A multi-heterogeneous unmanned aerial vehicle system cooperative task allocation method based on a multi-objective evolutionary algorithm considering conditional probability

Abstract

The application discloses a multi-heterogeneous unmanned aerial vehicle system cooperative task allocation method based on a multi-objective evolutionary algorithm considering conditional probability, which comprises the following steps: firstly, according to the enemy target information, the current battlefield situation, the combat risk and the combat resources of the self side, a target function is set based on conditional probability, constraint conditions are given, and a multi-objective optimization model of the heterogeneous unmanned aerial vehicle cooperative task allocation is established; secondly, corresponding genetic operators are selected based on the actual combat situation, and an improved multi-objective optimization algorithm is used to solve the multi-objective optimization model of the cooperative task allocation; and thirdly, a decision maker selects a solution from a Pareto solution set according to the preference for the target function, and the corresponding task allocation scheme is taken as an execution scheme. The application is suitable for the multi-heterogeneous unmanned aerial vehicle cooperative task allocation under the condition of multiple combat resources, can provide a task allocation scheme which is suitable for the actual combat situation for the pre-war task allocation, and has a certain significance for the research on the multi-objective optimization problem of the multi-heterogeneous unmanned aerial vehicle cooperative task allocation.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) mission planning and relates to a collaborative task allocation method for multi-heterogeneous UAV systems based on conditional probability and considering a multi-objective evolutionary algorithm. Background Technology

[0002] In today's complex battlefield environment, multi-heterogeneous UAV collaborative operations are a crucial means of coping with complex battlefield situations. To fully leverage the advantages of multi-heterogeneous UAV collaborative operations, it is essential to formulate efficient operational plans that meet operational requirements by pre-battle collaborative task allocation based on the current battlefield situation and operational resources, while comprehensively considering various operational risks (UAV damage, mission failure, etc.). The heterogeneity of UAVs, the complexity of targets, and the increasing number of UAVs and targets all contribute to a more complex and larger-scale problem model. For complex task allocation problems, swarm intelligence algorithms, with their low dependence on models and ability to quickly solve large-scale problems, are more advantageous than traditional exact solution methods in solving collaborative task allocation problems. However, previous studies have simplified the model by ignoring potential risks in actual combat, such as UAV destruction by the enemy or mission failure, and have assumed sufficient operational resources. These assumptions lead to task allocation schemes that deviate significantly from reality. Furthermore, chromosome lock-in is a challenging problem to address when using evolutionary algorithms, and the efficiency of the unlocking method directly affects the optimization efficiency of the algorithm. To comprehensively improve operational effectiveness and maximize operational gains while minimizing operational costs is of significant research value in the current complex and risky situation. Optimizing these conflicting indicators requires considering multi-objective optimization problems. Therefore, conducting multi-objective optimization for collaborative mission planning of heterogeneous unmanned aerial vehicles (UAVs) under the premise of comprehensively considering various risks and operational resource conditions is of great importance. Summary of the Invention

[0003] To address the challenge of multi-objective optimization-based cooperative task allocation for heterogeneous unmanned aerial vehicle (UAV) systems while considering various risks, this invention proposes a conditional probability-based method incorporating a multi-objective evolutionary algorithm for cooperative task allocation. This method introduces conditional probability and utilizes a multi-objective evolutionary algorithm capable of simultaneously optimizing multiple conflicting objectives to construct a multi-objective UAV system task planning approach that considers multiple risks. Based on the characteristics of heterogeneous UAV cooperative task allocation, this method employs a chromosome encoding method and constructs a genetic operator, enabling the improved multi-objective evolutionary algorithm to solve cooperative task allocation problems under various resource conditions. Furthermore, for complex chromosome-locked situations, a logical unlocking method is provided that maintains population randomness while rapidly unlocking at chromosome-locked points. This multi-objective optimization-based cooperative task allocation method for heterogeneous UAV systems, considering multiple risks, better reflects actual combat scenarios and is applicable to cooperative task allocation under various resource conditions. Moreover, the logical unlocking method significantly improves allocation efficiency.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A method for cooperative task allocation in heterogeneous unmanned aerial vehicle (UAV) systems based on conditional probability and incorporating a multi-objective evolutionary algorithm is proposed. First, based on reconnaissance of enemy targets, the current battlefield situation, operational risks, and available operational resources, an objective function and constraints are set according to conditional probability to establish a multi-objective optimization model for cooperative task allocation among heterogeneous UAVs. Then, the constructed multi-objective evolutionary algorithm is used to solve the model to obtain a Pareto solution set. Finally, a solution selection method is used to select a solution from the Pareto solution set, and its corresponding task allocation scheme is used as the execution scheme. The computational flowchart of this invention is shown below. Figure 1 As shown, it includes the following steps:

[0006] Step 1: Based on the detected enemy target information, the current battlefield situation, operational risks, and our own operational resources, a multi-objective optimization model for heterogeneous UAV collaborative task allocation is established by setting an objective function based on conditional probability and providing constraints, as follows:

[0007] Step 1-1: Collect basic data on enemy targets and our drones in the mission plan.

[0008] For ease of description, let The detected enemy targets include N T There are several objectives, with the j-th objective denoted as T. j The set of targets is denoted as Each objective comprises three sub-tasks: classification, attack, and evaluation, denoted as C, A, and V, respectively. Different enemy objectives possess different values ​​and risks; typically, high-value objectives are accompanied by high-risk objectives. Set objective T. j Value recorded as Let high-value targets be denoted as Hv, and low-value targets as Lv. To improve the success rate of operations, we will set up attacks on high-value targets twice and attacks on low-value targets once. Let A... i This represents the i-th attack mission, with the target... The set of tasks is denoted as

[0009]

[0010] set up and These represent the drone's interaction with target T. j The start execution times of subtasks C, A1, A2, and V. and These represent the drone's interaction with target T. j The execution time of subtasks C, A1, A2, and V.

[0011] The drone swarm is configured to contain N U There are 10 drones, and the i-th drone is denoted as U. i The collection of drones is denoted as The fleet is configured to include three heterogeneous UAV types: reconnaissance aircraft, fighter jets, and munitions aircraft, denoted as s, c, and m respectively. Type s UAVs can perform mission types C and V, type c UAVs can perform mission types C, A, and V, and type m UAVs can perform mission type A. Let UAV U... i Value recorded as Different drones have different success rates and survival rates when performing different tasks on different targets. (Settings) U-shaped drone i Execution target T j The success rate of task type k U-shaped drone i Execution target T j The mortality rate for task type k, where, k∈I3, k=1,2,3 represent task types C, A, and V, respectively. Based on the capabilities of the UAV, it can be known that when the UAV U... i When the type is s, there is When the drone U i When the type is m, there is Command the U drone i The set of tasks is denoted as in Indicates assignment to drone Ui The number of tasks, U-shaped drone i The i-th task in the task set that is executed.

[0012] Step 1-2: Construct the objective function of the model

[0013] For target T j Successful execution of target T means that all its subtasks are successfully executed. If any subtask fails to execute, then target T... j The mission was not successfully executed. There are two risks associated with drones performing missions: the drone may be destroyed, and the mission may not be successfully completed. To maximize operational gains with minimal operational costs, we consider maximizing the expected value of successfully executed objectives and minimizing the expected value of destroyed drones as objective functions. These two objective functions are constructed by introducing conditional probabilities. Let... U-shaped drone i From the target Flying towards target T j Execute T j The task type k takes a value of 1 to indicate execution, otherwise it is not executed. Where V... j Indicates target T j The location.

[0014] (i) Maximize the expected value of successfully executed objectives

[0015]

[0016] (ii) Minimize the expected value of destroyed drones.

[0017]

[0018] in, express The first task in the middle, make Based on the actual situation, there are

[0019]

[0020]

[0021] and

[0022]

[0023] The objective (i) is reduced to a minimum form as follows:

[0024] (iii) Minimize the expected value of objectives that are not successfully executed.

[0025]

[0026] Let f = (f1, f2) T , where f1 = J3, f2 = J2.

[0027] Steps 1-3: Constructing the constraints of the model

[0028] Based on the actual situation of collaborative task allocation and the above analysis, the following constraints are considered.

[0029] The payload capacity of drones is limited. Therefore, in actual combat mission allocation, the number of missions of type A assigned to each drone cannot exceed its payload capacity. Drones capable of performing mission types C and V are not subject to any limit on the number of missions of types C and V assigned to them.

[0030]

[0031] Among them, AM i U-shaped drone i The payload capacity. If the drone type is 's', then its payload capacity is 0.

[0032] To shorten the time the same drone stays on the same target and reduce the risk of the drone being detected and destroyed, we assume that the same drone attacks the same target no more than once.

[0033]

[0034] Therefore, different attack missions targeting high-value targets are assigned to different drones.

[0035] The tasks for each objective must be executed sequentially, meaning they must satisfy timing constraints. For tasks with the same objective, task type A can only be executed after task type C has been completed, and task type V can only be executed after all tasks of task type A have been completed. Therefore, task allocation must satisfy the following constraints.

[0036]

[0037] Where, if target T j For high-value goals, then otherwise

[0038] Each drone, when executing tasks within its task set, must do so in the order in which the tasks were assigned within the set. It can only be executed after its previous task has been completed.

[0039]

[0040] in, express

[0041] Steps 1-4: Construct a multi-objective optimization model for collaborative task allocation among heterogeneous UAVs.

[0042] Based on the objective function and constraints constructed above, the multi-objective optimization of multi-heterogeneous UAV collaborative task allocation is to minimize the vector function f while satisfying constraints (4)-(7).

[0043] min f=(f1,f2) T

[0044] st(4)-(7)

[0045] Step 2: Solve the multi-objective optimization model for cooperative task assignment using the constructed multi-objective evolutionary algorithm to obtain its Pareto solution set, as follows:

[0046] Step 2-1: Set the parameters of the improved multi-objective optimization algorithm

[0047] Define the population size S, the maximum number of iterations G, and the crossover probability P. c and the probability of mutation P m The value of .

[0048] Step 2-2: Initialize the population

[0049] The chromosome is encoded using genes, where each gene represents the assignment of a specific task. The information in each gene includes the target ID corresponding to the assigned task, the task type, and the drone ID executing the task. Genes corresponding to the same drone form a sub-chromosome, and all sub-chromosomes constitute the complete chromosome. A population P of size S is randomly generated. g (g=0). The following is the initialization process for generating an initial population containing S chromosomes according to the above encoding method.

[0050] Step 2-2-1: Input the parameters of the drone and the enemy target, and the population size.

[0051] Input the ammunition quantity A of all drones as an array. m Let TS be the set of enemy target IDs, AU be the set of UAV IDs of type c and m, and CVU be the set of UAV IDs of type s and c. Set the number of chromosomes in the population to 0.

[0052] Step 2-2-2: Determine the number of chromosomes in the current population.

[0053] If the number of chromosomes in the population equals S, then population initialization is complete, and the current initialization process exits. Otherwise, proceed to steps 2-2-3 to 2-2-4.

[0054] Step 2-2-3: Initialize an empty chromosome, denoted as Chromosome.

[0055] Step 2-2-4: Encoding the Chromosome

[0056] Step 2-2-4-1: Randomly select a target from TS, denoted as

[0057] Step 2-2-4-2: Select the target Subtask C and attack task A1 are assigned.

[0058] Initialize genes Gene1, Gene2, Gene3, and Gene4. First, randomly select a drone from the CVU, denoted as Gene1, Gene2, Gene3, and Gene4. make Chromosome = Chromosome∪Gene1. Then, randomly select a drone from the AU, denoted as... make Chromosome = Chromosome∪Gene2, and update A. m , that is to say

[0059] Step 2-2-4-3: If For high-value goals and A m If ≠0, then for the target Assign attack task A2; otherwise, proceed to step 2-2-4-4.

[0060] Under the premise of satisfying constraint (5), a drone is randomly selected from the AU, denoted as . make Chromosome = Chromosome ∪ Gene3. Update A m , that is to say

[0061] Step 2-2-4-4: Select the target Assign subtasks V

[0062] Randomly select a drone from the CVUs, denoted as . Remove the target from TS make Chromosome=Chromosome∪Gene4.

[0063] Step 2-2-4-5: Determine whether the current chromosome has completed encoding.

[0064] If the length of array TS is not 0, and A m If the vector is not zero, proceed to steps 2-2-4-1 to 2-2-4-5; otherwise, increment the number of chromosomes in the population by 1, exit the encoding process of the current chromosome, and proceed to step 2-2-2.

[0065] Steps 2-3: Calculate the population P based on objective functions (2) and (3). g fitness value of chromosome

[0066] Steps 2-4: Generate a subpopulation of size S

[0067] Step 2-4-1: Use the roulette wheel algorithm to select from population P g Select two cross-parents F1 and F2

[0068] Step 2-4-2: Based on the total resources and total task volume, select the appropriate crossover operator and perform the crossover operation according to the crossover probability.

[0069] Let L represent the number of high-value targets. Let RM i Represents parent generation F i The corresponding combination of drones with remaining ammunition, RT i Represents parent generation F i The set of unassigned enemy targets. Let p1 and p2 represent intersections randomly selected from the intersection parent. and Let p1 and p2 represent the sets of drones performing mission types C and V between the intersection points p1 and p2 of F1 and F2, respectively. and Let PT1 and PT2 represent the sets of UAVs performing task type A between the intersection points p1 and p2 of F1 and F2, respectively. Let PT1 and PT2 represent the sets of targets assigned in F1 and F2, respectively.

[0070] The crossover process is given using parent generation F1 as an example. If... That is, there is sufficient ammunition and some remaining. At this point, the crossover process is from step 2-4-2-1 to step 2-4-2-4.

[0071] Step 2-4-2-1: Input the information of the intersection point and the intersection parent, let i = p1-1

[0072] Step 2-4-2-2: Determine if the crossover process has ended.

[0073] If i satisfies i≤p2-1, then execute steps 2-4-2-3 to 2-4-2-4; otherwise, exit the crossover operation on parent generation F1.

[0074] Step 2-4-2-3: Perform a crossover operation on the gene at crossover point i.

[0075] When the task type in the gene at crossover point i is A, the following operation is performed. If and If removing the target gene at intersection i results in a non-empty set, then from... Randomly select a drone from the current gene; otherwise, randomly select a drone from the set of genes containing the target at intersection i in RM1. Replace the drone information in the current gene with the information of the selected drone, and update RM1.

[0076] The following operation is performed when the task type in the gene at crossover point i is C or V. If So randomly from One drone is randomly selected from the list; otherwise, it never belongs to any other drone. Randomly select one drone from those of type s or c. Replace the drone information in the current gene with the information from this selected drone.

[0077] Step 2-4-2-4: Let i = i + 1, then go to step 2-4-2-2

[0078] like That is, there is a sufficient amount of ammunition with no surplus. At this point, the crossover process is from step 2-4-2-5 to step 2-4-2-8.

[0079] Step 2-4-2-5: Input the information of the intersection point and the intersection parent, let i = p1-1

[0080] Step 2-4-2-6: Determine if the crossover process has ended.

[0081] If i satisfies i≤p2-1, then execute steps 2-4-2-7 to 2-4-2-8; otherwise, exit the crossover operation on parent generation F1.

[0082] Step 2-4-2-7: Perform a crossover operation on the gene at crossover point i.

[0083] Let the target genes corresponding to the crossover point i in the crossover parents F1 and F2 be denoted as ... like Then proceed to step 2-4-2-8; otherwise, execute the following process.

[0084] like If all genes are low-value or all are high-value, then the corresponding genes can be directly interchanged in F1. Information. If For high-value goals, If the target is of low value, then first swap the target numbers in the corresponding genes, and then randomly select one. The attack gene was modified, and the target number of that gene was changed. like For low-value targets, For high-value targets, first swap the target numbers in the corresponding genes, then randomly select one. The attack gene was modified, and the target number of that gene was changed.

[0085] Step 2-4-2-8: Let l be the number of genes updating information between crossover p1 and p2 in F1, and let i = i + l, then go to step 2-4-2-6.

[0086] like That is, there is insufficient ammunition. The crossover process at this point is from step 2-4-2-9 to step 2-4-2-13.

[0087] Step 2-4-2-9: Input the information of the intersection point and the intersection parent, let i = p1-1

[0088] Step 2-4-2-10: Determine if the crossover process has ended.

[0089] If i satisfies i≤p2-1, then execute steps 2-4-2-11 to 2-4-2-13; otherwise, exit the crossover operation on parent generation F1.

[0090] Step 2-4-2-11: Perform a crossover operation on the gene at crossover point i.

[0091] Let the target in the gene at the crossover point i be denoted as like Then, randomly select a target from RT1∩PT2; otherwise, randomly select a target from RT1. Let the selected target be denoted as .

[0092] when When the goal is low value, if If it is a low-value goal, then set the goal as follows: The target replacement in all corresponding genes is Otherwise, randomly select a low-value target from F1. target and The target replacement in all corresponding genes is

[0093] when When it is a high-value goal, if If the target is low-value, then first randomly select a target from RT1. target The target replacement in all corresponding genes is Then randomly select a target The attack gene, and replace the target of that gene with Finally, add a goal The genes for task types C and V; if For high-value goals, set the goals The target replacement in all corresponding genes is

[0094] Step 2-4-2-12: Update RT1 and PT1

[0095] Step 2-4-2-13: Let l be the number of genes updating information between crossover p1 and p2 in F1, and let i = i + l, then go to step 2-4-2-10.

[0096] Step 2-4-3: Perform mutation operations on the crossover offspring based on the mutation probability.

[0097] Let the two crossover offspring obtained from the crossover operation be denoted as O1 and O2. Taking O1 as an example, the mutation process is given. Let q1 and q2 be the start and end points of the gene segments to be mutated. Let the set of low-value targets in O1 be denoted as CT, and the set of C and V genes in O1 be denoted as G. cv The set of genes A in O1 is denoted as G. a O1 represents the set of drones with remaining ammunition and the set of unassigned targets, respectively denoted as... and

[0098] Step 2-4-3-1: Input the mutation point and O1, let i = q1-1

[0099] Step 2-4-3-2: Determine if the mutation process has ended.

[0100] If i satisfies i≤q2-1, then execute steps 2-4-3-3 to 2-4-3-4; otherwise, exit the mutation operation on O1.

[0101] Step 2-4-3-3: Perform mutation operation on the gene at mutation point i.

[0102] when If the task type in the gene at mutation point i of O1 is C or V, then remove the drone from the gene at mutation point i of O1 in the CVU, and then randomly select a drone from the CVU; if the task type in the gene at mutation point i of O1 is A, then remove the drone from the CVU. Remove the drone from the gene at mutation point i of O1, and then from... Randomly select a drone and update Replace the drone information in the gene at mutation point i of O1 with the information of the selected drone.

[0103] when Firstly, from Randomly select a target, denoted as T. m Then find all genes whose target is the same as the target of the gene at mutation point i of O1, and replace its target with T. m If T m If the target is a high-value target, while the target in the gene at mutation point i of O1 is a low-value target, then remove the target in the gene at mutation point i of O1 from CT. Then, randomly select a target from CT and delete all its genetic information. Finally, allocate the ammunition to T. m If T m If the target is low-value, while the target in the gene at mutation point i of O1 is high-value, then randomly delete T. m A gene for an attack mission.

[0104] when If the task type in the gene at mutation point i in O1 is C or V, then from G cv Remove the gene at mutation point i of O1 from G, and then randomly select from G. cv Choose one gene from G; otherwise, choose from G. a Remove the gene at mutation point i of O1 from G, and then randomly select from G. a Select one gene. Exchange the drone information of the selected gene with the gene at mutation point i of O1.

[0105] Step 2-4-3-4: Let l be the number of genes updating information between mutation points q1 and q2 of O1, and let i = i + l, then go to step 2-4-3-2.

[0106] Step 2-4-4: Repeat steps 2-4-1 to 2-4-3 until a subpopulation of size S is obtained.

[0107] Step 2-5: Determine whether the chromosomes in the subpopulation obtained in Step 2-4 are locked, and unlock the locked chromosomes.

[0108] The following describes the process for determining and unlocking a chromosome lockout.

[0109] Step 2-5-1: Transform the chromosome into a form composed of daughter chromosomes.

[0110] The genes in the chromosome are arranged in ascending order according to the drone number, and the tasks and execution order of each drone are obtained based on the sub-chromosomes.

[0111] Step 2-5-2: Establish a set representing completed tasks, denoted as CS.

[0112] Step 2-5-3: If the dimension of CS is equal to the dimension of chromosome, terminate the unlocking process; otherwise, proceed to steps 2-5-4 to 2-5-6.

[0113] Step 2-5-4: Determine whether the task in the current gene of each subchromosome can be executed.

[0114] Each drone starts execution from the first task in the task set. If the task type is C, it is executed directly, the task is removed from the task set, and added to the task list (CS). If the task type is A, it is checked whether the target's task type C is included in the CS. If it is, it is executed directly, the task is removed from the task set, and added to the CS. Otherwise, the task cannot be executed, and the drone is in a waiting state. If the task type is V, it is checked whether the target's task type C and all of the target's attack tasks are included in the CS. If they are, it is executed directly, the task is removed from the task set, and added to the CS. Otherwise, the task cannot be executed, and the drone is in a waiting state.

[0115] Step 2-5-5: Determine if chromosomes are locked.

[0116] Based on step 2-5-4, determine the current status of all drones. If all drones are in a waiting state, then the chromosome is locked.

[0117] Steps 2-5-6: Unlock at the locked point.

[0118] If the drone at the lock point is waiting to execute a task of type A, then randomly select a task of type C from the drone's remaining tasks and swap its order with the current task. Perform this operation on all drones in a waiting state. Proceed to step 2-5-3.

[0119] Steps 2-6: Combine population P g and A population Q of size 2S is obtained. g

[0120] Steps 2-7: Calculate the population Q based on objective functions (2) and (3). g fitness value of chromosome

[0121] Steps 2-8: Based on the non-dominated quicksort method and the elite preservation strategy, extract data from the population Q... g Select S chromosomes to form the parent population P for the next iteration. q

[0122] Steps 2-9: Let g = g + 1, q = g

[0123] Step 2-10: If g < G, go to step 2-3; otherwise, output the Pareto solution set.

[0124] Step 3: Select a solution from the Pareto solution set using the solution selection method, and use its corresponding task allocation scheme as the execution scheme.

[0125] make This represents the number of non-dominated solutions on the Pareto optimal front. Let represent the i-th non-dominated solution. First, calculate... Here, α1 and α2 are weights, and α1 + α2 = 1. The decision-maker assigns values ​​to these weights based on their preference for the objective function. Then, the calculated S... i Sort the results. Finally, select the smallest S. i The task allocation scheme for the non-dominated solution corresponding to the value.

[0126] The beneficial effects of this invention are as follows:

[0127] This invention provides a method for coordinating mission allocation among multiple heterogeneous UAV systems based on currently detected enemy target information and various friendly resource information. This method comprehensively considers various risks present in actual combat, such as UAV destruction or mission failure. Based on this, the method can collaboratively allocate missions to multiple heterogeneous UAV systems according to different combat resource conditions, simultaneously optimizing multiple conflicting objective functions, enabling UAV swarms to achieve more efficient and realistic mission execution schemes. Considering multiple risks and various combat resource conditions makes the allocation method more comprehensive. Furthermore, the logical unlocking method constructed for chromosome lock-in cases significantly improves the algorithm's optimization efficiency. Attached Figure Description

[0128] Figure 1 This is a flowchart of the calculation process of the present invention.

[0129] Figure 2 This is the optimal Pareto front end obtained through optimization in this embodiment of the invention.

[0130] Figure 3This shows how the optimal value of the objective function J3 changes with the number of iterations in each iteration in this embodiment of the invention.

[0131] Figure 4 This shows how the optimal value of the objective function J2 changes with the number of iterations in each iteration in this embodiment of the invention.

[0132] Figure 5 This represents the CPU runtime for 15 task allocation experiments conducted using the collaborative task allocation method in this embodiment of the invention.

[0133] Figure 6 This refers to the CPU runtime for 15 unlocking experiments using the constructed logic unlocking method in this embodiment of the invention. Detailed Implementation

[0134] The present invention will be further described below with reference to specific embodiments.

[0135] Assuming sufficient ammunition, five heterogeneous drones will conduct missions against five targets. The drones will... i The set of tasks is denoted as UAV i Value recorded as UAV i The payload is denoted as AM i Let high-value targets be denoted as Hv, low-value targets as Lv, and target T. j Value recorded as The set of tasks for the target is denoted as Command the U drone i Execution target T j The success rate of task type k is denoted as UAV i Execution target T j The mortality rate of task type k is denoted as

[0136] This invention provides a task allocation method for multi-heterogeneous unmanned aerial vehicle (UAV) systems based on conditional probability and considering a multi-objective evolutionary algorithm, comprising the following steps:

[0137] Step 1: Based on the detected enemy target information, the current battlefield situation, operational risks, and our own operational resources, establish a multi-objective optimization model for heterogeneous UAV collaborative task allocation by setting an objective function based on conditional probability and providing constraints.

[0138] Step 1-1: Collect basic data on enemy targets and our drones in the mission plan.

[0139] Collect the types of drones, AM i Types of targets and The information is shown in Table 1-3.

[0140] Table 1 Information on UAVs

[0141]

[0142] Table 2 Target Information

[0143]

[0144] Table 3 Success rate and destruction rate of UAVs during mission execution.

[0145]

[0146] Step 1-2: Construct the objective function of the model

[0147] make U-shaped drone i From the target Flying towards target T j Execute T j The task type k takes a value of 1 to indicate execution, otherwise it is not executed. Where V... j Indicates target T j The position. To obtain the maximum operational benefit with the minimum operational cost, the following objective function is constructed.

[0148] (i) Maximize the expected value of successfully executed objectives

[0149]

[0150] (ii) Minimize the expected value of destroyed drones.

[0151]

[0152] in, M represents Ui The first task in the middle, make Based on the actual situation, there are

[0153]

[0154]

[0155] and

[0156]

[0157] The objective (i) is reduced to a minimum form as follows:

[0158] (iii) Minimize the expected value of objectives that are not successfully executed.

[0159]

[0160] Let f = (f1, f2) T , where f1 = J3, f2 = J2.

[0161] Steps 1-3: Constructing the constraints of the model

[0162] Based on the characteristics of collaborative task allocation and the above analysis, the following constraints are considered.

[0163] The payload capacity of drones is limited. Therefore, in actual combat mission allocation, the number of missions of type A assigned to each drone cannot exceed its payload capacity. Drones capable of performing mission types C and V are not subject to any limit on the number of missions of types C and V assigned to them.

[0164]

[0165] Among them, AM i U-shaped drone i The payload capacity. If the drone type is 's', then its payload capacity is 0.

[0166] To shorten the time the same drone stays on the same target and reduce the risk of the drone being detected and destroyed, we assume that the same drone attacks the same target no more than once.

[0167]

[0168] Therefore, different attack missions targeting high-value targets are assigned to different drones.

[0169] The tasks for each objective must be executed sequentially, meaning they must satisfy timing constraints. For tasks with the same objective, task type A can only be executed after task type C is completed, and task type V can only be executed after task type A is completed. Therefore, task allocation must satisfy the following constraints.

[0170]

[0171] in,

[0172] Each drone, when executing tasks within its task set, must do so in the order in which the tasks were assigned within the set. It can only be executed after its previous task has been completed.

[0173]

[0174] in, express

[0175] Steps 1-4: Construct a multi-objective optimization model for collaborative task allocation among heterogeneous UAVs.

[0176] Based on the objective function and constraints constructed above, the multi-objective optimization of multi-heterogeneous UAV collaborative task allocation is to minimize the vector function f while satisfying constraints (11)-(14).

[0177] min f=(f1,f2) T

[0178] st(11)-(14)

[0179] Step 2: Solve the multi-objective optimization model of cooperative task assignment using the constructed multi-objective evolutionary algorithm to obtain its Pareto solution set.

[0180] Step 2-1: Set the parameters of the improved multi-objective optimization algorithm

[0181] Set the population size S = 100, the maximum number of iterations G = 200, and the crossover probability P. c =0.8 and the probability of mutation P m =0.2.

[0182] Step 2-2: Initialize the population

[0183] The chromosome is encoded using genes, where each gene represents the assignment of a specific task. The information in the gene includes the target number corresponding to the assigned task, the task type, and the ID of the drone executing the task. A population P of size 100 is randomly generated. g (g=0). The following describes the initialization process for generating an initial population containing 100 chromosomes according to the above encoding method.

[0184] Step 2-2-1: Input the parameters of the drone and the enemy target.

[0185] Input the ammunition quantity A of all drones as an array. m Let TS be the set of enemy target IDs, AU be the set of UAV IDs of type c and m, and CVU be the set of UAV IDs of type s and c. Set the number of chromosomes in the population to 0.

[0186] Step 2-2-2: Determine the number of chromosomes in the current population.

[0187] If the number of chromosomes in the population is 100, then population initialization is complete, and the current initialization process exits. Otherwise, proceed to steps 2-2-3 to 2-2-4.

[0188] Step 2-2-3: Initialize an empty chromosome, denoted as Chromosome.

[0189] Step 2-2-4: Encoding the Chromosome

[0190] Step 2-2-4-1: Randomly select a target from TS, denoted as

[0191] Step 2-2-4-2: Select the target Subtask C and attack task A1 are assigned.

[0192] Initialize genes Gene1, Gene2, Gene3, and Gene4. First, randomly select a drone from the CVU, denoted as Gene1, Gene2, Gene3, and Gene4. make Chromosome = Chromosome∪Gene1. Then, randomly select a drone from the AU, denoted as... make Chromosome = Chromosome∪Gene2, and update A. m , that is to say

[0193] Step 2-2-4-3: If For high-value goals and A m If ≠0, then for the target The attack task A2 is assigned; otherwise, proceed to step 2-2-4-4. Under the premise of satisfying constraint (12), a drone is randomly selected from AU, denoted as . make Chromosome = Chromosome ∪ Gene3. Update A m , that is to say

[0194] Step 2-2-4-4: Select the target Assign subtasks V

[0195] Randomly select a drone from the CVUs, denoted as . Remove the target from TS make Chromosome=Chromosome∪Gene4.

[0196] Step 2-2-4-5: Determine whether the current chromosome has completed encoding.

[0197] If the length of array TS is not 0, and A mIf the vector is not zero, proceed to steps 2-2-4-1 to 2-2-4-5; otherwise, increment the number of chromosomes in the population by 1, exit the encoding process of the current chromosome, and proceed to step 2-2-2.

[0198] Steps 2-3: Calculate the population P based on objective functions (10) and (9). g fitness value of chromosome

[0199] Steps 2-4: Generate a subpopulation of size 100

[0200] Step 2-4-1: Use the roulette wheel algorithm to select from population P g Select two cross-parents F1 and F2

[0201] Step 2-4-2: Based on the total resources and total task volume, select the appropriate crossover operator and perform the crossover operation with a crossover probability of 0.8.

[0202] RM i Represents parent generation F i The corresponding combination of drones with remaining ammunition, RT i Represents parent generation F i The set of unassigned enemy targets. Let p1 and p2 represent intersections randomly selected from the intersection parent. and Let p1 and p2 represent the sets of drones performing mission types C and V between the intersection points p1 and p2 of F1 and F2, respectively. and Let PT1 and PT2 represent the sets of UAVs performing task type A between the intersection points p1 and p2 of F1 and F2, respectively. Let PT1 and PT2 represent the sets of targets assigned in F1 and F2, respectively.

[0203] The crossover process is given using the parent generation F1 as an example. That is, there is sufficient ammunition and some remaining. At this point, the crossover process is from step 2-4-2-1 to step 2-4-2-5.

[0204] Step 2-4-2-1: Input the information of the intersection point and the intersection parent, let i = p1-1

[0205] Step 2-4-2-2: Determine if the crossover process has ended.

[0206] If i satisfies i≤p2-1, then execute steps 2-4-2-3 to 2-4-2-4; otherwise, exit the crossover operation on parent generation F1.

[0207] Step 2-4-2-3: Perform a crossover operation on the gene at crossover point i.

[0208] When the task type in the gene at crossover point i is A, the following operation is performed. If and If removing the target gene at intersection i results in a non-empty set, then from... Randomly select a drone from the current gene; otherwise, randomly select a drone from the set of genes containing the target at intersection i in RM1. Replace the drone information in the current gene with the information of the selected drone, and update RM1.

[0209] The following operation is performed when the task type in the gene at crossover point i is C or V. If So randomly from One drone is randomly selected from the list; otherwise, it never belongs to any other drone. Randomly select one drone from those of type s or c. Replace the drone information in the current gene with the information from this selected drone.

[0210] Step 2-4-2-4: Let i = i + 1, then go to step 2-4-2-2

[0211] Step 2-4-3: Perform mutation operation on the crossover offspring based on a mutation probability of 0.2.

[0212] Let the two crossover offspring obtained from the crossover operation be denoted as O1 and O2. Taking O1 as an example, the mutation process is given. Let q1 and q2 be the start and end points of the gene segments to be mutated. Let the set of low-value targets in O1 be denoted as CT, and the set of C and V genes in O1 be denoted as G. cv The set of genes A in O1 is denoted as G. a O1 represents the set of drones with remaining ammunition and the set of unassigned targets, respectively denoted as... and

[0213] Step 2-4-3-1: Input the mutation point and O1, let i = q1-1

[0214] Step 2-4-3-2: Determine if the mutation process has ended.

[0215] If i satisfies i≤q2-1, then execute steps 2-4-3-3 to 2-4-3-4; otherwise, exit the mutation operation on O1.

[0216] Step 2-4-3-3: Perform mutation operation on the gene at mutation point i.

[0217] If the task type in the gene at mutation point i of O1 is C or V, then remove the drone from the gene at mutation point i of O1 in the CVU, and then randomly select a drone from the CVU; if the task type in the gene at mutation point i of O1 is A, then remove the drone from the CVU. Remove the drone from the gene at mutation point i of O1, and then from... Randomly select a drone and update Replace the drone information in the gene at mutation point i of O1 with the information of the selected drone.

[0218] Step 2-4-3-4: Let l be the number of genes updating information between mutation points q1 and q2 of O1, and let i = i + l, then go to step 2-4-3-2.

[0219] Step 2-4-4: Repeat steps 2-4-1 to 2-4-3 until a subpopulation of size 100 is obtained.

[0220] Step 2-5: Determine whether the chromosomes in the subpopulation obtained in Step 2-4 are locked, and unlock the locked chromosomes.

[0221] The following describes the process for determining and unlocking a chromosome lockout.

[0222] Step 2-5-1: Transform the chromosome into a form composed of daughter chromosomes.

[0223] The genes in the chromosome are arranged in ascending order according to the drone number, and the tasks and execution order of each drone are obtained based on the sub-chromosomes.

[0224] Step 2-5-2: Construct an empty set of completed tasks, denoted as CS.

[0225] Step 2-5-3: If the dimension of CS is equal to the dimension of chromosome, terminate the unlocking process; otherwise, proceed to steps 2-5-4 to 2-5-6.

[0226] Step 2-5-4: Determine whether the task in the current gene of each subchromosome can be executed.

[0227] Each drone starts execution from the first task in the task set. If the task type is C, it is executed directly, the task is removed from the task set, and added to the task list (CS). If the task type is A, it is checked whether the target's task type C is included in the CS. If it is, it is executed directly, the task is removed from the task set, and added to the CS. Otherwise, the task cannot be executed, and the drone is in a waiting state. If the task type is V, it is checked whether the target's task type C and all of the target's attack tasks are included in the CS. If they are, it is executed directly, the task is removed from the task set, and added to the CS. Otherwise, the task cannot be executed, and the drone is in a waiting state.

[0228] Step 2-5-5: Determine if chromosomes are locked.

[0229] Based on step 2-5-4, determine the current status of all drones. If all drones are in a waiting state, then the chromosome is locked.

[0230] Steps 2-5-6: Unlock at the locked point.

[0231] If the drone at the lock point is waiting to execute a task of type A, then randomly select a task of type C from the drone's remaining tasks and swap its order with the current task. Perform this operation on all drones in a waiting state. Proceed to step 2-5-3.

[0232] Steps 2-6: Combine population P g and A population Q of size 200 was obtained. g

[0233] Steps 2-7: Calculate the population Q based on objective functions (10) and (9). g fitness value of chromosome

[0234] Steps 2-8: Based on the non-dominated quicksort method and the elite preservation strategy, extract data from the population Q... g Select 100 chromosomes to form the parent population P for the next iteration. q .

[0235] Steps 2-9: Let g = g + 1, q = g

[0236] Step 2-10: If g < 200, go to step 2-3; otherwise, output the Pareto solution set.

[0237] Step 3: Select a solution from the Pareto solution set using the solution selection method, and use its corresponding task allocation scheme as the execution scheme.

[0238] make This represents the number of non-dominated solutions on the Pareto optimal front. Let represent the i-th non-dominated solution. First, calculate... The weights are set to α1 = 0.5 and α2 = 0.5. Then, the calculated S... i Sort the results. Finally, select the smallest S. i The task allocation scheme for the non-dominated solution corresponding to the value.

[0239] The optimized Pareto front end is as follows: Figure 2 As shown, Figure 3 and Figure 4The optimal values ​​of the two objective functions change with the number of iterations in each iteration. Based on the solution selection strategy, the 15th non-dominated solution on the Pareto front is selected as the chosen solution, and its corresponding solution information and specific task allocation scheme are shown in Table 4. To test the computational efficiency of the method, 15 experiments were conducted, and the CPU runtime is as follows: Figure 5 As shown.

[0240] Table 4. Information on the 15th nondominated solution on the Pareto optimal front and the corresponding task assignment scheme.

[0241]

[0242] To verify the effectiveness of the constructed logical unlocking method, 100 chromosomes were randomly generated without considering timing constraints. The constructed logical unlocking method was used to perform 15 experiments on these 100 chromosomes, determining whether they were locked and unlocking them. The CPU runtime for unlocking is as follows: Figure 6 As shown in the figure, the logical unlocking method can efficiently resolve deadlock situations.

[0243] This invention proposes a cooperative task allocation method for multi-heterogeneous unmanned aerial vehicle (UAV) systems based on conditional probability and incorporating a multi-objective evolutionary algorithm. This method enables cooperative task allocation for heterogeneous UAV swarms under various combat resource situations, taking into account both the risk of UAVs being destroyed and the risk of missions not being completed, and can simultaneously optimize multiple conflicting objectives. Furthermore, the logical unlocking mechanism constructed in this method significantly improves the solution efficiency. This task allocation method for multi-heterogeneous UAV systems, considering operational risks and various combat scenarios, provides a more realistic task allocation scheme for pre-battle task allocation, effectively improving the solution efficiency of task allocation.

[0244] The above-described embodiments are merely illustrative of the implementation methods of the present invention, but should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the protection scope of the present invention.

Claims

1. A task allocation method for multi-heterogeneous unmanned aerial vehicle (UAV) systems based on conditional probability and incorporating a multi-objective evolutionary algorithm, characterized in that, Includes the following steps: First, based on the detected enemy target information, the current battlefield situation, operational risks, and our own operational resources, a multi-objective optimization model for heterogeneous UAV collaborative task allocation is established by setting objective functions and constraints according to conditional probability. Second, the constructed multi-objective evolutionary algorithm is used to solve the model to obtain the Pareto solution set. Finally, a solution selection method is used to select a solution from the Pareto solution set, and its corresponding task allocation scheme is used as the execution scheme. Includes the following steps: Step 1: Based on the detected enemy target information, the current battlefield situation, operational risks, and our own operational resources, a multi-objective optimization model for heterogeneous UAV collaborative task allocation is established by setting an objective function based on conditional probability and providing constraints, as follows: Step 1-1: Collect basic data on enemy targets and our drones in the mission plan; make ; Set the detected enemy targets to include The first goal, the... Each goal is denoted as The set of targets is denoted as Each objective comprises three sub-tasks: classification task, attack task, and evaluation task, denoted as follows: , , Different enemy targets have different values ​​and risks; setting targets Value recorded as ; Record high-value goals as Low-value goals are denoted as To improve the success rate of operations, the strategy is to attack high-value targets twice and low-value targets once; Indicates the first Second attack mission, target The set of tasks is denoted as: set up , , and These represent the drone's interaction with the target. subtasks , , and The start time of execution, , , and These represent the drone's interaction with the target. subtasks , , and Execution time; Setting up a drone swarm includes The first drone i Recording of drones The collection of drones is denoted as The fleet is configured to include three types of heterogeneous unmanned aerial vehicles (UAVs): reconnaissance aircraft, fighter jets, and munitions aircraft, denoted as follows: , , ; The types of tasks that this type of drone can perform are: and , The types of tasks that this type of drone can perform are: , and , The types of tasks that this type of drone can perform are: ; make drones Value recorded as Different drones have different success rates and survival rates when performing different tasks and targeting different objectives; setting Indicates drone Execution Objectives Task types The success rate Indicates drone Execution Objectives Task types The mortality rate, of which, , , , Representing task type , and ; Based on the capabilities of drones, it can be known that when drones The type is From time to time When drones The type is From time to time ; make drones The set of tasks is denoted as ,in Indicates allocation to drones The number of tasks, Indicates drone The first task in the task set One task to be executed; Step 1-2: Construct the objective function of the model; For the goal Successful execution means that all its subtasks are successfully executed. If any subtask fails to execute successfully, then the target... The mission was not successfully executed; there are two risks when a drone performs a mission: the drone may be destroyed and the mission may not be successfully completed. To achieve the maximum operational benefit with the minimum operational cost, we consider maximizing the expected value of successfully executed targets and minimizing the expected value of destroyed drones as objective functions. These two objective functions are constructed by introducing conditional probability. Let... Indicates drone From the target Flying towards the target implement Task types A value of 1 indicates execution; otherwise, execution is not performed. Indicate target Location; (i) Maximize the expected value of the successfully executed objectives; (1) (ii) Minimize the expected value of the destroyed drones; (2) in, express The Middle One task, ,make Based on the actual situation, we have: and The objective (i) is reduced to a minimum form as follows: (iii) Minimize the expected value of objectives that were not successfully executed; (3) make ,in, , ; Steps 1-3: Construct the constraints of the model; In actual combat mission allocation, the type assigned to each UAV is: The number of missions cannot exceed the drone's payload; the types of missions it can perform. and The types of tasks assigned to the drones and There is no limit to the number of tasks; (4) in, Indicates drone The payload capacity; if the type of drone is If so, its payload is 0; Assume that the same drone attacks the same target no more than once; (5) Therefore, different attack missions targeting high-value targets are assigned to different drones; Each objective's tasks need to be executed sequentially, meaning they must satisfy timing constraints; for tasks with the same objective, the task type... Only in task type Executed after completion, and the task type Only in task type Execution will begin once all tasks are completed; therefore, task allocation must satisfy the following constraints. (6) Among them, if the target For high-value goals, then ,otherwise ; Each drone, when executing tasks within its task set, must do so in the order in which the tasks were assigned within the set. It can only be executed after its previous task has been completed; (7) in, express ; Steps 1-4: Construct a multi-objective optimization model for collaborative task allocation among multiple heterogeneous UAVs; Based on the objective function and constraints constructed above, the multi-objective optimization of multi-heterogeneous UAV collaborative task allocation is to minimize the vector function while satisfying constraints (4)-(7). ; Step 2: Solve the multi-objective optimization model for cooperative task assignment using the constructed multi-objective evolutionary algorithm to obtain its Pareto solution set, as follows: Step 2-1: Set the parameters of the improved multi-objective optimization algorithm, including population size. Maximum number of iterations Crossover probability and mutation probability The value; Step 2-2: Initialize the population; The chromosome is encoded using genes, where each gene represents the assignment of a specific task. The information in each gene includes the target number corresponding to the assigned task, the task type, and the drone number executing the task. Genes corresponding to the same drone form a sub-chromosome, and all sub-chromosomes constitute a complete chromosome. A randomized dataset of size... population ; Steps 2-3: Calculate the population based on objective functions (2) and (3). Fitness values ​​of chromosomes; Steps 2-4: Generate a scale of Subpopulations; Step 2-5: Determine whether the chromosomes in the subpopulation obtained in Step 2-4 are locked, and unlock the locked chromosomes; Steps 2-6: Combine populations and The resulting scale is population ; Steps 2-7: Calculate the population based on objective functions (2) and (3). Fitness values ​​of chromosomes; Steps 2-8: Based on the non-dominated quicksort method and elite preservation strategy, from the population Select Each chromosome forms the parent population for the next iteration. ; Steps 2-9: Let ; Step 2-10: If If the condition is met, proceed to step 2-3; otherwise, output the Pareto solution set. Step 3: Select a solution from the Pareto solution set using the solution selection method, and use its corresponding task allocation scheme as the execution scheme, as follows: make This represents the number of non-dominated solutions on the Pareto optimal front. Indicates the first One non-dominated solution; first, calculate ,in and Let be the weight, and satisfy... Decision-makers assign weights based on their preference for the objective function; then, the calculated weights are... Sort the results; finally, select the smallest one. The task allocation scheme for the non-dominated solution corresponding to the value.

2. The task allocation method for a multi-heterogeneous unmanned aerial vehicle system based on conditional probability and considering a multi-objective evolutionary algorithm as described in claim 1, characterized in that, In step 2-2, the encoding method is used to generate a code containing... The initialization process of the initial population of each chromosome is as follows: Step 2-2-1: Input the parameters and population size of the drone and enemy targets; Input the ammunition quantity of all drones as an array. Enemy target number set Type is and drone number set Type is and drone number set Set the number of chromosomes in the population to 0. Step 2-2-2: Determine the number of chromosomes in the current population; If the number of chromosomes in the population is equal to If the initialization is complete, the population initialization process is exited. Otherwise, proceed to steps 2-2-3 to 2-2-4; Step 2-2-3: Initialize an empty chromosome, denoted as . ; Step 2-2-4: Encoding Chromosomes ; Step 2-2-4-1: From Randomly select a target, denoted as ; Step 2-2-4-2: Select the target subtasks and attack mission Distribute; Initialize genes First, from Randomly select one drone, denoted as ;make , Then, from Randomly select one drone, denoted as ;make , and update , that is to say ; Step 2-2-4-3: If For high-value goals and Then for the target attack mission Perform the allocation; otherwise, proceed to step 2-2-4-4. Under the premise of satisfying constraint (5) Randomly select one drone, denoted as ; make , ;renew , that is to say ; Step 2-2-4-4: Select the target subtasks Distribute; from Randomly select one drone, denoted as ;from Remove target ; make , ; Step 2-2-4-5: Determine whether the current chromosome has completed encoding; If array The length is not 0, and If the vector is not zero, proceed to steps 2-2-4-1 to 2-2-4-5; otherwise, increment the number of chromosomes in the population by 1, exit the encoding process of the current chromosome, and proceed to step 2-2-2.

3. The task allocation method for a multi-heterogeneous unmanned aerial vehicle system based on conditional probability and considering a multi-objective evolutionary algorithm as described in claim 1, characterized in that, Steps 2-4 produce a scale of The specific steps for creating a subpopulation are as follows: Step 2-4-1: Use the roulette wheel algorithm to select from the population Select two cross parent generations ; Step 2-4-2: Based on the total resources and total task volume, select the appropriate crossover operator and perform the crossover operation according to the crossover probability; make Indicates the number of high-value targets; let Indicates the parent generation The corresponding combination of drones with remaining ammunition, Indicates the parent generation A collection of unassigned enemy targets; and This represents a crossover point randomly selected from the crossover parent. and They represent and intersection and Execute task types between and A collection of drones, and They represent and intersection and Execute task types between A collection of drones; and They represent and The set of targets assigned to it; father generation Crossover process; if This means that there is sufficient ammunition and some remaining. At this point, the crossover process is from step 2-4-2-1 to step 2-4-2-4. Step 2-4-2-1: Input the information of the intersection point and the intersection parent, let ; Step 2-4-2-2: Determine if the crossover process has ended; like satisfy If successful, proceed to steps 2-4-2-3 to 2-4-2-4; otherwise, exit the parent generation process. Cross operations; Step 2-4-2-3: For the intersection point Genes at that location undergo crossover operations; Intersection The task type in the gene is Perform the following operations at this time; if and Remove intersections If the target set corresponding to the gene at a given location is not empty, then from... One drone is randomly selected from the list; otherwise, one is selected from the list. Remove intersections Randomly select a drone from the set following the target in the gene; replace the drone information in the current gene with the information of the selected drone, and update... ; Intersection The task type in the gene is or Perform the following operations at this time; if Then randomly from One drone is randomly selected from the list; otherwise, it never belongs to any other drone. The type is or Randomly select one drone from the available drones; replace the drone information in the current gene with the information from that drone. Step 2-4-2-4: Let Proceed to step 2-4-2-2; like That is, there is a sufficient amount of ammunition with no surplus. At this point, the crossover process is from step 2-4-2-5 to step 2-4-2-8. Step 2-4-2-5: Input the information of the intersection point and the intersection parent, let ; Step 2-4-2-6: Determine if the crossover process has ended; like satisfy If successful, proceed to steps 2-4-2-7 to 2-4-2-8; otherwise, exit the parent generation process. Cross operations; Step 2-4-2-7: For the intersection point Genes at that location undergo crossover operations; Cross parent generation intersection The targets corresponding to the genes at each location are respectively denoted as ;like If so, proceed to step 2-4-2-8; otherwise, execute the following process. like If all are low-value objectives or all are high-value objectives, then directly in Interchange of corresponding genes Information; if For high-value goals, If the target is of low value, then first swap the target numbers in the corresponding genes, and then randomly select one. The attack gene was modified, and the target number of that gene was changed. ;like For low-value targets, For high-value targets, first swap the target numbers in the corresponding genes, then randomly select one. The attack gene was modified, and the target number of that gene was changed. ; Step 2-4-2-8: Let intersection and The number of genes updating information between them is denoted as And let Proceed to step 2-4-2-6; like This means that there is insufficient ammunition. At this point, the crossover process is from step 2-4-2-9 to step 2-4-2-13; Step 2-4-2-9: Input the information of the intersection point and the intersection parent, let ; Step 2-4-2-10: Determine if the crossover process has ended; like satisfy If successful, proceed to steps 2-4-2-11 to 2-4-2-13; otherwise, exit the parent generation process. Cross operations; Step 2-4-2-11: For the intersection point Genes at that location undergo crossover operations; Let the intersection point The target in the gene is denoted as ;like So from Randomly select a target from the list; otherwise, select from the list. Randomly select a target; denote the selected target as... ; when When the goal is low value, if If it is a low-value goal, then set the goal as follows: The target replacement in all corresponding genes is Otherwise, from Randomly select a low-value target , target and The target replacement in all corresponding genes is ; when When it is a high-value goal, if If it's a low-value goal, then start with... Randomly select a target , target The target replacement in all corresponding genes is Then randomly select a target The attack gene, and replace the target of that gene with Finally, add targets. Task types and The genes; if For high-value goals, set the goals The target replacement in all corresponding genes is ; Step 2-4-2-12: Update and ; Step 2-4-2-13: Let intersection and The number of genes updating information between them is denoted as And let Proceed to step 2-4-2-10; Step 2-4-3: Perform mutation operations on the crossover offspring based on the mutation probability; Let the two cross-subsidiaries obtained from the crossover operation be denoted as... , ;by For example, the mutation process is given; let and Let be the start and end points of the gene segment to be mutated; The set of low- and medium-value objectives is denoted as , middle and The set of genes is denoted as , middle The set of genes is denoted as , The corresponding sets of drones with remaining ammunition and the sets of unassigned targets are denoted as follows: and ; Step 2-4-3-1: Input mutation points and ,make ; Step 2-4-3-2: Determine whether the mutation process has ended; like satisfy If successful, proceed to steps 2-4-3-3 to 2-4-3-4; otherwise, exit the process. The mutation operation; Step 2-4-3-3: Analyze the mutation points The gene at that location undergoes mutation manipulation; when ,like mutation points The task type in the gene is or So from Remove from middle mutation points The drone in the gene, and then from Randomly select one drone; if mutation points The task type in the gene is So from Remove from middle mutation points The drone in the gene, and then from Randomly select a drone and update Replace with information from the selected drone. mutation points Drone information in the genes at the location; when First from Randomly select a target, denoted as Then find all targets and mutation points The target of the gene at that location is the same gene, and its target is replaced with ;like For high-value goals, and mutation points If the target in the gene is a low-value target, then... mutation points The target in the gene is from Remove from the middle, then from Randomly select a target and delete all its genetic information, then distribute the ammunition to... ;like For low-value goals, mutation points If the target in the gene is a high-value target, then randomly delete it. A gene for an attack mission; when ,like mutation points The task type in the gene is or So from Remove from middle mutation points Genes at the location, and then randomly from Choose one gene from; otherwise, choose from... Remove from middle mutation points Genes at the location, and then randomly from Select one gene; combine the selected gene with... mutation points The drones exchange genetic information. Step 2-4-3-4: Let mutation points and The number of genes updating information between them is denoted as And let Proceed to step 2-4-3-2; Step 2-4-4: Repeat steps 2-4-1 to 2-4-3 until a size of [size missing] is obtained. subpopulation .

4. The task allocation method for a multi-heterogeneous unmanned aerial vehicle system based on conditional probability and considering a multi-objective evolutionary algorithm as described in claim 1, characterized in that, The process for determining and unlocking a chromosome lock-in in steps 2-5 is as follows: Step 2-5-1: Transform the chromosome into a form composed of daughter chromosomes; The genes in the chromosome are arranged in ascending order according to the drone number, and the tasks and execution order of each drone are obtained based on the sub-chromosomes; Step 2-5-2: Create a set representing completed tasks, denoted as . ; Step 2-5-3: If If the dimension is equal to the dimension of the chromosome, the unlocking process terminates; otherwise, proceed to steps 2-5-4 to 2-5-6. Step 2-5-4: Determine whether the task in the current gene of each subchromosome can be executed; Each drone starts execution from the first task in the task set. If the task type is... If so, execute directly, remove the task from the task set, and add it to the task list. In the middle; if the task type is Then determine the task type of the target corresponding to the task. Is it included? If the task is included in the task set, execute it directly, remove the task from the task set, and add it to the task list. Otherwise, the task cannot be executed, and the drone will be in a waiting state; if the task type is Then determine the task type of the target corresponding to the task. Does the target include all attack missions? If the task is included in the task set, execute it directly, remove the task from the task set, and add it to the task list. Otherwise, the mission cannot be executed, and the drone will be in a waiting state; Step 2-5-5: Determine if the chromosomes are locked together; Based on step 2-5-4, determine the current status of all drones. If all drones are in a waiting state, then the chromosome is locked. Steps 2-5-6: Unlock at the locked point; If the drone at the lock-point is waiting to perform a task of type Then, randomly select a type from the remaining tasks of the drone. Swap the order of the current task with the current task; perform this operation on all drones in a waiting state; proceed to step 2-5-3.

Images