Collaborative Multitasking Reallocation Method, Apparatus, Device, and Readable Storage Medium

Abstract

The present application relates to a collaborative multi-task reallocation method, device, equipment and readable storage medium, and relates to the technical field of unmanned aerial vehicle (UAV) mission planning. The heterogeneous UAV collaborative multi-task reallocation problem is evaluated by using two objective functions, namely the total flight distance of the UAVs and the mission completion time. This not only ensures that the entire UAV system can complete the entire combat mission with the least amount of resources, but also ensures that the number of tasks assigned to each UAV by the UAV system can maintain a certain balance, avoiding the problem that some UAVs are assigned more tasks due to the performance differences of heterogeneous UAVs. Moreover, based on the prior knowledge obtained through the KnCMPSO algorithm, the structure learning strategy and the perturbation local search strategy, as well as the non-participation in the reallocation constraint and the UAV availability constraint, the task assignment sequence is optimized to achieve multi-objective heterogeneous UAV collaborative task reallocation, which can effectively improve the search efficiency of the algorithm and reduce the solution time of the algorithm.

Description

Technical Field

[0001] The present application relates to the technical field of unmanned aerial vehicle mission planning, and particularly relates to a collaborative multi-task reallocation method, device, equipment and readable storage medium. Background Technique

[0002] The heterogeneous unmanned aerial vehicle collaborative multi-task reallocation model refers to that when an unexpected situation (such as unmanned aerial vehicle failure or newly discovered military target, etc.) occurs during the execution of the existing task allocation sequence by the unmanned aerial vehicle system, in order to ensure the smooth completion of the task, at this time, it is necessary to reallocate the task allocation sequences of some unmanned aerial vehicles in the unmanned aerial vehicle system to successfully cope with the unexpected situation. In order to overcome the uncertainty brought by the change of the battlefield environment, the task reallocation has higher requirements for the response time of the unmanned aerial vehicle system. Therefore, after an unexpected situation occurs, it is required that the unmanned aerial vehicle system re-plan an efficient and reasonable task allocation plan within a limited time.

[0003] The scenarios involved in the unmanned aerial vehicle task allocation problem are complex and there are many constraint conditions. It is a typical optimization NP (Non-Deterministic Polynomial Problems) problem, and it is difficult to solve. Currently, the commonly used solutions can be roughly divided into two types: optimization methods and heuristic algorithms. Among them, the optimization method has a simple structure and is easy to implement. However, due to its huge calculation amount and slow operation speed, it is not suitable for solving problems with a large scale. Different from the optimization method, the heuristic algorithm does not aim to obtain an exact solution, and can thus give a satisfactory solution to the problem within an acceptable time range. Therefore, it has a wide application range and low computational complexity.

[0004] In the related technologies, in the research on solving the unmanned aerial vehicle task allocation problem using heuristic algorithms, the algorithms that are more applied include genetic algorithms, ant colony algorithms and particle swarm algorithms. Among them, as a classic algorithm in swarm intelligence algorithms, the particle swarm algorithm has the characteristics of simple operation and strong convergence ability, and has strong application prospects in solving the unmanned aerial vehicle collaborative task allocation problem. However, the current particle swarm algorithm mainly focuses on the initial allocation stage of unmanned aerial vehicle tasks. As a result, when an unexpected situation occurs, it is unable to reallocate the task allocation sequences of some unmanned aerial vehicles with problems in the unmanned aerial vehicle system, thereby affecting the smooth completion of the task. In addition, most of the current particle swarm algorithms for solving the unmanned aerial vehicle task allocation problem are designed for single-objective models, and their constraint conditions are relatively simple and not suitable for the reallocation of multi-objective heterogeneous unmanned aerial vehicle collaborative tasks. Summary of the Invention

[0005] The present application provides a collaborative multi-task reallocation method, apparatus, device and readable storage medium to solve the problem in the related art that multi-target heterogeneous UAV collaborative task reallocation cannot be effectively achieved in case of emergencies.

[0006] In a first aspect, a collaborative multi-task reallocation method is provided, including the following steps:

[0007] Obtain an objective function and constraint conditions, where the objective function includes a total UAV flight distance function and a total task completion time function, and the constraint conditions include non-participation in reallocation constraints and UAV availability constraints;

[0008] Obtain an initial task allocation sequence obtained based on the KnCMPSO algorithm and create multiple empty task allocation sequences;

[0009] Perform initialization processing on multiple empty task allocation sequences based on the initial task allocation sequence and the constraint conditions to obtain multiple initialized task allocation sequences;

[0010] Allocate multiple initialized task allocation sequences to a distance task allocation sequence set for solving the total UAV flight distance function and a time task allocation sequence set for solving the total task completion time function;

[0011] Update each initialized task allocation sequence in the distance task allocation sequence set and the time task allocation sequence set respectively based on a structure learning strategy and a perturbation local search strategy to obtain an updated distance task allocation sequence set and an updated time task allocation sequence set;

[0012] Sort the task allocation sequences in the updated distance task allocation sequence set and the task allocation sequences in the updated time task allocation sequence set according to the Pareto dominance relationship to obtain a sorted queue; select the top N task allocation sequences from the sorted queue as the collaborative multi-task reallocation result, where N is a positive integer.

[0013] In a second aspect, a collaborative multi-task reallocation apparatus is provided, including:

[0014] An acquisition unit, which is used to acquire an objective function and constraint conditions, where the objective function includes a total UAV flight distance function and a total task completion time function, and the constraint conditions include non-participation in reallocation constraints and UAV availability constraints;

[0015] A creation unit, which is used to obtain an initial task allocation sequence obtained based on the KnCMPSO algorithm and create multiple empty task allocation sequences;

[0016] An initialization unit, which is configured to perform initialization processing on a plurality of empty task assignment sequences based on the initial task assignment sequence and the constraint conditions to obtain a plurality of initialized task assignment sequences; and allocate the plurality of initialized task assignment sequences to a distance task assignment sequence set for solving the total flight distance function of the unmanned aerial vehicle and a time task assignment sequence set for solving the total task completion time function.

[0017] An update unit, which is configured to update each of the initialized task assignment sequences in the distance task assignment sequence set and the time task assignment sequence set based on a structure learning strategy and a perturbation local search strategy respectively to obtain an updated distance task assignment sequence set and an updated time task assignment sequence set.

[0018] An allocation unit, which is configured to sort the task assignment sequences in the updated distance task assignment sequence set and the task assignment sequences in the updated time task assignment sequence set according to the Pareto dominance relationship to obtain a sorted queue; and select the top N task assignment sequences in the sorted queue as the collaborative multi-task reallocation result, where N is a positive integer.

[0019] In a third aspect, there is provided a collaborative multi-task reallocation device, including: a memory and a processor, where at least one instruction is stored in the memory, and the at least one instruction is loaded and executed by the processor to implement the foregoing collaborative multi-task reallocation method.

[0020] In a fourth aspect, there is provided a computer-readable storage medium, where a computer program is stored in the computer storage medium, and when the computer program is executed by a processor, the foregoing collaborative multi-task reallocation method is implemented.

[0021] The beneficial effects brought by the technical solution provided in this application include: it can effectively implement collaborative task reallocation of multi-object heterogeneous unmanned aerial vehicles in case of emergencies, improve the search efficiency of the algorithm, and reduce the solution time of the algorithm.

[0022] The present application provides a collaborative multi-task reallocation method, apparatus, device and readable storage medium. When encountering emergencies, the present application uses two objective functions, the total flight distance of the unmanned aerial vehicles (UAVs) and the task completion time, to evaluate the collaborative multi-task reallocation problem of heterogeneous UAVs. Among them, based on the evaluation index of the total flight distance of the UAVs, it can ensure that the entire UAV system can complete the entire combat mission with the least amount of resources. And based on the evaluation index of the total task completion time function, it can ensure that the number of tasks assigned to each UAV by the UAV system can maintain a certain balance, avoiding the problem that some UAVs are assigned more tasks due to the performance differences of heterogeneous UAVs. Moreover, based on the prior knowledge, structure learning strategy and perturbation local search strategy obtained by the KnCMPSO algorithm, as well as the non-participation in reallocation constraints and UAV availability constraints, the present application introduces an initialization strategy based on prior knowledge to optimize the task assignment sequence, so as to realize the collaborative task reallocation of multi-objective heterogeneous UAVs, which can effectively improve the search efficiency of the algorithm and reduce the solution time of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0024] Figure 1 It is a schematic flowchart of a collaborative multi-task reallocation method provided by an embodiment of the present application;

[0025] Figure 2 It is a schematic diagram of the initialization particle operation provided by an embodiment of the present application;

[0026] Figure 3 It is a schematic diagram of the particle update strategy provided by an embodiment of the present application;

[0027] Figure 4 It is a schematic structural diagram of a collaborative multi-task reallocation apparatus provided by an embodiment of the present application;

[0028] Figure 5 It is a schematic structural diagram of a collaborative multi-task reallocation device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are some, but not all, of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the scope of protection of this application.

[0030] The embodiments of this application provide a collaborative multi-task reallocation method, device, equipment, and readable storage medium, which can solve the problem in the related art that it is impossible to effectively achieve collaborative task reallocation of multi-target heterogeneous unmanned aerial vehicles (UAVs) in the event of emergencies.

[0031] Figure 1 The following is a collaborative multi-task reallocation method provided by the embodiments of this application, including the following steps:

[0032] Step S10: Obtain the objective function and constraint conditions. The objective function includes the total flight distance function of UAVs and the total task completion time function, and the constraint conditions include non-participation in reallocation constraints and UAV availability constraints;

[0033] Furthermore, the total flight distance function f1 of the UAVs and the total task completion time function f2 are respectively:

[0034]

[0035] In the formula, S represents the number of reconnaissance UAVs, A represents the number of attack UAVs, m represents the mth task, represents the task sequence of UAV U i , represents the flight distance of UAV U i from executing the mth task to executing the m+1th task, N represents the total number of targets, Evaluate represents the evaluation task, represents the time taken to complete the jth target.

[0036] Exemplarily, in this embodiment, based on the heterogeneous UAV collaborative multi-task allocation model and combined with the application actual situation, the original model is extended to a heterogeneous UAV collaborative multi-task reallocation model (Multi-constraint Cooperative Multiple Task Reallocation Problem, MCMTRP) to effectively cope with the dynamic changes in the battlefield environment and better conform to the actual combat application. Among them, the MCMTRP model includes the following objective function and constraint conditions:

[0037] Specifically, the objective function includes the total flight distance function f1 of the UAVs and the total task completion time function f2:

[0038]

[0039] Where S represents the number of reconnaissance UAVs (reconnaissance UAVs can be used for both reconnaissance tasks and assessment tasks), A represents the number of attack UAVs, m represents the m-th task, represents the task sequence of UAV U i ; represents the flight distance of UAV U i from the execution of the m-th task to the execution of the m+1-th task, N represents the total number of targets, Evaluate represents the assessment task, represents the time taken to complete the j-th target. For a target, it generally includes a reconnaissance task (i.e., Observe), a strike task (i.e., Attack), and an assessment task. That is, the time taken for the last Evaluate task in the j-th target to complete is the time taken for the j-th target. Therefore, f2 represents the time when the last task in the UAV system is completed, which is the maximum completion time of the task; while f1 represents the total flight distance of the UAVs, that is, the sum of the flight distances of all UAVs participating in the tasks in the UAV system.

[0040] The constraint conditions include:

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050] Where being 1 means that task m is executed by UAV U i ; being 0 means that task m is not assigned to UAV U i ; Denote the moment when the Observe task is completed for the j-th target, Denote the moment when the Attack task is completed for the j-th target, Denote the moment when the Evaluate task is completed for the j-th target; Denote the number of ammunitions or time consumed by the UAV U i when performing task m, Denote the maximum number of ammunitions carried by the attack UAV U i The attack UAV can only carry a certain number of ammunitions. Therefore, the number of ammunitions consumed by the attack UAV when performing the strike task should be less than its maximum carried ammunitions; The reconnaissance task and the evaluation task of the strike result for the same target need to be performed by at least one reconnaissance UAV, while the strike task needs to be assigned to at least one attack UAV, that is When k ∈ {Observe, Evaluate}, k represents the reconnaissance task or the evaluation task, i ∈ (1, 2, …, S), then Denote the time consumed by the UAV U i when performing task k, C k Denote the total time required to complete task k; When k ∈ {Attack}, k represents the strike task, i ∈ (S + 1, …, S + A), then Denote the time consumed by the UAV U i when performing task k, Denote the UAV U i 's maximum flight distance.

[0051] In addition, t m Denote the start time of task m, t represents the time when the emergency occurs. Therefore, T m being 1 means that task m is an unfinished task, T m being 0 means that task m is a completed task or a task being completed; U i being 1 means that the i-th UAV is not failed, U i being 0 means that the i-th UAV has failed. After the emergency occurs, all tasks on the targets can be divided into three types: completed tasks, tasks being completed, and unfinished tasks. Therefore, in this embodiment, the completed tasks, tasks being completed, and failed UAVs do not participate in the subsequent task allocation. Thus, it can be seen that the last two constraints T m and U i are the non-participation in reallocation constraint and UAV availability constraint respectively.

[0052] Step S20: Obtain the initial task allocation sequence obtained based on the KnCMPSO algorithm and create multiple empty task allocation sequences;

[0053] Exemplarily, the KnCMPSO algorithm is a collaborative multi-objective particle swarm optimization algorithm based on inflection points, mainly used to solve the problem of collaborative multi-task allocation for heterogeneous UAVs. Its basic idea is to first generate D sub-populations Pt of size B according to the number of objectives D, initialize the particles using an initialization strategy based on constraint handling, and update the individual optimal solutions of the particles and the global optimal solutions of each sub-population. Then, perform a fast non-dominated sorting on all particles to update the external archive set Archive, obtaining the elite solution set Qt. Next, update the particle information based on the structure learning strategy and the local search strategy to obtain the population Pt+1, update the elite individuals in Archive using the local search strategy to obtain the population S, update the elite individuals in Archive using the structure learning strategy based on inflection points to obtain the population R. After that, merge Qt, Pt+1, S, and R into a new population and perform a fast non-dominated sorting, thereby updating the external archive set Archive to obtain the elite solution set Qt+1, and repeat the loop until the end condition is met. However, since the tasks in the MCMTRP model constructed in this embodiment have multiple states, the initialization strategy based on constraint handling is no longer applicable to this embodiment. Nevertheless, the allocation sequence generated by the KnCMPSO algorithm during the task allocation stage has certain guiding value for solving the MCMTRP model. Therefore, in this embodiment, an initial task allocation sequence is first generated based on the KnCMPSO algorithm and multiple empty task allocation sequences are created, and then the particle initialization strategy and the particle update strategy are improved to form a collaborative multi-objective particle swarm algorithm based on learning (i.e., the LeCMPSO algorithm).

[0054] Step S30: Initialize multiple empty task allocation sequences based on the initial task allocation sequence and the constraint conditions to obtain multiple initialized task allocation sequences;

[0055] Further, the initializing multiple empty task allocation sequences based on the initial task allocation sequence and the constraint conditions to obtain multiple initialized task allocation sequences specifically includes:

[0056] Determine the completed tasks, tasks being executed, and unexecuted tasks in the initial task allocation sequence at the t-th moment based on the non-reassignment constraint;

[0057] Update the completed tasks, the UAV information corresponding to the execution of the completed tasks, and the resource consumption to the corresponding positions of the first empty task allocation sequence to obtain the first processed task allocation sequence;

[0058] Update the completed tasks among the tasks being executed and the UAV information and resource consumption for executing the completed tasks among the tasks being executed to the corresponding positions of the first processed task allocation sequence to obtain the second processed task allocation sequence;

[0059] Based on the constraints, allocate drones and resource consumption amounts for the unfinished tasks in the tasks being executed, and insert the unfinished tasks in the tasks being executed and the drones and resource consumption amounts corresponding to the unfinished tasks in the tasks being executed into the second processed task allocation sequence to obtain a third processed task allocation sequence;

[0060] Based on the constraints, allocate drones and resource consumption amounts for the unexecuted tasks, and insert the unexecuted tasks and the drones and resource consumption amounts corresponding to the unexecuted tasks into the third processed task allocation sequence to obtain an initialized task allocation sequence.

[0061] Exemplarily, in this embodiment, when initializing and generating particles, a hybrid coding strategy based on a three-dimensional matrix is adopted. The position vector P of each particle will be represented by a three-dimensional matrix P = (T, U, C); where T represents the target number of the task being executed. Each target number will appear three times in the first row of the particle, representing the reconnaissance task, the strike task, and the evaluation task of the strike result executed on this target in the order of appearance; U represents the drone number executing the task on this target. Since the order of appearance of the target number in the matrix represents different task types, the drone number U (drone type) needs to be set according to the task type represented by the target number T; C represents the resource consumption when the drone executes the task. For the reconnaissance drone, this consumption represents the reconnaissance time of the drone, and the resource consumption at this time is a continuous variable. For the attack drone, this consumption represents the ammunition usage of the drone, and the resource consumption at this time is a discrete variable.

[0062] After a sudden situation occurs, the set of tasks to be executed in the MCMTRP model includes three states: the task has been completed, the task is being executed, and the task has not been executed. It can be seen that the position vector contains some tasks that do not participate in reallocation. To ensure that the completed tasks in the task allocation do not participate in task reallocation and avoid the generation of infeasible solutions, this embodiment proposes a prior knowledge-based initialization strategy (PKIS) to initialize the particles and generate the position vectors of the particles: First, evaluate the task sequence to find all tasks that have been completed or are being executed at time t; then directly add the completed tasks to the corresponding coding positions and update the remaining resources of the drones; for the tasks being executed, first add the drone information that has completed the task to the corresponding coding positions, and then continue to allocate the remaining demand of the task until the task is allocated; and for the unexecuted tasks, use an initialization strategy based on constraint processing for initialization, and loop until all tasks are allocated.

[0063] Specifically, 1) Evaluate the initial task assignment sequence to find all tasks that have been completed or are being executed at time t; 2) For the completed tasks, directly add the task, the UAV number executing the task, and its resource consumption to the corresponding coding position of the particle in the empty task assignment sequence and update the remaining resources of the UAV; 3) For the tasks being executed, first add the task to the corresponding coding position of the particle, then add the UAV number that has completed the task and its resource consumption to the corresponding coding position of the particle, update the remaining resources of the UAV executing the task and the remaining resources required to complete the task, and then continue to assign the task; 4) For the tasks not yet executed, use an initialization strategy based on constraint handling for initialization. The above-mentioned initialization strategy based on prior knowledge has the following advantages: 1) The tasks that have been completed will not participate in reallocation, which satisfies the constraint of the reallocability of the algorithm. The tasks being executed partially participate in reallocation, which also satisfies the constraint of the reallocability of the algorithm and the cooperation constraint between UAVs; 2) For other constraints such as the timing constraint of tasks, the particles generated according to this strategy also satisfy, which provides a basis for subsequent particle update.

[0064] Step S40: Assign multiple initialized task assignment sequences to a distance task assignment sequence set for solving the total flight distance function of the UAV and a time task assignment sequence set for solving the total task completion time function.

[0065] Step S50: Update each initialized task assignment sequence in the distance task assignment sequence set and the time task assignment sequence set based on the structure learning strategy and the perturbation local search strategy (i.e., update particle information based on the structure learning strategy and the perturbation-based local search strategy) to obtain an updated distance task assignment sequence set and an updated time task assignment sequence set.

[0066] Exemplarily, since the MCMTRP model contains mixed variables and has many constraint conditions, it is difficult to generate particles that satisfy the constraints according to the update method of the standard particle swarm. Therefore, in this embodiment, the concept of particle velocity is no longer used, and the particle only updates its state by learning from excellent individuals during the update. The update formula is:

[0067]

[0068] In the formula, X i (t) represents the position of particle i in the t-th iteration, PBest i is the individual optimal value of particle i in the t-th iteration, E is the individual in the external archive set, GBest tis the global optimum in the t-th iteration, w is the weight coefficient, F1 represents the result based on the individual best learning of the particle, and F2 represents the result based on the learning of the better individuals among E and GBest t among the better individuals

[0069] competition represents the opportunity for the particles E and GBest t to compete with each other as the global learning object of the particle X. The specific method is as follows:

[0070] Suppose the individual to be updated is H (i.e., the task assignment sequence to be updated. In this embodiment, each individual is a task assignment sequence). First, randomly select an elite individual I from the external archive set Archive, calculate the cosine similarity between the individual I and H, and then compare it with the cosine similarity between GBest t and H. When the cosine similarity between I and H is smaller, it means that the guiding ability of the individual I to the individual H is stronger and more conducive to the convergence of the algorithm. Therefore, select I as the global learning object of H. Otherwise, select GBest t as the global learning object of H.

[0071] F1 and F2 represent the learning process of the particle towards the better individuals based on the structure learning method, which is described in detail as follows: Suppose lH represents the individual to be learned, cH represents the current individual, and nH represents the new individual generated after the learning method. The entire learning process is as follows: First, initialize an empty individual nH, and sequentially select the target numbers in the MCMTRP model. Select the current target number for learning operations with a certain probability. Suppose the currently selected target number is 2. Then, insert the values at the three positions where the first row of the position vector of the individual to be learned lH is equal to 2 into the corresponding positions of the position vector of the new individual nH in the order in which they appear in the position vector of lH. At the same time, insert the corresponding drones and the drone resource consumption amounts for each task into the corresponding positions of the position vector of nH. After traversing the target numbers once, the information of the individual to be learned lH is saved in the non-empty positions of the new individual nH. Then, fill the targets, the corresponding drones, and the drone resource usage amounts in the original individual cH into the empty positions of the position vector of the new individual from left to right, but skip the targets that already exist in the new individual nH.

[0072] In addition, in order to enhance the diversity of the population and the search ability of the algorithm in this embodiment, a disturbance-based local search strategy (DLS) is adopted, which is described in detail as follows: Randomly select one of the uncompleted tasks m on the current target T corresponding to the current particle, and remove the UAV sequence and the UAV resource consumption sequence corresponding to task m on the current particle position vector; Randomly select a UAV from the UAV set according to the nature of task m and randomly generate resource consumption, and loop until task m is completely assigned.

[0073] Step S60: Sort the task assignment sequences in the updated distance task assignment sequence set and the task assignment sequences in the updated time task assignment sequence set according to the Pareto dominance relationship to obtain a sorted queue; Select the top N task assignment sequences in the sorted queue as the collaborative multi-task reallocation result, where N is a positive integer, and the specific value of N can be determined according to actual needs and is not limited here.

[0074] Exemplarily, in this embodiment, through the Pareto dominance relationship, the mutual dominance relationships (i.e., dominant solutions and non-dominant solutions) between each task assignment sequence can be obtained, and then the task assignment sequences are sorted. Finally, the top N task assignment sequences can be selected as the collaborative multi-task reallocation result.

[0075] Further, the sorting the task assignment sequences in the updated distance task assignment sequence set and the task assignment sequences in the updated time task assignment sequence set according to the Pareto dominance relationship to obtain a sorted queue specifically includes:

[0076] Update the task assignment sequences in the external archive set based on the disturbance local search strategy (i.e., update the elite individual information in the external archive set based on the disturbance local search strategy) to obtain an updated external archive set;

[0077] Sort the task assignment sequences in the updated distance task assignment sequence set, the task assignment sequences in the updated time task assignment sequence set, and the task assignment sequences in the updated external archive set according to the Pareto dominance relationship to obtain a sorted queue.

[0078] Further, the updating the task assignment sequences in the external archive set based on the disturbance local search strategy to obtain an updated external archive set specifically includes:

[0079] Randomly select a first task from the first task assignment sequence in the external archive set;

[0080] Remove the UAV sequence and the UAV resource consumption sequence corresponding to the first task from the first task assignment sequence;

[0081] Randomly select at least one first drone from the set of drones that is the same as the first task category and randomly generate a first resource consumption for performing the first task;

[0082] Update the first task assignment sequence based on the first drone and the first resource consumption to obtain an updated external archive set.

[0083] Exemplarily, each particle includes a string of task sets, and each task has a corresponding drone and its resource consumption. The detailed operation of updating the external archive set based on the perturbation local search strategy in this embodiment is described as follows: 1) First, randomly select an unfinished task m on any task assignment sequence in the external archive set; 2) Remove the drone sequence and the drone resource consumption sequence corresponding to task m from this task assignment sequence; 3) Randomly select a drone from the corresponding set of drones according to the category of task m and randomly generate the resource consumption, and loop until task m is assigned completely to obtain an updated external archive set; then participate the task assignment sequences in the updated external archive set in the sorting, which can avoid the generation of infeasible solutions, and at the same time increase the population diversity and the search ability of the algorithm.

[0084] Further, sorting the task assignment sequences of the updated distance task assignment sequence set, the task assignment sequences of the updated time task assignment sequence set, and the task assignment sequences of the updated external archive set according to the Pareto dominance relationship to obtain a sorted queue, specifically including:

[0085] Update the initial task assignment sequence based on the drone availability constraint to obtain an updated task assignment sequence;

[0086] Sort the task assignment sequences of the updated distance task assignment sequence set, the task assignment sequences of the updated time task assignment sequence set, the task assignment sequences of the updated external archive set, and the updated task assignment sequence according to the Pareto dominance relationship to obtain a sorted queue.

[0087] Further, the constraint condition further includes the completion time sequence constraint of different types of tasks. The updating the initial task assignment sequence based on the drone availability constraint to obtain an updated task assignment sequence specifically includes:

[0088] When it is detected that there is a first unfinished task in the initial task assignment sequence and the first unfinished task is executed by a failed drone, then initialize an empty first task assignment sequence;

[0089] Update each task in the initial task assignment sequence, the UAV information corresponding to the execution of each task, and the resource consumption information corresponding to each UAV to the empty first task assignment sequence to obtain a second task assignment sequence;

[0090] Remove the failed UAVs and the resource consumption information corresponding to the failed UAVs from the second task assignment sequence, and update the remaining resource amount required to execute the first unfinished task to obtain a third task assignment sequence;

[0091] Randomly select at least one second UAV from the set of UAVs of the same type as the first unfinished task based on the remaining resource amount, and randomly generate a second resource consumption amount for executing the first unfinished task;

[0092] Determine the insertable position range of the first unfinished task from the third task assignment sequence based on the completion time sequence constraints of different types of tasks;

[0093] Insert the first unfinished task, the second UAV corresponding to the first unfinished task, and the second resource consumption amount into any position within the insertable position range of the first unfinished task in the third task assignment sequence to obtain an updated task assignment sequence.

[0094] Exemplarily, in this embodiment, the triggering conditions for the UAV system to perform reallocation can be roughly divided into three types, namely, adding a new target point, changing the target point position, and UAV failure (i.e., the availability of the UAV is unavailable). However, regardless of which of the above situations, when performing task reallocation for the entire UAV system, only a small part of the UAV task sequence needs to be updated. Therefore, it is unreasonable to regenerate the task sequence of the entire UAV cluster, which will inevitably cause waste of resources and the algorithm timeliness cannot meet the requirements. In addition, the assignment sequence generated during the task assignment stage has certain guiding value for solving the task reallocation problem. Making good use of this part of information helps to accelerate the convergence speed of the algorithm.

[0095] Therefore, this embodiment proposes a particle update strategy based on historical information learning (i.e., the HILPUS strategy) to update particle information. Taking the failure of an unmanned aerial vehicle (UAV) as an example, it is assumed that at time t, UAV U1 fails. After that, the tasks that UAV U1 participates in after time t in the original allocation sequence need to be reassigned to other UAVs of the same type as UAV U1. Therefore, the HILPUS strategy first extracts the unfinished task T that contains UAV U1 in the original allocation sequence, removes UAV U1 corresponding to this task T and its corresponding resource consumption C1, randomly selects a UAV U2 of the same type as UAV U1 that meets the constraints, sets its resource usage, and repeats this process until task T is completed. After meeting the task completion requirements, the newly added UAV and its resource usage are inserted into the corresponding position in the particle encoding. In addition, since the order of tasks determines the order in which each UAV executes tasks, to increase the diversity of the population, an unfinished task is randomly selected. First, the maximum insertable position range is found, and then this task is randomly inserted within this range. The individual generated in this way is still an executable combat sequence that meets the constraints, and the newly generated individual not only retains the excellent characteristics of the original allocation sequence but also increases the diversity of the entire population to a certain extent.

[0096] Specifically, taking the failure of an unmanned aerial vehicle as an example, it is assumed that at time t, UAV U1 fails. The detailed operation description is as follows: Initialize an empty particle nS, and assign each task, the UAV information corresponding to the execution of each task, and the resource consumption information corresponding to each UAV in the initial task allocation sequence to nS; Select the m-th task on the position vector of nS. If task m is not completed and contains UAV U1, then delete UAV U1 corresponding to task m and its resource consumption from its encoding position, and update the remaining resource amount required to execute task m. Randomly select a UAV of the same type to continue allocating task m until task m is allocated; Then randomly generate a number r, r ∈ [0, 1]. If r is less than the initial random number L, then randomly select a task O that is completed after time t, find the insertable position range [startPos, endPos] of task O in the position vector of nS, and randomly select an insert position insertPos between [startPos, endPos], insert task O into insertPos in the position vector, and insert the tasks at the remaining positions into [startPos, endPos] of nS in turn, so as to obtain an updated task allocation sequence. Then, participate the updated task allocation sequence in the sorting, which can further avoid the generation of infeasible solutions, improve the search efficiency of the algorithm, and reduce the solution time of the algorithm.

[0097] The above-mentioned update strategy based on historical information learning can generate a reallocated operation sequence as fast as possible without changing most of the existing UAV operation task sequences. In addition, since the existing allocation sequences themselves have good properties, most of the individuals generated by the HILPUS update strategy are relatively optimal solutions, which speeds up the convergence rate of the algorithm. This enables the LeCMPSO algorithm to meet the requirements of real-time task reallocation and rapid response in the battlefield environment while meeting the combat requirements, avoiding the generation of infeasible solutions, improving the search efficiency of the algorithm, and effectively reducing the solution time of the algorithm.

[0098] It can be seen that when encountering emergencies, this embodiment uses two objective functions, the total flight distance of UAVs and the task completion time, to evaluate the heterogeneous UAV cooperative multi-task reallocation problem. Among them, based on the evaluation index of the total flight distance of UAVs, it can ensure that the entire UAV system can complete the entire combat mission with the least amount of resources. And based on the evaluation index of the total task completion time function, it can ensure that the number of tasks assigned to each UAV by the UAV system can maintain a certain balance, avoiding the problem that some UAVs are assigned more tasks due to the performance differences of heterogeneous UAVs. Moreover, this embodiment optimizes the task allocation sequence based on the prior knowledge, structure learning strategy, and perturbation local search strategy obtained through the KnCMPSO algorithm, as well as the non-participation in reallocation constraints and UAV availability constraints, to achieve multi-objective heterogeneous UAV cooperative task reallocation, which can effectively improve the search efficiency of the algorithm and reduce the solution time of the algorithm.

[0099] The working principle of the cooperative multi-task reallocation method involved in this embodiment will be explained below in combination with specific cases.

[0100] Take an example where the instance includes 20 UAVs and 24 military task objectives. The task objectives are randomly set at a fixed location, and each task objective contains three attributes, which respectively represent the amount of resources required to complete a certain military task (reconnaissance task, strike task, and assessment task corresponding to the strike result) on this objective. The UAVs are also randomly set at a fixed location, and each UAV includes flight speed, the maximum number of ammunitions carried, and the farthest flight distance. Among them, Table 1 and Table 2 respectively show the target attribute values and UAV attribute values.

[0101] Table 1 Target Attribute Values

[0102]

[0103] Table 2 UAV Attribute Values

[0104]

[0105]

[0106] The KnCMPSO algorithm is used to perform task allocation for this example. A task allocation result is randomly selected from the solution set obtained by the KnCMPSO algorithm. The total execution time of the corresponding tasks is 3,683 seconds, and the total flight distance of the UAVs is 4,854 kilometers. The task sequences of each UAV are shown in Table 3.

[0107] Table 3 Task Sequences of Each UAV

[0108]

[0109]

[0110] Assume that at time t = 500 seconds, UAV U8 fails. Based on the above data, the specific implementation steps of this embodiment are as follows:

[0111] The first step: Define a heterogeneous UAV cooperative multi-task reallocation model.

[0112] The total flight distance and task completion time of the UAVs in the reallocation model are calculated according to the formulas described above, and are used as the objective function of the model for the following solution steps.

[0113] The second step: Solve the model.

[0114] Use the LeCMPSO algorithm to implement the solution of the heterogeneous UAV cooperative multi-task reallocation model. The specific implementation steps are as follows: See Figure 2 shown ( Figure 2 the italic font indicates the tasks participating in the reallocation), evaluate the task allocation sequence based on the initialization strategy of prior knowledge, find all the tasks being executed at time t = 500. At this time, the remaining task set in the combat sequence of UAV U8 is {T12(O)-T22(O)-T19(O)-T5(E)-T10(E)-T2(E)-T4(E)}. Add the tasks that have been completed and are being executed at t = 500 directly to the corresponding coding positions of the particle position vector, and add the UAVs that have completed the tasks under this task and their resource consumption to the corresponding coding positions. Update the remaining resource amount of the UAVs and the remaining resource amount required to complete the tasks, and use the constraint handling mechanism to reallocate the uncompleted tasks.

[0115] Particle update strategy based on historical information learning: See Figure 3As shown in the figure, first, take out the unfinished task T that contains the UAV 8 in the original task assignment sequence, remove the UAV U1 corresponding to this task T and its corresponding resource consumption C1, randomly select a UAV U2 of the same type as UAV U1 that meets the constraints, set its resource usage, and repeat this process until the task T is completed. After meeting the task completion requirements, insert the newly added UAV and its resource usage into the corresponding position in the particle encoding. Then randomly select an unfinished task, find the maximum insertable position range, and randomly insert this task within this range. Based on the above operations, the LeCMPSO algorithm is used for solving, and the algorithm parameters are set as follows: the number of sub-populations is 2, the initial population size is set to 300, the number of algorithm iterations is set to 300, and the size of the external archive set is half of the population size, which is 150. The non-dominated solution set shown in the following table can be obtained.

[0116] Table 4 Partial results after 300 iterations

[0117] Pareto point Total flight distance Task completion time 1 4774 3334 2 4784 3333 3 4790 3212 4 4829 3210 ... ... ...

[0118] As can be seen from Table 4, there is a certain conflict between the total flight distance of the UAVs and the task completion time, and it is difficult to minimize both at the same time. The LeCMPSO algorithm proposed in this embodiment shows good performance in both convergence and diversity, demonstrating the effectiveness and timeliness in solving the MCMTRP model.

[0119] It can be seen that after the sudden situation occurs in this embodiment, under the condition of meeting various complex constraints, the task assignment sequence of some UAVs is reallocated, and an efficient and reasonable allocation plan is quickly planned; in this embodiment, an MCMTRP model is established, with the UAV failure as the trigger scenario for reallocation, and the total flight distance of the UAVs and the task completion time as the optimization objectives, and a learning-based cooperative multi-objective particle swarm algorithm is proposed to solve the above MCMTRP model; at the same time, this embodiment also proposes an initialization strategy based on prior knowledge and a particle update strategy based on historical information learning. While improving the search efficiency of the algorithm, it can significantly reduce the solution time of the algorithm and achieve reasonable task allocation, improve the combat efficiency and save the UAV resource cost, thereby enhancing the combat effectiveness of the UAVs.

[0120] See Figure 4 As shown in the figure, an embodiment of the present application provides a cooperative multi-task reallocation device, including:

[0121] An acquisition unit, which is used to acquire the objective function and constraint conditions, the objective function includes the total flight distance function of the UAVs and the total task completion time function, and the constraint conditions include the non-reallocation constraint and the UAV availability constraint;

[0122] A creation unit, which is used to obtain an initial task assignment sequence obtained based on the KnCMPSO algorithm and create multiple empty task assignment sequences;

[0123] An initialization unit, which is used to perform initialization processing on multiple empty task assignment sequences based on the initial task assignment sequence and the constraint conditions to obtain multiple initialized task assignment sequences; and allocate the multiple initialized task assignment sequences to a distance task assignment sequence set for solving the total flight distance function of the unmanned aerial vehicles and a time task assignment sequence set for solving the total task completion time function;

[0124] An update unit, which is used to update each initialized task assignment sequence in the distance task assignment sequence set and the time task assignment sequence set respectively based on a structure learning strategy and a perturbation local search strategy to obtain an updated distance task assignment sequence set and an updated time task assignment sequence set;

[0125] An allocation unit, which is used to sort the task assignment sequences in the updated distance task assignment sequence set and the task assignment sequences in the updated time task assignment sequence set according to the Pareto dominance relationship to obtain a sorting queue; and select the task assignment sequences ranked in the top N positions from the sorting queue as the collaborative multi-task reallocation result, where N is a positive integer.

[0126] It can be seen that when encountering emergencies, this embodiment evaluates the heterogeneous unmanned aerial vehicle collaborative multi-task reallocation problem by using two objective functions, namely the total flight distance of the unmanned aerial vehicles and the total task completion time. Among them, based on the evaluation index of the total flight distance of the unmanned aerial vehicles, it can be ensured that the entire unmanned aerial vehicle system can complete the entire combat mission with the least amount of resources, and based on the evaluation index of the total task completion time function, it can be ensured that the number of tasks assigned to each unmanned aerial vehicle by the unmanned aerial vehicle system can maintain a certain balance, avoiding the problem that some unmanned aerial vehicles are assigned more tasks due to the performance differences of heterogeneous unmanned aerial vehicles; and this embodiment optimizes the task assignment sequence based on the prior knowledge obtained by the KnCMPSO algorithm, the structure learning strategy and the perturbation local search strategy, as well as the non-participation in the reallocation constraint and the unmanned aerial vehicle availability constraint, so as to realize the collaborative task reallocation of multi-objective heterogeneous unmanned aerial vehicles, which can effectively improve the search efficiency of the algorithm and reduce the solution time of the algorithm.

[0127] Further, the allocation unit is specifically used for:

[0128] Updating the task assignment sequence of the external archive set based on the perturbation local search strategy to obtain an updated external archive set;

[0129] Sort the task assignment sequences in the updated distance task assignment sequence set, the task assignment sequences in the updated time task assignment sequence set, and the task assignment sequences in the updated external archive set according to the Pareto dominance relationship to obtain a sorted queue.

[0130] Further, the allocation unit is specifically further configured to:

[0131] Randomly select a first task from the first task assignment sequence in the external archive set;

[0132] Remove the UAV sequence and the UAV resource consumption sequence corresponding to the first task from the first task assignment sequence;

[0133] Randomly select at least one first UAV from the set of UAVs of the same type as the first task and randomly generate a first resource consumption amount for executing the first task;

[0134] Update the first task assignment sequence based on the first UAV and the first resource consumption amount to obtain an updated external archive set.

[0135] Further, the allocation unit is specifically further configured to:

[0136] Update the initial task assignment sequence based on the UAV availability constraint to obtain an updated task assignment sequence;

[0137] Sort the task assignment sequences in the updated distance task assignment sequence set, the task assignment sequences in the updated time task assignment sequence set, the task assignment sequences in the updated external archive set, and the updated task assignment sequence according to the Pareto dominance relationship to obtain a sorted queue.

[0138] Further, the constraint condition further includes a completion time sequence constraint for different types of tasks, and the allocation unit is specifically further configured to:

[0139] When it is detected that there is a first unfinished task in the initial task assignment sequence and the first unfinished task is executed by a failed UAV, initialize an empty first task assignment sequence;

[0140] Update each task, the UAV information corresponding to each task, and the resource consumption information corresponding to each UAV in the initial task assignment sequence to the empty first task assignment sequence to obtain a second task assignment sequence;

[0141] Remove the failed UAV and the resource consumption information corresponding to the failed UAV from the second task assignment sequence, and update the remaining resource amount required to execute the first unfinished task to obtain a third task assignment sequence;

[0142] Randomly select at least one second unmanned aerial vehicle (UAV) from the set of UAVs of the same type as the first unfinished task based on the remaining resource amount, and randomly generate a second resource consumption amount for executing the first unfinished task;

[0143] Determine the insertable position range of the first unfinished task from the third task assignment sequence based on the completion time sequence constraints of different types of tasks;

[0144] Insert the first unfinished task, the corresponding second UAV, and the second resource consumption amount into any position within the insertable position range of the first unfinished task in the third task assignment sequence to obtain an updated task assignment sequence.

[0145] Further, the initialization unit is specifically configured to:

[0146] Determine the completed tasks, the tasks being executed, and the unexecuted tasks in the initial task assignment sequence at the t-th moment based on the non-participation in reallocation constraints;

[0147] Update the completed tasks, the UAV information, and the resource consumption amount corresponding to the execution of the completed tasks to the corresponding positions in the first empty task assignment sequence to obtain a first processed task assignment sequence;

[0148] Update the completed tasks among the tasks being executed, the UAV information, and the resource consumption amount corresponding to the execution of the completed tasks among the tasks being executed to the corresponding positions in the first processed task assignment sequence to obtain a second processed task assignment sequence;

[0149] Allocate UAVs and resource consumption amounts for the unfinished tasks among the tasks being executed based on the constraint conditions, and insert the unfinished tasks among the tasks being executed and the corresponding UAVs and resource consumption amounts into the second processed task assignment sequence to obtain a third processed task assignment sequence;

[0150] Allocate UAVs and resource consumption amounts for the unexecuted tasks based on the constraint conditions, and insert the unexecuted tasks and the corresponding UAVs and resource consumption amounts into the third processed task assignment sequence to obtain an initialized task assignment sequence.

[0151] Further, the total UAV flight distance function f1 and the total task completion time function f2 are respectively:

[0152]

[0153] In the formula, S represents the number of reconnaissance UAVs, A represents the number of attack UAVs, m represents the m-th task, represents the task sequence of UAV U i of, denotes the drone U i The flight distance from the execution of the m-th task to the execution of the (m + 1)-th task, N denotes the total number of tasks, and Evaluate denotes the evaluation task.

[0154] It should be noted that those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the above-described device and each unit can refer to the corresponding processes in the foregoing embodiments of the collaborative multi-task reallocation method, and will not be elaborated herein.

[0155] The device provided in the above embodiment can be implemented in the form of a computer program, and this computer program can run on the Figure 5 collaborative multi-task reallocation device shown as follows.

[0156] The embodiment of the present application further provides a collaborative multi-task reallocation device, including: a memory, a processor, and a network interface connected through a system bus. At least one instruction is stored in the memory, and at least one instruction is loaded and executed by the processor to implement all or part of the steps of the foregoing collaborative multi-task reallocation method.

[0157] Among them, the network interface is used for network communication, such as sending the allocated tasks, etc. Those skilled in the art can understand that Figure 5 the structure shown in is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0158] The processor may be a CPU, or may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or the processor may also be any conventional processor, etc. The processor is the control center of the computer device, and connects various parts of the entire computer device through various interfaces and lines.

[0159] The memory can be used to store computer programs and / or modules. By running or executing the computer programs and / or modules stored in the memory, and invoking the data stored in the memory, the processor realizes various functions of the computer device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store the operating system, application programs required for at least one function (such as video playback function, image playback function, etc.); the data storage area can store the data created according to the use of the mobile phone (such as video data, image data, etc.). In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one magnetic disk storage device, flash memory device or other volatile solid-state storage devices.

[0160] An embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, all or part of the steps of the foregoing cooperative multi-task reallocation method are realized.

[0161] To implement all or part of the foregoing processes in the embodiments of the present application, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above various methods can be realized. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device that can carry computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0162] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, servers, or computer program products. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories and optical memories, etc.) containing computer-usable program code.

[0163] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices generate means for implementing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0164] It should be noted that in this article, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article, or system. Without further limitations, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article, or system including that element.

[0165] The above are only the specific embodiments of the present application, enabling those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features claimed herein.

Claims

1. A collaborative multi-task reallocation method, characterized in that Including the following steps: Obtain the objective function and constraints, where the objective function includes the total flight distance function of the unmanned aerial vehicle (UAV) and the total task completion time function, and the constraints include the non-participation in reallocation constraint and the UAV availability constraint; Obtain the initial task assignment sequence obtained based on the KnCMPSO algorithm and create multiple empty task assignment sequences; Perform initialization processing on the multiple empty task assignment sequences based on the initial task assignment sequence and the constraints to obtain multiple initialized task assignment sequences; Assign the multiple initialized task assignment sequences to a distance task assignment sequence set for solving the total flight distance function of the UAV and a time task assignment sequence set for solving the total task completion time function; Update each of the initialized task assignment sequences in the distance task assignment sequence set and the time task assignment sequence set based on the structure learning strategy and the perturbation local search strategy to obtain an updated distance task assignment sequence set and an updated time task assignment sequence set; Sort the task assignment sequences in the updated distance task assignment sequence set and the task assignment sequences in the updated time task assignment sequence set according to the Pareto dominance relationship to obtain a sorted queue; select the top N task assignment sequences from the sorted queue as the collaborative multi-task reallocation result, where N is a positive integer.

2. The collaborative multi-task reallocation method according to claim 1, wherein, The sorting the task assignment sequences in the updated distance task assignment sequence set and the task assignment sequences in the updated time task assignment sequence set according to the Pareto dominance relationship to obtain a sorted queue specifically includes: Update the task assignment sequences in the external archive set based on the perturbation local search strategy to obtain an updated external archive set; Sort the task assignment sequences in the updated distance task assignment sequence set, the task assignment sequences in the updated time task assignment sequence set, and the task assignment sequences in the updated external archive set according to the Pareto dominance relationship to obtain a sorted queue.

3. The collaborative multi-task reallocation method according to claim 2, wherein The updating the task assignment sequences in the external archive set based on the perturbation local search strategy to obtain an updated external archive set specifically includes: Randomly select a first task from the first task assignment sequence in the external archive set; Remove the UAV sequence and the UAV resource consumption sequence corresponding to the first task from the first task assignment sequence; Randomly select at least one first UAV from the UAV set of the same category as the first task and randomly generate a first resource consumption amount for executing the first task; Update the first task assignment sequence based on the first UAV and the first resource consumption amount to obtain an updated external archive set.

4. The collaborative multi-task reallocation method according to claim 2, characterized in that, The sorting the task assignment sequences in the updated distance task assignment sequence set, the task assignment sequences in the updated time task assignment sequence set, and the task assignment sequences in the updated external archive set according to the Pareto dominance relationship to obtain a sorted queue specifically includes: Update the initial task assignment sequence based on the UAV availability constraint to obtain an updated task assignment sequence; Sort the task assignment sequences in the updated distance task assignment sequence set, the task assignment sequences in the updated time task assignment sequence set, the task assignment sequences in the updated external archive set, and the updated task assignment sequences according to the Pareto dominance relationship to obtain a sorted queue.

5. The collaborative multi-task reallocation method according to claim 4, wherein The constraint conditions further include the completion time sequence constraints for different types of tasks. Updating the initial task assignment sequence based on the UAV availability constraints to obtain an updated task assignment sequence specifically includes: When it is detected that there is a first unfinished task in the initial task assignment sequence and the first unfinished task is executed by a failed UAV, initialize an empty first task assignment sequence; Update each task, the UAV information corresponding to the execution of each task, and the resource consumption information corresponding to each UAV in the initial task assignment sequence to the empty first task assignment sequence to obtain a second task assignment sequence; Remove the failed UAV and the resource consumption information corresponding to the failed UAV from the second task assignment sequence, and update the remaining resource amount required to execute the first unfinished task to obtain a third task assignment sequence; Randomly select at least one second UAV from the set of UAVs of the same type as the first unfinished task based on the remaining resource amount and randomly generate a second resource consumption amount for executing the first unfinished task; Determine the insertable position range of the first unfinished task from the third task assignment sequence based on the completion time sequence constraints for different types of tasks; Insert the first unfinished task, the second UAV corresponding to the first unfinished task, and the second resource consumption amount into any position within the insertable position range of the first unfinished task in the third task assignment sequence to obtain an updated task assignment sequence.

6. The collaborative multi-task reallocation method according to claim 1, characterized in that, Initializing multiple empty task assignment sequences based on the initial task assignment sequence and the constraint conditions to obtain multiple initialized task assignment sequences specifically includes: Determine the completed tasks, tasks being executed, and unexecuted tasks in the initial task assignment sequence at the t-th moment based on the non-participation in reallocation constraints; Update the completed tasks, the UAV information corresponding to the execution of the completed tasks, and the resource consumption amount to the corresponding positions in the first empty task assignment sequence to obtain a first processed task assignment sequence; Update the completed tasks among the tasks being executed and the UAV information and resource consumption amount for executing the completed tasks among the tasks being executed to the corresponding positions in the first processed task assignment sequence to obtain a second processed task assignment sequence; Allocate UAVs and resource consumption amounts for the unfinished tasks among the tasks being executed based on the constraint conditions, and insert the unfinished tasks among the tasks being executed and the UAVs and resource consumption amounts corresponding to the unfinished tasks among the tasks being executed into the second processed task assignment sequence to obtain a third processed task assignment sequence; Allocate drones and resource consumption for unexecuted tasks based on the constraints, and insert the unexecuted tasks, as well as the corresponding drones and resource consumption, into the third processed task allocation sequence to obtain the initialized task allocation sequence.

7. The collaborative multi-task reallocation method according to claim 1, wherein The total drone flight distance function f1 and the total task completion time function f2 are respectively: Wherein, S represents the number of reconnaissance UAVs, A represents the number of attack UAVs, m represents the m-th mission, represents the UAV U i 's mission sequence, represents the UAV U i 's flight distance from the execution of the m-th mission to the execution of the m+1-th mission, N represents the total number of missions, Evaluate represents the evaluation mission, represents the time taken to complete the j-th target.

8. A collaborative multi-task reallocation device, characterized in that, Including: An acquisition unit, which is used to acquire the objective function and the constraints. The objective function includes the total drone flight distance function and the total task completion time function, and the constraints include the non-participation in reallocation constraint and the drone availability constraint; A creation unit, which is used to acquire the initial task allocation sequence obtained based on the KnCMPSO algorithm and create multiple empty task allocation sequences; An initialization unit, which is used to perform initialization processing on the multiple empty task allocation sequences based on the initial task allocation sequence and the constraints to obtain multiple initialized task allocation sequences; allocate the multiple initialized task allocation sequences to the distance task allocation sequence set for solving the total drone flight distance function and the time task allocation sequence set for solving the total task completion time function; An update unit, which is used to update each of the initialized task allocation sequences in the distance task allocation sequence set and the time task allocation sequence set based on the structure learning strategy and the perturbation local search strategy respectively to obtain the updated distance task allocation sequence set and the updated time task allocation sequence set; An allocation unit, which is used to sort the task allocation sequences in the updated distance task allocation sequence set and the task allocation sequences in the updated time task allocation sequence set according to the Pareto dominance relationship to obtain a sorted queue; select the top N task allocation sequences from the sorted queue as the collaborative multi-task reallocation result, where N is a positive integer.

9. A collaborative multi-task reallocation device, characterized in that, Including: A memory and a processor. At least one instruction is stored in the memory, and the at least one instruction is loaded and executed by the processor to implement the collaborative multi-task reallocation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer storage medium stores a computer program, which, when executed by the processor, implements the collaborative multi-task reallocation method according to any one of claims 1 to 7.

Images