Multi-target unmanned aerial vehicle task allocation method and system based on quantum flying fox mechanism

Abstract

The invention discloses a multi-target unmanned aerial vehicle task allocation method and system based on a quantum flying fox mechanism, relates to the technical field of unmanned aerial vehicle task allocation, and is used for solving the technical problem that multi-target unmanned aerial vehicle task allocation cannot be solved in a traditional method. The method is technically characterized by comprising the following steps: establishing a multi-unmanned aerial vehicle task allocation model, and further setting a target function of multi-target unmanned aerial vehicle task allocation; taking the target function as a fitness function, and performing optimization solution on the target function by using a quantum flying fox swarm algorithm to obtain an optimal task allocation matrix; complicated parameters in an original flying fox algorithm are simplified, the defects that the flying fox algorithm is prone to falling into a local optimal solution and cannot solve a multi-target engineering problem are overcome, and compared with other schemes, the obtained non-dominated solution set can meet a wider distribution scene, and the application range is expanded. According to the invention, multi-target unmanned aerial vehicle task allocation is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle task allocation, and in particular to a multi-target unmanned aerial vehicle task allocation method and system based on a quantum flying fox mechanism. Background Art

[0002] UAV task allocation is an important part of the UAV system. In the task allocation process, UAVs are automatically planned, decided and scheduled according to the nature of the task, UAV performance and environmental conditions. The task allocation algorithm needs to comprehensively consider various complex factors, including: task requirements, UAV capabilities, environmental factors, communication and control, collaborative operations, task priorities and dynamic adjustments. There are three main task allocation methods today: centralized control, distributed control and hybrid control. However, multi-objective UAV task allocation with joint optimization of multiple conflicting objectives is a difficult technical problem to solve. Its successful solution can break through the application limitations and engineering limitations of existing UAV task allocation.

[0003] Through the search of existing literature, it is found that the task allocation problem has existed since the first drone was manufactured in 1917. After gradual evolution and accumulation, many research institutions, scholars and engineers have continuously experimented and innovated, and the task allocation method has been fully developed. Qi Xiaogang et al. pointed out in the "Review of Task Planning of Multi-UAVs under Multiple Constraints" published in the Journal of Intelligent Systems (2020, 15(2): 204-217.) that the challenges currently faced in the field of multi-UAV task allocation include: how to build a unified framework to coordinate the collaborative target allocation of multiple UAVs; how to select and quantify the constraints in collaborative tasks; how to effectively handle the path planning of multi-UAV collaborative operations to ensure the smoothness and rationality of the path; and how to quickly respond and replan the flight path of multiple UAVs when encountering serious threats. The above problems are difficult to solve in actual engineering. Therefore, in actual engineering, it is necessary to fully consider the possible situations, continuously simplify the constraints and optimize the UAV model. Zhang Ruipeng et al. designed a hybrid particle swarm algorithm in the article “Hybrid Particle Swarm Algorithm for Collaborative Task Assignment of Multiple UAVs” published in Acta Aeronautica Sinica (2022, 43(12): 326011-326011). This algorithm successfully solved the problem that traditional particle swarm algorithms are prone to local convergence and can effectively solve the problem of UAV task assignment. However, due to the limitations of the single-objective particle swarm algorithm itself and limited application scenarios, it cannot meet the joint optimization of multiple objectives, and the effect achieved is not very good. Konstantinos Zervoudakis and Stelios Tsafarakis proposed the Flying Foxes Optimization (FFO) in “A global optimizer inspired from the survival strategies offlying foxes” published in Engineering with Computers (2023, 39(2): 1583-1616). The Flying Foxes Optimization (FFO) algorithm was inspired by the survival strategy of flying foxes in heat waves. This algorithm solves the optimization problem by simulating the behavior of flying foxes, which provides a new idea for solving the drone allocation problem. However, it can only solve single-objective problems and cannot effectively solve the Pareto front solution set of multi-objective joint optimization.

[0004] In summary, there are still many problems in the existing technology in the direction of UAV task allocation, especially the common task allocation methods such as exhaustive method, graph theory method, genetic algorithm, ant colony optimization algorithm and threshold response method, although they have certain effects on UAV task allocation, these methods have the disadvantages of complex algorithms, highly random results, and extremely high instability. Although the new optimization algorithms proposed in recent years, such as the Feihu algorithm, have improved these problems, they cannot solve the multi-objective problem. Summary of the invention

[0005] In view of this, the present invention proposes a multi-target UAV task allocation method and system based on the quantum flying fox mechanism to solve the technical problems of multi-target UAV task allocation that cannot be solved in the prior art.

[0006] According to one aspect of the present invention, a multi-target UAV task allocation method based on quantum flying fox mechanism is proposed, comprising the following steps:

[0007] Step 1: Establish a multi-UAV task allocation model, and then set the objective function of multi-objective UAV task allocation;

[0008] Step 2: Taking the objective function as the fitness function, optimizing and solving the objective function using the quantum flying fox swarm algorithm to obtain the optimal task allocation matrix; including:

[0009] Step 21: Initialize the quantum flying fox group;

[0010] Step 22: Calculate the fitness function value of each quantum flying fox in the quantum flying fox group;

[0011] Step 2 and 3: Use the fitness function values ​​of all quantum flying foxes to obtain the non-dominated levels and crowding distances of all quantum flying foxes;

[0012] Step 24: Use the quantum flying fox death mechanism and the density of the quantum flying fox to update the quantum flying fox;

[0013] Step 25: Use the population evolution strategy of the quantum flying fox group to update the quantum position of the quantum flying fox group;

[0014] Step 26: Determine whether the upper limit of the number of iterations has been reached. If so, terminate the iteration and proceed to step 27; if not, return to step 24;

[0015] Step 27: Convert the quantum position information of the quantum flying fox with the highest non-dominated level into a task allocation matrix.

[0016] Furthermore, the objective function in step 1 is:

[0017]

[0018] In the formula, min means minimization; is the combat benefit fitness function; is the loss cost fitness function; is the time cost fitness function;

[0019]

[0020]

[0021] Where N is the number of drones; M is the number of targets; A = {a n,m |a n,m ∈{0,1}} N×M represents the task allocation matrix, where n = 1, 2, ..., N, m = 1, 2, ..., M, a n,m =1 means that the nth UAV is responsible for carrying out the combat mission against the mth enemy target, a n,m = 0 means that the nth UAV is not carrying out combat missions against the mth enemy target; P′ n,m represents the probability of drone n attacking target m; P″ n,m represents the survival probability of drone n after attacking target m; V′ m represents the value of the mth target; The maximum number of tasks to be performed by the nth UAV; S n is the total flight distance of the nth UAV’s mission, is the maximum range of the nth UAV; represents the moving average speed of the nth UAV; max(·) is the maximum value function; |·| is the absolute value function; χ 1 , χ 2 and χ 3 are the weight factors of the mission constraint penalty term, attack constraint penalty term and range constraint penalty term respectively.

[0022] Further, the initialization of the quantum flying fox group in step 21 includes:

[0023] Set the number of individuals in the quantum flying fox group to The maximum number of iterations is T. Each dimension of the quantum position of the quantum flying fox is randomly initialized to a random number between [0,1]. The quantum position of the ith quantum flying fox of the tth generation is recorded as where D represents the maximum dimension of the quantum position vector, which matches the target number, d=1,2,...D; the quantum position of the ith quantum flying fox of the tth generation corresponds to the position This position is mapped by Income, and are the maximum and minimum values ​​of the d-th dimension position in the search space respectively; the fitness vector of the ith quantum flying fox of the t-th generation is recorded as t is the number of iterations; initialize the survival list of the substitute quantum flying fox: randomly add the first generation of quantum flying foxes to the survival list, fill the survival list, and the size is where β is a constant.

[0024] Further, the process of step 22 includes: the quantum position of the ith quantum flying fox of the tth generation Mapping to Location Then map the position of the quantum flying fox into a task allocation matrix Then the fitness function vector of the ith quantum flying fox of the tth generation is: in, t=1,2,...,T.

[0025] Furthermore, the process of steps 2 and 3 includes:

[0026] Calculate the non-dominated level: determine the dominance relationship between quantum flying foxes, the dominance relationship includes the non-dominated situation between quantum flying foxes and the non-dominated relationship between two quantum flying foxes; the calculation process of the non-dominated situation between quantum flying foxes includes: there are two quantum flying foxes in the tth generation, numbered i and i respectively 1 and i 2 , if the fitness values ​​of the two quantum flying foxes have the same relationship

[0027] And they do not take the equal sign at the same time, then it is called quantum flying fox i 2 Non-dominated quantum flying fox i 1 , namely Quantum Flying Fox i 2 Inferior to Quantum Flying Fox i 1 ; The calculation process of the two quantum flying foxes having no domination relationship includes: according to the non-dominated situation between the quantum flying foxes, the quantum flying foxes that are not dominated by any quantum flying foxes are stored in the first level, and the non-dominated level is 1; the quantum flying foxes that are dominated by the quantum flying foxes in the first level and not dominated by any other quantum flying foxes are stored in the second level, and the non-dominated level is 2, and the calculation is continued until the non-dominated level of all quantum flying foxes in the tth generation is obtained;

[0028] Calculate the crowding distance by target fitness function: First, suppose there are μ quantum flying foxes in the first level of the tth generation of quantum flying foxes in the first level, and the fitness vector is k=1,2,...,μ, sort the quantum flying foxes in the first level in ascending order according to the first fitness value, and the crowding distance between the two boundaries after sorting and Set to infinity, the crowding distance of the first fitness function of the kth quantum flying fox is:

[0029]

[0030] Among them, k is the number of cells sorted in ascending order by the first fitness function. The corresponding unsorted number, is the maximum fitness value of the first fitness function, is the minimum fitness value of the first fitness function,

[0031] Then, the crowding distance of the second and third fitness functions is calculated and Then the total crowding distance of the kth quantum flying fox is

[0032] Finally, arrange the μ quantum flying foxes in descending order according to the crowding distance.

[0033] Furthermore, the process of step 24 includes:

[0034] Density calculation of quantum flying foxes: taking the quantum position of the quantum flying foxes and the target fitness function as input, firstly calculating the distance between the quantum flying foxes, then setting the distance threshold according to the distance between the quantum flying foxes and the average distance of the group, and finally obtaining the density according to the distance threshold and the distance between the quantum flying foxes;

[0035] According to the density of the quantum flying foxes, some suffocated quantum flying foxes are selected and supplemented with quantum flying foxes in the survival list; the survival list is excellent individuals selected from the previous generation of quantum flying foxes, and the size is about β times the number of groups. After each iteration, the better quantum flying foxes in the current population are added to the survival list, and the redundant solutions in the list are deleted.

[0036] Furthermore, the process of step 25 is:

[0037] Crossover: Exclude quantum flying foxes with non-dominated level 1 and randomly select different quantum flying foxes for multi-point crossover. The number of crossover points is related to the dimension of the quantum flying fox position vector. The formula is in is the floor rounding function;

[0038] Quantum flying fox update: The quantum rotation angle update formula of the i-th quantum flying fox is in is the quantum position of the jth dimension of the quantum flying fox labeled b randomly selected from the tth generation Pareto optimal solution set, α t is the distance exploration factor, according to the current individual fitness and the Pareto optimal solution fitness, α t The update amplitude will be automatically adjusted; α t The update formula is in is the kth fitness value of the quantum flying fox labeled b randomly selected from the tth generation Pareto optimal solution set, is the kth fitness value of the ith quantum flying fox, c 1 and c 2 is a constant; the update formula for the j-dimensional quantum position of the ith quantum flying fox is: in is the j-th dimensional quantum position of the i-th quantum flying fox before updating;

[0039] The parent generation quantum flying fox population is The offspring population is Will and Synthesized into At this time, perform Pareto non-dominated sorting and calculate the crowding distance. Arrange in ascending order according to the non-dominated level. Arrange the quantum flying foxes of the same non-dominated level in descending order according to the crowding distance. Finally, select from top to bottom. Only quantum flying foxes are placed into the new parent population middle.

[0040] According to another aspect of the present invention, a multi-target UAV task allocation system based on quantum flying fox mechanism is proposed, the system comprising:

[0041] a model building module configured to establish a multi-UAV task allocation model and then set an objective function for multi-objective UAV task allocation;

[0042] The optimal task allocation solution module is configured to use the objective function as a fitness function, optimize and solve the objective function using a quantum flying fox group algorithm, and obtain an optimal task allocation matrix; including: initializing a quantum flying fox group; calculating the fitness function value of each quantum flying fox in the quantum flying fox group; using the fitness function values ​​of all quantum flying foxes to obtain the non-dominated level and crowding distance of all quantum flying foxes; using the quantum flying fox death mechanism and the density of quantum flying foxes to update the quantum flying foxes; using the population evolution strategy of the quantum flying fox group to update the quantum position of the quantum flying fox group; judging whether the upper limit of the number of iterations is reached, if reached, terminating the iteration, and converting the quantum position information of the quantum flying fox with the highest non-dominated level into a task allocation matrix; if not reached, continuing to update.

[0043] Furthermore, the objective function in the model building module is:

[0044]

[0045] In the formula, min means minimization; is the combat benefit fitness function; is the loss cost fitness function; is the time cost fitness function;

[0046]

[0047] Where N is the number of drones; M is the number of targets; A = {a n,m |a n,m ∈{0,1}}N×M represents the task allocation matrix, where n = 1, 2, ..., N, m = 1, 2, ..., M, a n,m =1 means that the nth UAV is responsible for carrying out the combat mission against the mth enemy target, a n,m =0 means that the nth UAV is not on a combat mission against the mth enemy target; represents the probability of drone n attacking target m; P″ n,m represents the survival probability of drone n after attacking target m; V′ m represents the value of the mth target; The maximum number of tasks to be performed by the nth UAV; S n is the total flight distance of the nth UAV’s mission, is the maximum range of the nth UAV; represents the moving average speed of the nth UAV; max(·) is the maximum value function; |·| is the absolute value function; χ 1 , χ 2 and χ 3 are the weight factors of the mission constraint penalty term, attack constraint penalty term and range constraint penalty term respectively.

[0048] The beneficial technical effects of the present invention are:

[0049] The present invention establishes a multi-UAV task allocation model, the purpose of which is to obtain a multi-target multi-UAV task allocation scheme, provide a new method for solving practical engineering problems, and use the quantum flying fox mechanism to simplify the scheme steps and improve the stability, feasibility and effectiveness of the scheme. Specifically, based on the target fitness function of the model, the quantum flying fox swarm algorithm is used to optimize and solve the target fitness function, and the optimal quantum position of the quantum flying fox is obtained, and then the optimal fitness value of the multi-target UAV task allocation method is obtained; among them, the present invention simplifies the complex parameters in the original flying fox algorithm, improves the shortcomings of the flying fox algorithm that is easy to fall into the local optimal solution and cannot solve multi-target engineering problems. Compared with other schemes, the obtained non-dominated solution set can meet a wider range of allocation scenarios and expand the scope of application.

[0050] The present invention proposes to combine quantum computing theory with the Feihu algorithm, and improves the original algorithm from multiple aspects. The traditional Feihu algorithm adopts a fuzzy self-adjustment method to avoid parameter dependence, but the method has a complex structure and involves multiple operations and parameters, which makes the Feihu algorithm converge slowly; the original Feihu algorithm cannot solve the multi-target problem; there are too many constraints in the UAV task allocation model. The quantum Feihu swarm algorithm of the present invention introduces a quantum revolving door to update the quantum position by updating the quantum rotation angle in terms of Feihu position update, improves the global search ability, and makes it easier for Feihu to jump out of the local optimum; secondly, the two generations of quantum Feihu swarms are merged together for special processing, so that the excellent individuals that appear in the iterative process are always retained. The quantum Feihu swarm algorithm of the present invention innovatively overcomes the limitations of the existing quantum fox swarm algorithm in processing multi-target tasks, and can obtain the Pareto optimal solution, thereby efficiently solving the problem of UAV task allocation when facing multiple optimization targets. Finally, the penalty term added to the quantum Feihu swarm algorithm of the present invention simplifies the problem of multiple constraints, provides more allocation schemes for the problem of multi-target task allocation, and has a broader application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention can be better understood by referring to the description given below in conjunction with the accompanying drawings, which together with the following detailed description are included in this specification and form a part of this specification, and are used to further illustrate the preferred embodiments of the present invention and explain the principles and advantages of the present invention.

[0052] Figure 1 This is a flow chart of a multi-target UAV task allocation method based on the quantum flying fox mechanism described in the present invention.

[0053] Figure 2 This is a schematic diagram of a multi-target UAV task allocation method based on the quantum flying fox mechanism of the present invention.

[0054] Figure 3 This is the initial position distribution map of the UAV and the target in the present invention.

[0055] Figure 4 This is a comparison chart of the Pareto optimal solution and non-dominated solution obtained by different methods in the present invention. DETAILED DESCRIPTION

[0056] In order to enable those skilled in the art to better understand the scheme of the present invention, exemplary implementations or embodiments of the present invention will be described below in conjunction with the accompanying drawings. Obviously, the described implementations or embodiments are only implementations or embodiments of a part of the present invention, not all of them. Based on the implementations or embodiments of the present invention, all other implementations or embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0057] The embodiment of the present invention proposes a multi-target UAV task allocation method based on the quantum flying fox mechanism, such as Figure 1-2 As shown, the method comprises the following steps:

[0058] Step 1: Set the objective function of multi-objective UAV task allocation;

[0059] Step 2: Taking the objective function as the fitness function, optimizing and solving the objective function using the quantum flying fox swarm algorithm to obtain the optimal task allocation matrix; including:

[0060] Step 21: Initialize the quantum flying fox group;

[0061] Step 22: Calculate the fitness function value of each quantum flying fox in the quantum flying fox group;

[0062] Step 2 and 3: Use the fitness function values ​​of all quantum flying foxes to obtain the non-dominated levels and crowding distances of all quantum flying foxes;

[0063] Step 24: Use the quantum flying fox death mechanism and the density of the quantum flying fox to update the quantum flying fox;

[0064] Step 25: Use the population evolution strategy of the quantum flying fox group to update the quantum position of the quantum flying fox group;

[0065] Step 26: Determine whether the upper limit of the number of iterations has been reached. If so, terminate the iteration and proceed to step 27; if not, return to step 24;

[0066] Step 27: Convert the quantum position information of the quantum flying fox with the highest non-dominated level into a task allocation matrix.

[0067] The method starts from step 1. In step 1, a multi-UAV task allocation model is established, and then the objective function of multi-objective UAV task allocation is set.

[0068] The specific steps are as follows: Assume that there are M targets that need to be attacked by drones, and there is a drone fleet consisting of N drones. The core of task allocation is to create a correspondence table between drones and targets. This correspondence table is defined as a task allocation matrix. The task allocation matrix is ​​A = {a n,m|a n,m ∈{0,1}} N×M , n=1,2,...,N,m=1,2,...,M,the element a in the matrix n,m Can be defined as The present invention considers three optimization objectives: maximizing the total value of the accumulated destruction of targets by drones, minimizing the cost of drone losses, and minimizing the task completion time.

[0069] The total value of the targets destroyed by the drone can be expressed as Where P′ n,m is the probability of drone n attacking target m. The better the performance of the drone, the greater the attack probability. P″ n,m is the survival probability of drone n after attacking m. The stronger the target’s defense, the smaller the survival probability. V′ m Represents the value of the mth target.

[0070] The loss cost of the drone can be expressed as 1-P″ n,m is the damage probability of the drone attacking the target.

[0071] The maximum time cost of a drone attacking a target is Where S n is the total flight distance of the nth UAV performing the mission, is the average flight speed of the nth UAV, assuming that the time spent by the UAV to perform the task is negligible.

[0072] In addition, the target and the UAV mission should also satisfy the following constraints:

[0073] 1. Regarding the limit on the number of times a target can be attacked: Each target can only be attacked once at most.

[0074] 2. Constraints on drones: First, due to the limitation on the number of ammunition that drones can carry, each drone can only attack a limited number of targets; second, drones have limited fuel, and their attack range cannot exceed their maximum range.

[0075] Therefore, the penalty term is introduced, the constraints are converted into penalty terms and substituted into the objective function to transform the multi-UAV task allocation model.

[0076] In general, there are N drones that need to perform tasks and the number of targets is M. Then the task allocation matrix between drones and targets is: A = {a n,m |a n,m ∈{0,1}} N×M , where n = 1, 2, ..., N, m = 1, 2, ..., M, a n,m=1 means that the nth UAV is responsible for carrying out the combat mission against the mth enemy target, a n,m =0 means that the nth drone is not on a combat mission for the mth enemy target. The multi-drone task allocation model constructed by the present invention has three objective functions and three constraints. In order to simplify the calculation, a penalty term is introduced, and the constraints are converted into penalty terms and substituted into the objective function. Among them, the target optimization involves three core aspects: the overall value of the target destroyed by the drone, the loss cost of the drone in the process of executing the mission, and the time required to complete the entire mission; the penalty terms are: penalty for exceeding the number of times the target is attacked, penalty for exceeding the number of drone attacks, and penalty for exceeding the drone range. The multi-drone task allocation model is:

[0077]

[0078] Where N is the number of drones; M is the number of targets; A = {a n,m |a n,m ∈{0,1}} N×M , represents the task allocation matrix, where n = 1, 2, ..., N, m = 1, 2, ..., M, a n,m =1 means that the nth UAV is responsible for carrying out the combat mission against the mth enemy target, a n,m = 0 means that the nth UAV is not carrying out combat missions against the mth enemy target; P′ n,m represents the probability that drone n attacks target m, P″ n,m represents the survival probability of drone n after attacking target m; V′ m represents the value of the mth target, P d is the two-dimensional coordinate of the UAV base, P t is the two-dimensional coordinate of the target; the maximum moving distance of the drone is D max ; The maximum number of tasks to be performed by the nth UAV; S n is the total flight distance of the nth UAV’s mission, is the maximum range of the nth UAV; represents the average moving speed of the nth UAV; Assigning a model fitness function to the multi-UAV mission; is the combat benefit fitness function; is the loss cost fitness function; is the time cost fitness function; max(·) is the maximum value function; |·| is the absolute value function; χ 1 , χ 2 and χ 3 are the weight factors of the mission constraint penalty term, attack constraint penalty term and range constraint penalty term respectively.

[0079] Then, step 2 is executed. In step 2, the objective function is used as the fitness function, and the objective function is optimized and solved using the quantum flying fox swarm algorithm to obtain the optimal task allocation matrix.

[0080] In this step 2, first execute step 21 to initialize the quantum flying fox group.

[0081] According to an embodiment of the present invention, Figure 3 As shown in the figure, the initial task graph of UAV and tasks is given, based on the UAV task allocation model and Figure 3 , initialize the quantum flying fox group as follows: Set the number of individuals in the quantum flying fox group to The maximum number of iterations is T. Each dimension of the quantum position of the quantum flying fox is randomly initialized to a random number between [0,1]. The quantum position of the ith quantum flying fox of the tth generation is recorded as Here, D represents the maximum dimension of the quantum position vector, which matches the number of targets. d=1,2,...D; the quantum position of the ith quantum flying fox of the tth generation corresponds to the position This position is mapped through Income, and are the maximum and minimum values ​​of the d-th dimension position in the search space respectively. The fitness vector of the ith quantum flying fox of the tth generation is denoted as t is the number of iterations, and t=1 is taken at the beginning; initialize the survival list of the substitute quantum flying foxes: randomly add the first generation of quantum flying foxes to the survival list, fill the survival list, and the size is

[0082] Then execute step 22 to calculate the fitness function value of each quantum flying fox in the quantum flying fox group.

[0083] According to an embodiment of the present invention, the calculation process is as follows: the quantum position of the ith quantum flying fox of the tth generation is Mapping to Location Then map the position of the quantum flying fox into a task allocation matrix Then the fitness function vector of the ith quantum flying fox of the tth generation is in, t=1,2,...,T.

[0084] Then execute steps two and three, and use the fitness function values ​​of all quantum flying foxes to obtain the non-dominated levels and crowding distances of all quantum flying foxes.

[0085] According to an embodiment of the present invention, the calculation of the non-dominated level is as follows: Non-dominated situation between quantum flying foxes: There are two quantum flying foxes in the tth generation, numbered i and i respectively. 1 and i2 , if its fitness value has the relationship And they do not take the equal sign at the same time, then it is called quantum flying fox i 2 Non-dominated quantum flying fox i 1 , namely Quantum Flying Fox i 2 Inferior to Quantum Flying Fox i 1 , except that, there is no domination relationship between the two quantum flying foxes. According to the non-dominance situation between quantum flying foxes: quantum flying foxes that are not dominated by any quantum flying fox are stored in the first level, and the non-dominance level is 1; quantum flying foxes that are dominated by quantum flying foxes in the first level but not dominated by any other quantum flying foxes are stored in the second level, and the non-dominance level is 2. Similarly, the non-dominance level of all quantum flying foxes in the tth generation can be calculated.

[0086] The calculation of crowding distance is as follows: Taking the quantum flying fox in the first level as an example, assuming that there are μ quantum flying foxes in the first level of the tth generation, the fitness vector is k=1,2,...,μ, first sort the quantum flying foxes in the first level in ascending order according to the first fitness value, and then sort the crowding distance between the two boundaries and Set to infinity, then the crowding distance of the first fitness function of the kth quantum flying fox for Among them, label k is the number of fitness functions after being arranged in ascending order. The corresponding unsorted number, is the maximum fitness value of the first fitness function, is the minimum fitness value of the first fitness function, Follow the above steps to calculate the crowding distance of the second and third fitness functions and Then the total crowding distance of the kth quantum flying fox is Finally, arrange the μ quantum flying foxes in descending order according to the crowding distance.

[0087] Then execute step 24 to update the quantum flying fox using the quantum flying fox death mechanism and the density of the quantum flying fox.

[0088] According to an embodiment of the present invention, the quantum position of the quantum flying fox is updated according to the density of each quantum flying fox by the death mechanism of the quantum flying fox. Density calculation of quantum flying fox: first calculate the distance between the quantum flying foxes, then set the distance threshold according to the obtained distance and the average distance of the group, and finally obtain the density according to the distance threshold and the distance between the quantum flying foxes. Suffocation: When the density of quantum flying foxes in a certain area is too large, the quantum flying fox will suffocate. According to the previously obtained density, a part of the flying foxes will be selected to suffocate, and the quantum flying foxes in the survival list will be supplemented. The survival list is an excellent individual selected from the previous generation of quantum flying foxes, and the size is β times the number of individuals. After each iteration, the survival list is updated, the better quantum flying foxes in the current population are added to the survival list, and the redundant solutions in the list are deleted.

[0089] Then execute step 25 and use the population evolution strategy of the quantum flying fox group to update the quantum position of the quantum flying fox group.

[0090] According to an embodiment of the present invention, updating the quantum position of a quantum flying fox group includes:

[0091] Crossover: Randomly select different quantum flying foxes for multi-point crossover. The number of crossover points is related to the dimension D of the quantum flying fox position vector. in is the floor function.

[0092] The quantum flying fox update process is as follows: The quantum rotation angle update formula of the i-th quantum flying fox is in is the quantum position of the jth dimension of the quantum flying fox labeled b randomly selected from the tth generation Pareto optimal solution set, α t is the distance exploration factor, according to the current individual fitness and the Pareto optimal solution fitness, α t The update amplitude will be adjusted automatically. t The update formula is in is the kth fitness value of the quantum flying fox labeled b randomly selected from the tth generation Pareto optimal solution set, is the kth fitness value of the ith quantum flying fox, c 1 and c 2 is a constant. The formula for updating the j-dimensional quantum position of the ith quantum flying fox is: in is the j-th dimensional quantum position of the i-th quantum flying fox before updating.

[0093] The parent generation quantum flying fox population is The offspring population is Will and Synthesized into At this time, the Pareto optimal solution is sorted non-dominated and the crowding distance is calculated. The quantum flying foxes with the same non-dominated level are sorted in descending order according to the crowding distance. Finally, the quantum flying foxes are selected from top to bottom. Only quantum flying foxes are placed into the new parent population middle.

[0094] Then execute step 26 to determine whether the upper limit of the number of iterations has been reached. If so, terminate the iteration and proceed to step 27. If not, return to step 24.

[0095] According to the embodiment of the present invention, specifically, first check whether the preset iteration upper limit T has been reached. If it has been reached, the iteration is terminated and the next step is entered; if it has not been reached, t=t+1 is set and the fifth step is returned to continue execution.

[0096] Finally, step 27 is executed to convert the quantum position information of the quantum flying fox with the highest non-dominated level, i.e., level 1, into a task allocation matrix, thereby obtaining the final task allocation plan.

[0097] According to an embodiment of the present invention, the position information of the quantum fox with the highest non-dominated level (level 1) is converted into a task allocation matrix, the quantum position of the quantum flying fox in the Pareto optimal solution set is mapped to a position, and then the position of the quantum flying fox is mapped to a task allocation matrix to obtain the final task allocation solution.

[0098] According to another embodiment of the present invention, a multi-target UAV task allocation system based on the quantum flying fox mechanism is proposed, and the system includes:

[0099] a model building module configured to set an objective function for multi-objective UAV task assignment;

[0100] The optimal task allocation solution module is configured to use the objective function as a fitness function, optimize and solve the objective function using a quantum flying fox group algorithm, and obtain an optimal task allocation matrix; including: initializing a quantum flying fox group; calculating the fitness function value of each quantum flying fox in the quantum flying fox group; using the fitness function values ​​of all quantum flying foxes to obtain the non-dominated level and crowding distance of all quantum flying foxes; using the quantum flying fox death mechanism and the density of quantum flying foxes to update the quantum flying foxes; using the population evolution strategy of the quantum flying fox group to update the quantum position of the quantum flying fox group; judging whether the upper limit of the number of iterations is reached, if reached, terminating the iteration, and converting the quantum position information of the quantum flying fox with the highest non-dominated level into a task allocation matrix; if not reached, continuing to update.

[0101] In this embodiment, preferably, the objective function in the model building module is:

[0102]

[0103] In the formula, min means minimization; is the combat benefit fitness function; is the loss cost fitness function; is the time cost fitness function;

[0104]

[0105] Where N is the number of drones; M is the number of targets; A = {a n,m |a n,m ∈{0,1}} N×M represents the task allocation matrix, where n = 1, 2, ..., N, m = 1, 2, ..., M, a n,m =1 means that the nth UAV is responsible for carrying out the combat mission against the mth enemy target, a n,m = 0 means that the nth UAV is not on a combat mission for the mth enemy target; P n ' ,m represents the probability of drone n attacking target m; P″ n,m represents the survival probability of drone n after attacking target m; V′ m represents the value of the mth target; The maximum number of tasks to be performed by the nth UAV; S n is the total flight distance of the nth UAV’s mission, is the maximum range of the nth UAV; represents the moving average speed of the nth UAV; max(·) is the maximum value function; |·| is the absolute value function; χ 1 , χ 2 and χ 3 are the weight factors of the mission constraint penalty term, attack constraint penalty term and range constraint penalty term respectively.

[0106] The technical effect of the present invention is further verified through experiments.

[0107] The experimental parameters are set as follows. Set the population size The proportion of flying fox suffocation p s = 0.1, maximum number of iterations T = 200, size of the survival list Where β = 0.3, the constant c in the distance exploration factor 1 and c 2 Take 0.2 and 0.6 respectively. This paper adopts the 0-1 planning model to solve the multi-UAV task allocation problem. The upper bound of the d-th dimension position in the search space is The Nether Assume that the total number of drones is N = 8, the total number of targets to be attacked is M = 10, the success probability of a drone attacking a target is P′, the probability of a drone surviving after the attack is P″, the value of each target is V′, and the two-dimensional coordinates of the drone base are P d , the two-dimensional coordinates of the target are P t , the maximum number of tasks performed by the drone is N max , the maximum range of the drone is S max The average speed of the drone is The parameters are as follows:

[0108]

[0109] V′=[1.1 2.0 1.5 1.3 1.4 1.3 1.5 1.5 1.7 1.6] 1×10

[0110]

[0111] N max =[2 3 3 2 3 2 4 5] 1×8

[0112] S max =[3 2 4 3 2.6 3.7 4.5 2.4] 1×8

[0113]

[0114] Table 1 is the multi-target UAV task allocation scheme finally obtained by the present invention. 1 ~U 8 Represents the drone number, T 1 ~T 10 represents the mission number. Objective functions 1, 2 and 3 represent the UAV attack benefit, UAV loss and mission completion time respectively. Objective function 1 is the minimum optimization target obtained by the inverse operation.

[0115] It can be seen from Table 1 that Table 1 shows three complete task allocation results. The three task allocation schemes are all transformed from the Pareto optimal solution. When considering the three optimization objectives at the same time, the task allocation scheme obtained by the present invention meets the requirements of multi-objective optimization solution and effectively solves the problem of multi-objective and multi-UAV task allocation.

[0116] Table 1

[0117]

[0118] like Figure 4As shown in the figure, the Pareto front diagrams obtained by different methods are given. The multi-objective quantum flying fox optimization (MOFFO) method proposed in the present invention is compared with other optimization methods, the multi-objective particle swarm optimization (MOPSO) method and the population diversity-based particle swarm optimization (PDPSO) method. The results show that the Pareto optimal solution front solution obtained by the MOFFO method proposed in the present invention has better convergence, more uniform distribution and better performance.

[0119] Although the present invention has been described according to a limited number of embodiments, it will be apparent to those skilled in the art, with the benefit of the above description, that other embodiments are contemplated within the scope of the invention thus described. The disclosure of the present invention is intended to be illustrative rather than restrictive of the scope of the invention, which is defined by the appended claims.

Claims

1. A multi-target UAV task allocation method based on quantum flying fox mechanism, characterized in that: The following steps are involved: Step 1: Establish a multi-UAV task allocation model, and then set the objective function of multi-objective UAV task allocation; Step 2: Taking the objective function as the fitness function, optimizing and solving the objective function using the quantum flying fox swarm algorithm to obtain the optimal task allocation matrix; including: Step 21: Initialize the quantum flying fox group; Step 22: Calculate the fitness function value of each quantum flying fox in the quantum flying fox group; Step 2 and 3: Use the fitness function values ​​of all quantum flying foxes to obtain the non-dominated levels and crowding distances of all quantum flying foxes; Step 24: Use the quantum flying fox death mechanism and the density of the quantum flying fox to update the quantum flying fox; Step 25: Use the population evolution strategy of the quantum flying fox group to update the quantum position of the quantum flying fox group; Step 26: Determine whether the upper limit of the number of iterations has been reached. If so, terminate the iteration and proceed to step 27; if not, return to step 24; Step 27: Convert the quantum position information of the quantum flying fox with the highest non-dominated level into a task allocation matrix.

2. According to a multi-target UAV task allocation method based on quantum flying fox mechanism according to claim 1, it is characterized in that: The objective function in step 1 is: In the formula, min means minimization; is the combat benefit fitness function; is the loss cost fitness function; is the time cost fitness function; Where N is the number of drones; M is the number of targets; A = {a n,m |a n,m ∈{0,1}} N×M represents the task allocation matrix, where n = 1, 2, ..., N, m = 1, 2, ..., M, a n,m =1 means that the nth UAV is responsible for carrying out the combat mission against the mth enemy target, a n,m = 0 means that the nth UAV is not on a combat mission for the mth enemy target; P n ' ,m represents the probability of drone n attacking target m; P n ' , ' m represents the survival probability of drone n after attacking target m; V m ′ represents the value of the mth target; The maximum number of tasks to be performed by the nth UAV; S n is the total flight distance of the nth UAV’s mission, is the maximum range of the nth UAV; represents the moving average speed of the nth UAV; max(·) is the maximum value function; |·| is the absolute value function; χ1, χ2 and χ3 are the weight factors of the mission constraint penalty term, attack constraint penalty term and range constraint penalty term, respectively.

3. A multi-target UAV task allocation method based on quantum flying fox mechanism according to claim 2, characterized in that: Step 21 of initializing the quantum flying fox group includes: Set the number of individuals in the quantum flying fox group to The maximum number of iterations is T. Each dimension of the quantum position of the quantum flying fox is randomly initialized to a random number between [0,1]. The quantum position of the ith quantum flying fox of the tth generation is recorded as where D represents the maximum dimension of the quantum position vector, which matches the target number, d=1,2,...D; the quantum position of the ith quantum flying fox of the tth generation corresponds to the position This position is mapped by Income, and are the maximum and minimum values ​​of the d-th dimension position in the search space respectively; the fitness vector of the ith quantum flying fox of the t-th generation is recorded as t is the number of iterations; initialize the survival list of the substitute quantum flying fox: randomly add the first generation of quantum flying foxes to the survival list, fill the survival list, and the size is where β is a constant.

4. A multi-target UAV task allocation method based on quantum flying fox mechanism according to claim 3, characterized in that: The process of step 22 includes: changing the quantum position of the ith quantum flying fox of the tth generation to Mapping to Location Then map the position of the quantum flying fox into a task allocation matrix Then the fitness function vector of the ith quantum flying fox of the tth generation is: in, t=1,2,...,T。 5. A multi-target UAV task allocation method based on quantum flying fox mechanism according to claim 4, characterized in that: The process of steps two and three includes: Calculate the non-dominated level: determine the dominance relationship between quantum flying foxes, the dominance relationship includes the non-dominated situation between quantum flying foxes and the non-dominated relationship between the two quantum flying foxes; the calculation process of the non-dominated situation between the quantum flying foxes includes: there are two quantum flying foxes in the tth generation, numbered i1 and i2 respectively, if the fitness values ​​of the two quantum flying foxes exist at the same time, the relationship includes and If the equal signs are not taken at the same time, then the quantum flying fox i2 is said to be non-dominated by the quantum flying fox i1, that is, the quantum flying fox i2 is inferior to the quantum flying fox i1; the calculation process of the two quantum flying foxes having no domination relationship includes: according to the non-dominated situation between the quantum flying foxes, the quantum flying foxes that are not dominated by any quantum flying fox are stored in the first level, and the non-dominated level is 1, and the quantum flying foxes that are dominated by the quantum flying foxes in the first level and not dominated by any other quantum flying foxes are stored in the second level, and the non-dominated level is 2, and the calculation is continued until the non-dominated level of all quantum flying foxes in the tth generation is obtained; Calculate the crowding distance by target fitness function: First, suppose there are μ quantum flying foxes in the first level of the tth generation of quantum flying foxes in the first level, and the fitness vector is Arrange the quantum flying foxes in the first level in ascending order according to the first fitness value, and then the crowding distance between the two boundaries after sorting and Set to infinity, the crowding distance of the first fitness function of the kth quantum flying fox is: Among them, k is the number of cells sorted in ascending order by the first fitness function. The corresponding unsorted number, is the maximum fitness value of the first fitness function, is the minimum fitness value of the first fitness function, Then, the crowding distance of the second and third fitness functions is calculated and Then the total crowding distance of the kth quantum flying fox is Finally, arrange the μ quantum flying foxes in descending order according to the crowding distance.

6. A multi-target UAV task allocation method based on quantum flying fox mechanism according to claim 5, characterized in that: The process of step 24 includes: Density calculation of quantum flying foxes: taking the quantum position of the quantum flying foxes and the target fitness function as input, firstly calculating the distance between the quantum flying foxes, then setting the distance threshold according to the distance between the quantum flying foxes and the average distance of the group, and finally obtaining the density according to the distance threshold and the distance between the quantum flying foxes; According to the density of the quantum flying foxes, some suffocated quantum flying foxes are selected and supplemented with quantum flying foxes in the survival list; the survival list is excellent individuals selected from the previous generation of quantum flying foxes, and the size is about β times the number of groups. After each iteration, the better quantum flying foxes in the current population are added to the survival list, and the redundant solutions in the list are deleted.

7. A multi-target UAV task allocation method based on quantum flying fox mechanism according to claim 6, characterized in that: The process of step 25 is: Crossover: Exclude quantum flying foxes with non-dominated level 1 and randomly select different quantum flying foxes for multi-point crossover. The number of crossover points is related to the dimension of the quantum flying fox position vector. The formula is in is the floor rounding function; Quantum flying fox update: The quantum rotation angle update formula of the i-th quantum flying fox is in is the quantum position of the jth dimension of the quantum flying fox labeled b randomly selected from the tth generation Pareto optimal solution set, α t is the distance exploration factor, according to the current individual fitness and the Pareto optimal solution fitness, α t The update amplitude will be automatically adjusted; α t The update formula is in is the kth fitness value of the quantum flying fox labeled b randomly selected from the tth generation Pareto optimal solution set, is the kth fitness value of the ith quantum flying fox, c1 and c2 are constants; the update formula for the jth dimension quantum position of the ith quantum flying fox is: in is the j-th dimensional quantum position of the i-th quantum flying fox before updating; The parent generation quantum flying fox population is The offspring population is Will and Synthesized into At this time, perform Pareto non-dominated sorting and calculate the crowding distance. Arrange in ascending order according to the non-dominated level. Arrange the quantum flying foxes of the same non-dominated level in descending order according to the crowding distance. Finally, select them from top to bottom. Only quantum flying foxes are placed into the new parent population middle.

8. A multi-target UAV task allocation system based on quantum flying fox mechanism, characterized in that: include: a model building module configured to establish a multi-UAV task allocation model and then set an objective function for multi-objective UAV task allocation; The optimal task allocation solution module is configured to use the objective function as a fitness function, optimize and solve the objective function using a quantum flying fox group algorithm, and obtain an optimal task allocation matrix; including: initializing a quantum flying fox group; calculating the fitness function value of each quantum flying fox in the quantum flying fox group; using the fitness function values ​​of all quantum flying foxes to obtain the non-dominated level and crowding distance of all quantum flying foxes; using the quantum flying fox death mechanism and the density of quantum flying foxes to update the quantum flying foxes; using the population evolution strategy of the quantum flying fox group to update the quantum position of the quantum flying fox group; judging whether the upper limit of the number of iterations is reached, if reached, terminating the iteration, and converting the quantum position information of the quantum flying fox with the highest non-dominated level into a task allocation matrix; if not reached, continuing to update.

9. A multi-target UAV task allocation system based on quantum flying fox mechanism according to claim 8, characterized in that: The objective function in the model building module is: In the formula, min means minimization; is the combat benefit fitness function; is the loss cost fitness function; is the time cost fitness function; Where N is the number of drones; M is the number of targets; A = {a n,m |a n,m ∈{0,1}} N×M represents the task allocation matrix, where n = 1, 2, ..., N, m = 1, 2, ..., M, a n,m =1 means that the nth UAV is responsible for carrying out the combat mission against the mth enemy target, a n,m = 0 means that the nth UAV is not carrying out combat missions against the mth enemy target; P′ n,m represents the probability that drone n attacks target m; P″ n,m represents the survival probability of drone n after attacking target m; V′ m represents the value of the mth target; The maximum number of tasks to be performed by the nth UAV; S n is the total flight distance of the nth UAV’s mission, is the maximum range of the nth UAV; represents the moving average speed of the nth UAV; max(·) is the maximum value function; |·| is the absolute value function; χ1, χ2 and χ3 are the weight factors of the mission constraint penalty term, attack constraint penalty term and range constraint penalty term, respectively.

Images