A method for task allocation of a UAV cluster for low-altitude economy

Abstract

The application is a kind of unmanned aerial vehicle cluster task allocation method for low altitude economy.The application relates to the technical field of unmanned aerial vehicle cluster, and the application establishes CMOTSW model, divides decision variables, carries out model constraint, establishes objective function, carries out non-dominated sorting based on elite strategy, and selects optimization strategy based on non-dominated rank.The application proposes to use improved multi-objective artificial rabbit optimization algorithm to carry out systematic research on multi-unmanned aerial vehicle cooperative task allocation problem, so as to provide method support for efficient use of unmanned aerial vehicle in application scenarios such as emergency rescue and public service.Experimental results show that the improved artificial rabbit optimization algorithm effectively improves the task execution efficiency.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) swarm technology, and is a UAV swarm task allocation method for the low-altitude economy. Background Technology

[0002] In recent years, unmanned aerial vehicles (UAVs) have demonstrated irreplaceable value in public services and emergency response, such as disaster search and rescue and environmental monitoring. With the increasing complexity and dynamism of application environments, various new collaborative operation modes have emerged, including multi-UAV collaborative operations and cross-domain UAV swarm collaborative operations. Against this backdrop, task assignment (TA) has gradually become a core issue in UAV and UAV swarm research, and a crucial link in achieving efficient execution of single-type tasks. How to rationally assign UAVs to complete corresponding tasks under complex environments and multiple constraints to improve overall task completion efficiency and system performance has become a key issue in current unmanned system application research.

[0003] In practical applications, the essence of multi-UAV task allocation lies in combining individual UAV attributes with task requirements to rationally plan the execution sequence of each UAV, thereby achieving optimal task efficiency or allocation results that meet application needs. Meanwhile, the task allocation process is often influenced by various constraints, such as inter-UAV coordination constraints, individual UAV performance limitations, and environmental uncertainties. These factors not only directly determine the feasibility of the task allocation scheme but may also affect the overall security and robustness of the UAV system. In multi-UAV scheduling systems, target allocation is a key aspect, involving the formulation of search and rescue task allocation strategies based on UAV platform attributes and target requirements. Currently, this type of problem has been abstracted into various mathematical models for solution, common models including: Mixed Integer Linear Programming (MILP), Weapon Target Assignment (WTA), Multiple Traveling Salesman Problem (MTSP), Network Flow Optimization (NFO), Vehicle Routing Problem (VRP), and Cooperative Multi-Task Assignment (CMTAP).

[0004] Traditional optimization methods, such as linear programming and weight- or constraint-based optimization methods, perform well in small-scale problems, but their computational efficiency and solution quality are limited when faced with complex constraints and large-scale objective tasks. In contrast, intelligent optimization algorithms, with their strong global search capabilities and good adaptability, have attracted widespread attention in objective allocation problems. Existing research has applied various intelligent optimization strategies, such as evolutionary algorithms (EA), genetic algorithms (GA), ant colony optimization (ACO), and artificial bee colony optimization (ABC), to multi-objective task modeling and solving. These algorithms can simultaneously optimize multiple objective functions such as task completion time, resource consumption, and damage effects, exhibiting good adaptability and scalability. In recent years, some scholars have further proposed improved versions of multi-objective optimization algorithms to enhance solution accuracy and diversity. For example, Wang et al. constructed a multi-UAV cooperative delivery and routing model with time windows (MVTARPSDPTW) and proposed an improved brainstorming optimization algorithm based on Pareto advantage (MIBSO). Lu et al. proposed a multi-objective multi-UAV cooperative reconnaissance scheme combining a multi-task constrained multi-objective optimization framework (MTCOM) and an enhanced auction distributed algorithm. CHEN et al. constructed the MCATAP model and proposed a dual-population co-evolutionary immune algorithm (BCIA), significantly enhancing global and local search capabilities. Li et al. treated sensor node data collection as a multiple traveling salesman problem and used a genetic algorithm to solve it. GAO et al. proposed a unified optimization method combining the NSGA-II multi-objective genetic algorithm and the Contract Network Protocol (CNP) for the multi-UAV collaborative multi-task allocation and reassignment problem. Zhu et al. proposed the constraint-tolerant NSGA-II for UAV swarm task allocation under multiple constraints, significantly improving the convergence and diversity of the Pareto solution set. Wang et al. proposed a deep Q-learning algorithm based on improved multi-objective artificial bee colony (MOADQN) to solve the multi-weapon target allocation problem in complex environments. Li et al. emphasized the data collection process, decomposing the problem into three sub-problems, including task allocation, and solved them using a minimum-maximum multiple traveling salesman problem algorithm. Although the above methods have achieved certain results in specific situations, current multi-objective task allocation algorithms still have technical bottlenecks such as heavy computational burden, weak adaptability, and insufficient solution space exploration under heterogeneous constraints, multi-objective coupling, and complex battlefield scenarios. There is an urgent need to propose more robust and practical allocation optimization methods. Summary of the Invention

[0005] This invention addresses the problem of multi-UAV cooperative target allocation by establishing a CMOTSW multi-UAV cooperative target allocation model. Based on the established model, an improved multi-target artificial rabbit optimization algorithm is proposed. The proposed algorithm is then simulated on the CMOTSW model, and the results show that the improved artificial rabbit optimization algorithm effectively improves task execution efficiency.

[0006] This invention provides the following technical solutions: A method for task allocation in unmanned aerial vehicle (UAV) swarms for the low-altitude economy, the method comprising the following steps: Step 1: Build the CMOTSW model; Step 2: Define decision variables; Step 3: Apply model constraints; Step 4: Establish the objective function; Step 5: Non-dominated ranking based on elite strategy; Step 6: Optimization strategy selection based on non-dominated rank.

[0007] Preferably, it is provided with Drones are used to coordinate search and rescue operations within the designated area. Given several known task points, with each task point considered an independent task, each drone can accept multiple tasks, and each task must be completed exactly once. The CMOTSW model can be represented as a quadruple. ,in It is a multi-drone system, using a collection This represents a set where each element includes information such as the drone's position, flight altitude, and maximum flight distance. The task represents all the task points that need to be completed, denoted as a set. This includes information such as the location, altitude, and mission value of each mission point; The drones are configured to take off from the same location at the same time and execute tasks sequentially according to the assigned order. Indicates drone The sequence of tasks to be executed.

[0008] Preferably, the feasible solution for task allocation includes two parts: a set of drone search and rescue task allocations. and search and rescue execution sequence set In terms of form, the solution for multi-UAV target allocation can be derived from binary combinations. The decision variables are defined as follows: Unmanned aerial vehicle (UAV) search and rescue mission allocation set; Multiple drones were assigned to perform different search and rescue missions. The set of tasks assigned to a drone is represented as follows: (1) in Indicates assignment to the first The number of missions per drone; mission allocation determines the mission points covered by each drone in a mission; Unmanned aerial vehicle (UAV) search and rescue execution sequence set; After the tasks are assigned, each drone must execute its assigned tasks in a specific order. The mission execution sequence of the drone is represented as follows: (2) in, express The first drone to perform The index position of each task in the task set, the sequence length is equal to the number of tasks, that is... This order reflects the first Mission scheduling scheme for drones.

[0009] Preferably, the task coordination constraint is that all tasks need to be executed, and each task only needs to be executed once: (3) Mission allocation constraints: each UAV must conduct at least one search and rescue operation. Task points: (4) Resource utilization constraints: Each UAV can execute a maximum of no more than its energy capacity load, and must execute at least one mission. (5) Range constraints, flight distance of each UAV Not exceeding its maximum range : (6)

[0010] Preferably, drones are deployed. For task points The success rate of execution is used The goal is to minimize the residual value of mission points and maximize overall mission revenue. Therefore, the total revenue obtained by all drones after completing their missions is expressed as: (7) Transform it into a minimization problem, using This indicates the remaining strategic value of the mission point after the mission is completed. (8) For a certain task allocation scheme The total flight distance of the drone is: (9) drones of The calculation formula is: (10) Using Euclidean distance , The distances of the two flights are calculated as follows: (11) (12) The two objective functions are transformed into a multi-objective optimization problem. The objective function of the CMOTSW model is as follows: (13)

[0011] .

[0012] Preferably, the first step is to include the parental group. and offspring group Combining to form a group ,Right now ; Step 2: Individuals in the merged population are stratified according to Pareto dominance, and based on this hierarchy... Divided into various sets, that is, . It is the first best non-dominated set. This is the last non-dominated set; The third step is to select individuals from different non-dominated sets to form a new population based on an elite strategy. The size is smaller than Scale, then choose All members within become the next group. The next population The remaining members are selected from the subsequent set in order of their rank; When Indicates the th element in the sorted population. The individual in the first Given the objective function values ​​on each objective function, the formula for calculating the crowding distance is: (14) Crowding distance reflects the aggregation density of the solution set. In the population selection process, individuals with larger crowding distances are given priority in the next generation of individuals, that is, individuals with lower aggregation density are selected as the next generation of individuals.

[0013] Preferably, the energy factor calculation formula is as follows: (15) in, , , , is the The probability that the non-dominated rank of an individual in a population, after correction, is greater than 1 increases as the non-dominated rank decreases. The maximum number of iterations for the population. For the current algebra; If the rabbit's non-dominant rank drops below the second rank, then... The probability increases directly to 1, indicating that Pareto fronts below the second rank will use Equation (14) for global search optimization.

[0014] (16) For rabbit individuals with high non-dominated rank, local search optimization will be performed using equation (15): (17)

[0015] Preferably, the parent population is initialized, the fitness of individuals in the population is calculated, and a non-dominated sort is performed; Calculate the energy factor and select a search strategy; Generate a progeny population; Merge the offspring with the parent population; Elite strategies and non-dominant sorting are used to update the population; Remove redundant individuals from the population; A portion will be retained for shuffling and updated into the next generation of parent populations.

[0016] A computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement a method for allocating unmanned aerial vehicle (UAV) swarm tasks for the low-altitude economy.

[0017] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement a method for allocating unmanned aerial vehicle (UAV) swarm tasks for the low-altitude economy.

[0018] The present invention has the following beneficial effects: Based on this, this invention proposes to systematically study the problem of multi-UAV collaborative task allocation using an improved multi-objective artificial rabbit optimization algorithm, in order to provide methodological support for the efficient use of UAVs in emergency rescue and public service application scenarios. Attached Figure Description

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 The diagram shown is an illustration of the elite strategy of the present invention. Figure 2 The diagram shown is a schematic representation of the congestion distance according to the present invention. Figure 3 Displayed as A diagram illustrating probability; Figure 4 The search mechanism appears to depend on energy factors; Figure 5 Displayed as a diagram showing the correspondence between models and algorithms; Figure 6 The flowchart is shown as the ENS-ARMO algorithm. Figure 7 The results are displayed as a task allocation chart. Figure 8 The results are displayed as a graph showing the results of three indicators at two different scales. Detailed Implementation

[0021] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0022] The present invention will be described in detail below with reference to specific embodiments. Specific Implementation Example 1: according to Figures 1 to 8 As shown, the specific optimized technical solution adopted by the present invention to solve the above-mentioned technical problems is: The present invention relates to a method for allocating unmanned aerial vehicle (UAV) swarm tasks for low-altitude economy.

[0024] This invention provides a method for task allocation in unmanned aerial vehicle (UAV) swarms for the low-altitude economy, the method comprising the following steps: A method for task allocation in unmanned aerial vehicle (UAV) swarms for the low-altitude economy, the method comprising the following steps: Step 1: Build the CMOTSW model; Set with Drones are used to coordinate search and rescue operations within the designated area. Given several known task points, with each task point considered an independent task, each drone can accept multiple tasks, and each task must be completed exactly once. The CMOTSW model can be represented as a quadruple. ,in It is a multi-drone system, using a collection This represents a set where each element includes information such as the drone's position, flight altitude, and maximum flight distance. The task represents all the task points that need to be completed, denoted as a set. This includes information such as the location, altitude, and mission value of each mission point; The drones are configured to take off from the same location at the same time and execute tasks sequentially according to the assigned order. Indicates drone The sequence of tasks to be executed.

[0025] Step 2: Define decision variables; The feasible solution for task allocation consists of two parts: the drone search and rescue task allocation set. and search and rescue execution sequence set In terms of form, the solution for multi-UAV target allocation can be derived from binary combinations. The decision variables are defined as follows: Unmanned aerial vehicle (UAV) search and rescue mission allocation set; Multiple drones were assigned to perform different search and rescue missions. The set of tasks assigned to a drone is represented as follows: (1) in Indicates assignment to the first The number of missions per drone; mission allocation determines the mission points covered by each drone in a mission; Unmanned aerial vehicle (UAV) search and rescue execution sequence set; After the tasks are assigned, each drone must execute its assigned tasks in a specific order. The mission execution sequence of the drone is represented as follows: (2) in, express The first drone to perform The index position of each task in the task set, the sequence length is equal to the number of tasks, that is... This order reflects the first Mission scheduling scheme for drones.

[0026] Step 3: Apply model constraints; Task coordination constraints: all tasks must be executed, and each task only needs to be executed once. (3) Mission allocation constraints: each UAV must conduct at least one search and rescue operation. Task points: (4) Resource utilization constraints: Each UAV can execute a maximum of no more than its energy capacity load, and must execute at least one mission. (5) Range constraints, flight distance of each UAV Not exceeding its maximum range : (6)

[0027] Step 4: Establish the objective function; Setting up a drone For task points The success rate of execution is used The goal is to minimize the residual value of mission points and maximize overall mission revenue. Therefore, the total revenue obtained by all drones after completing their missions is expressed as: (7) Transform it into a minimization problem, using This indicates the remaining strategic value of the mission point after the mission is completed. (8) For a certain task allocation scheme The total flight distance of the drone is: (9) drones of The calculation formula is: (10) Using Euclidean distance , The distances of the two flights are calculated as follows: (11) (12) The two objective functions are transformed into a multi-objective optimization problem. The objective function of the CMOTSW model is as follows: (13)

[0028] .

[0029] Step 5: Non-dominated ranking based on elite strategy; Step 1: Include the parent group and offspring group Combining to form a group ,Right now ; Step 2: Individuals in the merged population are stratified according to Pareto dominance, and based on this hierarchy... Divided into various sets, that is, . It is the first best non-dominated set. This is the last non-dominated set; The third step is to select individuals from different non-dominated sets to form a new population based on an elite strategy. The size is smaller than Scale, then choose All members within become the next group. The next population The remaining members are selected from the subsequent set in order of their rank; When Indicates the th element in the sorted population. The individual in the first Given the objective function values ​​on each objective function, the formula for calculating the crowding distance is: (14) Crowding distance reflects the aggregation density of the solution set. In the population selection process, individuals with larger crowding distances are given priority in the next generation of individuals, that is, individuals with lower aggregation density are selected as the next generation of individuals.

[0030] The formula for calculating the energy factor is as follows: (15) in, , , , is the The probability that the non-dominated rank of an individual in a population, after correction, is greater than 1 increases as the non-dominated rank decreases. The maximum number of iterations for the population. For the current algebra; If the rabbit's non-dominant rank drops below the second rank, then... The probability increases directly to 1, indicating that Pareto fronts below the second rank will use Equation (14) for global search optimization.

[0031] (16) For rabbit individuals with high non-dominated rank, local search optimization will be performed using equation (15): (17)

[0032] Step 6: Optimization strategy selection based on non-dominated rank.

[0033] Preferably, the parent population is initialized, the fitness of individuals in the population is calculated, and a non-dominated sort is performed; Calculate the energy factor and select a search strategy; Generate a progeny population; Merge the offspring with the parent population; Elite strategies and non-dominant sorting are used to update the population; Remove redundant individuals from the population; A portion will be retained for shuffling and updated into the next generation of parent populations.

[0034] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement a method for allocating unmanned aerial vehicle (UAV) swarm tasks for the low-altitude economy.

[0035] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a method for allocating unmanned aerial vehicle (UAV) swarm tasks for low-altitude economy. Specific Implementation Example 2: The only difference between Embodiment 2 and Embodiment 1 of this application is that: CMOTSW Model Establishment Assuming there is an environment in the disaster area Drones are used to coordinate search and rescue operations within the designated area. There are several known task points, and execution at a single task point is considered an independent task. Each drone can accept multiple tasks, and each task must be completed exactly once. The CMOTSW model can be represented as a quadruple. ,in It is a multi-drone system that can be used in combination. This represents a set where each element includes information such as the drone's position, flight altitude, and maximum flight distance. A task represents all the task points that need to be completed, and can be represented as a set. This includes information such as the location, altitude, and mission value of each task point. The drones are programmed to take off from the same location at the same time and execute tasks sequentially according to the assigned order. Indicates drone The sequence of tasks to be executed. Each drone must complete all assigned tasks. During task execution, drones may fail due to environmental conditions (such as insufficient energy, severe weather, etc.), and the probability of failure varies among different drones. See Table 1 for the specific meanings of each parameter.

[0037] Table 1. Multi-UAV Collaborative Multi-Task Allocation Parameters

[0038] Division of decision variables In this model, the feasible solution for task allocation mainly includes two parts: the UAV search and rescue task allocation set. and search and rescue execution sequence set Formally, the solution for multi-UAV target allocation can be derived from binary combinations. The decision variables are defined as follows: (1) Unmanned Aerial Vehicle Search and Rescue Mission Allocation Set Multiple drones were assigned to perform different search and rescue missions. The set of tasks assigned to a drone is represented as (1) in Indicates assignment to the first The number of missions per drone. This mission allocation determines the mission points covered by each drone during the mission.

[0039] (2) Unmanned aerial vehicle (UAV) search and rescue execution sequence set After task assignment is completed, each drone must execute its assigned tasks in a specific order. The mission execution sequence of the drone is represented as follows: (2) in express The first drone to perform The index position of each task in the task set. The length of this sequence is equal to the number of tasks, i.e. This order reflects the first... Mission scheduling scheme for drones.

[0040] Model constraints Drones have limited energy capacity, making it impossible to assign each drone tasks beyond its energy capacity. Tasks have different values, categorized as high-value and low-value tasks. In this study, the value of each task point is known in advance. For high-value task points, it may be necessary to allocate multiple drones to execute them simultaneously to improve the reliability of task completion. To reduce the time drones spend in complex environments or high-risk areas, the following constraints need to be established: (1) Task coordination constraints. All tasks need to be executed, and each task only needs to be executed once: (3) (2) Task allocation constraints. Each UAV must conduct at least one search and rescue operation. Task points: (4) (3) Resource utilization constraints. Each UAV may execute at most one mission, provided that the number of missions it can perform does not exceed its energy capacity load. (5) (4) Flight range constraints. Flight distance for each UAV. Not exceeding its maximum range : (6) objective function When multiple UAVs collaborate to perform search and rescue or monitoring missions, it is generally considered that the task allocation scheme with the lower the remaining task value in the uncovered areas and the lower the cost required to complete the task, the better. Here, the cost of task execution can be represented by the total range or total energy consumption required for the UAVs to complete the task. Based on this, the objective function of the CMOTSW model is constructed as follows: 1) Consider the residual value of the mission point after the drone mission is completed.

[0041] Setting up a drone For the task point The success rate of execution is used This means that if the goal is to minimize the residual value of mission points and maximize overall mission revenue, then the total revenue obtained by all drones after completing the mission can be expressed as: (7) Transform it into a minimization problem, using This indicates the remaining strategic value of the mission point after the mission is completed.

[0042] (8) 2) Consider the path cost required to execute the task. For a certain task allocation scheme The total flight distance of the drone is: (9) drones of The calculation formula is: (10) In this study, Euclidean distance was used. , The distances of the two flights are calculated.

[0043] (11) (12) The meanings of the parameters involved in the objective function are shown in Table 2.

[0044] Table 2. Meaning of each parameter in the objective function

[0045] In summary, this study transforms the two objective functions into a multi-objective optimization problem. The improved objective function of the CMOTSW model is as follows: (13)

[0046]

[0047] Improved multi-objective artificial rabbit optimization algorithm – ENS-ARMO Based on the multi-objective task allocation problem in this study, an Elite Non-dominated Sorting Artificial Rabbit Multi-objective Optimization (ENS-ARMO) algorithm is proposed. First, in the initialization phase, the dimension of the optimization objective is expanded to 2, and a new encoding method is used for both the UAV and the target. Second, the Elite strategy and Fastest Dominant Sorting (FSN) are incorporated into the artificial rabbit optimization algorithm, and the calculation method of the current energy factor is modified to dynamically change according to the current non-dominated sorting level of the rabbit. Finally, the two continuous position update operators in the original artificial rabbit optimization algorithm are replaced with discrete crossover operators and local mutation crossover operators, thereby achieving discrete encoding population update.

[0048] Non-dominant ranking based on elite strategy The specific process of this strategy is as follows: Step 1: Include the parent group and offspring groups Combining to form a group ,Right now .

[0049] Step 2: Individuals in the merged population are stratified according to Pareto dominance. Based on this hierarchical system... Divided into various sets, that is, . It is the first best non-dominated set. This is the last non-dominated set.

[0050] The third step involves selecting individuals from different non-dominated sets to form a new population, based on an elite strategy. Specifically, if the set The size is smaller than Scale, then choose All members within become the next group. The next population The remaining members are selected from the subsequent set in order of their ranking.

[0051] In the process of selecting new populations, the following may occur: Figure 2 As shown in the F3 layer, a portion of the cases need to be eliminated, so the concept of crowding distance is introduced.

[0052] If we take Indicates the th element in the sorted population. The individual in the first Given the objective function values ​​on each objective function, the formula for calculating the crowding distance is: (14) Crowding distance reflects the clustering density of the solution set. During population selection, individuals with larger crowding distances are prioritized for the next generation, meaning individuals with lower clustering densities are selected. For the first and last individuals in the objective function ranking, since the distance between them and their adjacent individuals cannot be calculated, their crowding distance is set to infinity. This ensures that individuals located on either side are prioritized during the selection process.

[0053] Optimization strategy selection based on non-dominated rank To enable the ENS-ARMO algorithm to efficiently search for the optimal non-dominated solution, the energy factor calculation method is dynamically adjusted based on the current non-dominated level of the rabbits, ensuring that individuals at different Pareto fronts receive a reasonable bias between global and local searches. The energy factor calculation formula is shown below: (15) in , , , is the The non-dominated rank of an individual in a population. The probability that the corrected energy factor is greater than 1 increases as the non-dominated rank decreases. The maximum number of iterations for the population. For the current algebra.

[0054] As can be seen from Equation 13, The smaller the value, the closer the solution is to the Pareto front. The higher the probability, the better. Furthermore, once the current rabbit's non-dominant rank drops below the second rank, then... The probability increases directly to 1, indicating that Pareto fronts below the second rank will use equation (14) for global search optimization.

[0055] (16) Conversely, for rabbit individuals with higher non-dominated ranks, Equation 15 will be used for local search optimization.

[0056] (17) like Figure 3 As shown, this adjustment improves the correlation between the choice of population update method and non-dominated sorting, thus facilitating the search for the optimal non-dominated solution.

[0057] Figure 4 The diagram depicts a search mechanism dependent on energy factors. In the figure, "Level 1," "Level 2," and "Level 3" represent groups at different levels of non-dominance. Rightward arrows indicate individuals in the population performing a global search along the horizontal axis, while upward arrows indicate individuals in the population performing a local search along the vertical axis.

[0058] The specific steps of the improved multi-objective artificial rabbit optimization algorithm are as follows: Step 1: Initialize the parent population, calculate the fitness of individuals in the population, and perform non-dominated sorting. Step 2: Calculate the energy factor and select a search strategy; Step 3: Generate the offspring population; Step 4: Merge the offspring with the parent population; Step 4: Elite strategy and non-dominant sorting to update the population; Step 5: Remove redundant individuals from the population; Step 6: Retain a portion of the mixture and update it as the next generation of parent population.

[0059] Experimental parameter settings The proposed ENS-ARMO algorithm is compared with three classic multi-objective algorithms—MOGA, MOACO, and NSGA-II—under the model established in this study. This research conducts simulation experiments on multi-UAV collaborative data acquisition task allocation using four algorithms for two different UAV-enemy target scale scenarios, and evaluates the algorithm's performance from three metrics: runtime, IGD, and HV.

[0060] To verify the feasibility and effectiveness of the model and algorithm proposed in this study, a simulation experiment was conducted in a 20km × 20km area. The takeoff positions of the UAVs were set at (0km, 2km), and two experimental scenarios were set according to the number of UAVs N and the number of targets M: a small-scale scenario (N=5, M=20) and a medium-scale scenario (N=10, M=40).

[0061] Four types of wireless sensor network data values ​​were randomly generated between [5, 10]. The success rates of different UAVs in executing tasks on different types of targets varied, as shown in Table 3.

[0062] Table 3 Task Execution Success Rate

[0063] To ensure the fairness of the algorithm comparison, the population size for all algorithms was set to 500, and the maximum number of iterations was set to 8000. Each algorithm was run independently 30 times under both scales. The final experimental result was the average of the 30 runs.

[0064] The above description is merely a preferred embodiment of a drone swarm task allocation method for low-altitude economy. The scope of protection for such a method is not limited to the above embodiments; all technical solutions falling within this framework are within the scope of protection of this invention. It should be noted that for those skilled in the art, any improvements and variations made without departing from the principles of this invention should also be considered within the scope of protection of this invention.

Claims

1. A method for task allocation in UAV swarms for the low-altitude economy, characterized by: The method includes the following steps: Step 1: Build the CMOTSW model; Step 2: Define decision variables; Step 3: Apply model constraints; Step 4: Establish the objective function; Step 5: Non-dominated ranking based on elitist strategy; Step 6: Optimization strategy selection based on non-dominated rank.

2. The method according to claim 1, characterized in that: Set with Drones are used to coordinate search and rescue operations within the designated area. Given several known task points, with each task point considered an independent task, each drone can accept multiple tasks, and each task must be completed exactly once. The CMOTSW model can be represented as a quadruple. ,in It is a multi-drone system, using a collection This represents a set where each element includes information such as the drone's position, flight altitude, and maximum flight distance. The task represents all the task points that need to be completed, denoted as a set. This includes information such as the location, altitude, and mission value of each mission point; The drones are configured to take off from the same location at the same time and execute tasks sequentially according to the assigned order. Indicates drone The sequence of tasks to be executed.

3. The method according to claim 2, characterized in that: The feasible solution for task allocation consists of two parts: the drone search and rescue task allocation set. and search and rescue execution sequence set In terms of form, the solution for multi-UAV target allocation can be derived from binary combinations. The decision variables are defined as follows: Unmanned aerial vehicle (UAV) search and rescue mission allocation set; Multiple drones were assigned to perform different search and rescue missions. The set of tasks assigned to a drone is represented as follows: (1) in Indicates assignment to the first The number of missions per drone; mission allocation determines the mission points covered by each drone in a mission; Unmanned aerial vehicle (UAV) search and rescue execution sequence set; After the tasks are assigned, each drone must execute its assigned tasks in a specific order. The mission execution sequence of the drone is represented as follows: (2) in, express The first drone to perform The index position of each task in the task set, the sequence length is equal to the number of tasks, that is... This order reflects the first Mission scheduling scheme for drones.

4. The method according to claim 3, characterized in that: Task coordination constraints: all tasks must be executed, and each task only needs to be executed once. (3) Mission allocation constraints: each UAV must conduct at least one search and rescue operation. Task points: (4) Resource utilization constraints: Each UAV can execute a maximum of no more than its energy capacity load, and must execute at least one mission. (5) Range constraints, flight distance of each UAV Not exceeding its maximum range : (6)。 5. The method according to claim 4, characterized in that: Setting up a drone For the task point The success rate of execution is used The goal is to minimize the residual value of mission points and maximize overall mission revenue. Therefore, the total revenue obtained by all drones after completing their missions is expressed as: (7) Transform it into a minimization problem, using This indicates the remaining strategic value of the mission point after the mission is completed. (8) For a certain task allocation scheme The total flight distance of the drone is: (9) drones of The calculation formula is: (10) Using Euclidean distance , The distances of the two flights are calculated as follows: (11) (12) The two objective functions are transformed into a multi-objective optimization problem. The objective function of the CMOTSW model is as follows: (13) 。 6. The method according to claim 5, characterized in that: Step 1: Include the parent group and offspring group Combining to form a group ,Right now ; Step 2: Individuals in the merged population are stratified according to Pareto dominance, and based on this hierarchy... Divided into various sets, that is, , It is the first best non-dominated set. This is the last non-dominated set; The third step is to select individuals from different non-dominated sets to form a new population based on an elite strategy. The size is smaller than Scale, then choose All members within become the next group. The next population The remaining members are selected from the subsequent set in order of their rank; When Indicates the th element in the sorted population. The individual in the first Given the objective function values ​​on each objective function, the formula for calculating the crowding distance is: (14) Crowding distance reflects the aggregation density of the solution set. In the population selection process, individuals with larger crowding distances are given priority in the next generation of individuals, that is, individuals with lower aggregation density are selected as the next generation of individuals.

7. The method according to claim 6, characterized in that: The formula for calculating the energy factor is as follows: (15) in, , , , is the The probability that the non-dominated rank of an individual in a population, after correction, is greater than 1 increases as the non-dominated rank decreases. The maximum number of iterations for the population. For the current algebra; If the rabbit's non-dominant rank drops below the second rank, then... The probability increases directly to 1, indicating that Pareto fronts below the second rank will use Equation (14) for global search optimization; (16) For rabbit individuals with high non-dominated rank, local search optimization will be performed using equation (15): (17)。 8. The method according to claim 7, characterized in that: Initialize the parent population, calculate the fitness of individuals in the population, and perform non-dominated sorting; Calculate the energy factor and select a search strategy; Generate a progeny population; Merge the offspring with the parent population; Elite strategies and non-dominant sorting are used to update the population; Remove redundant individuals from the population; A portion will be retained for shuffling and updated into the next generation of parent populations.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the method as claimed in any one of claims 1-8.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the method of any one of claims 1-8.

Images