Modeling 1.1 Modeling of Unmanned Cluster Systems single platform Use the quintuple This indicates that the state set Describe the platform's internal state at time t, including the platform's current two-dimensional coordinates. Movement speed Type identifier Resource reserves Cost matrix Task queue ; It is a set of executable actions, including functions such as status refresh, task processing, and cost estimation; Store shared information, including global task sets. In centralized control mode, global data synchronization can be achieved through a central server, including neighborhood platform status summaries and reconnaissance areas. Representative decision-making mechanism, based on and Choose the optimal operation; It is the state transition function, defined as The entire cluster consists of The platform consists of several components, denoted as... The above modeling method enables a systematic description of the platform's state, behavioral decisions, and information interaction.

[0026] 1.2 Task Requirements and Constraint Modeling This invention constructs a formalized and standardized mission modeling framework based on the reconnaissance-strike-assessment process in modern military operations. Each operational target... Characterized as containing spatial coordinates Type identifier Resource requirements during the reconnaissance / strike / assessment phase The set of prerequisite targets for task execution and completion status A tuple. And the following constraints apply: Coordination constraints: in, This indicates whether no one (i) participated in the m-stage of task j. The value is 1 if the allocated resources are greater than 0, and 0 otherwise, ensuring that at least one person (i) participated. ≥ The unmanned platforms work together, with R being the minimum threshold number of collaborative platforms.

[0027] Target dependency constraints: in, This indicates the start time of objective j in task phase m. Table 1 The completion time of each objective is specified, and objective j can only begin after all its dependent objective tasks have been completed.

[0028] Resource constraints: in, This represents the amount of resources allocated by the i-th unmanned platform to task j in stage m, ensuring that the task's resource requirements are met. Task timing constraints: This formula represents the timing requirements of mission phases, with reconnaissance, strike, and assessment missions for the same target being carried out sequentially.

[0029] This model depicts the mission logic and execution specifications of unmanned swarm collaborative operations, providing a theoretical foundation and verification framework for the design of subsequent intelligent decision-making algorithms, and possesses good scalability and engineering feasibility.

[0030] 1.3 Modeling of Complex Communication-Constrained Environments The task execution environment addressed in this embodiment of the invention is one where approximate map information is available, but the location of the task target is unknown. Because representing obstacles using rectangular combinations is universal, and by discretizing through a grid, obstacles of arbitrary shapes can be transformed into a set of standard rectangular units. This significantly reduces computational complexity while maintaining geometric accuracy, facilitating the implementation of subsequent path planning algorithms. Therefore, the obstacle set is represented as: Each obstacle is defined as: ,in The coordinates of the obstacle's center are... The length and width dimensions of the obstacle. The angle of rotation of the obstacle. These are the physical characteristic parameters of the obstacle.

[0031] Meanwhile, randomly distributed communication interference areas are present in this environment. By parameterizing the regional interference, the interference characteristics in real wireless communication are simulated. In equation (5), the phenomenon of linear attenuation of electromagnetic wave intensity with distance is simulated: in, This indicates the actual interference intensity at the location of the unmanned platform. Let be the distance between the unmanned platform i and the interference source w with an effective interference radius of r; The maximum value obtained by superimposing multiple regions reflects the "dominant effect" of electromagnetic interference: In the formula, K represents the number of interference zones where the unmanned platform i is located.

[0032] 2. Task pre-assignment: 2.1 Flexible and Dynamic Three-Tier Leadership Decision-Making Framework Drawing inspiration from the α / β / δ hierarchical mechanism of the Grey Wolf Optimization Algorithm, this invention proposes a hierarchical fault-tolerant leader election architecture to enhance the robustness and continuity of unmanned swarms in constrained communication environments. The system employs a dual-loop control mechanism of "monitoring-election": during operation, it continuously monitors node communication and neighbor distribution, and executes leader switching when trigger conditions are met, thereby avoiding single points of failure and ensuring the stability of the decision chain.

[0033] In the leadership election process, nodes that meet both communication activity and quality standards are first selected from the neighbor table. Then, a comprehensive scoring model is constructed using communication quality and packet loss rate, and the top three nodes are selected based on their scores to assume leadership roles at levels α, β, and δ. β and δ serve as first- and second-level backups of α, respectively, forming a redundancy and fault-tolerance mechanism. When α fails, β automatically becomes the primary leader, and δ takes over as the secondary leader. The constraints for candidate node selection are as follows: in, Indicates its communication quality, Represents a node The number of times you interact with the network. This is the minimum communication quality threshold.

[0034] The comprehensive scoring function is defined as follows: in, Normalized value for communication quality: in, For window internal nodes Compared to the average packet loss rate of its neighbors, , which represents the weight used to balance communication quality and packet loss rate.

[0035] During the leadership transition process, the system introduces three criteria: 1) The main leader is out of contact: heartbeat signal is lost or interrupted; 2) Local communication quality advantage: There exists a node whose communication quality is significantly better than the current α node (the leader), defined by a threshold. : 3) Potential Interference Zone Identification: If the communication quality of node α (the leader) shows a downward trend within the sliding window, and neighboring nodes experience abnormally high packet loss rates, then it is considered to be in an interference environment. The rate of decline is defined as: Majority voting rule: Simultaneously, a sliding window threshold adjustment mechanism is introduced to avoid frequent switching due to short-term fluctuations, and the cooling period is dynamically adjusted in conjunction with the network topology change rate. To further balance stability and response speed, a non-linear decaying convergence factor is used to control the cooling time. in, As the initial value, As the lower limit, Control the decay rate, This standardizes the time ratio. This mechanism makes the system more sensitive in the early stages for rapid response, and more stable in the later stages to avoid excessive switching.

[0036] The pseudocode is shown in Table 1: Table 1 2.2 Task Assignment and Scheduling Strategies This invention, based on a biomimetic gray wolf hierarchical architecture, assigns an elected α leader node to be responsible for cluster task assignment and scheduling decisions. Its core idea is to achieve optimal allocation of unmanned platforms to combat missions by constructing a multi-dimensional cost function and establishing a mixed-integer optimization model under formal constraints.

[0037] For N available unmanned platforms { } and Q target tasks { Considering the three stages of reconnaissance, strike, and assessment, a cost matrix is ​​constructed. ,in The cost function is defined as: The meanings of each component are consistent with those in the previous text: For the normalization path cost, For the cost of communication links, At the cost of environmental threats, For ability compatibility, For collaborative gain. All components are normalized to the interval [0,1], and the weights are... Supports dynamic adjustments to enhance adaptability.

[0038] During the task assignment phase, the optimization objective is to minimize the total cost and satisfy the coordination constraints, target dependency constraints, and resource constraints in equations (1), (2), and (3) to ensure that the unmanned swarm simultaneously satisfies the operational logic of coordinated execution, target dependency, and resource supply in task allocation. Where N is the number of available unmanned platforms; Q is the number of target tasks to be executed; For unmanned platforms Execute target task The overall cost under task phase m is defined by formula (14); To assign variables, if there is no platform The assigned task phase m executes the task. ,but If the value is 1, then it is 0 otherwise; J is the total cost objective function, which needs to be minimized and solved using the Hungarian function.

[0039] After assignment is completed, the task timing constraints in set equation (4) and the aforementioned target dependency constraints are used to sort and allocate time slots for the assigned tasks. The goal of scheduling is to minimize task completion time and ensure that dependency logic is not violated. in, For the task The scheduling results ensure the temporal consistency of dependencies between tasks and the platform's execution capabilities, thereby forming an executable task sequence.

[0040] 3. Task redistribution strategy based on time-event triggering mechanism In harsh communication and dynamic task environments, cluster task execution may encounter unexpected events such as node failure, task failure, interference changes, and link interruptions. Using only fixed-period scheduling often results in delayed responses; relying solely on event triggers can lead to frequent reassignments and excessive communication load. Therefore, this invention proposes an adaptive reassignment mechanism combining time-triggered and event-triggered mechanisms, improving flexibility and agility while ensuring system stability. Task reassignment trigger set Defined as: Among them, time triggering is every Periodically check task and platform status; event trigger is when a node goes offline. Task failed or node interference strength Exceeding the threshold If this occurs, a reallocation process will be initiated immediately.

[0041] In terms of objective optimization, the redistribution process simultaneously considers task completion, platform load balancing, and communication costs, constructing a joint optimization objective and solving it using a Hungarian function: in, Penalty based on the importance of the uncompleted task; This incurs additional communication costs; The cost of load balancing is represented by the mean squared error of resource utilization: , , They represent , , The corresponding weighting coefficients; Let be the resource utilization rate of unmanned platform i, and ... Task redistribution must satisfy the coordination constraints, goal dependency constraints, resource constraints, and task timing constraints in the construction of the task requirements and constraint model.

[0042] Through a dual triggering mechanism of time and event, the redistribution can respond quickly when the environment changes, while avoiding excessively frequent scheduling; the joint optimization objective ensures that the system can still achieve high mission completion rate, balanced platform load, and controllable communication costs under conditions of limited resources and communication interference.

[0043] 4. Effect Verification To verify the effectiveness of the method described in this invention, a comparative verification experiment was conducted. A simulation experimental platform was built based on the Ubuntu 16.04 operating system and the ROS Noetic framework, with an Intel i7-9750H processor (2.60GHz) and 16GB of memory. The experiment focused on an unmanned swarm consisting of 10 unmanned platforms, designing three scenarios with target sizes of 10, 20, and 30, primarily evaluating the multi-task collaborative performance under communication-constrained environments. The experimental task required the unmanned platform swarm to complete the entire process of reconnaissance, engagement, and assessment of randomly distributed targets within a time limit. The simulation environment was set as a rectangular area of ​​100m × 57m, with the origin of the coordinate system located at the upper left corner. Figure 2 As shown, the experimental scenario includes two key areas: polygonal areas representing impassable obstacles; and circular areas simulating communication interference zones with a radius of R. The communication interference model employs a linear attenuation design, with a packet loss rate of 100% at the center, decreasing to M% at the edges, and the packet loss rate in the middle area decreasing by [missing information]. Formula calculation, where The distance from the current position to the center of the circle is represented by . The visualization system uses specific symbols to mark key elements: the location of the unmanned platform is represented by "x", initially deployed behind a sloping wall in the lower right corner of the map, and the target is represented by a dot. As a control, the Lazy Theta* path planning method is selected as the control group. Experimental data show that the proposed algorithm improvement scheme significantly improves the task execution efficiency of the unmanned swarm system.

[0044] like Figure 3 As shown in Table 2, in the comparative experiment with a target size of 10, the comparative method exhibited significant performance limitations: the task success rate was only 50%, the average time was 97.1 seconds, and the target discovery rate fluctuated significantly (±40%). In contrast, our proposed method increased the success rate to 100% and reduced the time by 59.3%.

[0045] Table 2 Quantitative Comparison and Analysis of Algorithm Performance In summary, this invention proposes a dynamic task scheduling method for unmanned swarms mimicking gray wolf collaborative decision-making under harsh communication environments. The results show that even under complex obstacles and high packet loss communication interference, this method can still maintain stable system operation, improve the task success rate by 50% compared with the comparative method, and achieve 100% reliable execution. The research results provide a scalable technical framework for military, logistics and other applications.

[0046] Another embodiment of the present invention provides a dynamic scheduling system for unmanned swarm mimicking the Grey Wolf troop under adverse communication conditions, used to execute a dynamic scheduling method for unmanned swarm mimicking the Grey Wolf troop under adverse communication conditions, comprising: The system and environment modeling module is used to construct unmanned swarm system models, task requirement and constraint models, and complex communication-constrained environment models, providing a theoretical modeling basis for dynamic task scheduling. Among them, the task requirement and constraint model has multiple pre-set constraints. The task pre-assignment module is used for dynamic leader election, task assignment, and task scheduling based on the various constructed models. Specifically, it includes: Dynamic leadership election unit: Based on the gray wolf-like dynamic cluster hierarchical architecture, a three-level dynamic leadership decision-making architecture is established through candidate node screening, leadership hierarchy initialization and dynamic switching control, and a dynamic election and assignment center is dynamically elected. Task assignment unit: The assignment center, which is elected, uses the Hungarian algorithm to solve the cluster task assignment scheme and assign tasks. Task scheduling unit: For each unmanned platform's task set, it sequentially verifies the satisfaction of preset constraints. Through an iterative task filtering mechanism, tasks that meet the constraints are included in the execution sequence, while tasks that do not meet the preset constraints are returned to the waiting queue. The task redistribution module is used to initiate redistribution based on a time-event dual-trigger mechanism, and uses a load balancing strategy to build a joint optimization objective to achieve dynamic task adjustment under the condition of satisfying constraints.

[0047] Furthermore, the present invention also provides an electronic device comprising one or more processors; a memory; and one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, and the one or more application programs are configured to perform a dynamic scheduling method for unmanned swarm mimicking gray wolf missions under adverse communication conditions as described above.

[0048] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The storage medium can be a memory, magnetic disk, optical disk, etc.

[0049] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A dynamic scheduling method for unmanned swarm tasks mimicking the Grey Wolf archetype under poor communication conditions, characterized in that, Includes the following steps: S1, System and Environment Modeling: Construct an unmanned swarm system model, a task requirement and constraint model, and a complex communication-constrained environment model to provide a theoretical modeling basis for dynamic task scheduling. Among them, the task requirement and constraint model pre-sets multiple constraints. S2, Task Pre-allocation: Based on the models constructed in step S1, dynamic leader election, task assignment, and task scheduling are performed, specifically including: S201, Dynamic Leadership Election: Based on the gray wolf-like dynamic cluster hierarchical architecture, a three-level dynamic leadership decision-making architecture is established through candidate node screening, leadership hierarchy initialization and dynamic switching control, and a dynamic election and assignment center is dynamically elected. S202, Task Assignment: The assignment center, which is elected, uses the Hungarian algorithm to solve the cluster task assignment scheme and performs task assignment. S203, Task Scheduling: For each unmanned platform's task set, the satisfaction of preset constraints is verified sequentially. Through an iterative task filtering mechanism, tasks that meet the constraints are included in the execution sequence, while tasks that do not meet the preset constraints are returned to the waiting queue. S3, Task Reassignment: Reassignment is initiated based on a time-event dual-trigger mechanism, and a joint optimization objective is constructed using a load balancing strategy to achieve dynamic adjustment of tasks under the condition of satisfying constraints.

2. The method for dynamic scheduling of unmanned swarm-based gray wolf-like tasks under adverse communication conditions as described in claim 1, characterized in that, The method for constructing the unmanned swarm system model in S1 specifically includes: A single unmanned platform Use the quintuple This indicates that the state set Describe the platform's intrinsic state at time t, including the platform's current two-dimensional coordinates x. i ∈R 2 Speed ​​of motion v i Type identifier Resource reserves Cost matrix Task queue ; It is a set of executable actions, including state refresh, task processing, and cost estimation functions; Store shared information, including global task sets. The status summary of neighboring platforms and the reconnaissance area are synchronized globally through a central server in centralized control mode. Representative decision-making mechanism, based on and Choose the optimal operation; It is the state transition function, defined as ; The entire cluster consists of The platform consists of several components, denoted as... .

3. The method for dynamic scheduling of unmanned swarm tasks mimicking the Grey Wolf under adverse communication conditions as described in claim 1, characterized in that, In S1, multiple constraints include synergy constraints, goal dependency constraints, resource constraints, and task timing constraints. The method for constructing the task requirements and constraint model specifically includes: Each combat target Characterized as containing spatial coordinates Type identifier Resource requirements during the reconnaissance / strike / assessment phase The set of prerequisite targets for task execution and completion status The tuple exists; and the following constraints apply: Coordination constraints: in, This indicates whether the i-th unmanned platform participated in the m-stage of the objective j. The value is 1 if the allocated resources are greater than 0, and 0 otherwise, ensuring that at least one unmanned platform participated. ≥ The unmanned platforms coordinate to execute, where R is the minimum threshold number of cooperating platforms; Target dependency constraints: in, This indicates the start time of objective j in task phase m. Table 1 The completion time of each objective is such that objective j can only begin after all its dependent objective tasks have been completed. Resource constraints: in, This represents the amount of resources allocated to target j by the i-th unmanned platform during task phase m, ensuring that the task's resource requirements are met. Task timing constraints: Equation (4) represents the timing requirements of the mission phases, with reconnaissance, strike, and assessment missions for the same target being carried out sequentially.

4. The method for dynamic scheduling of unmanned swarm-based gray wolf-like tasks under adverse communication conditions as described in claim 1, characterized in that, The method for constructing the complex communication-constrained environment model in S1 specifically includes: The set of obstacles is represented as: Each obstacle is defined as: ,in The coordinates of the obstacle's center are... The length and width dimensions of the obstacle. The angle of rotation of the obstacle. These are the physical characteristic parameters of the obstacle; Randomly distributed communication interference areas are used to simulate the interference characteristics in real wireless communication by parameterizing the regional interference. Equation (5) is used to simulate the phenomenon of linear attenuation of electromagnetic wave intensity with distance: in, This indicates the actual interference intensity at the location of the unmanned platform. Let be the distance between the unmanned platform i and the interference source w with an effective interference radius of r; The maximum value obtained by superimposing multiple regions reflects the "dominant effect" of electromagnetic interference: In the formula, K represents the number of interference zones where the unmanned platform i is located.

5. The method for dynamic scheduling of unmanned swarm-based gray wolf-like tasks under adverse communication conditions as described in claim 1, characterized in that, In S201, the constraints for candidate node selection in dynamic leader election are as follows: in, Represents a node Communication quality, Represents a node The number of times you interact with the network. This is the minimum communication quality threshold; The comprehensive scoring function for initializing leadership levels is defined as follows: in, This is a normalized value for communication quality. For window internal nodes Compared to the average packet loss rate of its neighbors, This represents the weight used to balance communication quality and packet loss rate.

6. The method for dynamic scheduling of unmanned swarm tasks mimicking the Grey Wolf under adverse communication conditions as described in claim 5, characterized in that, In S201, during dynamic leadership election, the three criteria for dynamic switching control include: The main leader is out of contact: heartbeat signal lost or interrupted; Local communication quality advantage: There exists a node whose communication quality is significantly better than that of the current node α, defined by a threshold. : Potential interference zone identification: If the communication quality of node α shows a downward trend within the sliding window, and neighboring nodes show an abnormally high packet loss rate, then it is considered to be in an interference environment.

7. The method for dynamic scheduling of unmanned swarm tasks mimicking the Grey Wolf under adverse communication conditions as described in claim 3, characterized in that, S202, task assignment, specifically includes: For N available unmanned platforms and Q target missions, a cost matrix is ​​constructed, considering the three phases of reconnaissance, strike, and assessment. ,in The cost function is defined as: in, For the normalization path cost, For the cost of communication links, At the cost of environmental threats, For ability compatibility, For collaborative gain; all components are normalized to the interval [0,1], weighted... Supports dynamic adjustments to enhance adaptability; The optimization objective of task assignment is to minimize the total cost and satisfy the synergy constraints, objective dependency constraints, and resource constraints in equations (1), (2), and (3): Where N represents the number of available unmanned platforms; Q represents the number of target tasks to be executed; For unmanned platforms Execute target task The overall cost under task phase m is defined by formula (14); To assign variables, if there is no platform The assigned task phase m executes the task. ,but If the value is 1, then it is 0 otherwise; J is the total cost objective function, which needs to be minimized and solved using the Hungarian function.

8. The method for dynamic scheduling of unmanned swarm-based gray wolf-like tasks under adverse communication conditions as described in claim 7, characterized in that, S203, task scheduling, specifically includes: After task assignment is completed, the task timing constraints in set equation (4) and the aforementioned target dependency constraints are used to sort and allocate time slots for the assigned tasks; the goal of task scheduling is to minimize task completion time and ensure that dependency logic is not violated. in, For the target task The completion time; the task scheduling result ensures the temporal consistency of the dependencies between tasks and the platform's execution capabilities, thereby forming an executable task sequence.

9. The method for dynamic scheduling of unmanned swarm tasks mimicking the Grey Wolf under adverse communication conditions as described in claim 1, characterized in that, S3, task redistribution, specifically includes: Task reassignment trigger set Defined as: Among them, time triggering is every Periodically check task and platform status; event trigger is when a node goes offline. Task failed or node interference strength Exceeding the threshold Immediately initiate reallocation; Task redistribution considers task completion, platform load balancing, and communication costs simultaneously, constructing a joint optimization objective and solving it using a Hungarian function: in, Penalty based on the importance of the uncompleted task; This incurs additional communication costs; The cost of load balancing is represented by the mean squared error of resource utilization: , , They represent , , The corresponding weighting coefficients; Let be the resource utilization rate of unmanned platform i, and ... Task redistribution must meet the pre-defined constraints of collaboration, goal dependency, resources, and task timing in the task requirements and constraint model.

10. An electronic device, characterized in that, It includes one or more processors; memory; and one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to perform the unmanned swarm gray wolf-like task dynamic scheduling method under adverse communication conditions as described in any one of claims 1-9.

Images