A method and apparatus for UAV task allocation with adaptive matching mode switching

Abstract

This invention proposes an adaptive matching mode switching method and device for UAV task allocation. The matching mode switching mainly includes two parts: (1) switching between static and dynamic value calculation functions, based on historical task information; and (2) switching between task type matching, based on task analysis and task allocation type, switching to the corresponding fast task allocation algorithm. This invention provides a method for switching between static and dynamic value functions for task allocation, used for constructing a matching value matrix. It provides methods for defining static and dynamic value functions, used for determining the value of task allocation schemes. It achieves accurate matching of tasks and UAVs under different circumstances. This invention provides rules for selecting task allocation algorithms when the number of UAVs and tasks differs, and determines the allocation relationship between UAVs and tasks based on bipartite graph allocation algorithm and consistent auction algorithm, achieving rapid solution of allocation schemes.

Description

Technical Field

[0001] This invention belongs to the field of UAV swarm combat mission allocation, and involves the design of static value function, dynamic value function and matching algorithm selection rules for UAV mission allocation. In particular, it relates to a UAV mission allocation method and device with adaptive matching mode switching. Background Technology

[0002] With the development of swarm intelligence technology, the scale of drone swarms has expanded from a few to thousands.

[0003] When carrying out missions in a highly dynamic operational space, there is a contradiction between solution time and matching accuracy.

[0004] Therefore, it is necessary to select a suitable matching mode to achieve accurate and rapid matching. Through reasonable resource allocation, operational efficiency can be maximized. Summary of the Invention

[0005] This invention proposes a method and apparatus for adaptive matching mode switching of UAV mission allocation, which realizes adaptive switching of UAV mission matching mode in highly dynamic combat environments.

[0006] The first aspect of this invention provides a method for drone task allocation with adaptive matching mode switching, comprising:

[0007] Upon receiving the task set M, determine whether there is historical task execution data of UAV swarm A performing the same type of task;

[0008] If so, then the state value function is trained using reinforcement learning based on historical task execution data. The dynamic reward matrix of the task set M executed by the UAV swarm A is obtained based on the trained state value function. Based on the dynamic reward matrix, the task set M is assigned to the UAV swarm A using a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus coalition auction algorithm.

[0009] If not, calculate the static reward function of the drone swarm A executing the task set M. Based on the task type of each task in the task set M and the sufficiency of drone resources, use the bipartite graph allocation algorithm, the consensus auction algorithm, or the consensus federation auction algorithm to obtain the allocation scheme with the maximum value of the static reward function as the drone task allocation scheme. The task types include: point tasks and area tasks.

[0010] Optional, static payoff function

[0011] In the formula, R am The reward for drone a to complete task m. v importance Task importance; J = {J1, J2, ..., J...}N To fulfill the required capabilities of unmanned aerial vehicles (UAVs), at a minimum, they must include reconnaissance capabilities, attack capabilities, endurance capabilities, production capabilities, electronic jamming capabilities, and maneuverability. j The capability weighting coefficient satisfies a j The numerical representation of the capability j of drone a; m j This represents the numerical expression of the capability j required to complete task m; θ is the capability verification factor, which is the factor used to verify the capability j if it is required to satisfy a. j ≥m j Only drone a can perform task m, then when a j ≥m j When θ = 1, otherwise θ = 0;

[0012] X am Whether drone a completes task m is a decision variable, represented as:

[0013] Optionally, based on the task type of each task in the task set M and the sufficiency of UAV resources, a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus federation auction algorithm is used to obtain the allocation scheme that maximizes the static reward function as the UAV task allocation scheme, including:

[0014] For task set M, when there are only point tasks in task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme.

[0015] For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme.

[0016] For a task set M, when there are surface tasks in the task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme.

[0017] Optionally, the state-value function can be trained using reinforcement learning methods based on historical task execution data, including:

[0018] Based on historical mission execution data, obtain the mission completion status of a certain drone performing a certain mission in a time series, as well as the evaluation of the completion status, and establish a reinforcement learning problem model.

[0019] Based on the reinforcement learning problem model, obtain the state-value function;

[0020] The state-value function is trained using reinforcement learning methods;

[0021] Once the state value function is trained, the matching value V for assigning a task m to a drone a can be calculated.πd (S am ), where S am The state space represents the attributes of drone a and task m. By traversing the task set M and drone swarm A, the dynamic reward matrix of drone swarm A executing task set M can be obtained.

[0022] The reinforcement learning problem model includes: state space, action space, state transition, and immediate reward; the state space includes: current time segment information, UAV position attribute information, payload attribute information, weapon attribute information and platform capability information, mission spatial geographic information, mission type information, and mission capability requirement information; the platform capability information includes: reconnaissance capability, attack capability, endurance capability, generation capability, electronic jamming capability, and maneuverability capability;

[0023] The action space includes: when a drone is assigned a task and when a drone is not assigned a task in the current time slice;

[0024] Instant rewards γ is the discount factor, taking a decimal value between 0 and 1; T is the discretized time series T = {t1,...,t...} all}, where t∈T is any time point in the time series, and is a discrete quantity. R evaluate T is a numerical evaluation of a certain drone's completion of a certain task in the allocation scheme. execute The mission execution time includes the time it takes for the UAV to travel from its current location to the mission location, as well as the time it takes to complete the mission. Numerical evaluation includes: mission execution time, resource utilization, energy consumption, damage effect, and mission priority.

[0025] Optionally, based on the dynamic payoff matrix, a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus federation auction algorithm are used to allocate the task set M to the drone swarm A, including:

[0026] Based on the dynamic revenue matrix, for a task set M, when there are only point tasks in the task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme.

[0027] For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme.

[0028] For task set M, when there are surface tasks in task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme.

[0029] In bipartite graph assignment methods, edge reference weights and in consensus-based consortium auction algorithms, the drone's bid value w am =Ram +β(V πd (S am )); where R am Here, β is the static return function, and β is the weighting factor.

[0030] A second aspect of the present invention provides a drone task allocation device with adaptive matching mode switching, comprising:

[0031] The judgment module is used to determine, upon receiving the task set M, whether there is historical task execution data of the UAV swarm A performing the same type of task;

[0032] The first allocation module is used to train a state value function based on historical task execution data of the same type of task performed by UAV swarm A, and to obtain the dynamic reward matrix of task set M performed by UAV swarm A based on the trained state value function. Based on the dynamic reward matrix, the task set M is allocated to UAV swarm A using a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus coalition auction algorithm.

[0033] The second allocation module is used to calculate the static revenue function of the task set M executed by UAV swarm A when there is no historical task execution data of the same type of task in UAV swarm A. Based on the task type of each task in task set M and the sufficiency of UAV resources, the module uses a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus coalition auction algorithm to obtain the allocation scheme with the maximum value of the static revenue function as the UAV task allocation scheme. The task types include point tasks and area tasks.

[0034] Optional, static payoff function

[0035] In the formula, R am The reward for drone a to complete task m. v importance Task importance; J = {J1, J2, ..., J...} N To fulfill the required capabilities of unmanned aerial vehicles (UAVs), at a minimum, they must include reconnaissance capabilities, attack capabilities, endurance capabilities, production capabilities, electronic jamming capabilities, and maneuverability. j The capability weighting coefficient satisfies a j The numerical representation of the capability j of drone a; m j This represents the numerical expression of the capability j required to complete task m; θ is the capability verification factor, which is the factor used to verify the capability j if it is required to satisfy a. j ≥m j Only drone a can perform task m, then when a j ≥m j When θ = 1, otherwise θ = 0;

[0036] X am Whether drone a completes task m is a decision variable, represented as:

[0037] Optional, the second allocation module is specifically used for:

[0038] For task set M, when there are only point tasks in task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme.

[0039] For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme.

[0040] For a task set M, when there are surface tasks in the task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme.

[0041] Optional, the first allocation module is specifically used for:

[0042] Based on historical mission execution data, obtain the mission completion status of a certain drone performing a certain mission in a time series, as well as the evaluation of the completion status, and establish a reinforcement learning problem model.

[0043] Based on the reinforcement learning problem model, obtain the state-value function;

[0044] The state-value function is trained using reinforcement learning methods;

[0045] Once the state value function is trained, the matching value V for assigning a task m to a drone a can be calculated. πd (S am ), where S am The state space represents the attributes of drone a and task m. By traversing the task set M and drone swarm A, the dynamic reward matrix of drone swarm A executing task set M can be obtained.

[0046] The reinforcement learning problem model includes: state space, action space, state transition, and immediate reward; the state space includes: current time segment information, UAV position attribute information, payload attribute information, weapon attribute information and platform capability information, mission spatial geographic information, mission type information, and mission capability requirement information; the platform capability information includes: reconnaissance capability, attack capability, endurance capability, generation capability, electronic jamming capability, and maneuverability capability;

[0047] The action space includes: when a drone is assigned a task and when a drone is not assigned a task in the current time slice;

[0048] Instant rewards γ is the discount factor, taking a decimal value between 0 and 1; T is the discretized time series T = {t1,...,t...} all}, where t∈T is any time point in the time series, and is a discrete quantity. R evaluate T is a numerical evaluation of a certain drone's completion of a certain task in the allocation scheme. execute The mission execution time includes the time it takes for the UAV to travel from its current location to the mission location, as well as the time it takes to complete the mission. Numerical evaluation includes: mission execution time, resource utilization, energy consumption, damage effect, and mission priority.

[0049] Optional, the first allocation module is specifically used for:

[0050] Based on the dynamic revenue matrix, for a task set M, when there are only point tasks in the task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme.

[0051] For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme.

[0052] For task set M, when there are surface tasks in task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme.

[0053] In bipartite graph assignment methods, edge reference weights and in consensus-based consortium auction algorithms, the drone's bid value w am =R am +β(V πd (S am )); where R am Here, β is the static return function, and β is the weighting factor.

[0054] This invention proposes an adaptive matching mode switching method and device for UAV task allocation. The matching mode switching mainly includes two parts: (1) switching between static and dynamic value calculation functions, based on historical task information; and (2) switching between task type matching, based on task analysis and task allocation type, switching to the corresponding fast task allocation algorithm. This invention provides a method for switching between static and dynamic value functions for task allocation, used for constructing a matching value matrix. It provides methods for defining static and dynamic value functions, used for determining the value of task allocation schemes. It achieves accurate matching of tasks and UAVs under different circumstances. This invention provides rules for selecting task allocation algorithms when the number of UAVs and tasks differs, and determines the allocation relationship between UAVs and tasks based on bipartite graph allocation algorithm and consistent auction algorithm, achieving rapid solution of allocation schemes. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the construction of the matching value matrix. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] The features and illustrative embodiments of various aspects of the present invention will now be described in detail. Numerous specific details are set forth in the following detailed description to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention may be practiced without requiring some of these specific details. The following description of embodiments is merely intended to provide a better understanding of the invention by illustrating examples of the invention. The invention is by no means limited to any specific setups and methods set forth below, but covers any improvements, substitutions, and modifications to structures, methods, and devices without departing from the spirit of the invention. Well-known structures and techniques are not shown in the drawings and the following description to avoid unnecessarily obscuring the invention.

[0058] It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other, and the various embodiments can be referenced and cited in each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0059] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0060] The task allocation method proposed in this invention, which switches between matching modes, adjusts the resource matching mode used under different task conditions based on task type and battlefield resource distribution, achieving rapid matching of resources and task execution units. When there is no historical task completion data, a static reward function is used, and different optimization algorithms are selected to solve the matching scheme for different resource and task correspondences. When there is some historical task completion data, a dynamic value function is learned based on historical task completion data. After constructing the dynamic value matrix, different optimization algorithms are selected to solve the matching scheme for different resource and task correspondences. When using a bipartite graph solution algorithm, the edge weight calculation method of the existing bipartite allocation algorithm is improved.

[0061] The present invention has the following main technical features:

[0062] (1) A matching value matrix between execution units and tasks is generated by combining static revenue functions and dynamic reward functions.

[0063] The task matching value matrix describes the degree of matching between a task and the drone executing that task. It comprehensively considers the characteristics of various task types and the attributes of available drones, such as the drone's current position, maximum speed, endurance, and relative position to the task. Key indicators are selected to measure the effectiveness of the matching and scheduling scheme, such as task execution time, resource utilization, energy consumption, damage effect, and task priority. These indicators are transformed into a comprehensive evaluation function through a mathematical model to quantify the overall effectiveness of different schemes, thereby supporting the intelligent matching algorithm in evaluating and comparing various matching and scheduling schemes. Based on task characteristics and priorities, it provides objective quantitative references for matching and scheduling in different scenarios to optimize resource allocation and task arrangement decisions.

[0064] (2) Assign tasks based on the established matching value matrix.

[0065] For task set M, when there are only point tasks in task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme.

[0066] For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme.

[0067] For a task set M, when there are surface tasks in the task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme.

[0068] For example, the static revenue function definition method includes the following:

[0069] Assuming the number of drones is A and the task set is M, the static revenue function is:

[0070]

[0071] In the formula, R am The static value function for drone a to complete task m:

[0072]

[0073] In the formula, v importance Task importance; J = {J1, J2, ..., J...} N To fulfill the required capabilities of unmanned aerial vehicles (UAVs), the following categories are included: reconnaissance capabilities, attack capabilities, endurance capabilities, production capabilities, electronic jamming capabilities, and maneuverability. jThe capability weighting coefficient satisfies a j The numerical representation of the capability j of drone a; m j This represents the numerical expression of the capability j required to complete task m; θ is the capability verification factor, which is the factor used to verify the capability j if it is required to satisfy a. j ≥m j Only drone a can perform task m, then when a j ≥m j When θ = 1, otherwise θ = 0;

[0074] X am Whether drone a completes task m is a decision variable, represented as:

[0075]

[0076] For example, a method for defining a dynamic value function includes the following:

[0077] Given the existing successful task execution history, the task allocation problem is defined as a sequential decision problem, that is, in a relatively long time period T, orders are allocated in rounds, and the goal is to maximize the total revenue within that time period T.

[0078] The drone task allocation problem can be modeled as a Markov Decision Process (MDP) problem, which includes four basic elements: state space, action space, reward function, and state transition matrix. Solving the MDP problem means determining which task a drone should be assigned to in a given state to maximize its reward.

[0079] In the drone task allocation problem, the MDP is defined as follows:

[0080] 1) State Space: Numerical description of UAV state and mission state. The Cartesian product of UAV state and mission state is the set of all states S. The state includes: 1) Current time segment information; 2) UAV position attribute information, payload attribute information, weapon attribute information and platform capability information (including reconnaissance capability, attack capability, endurance capability, generation capability, electronic jamming capability, maneuverability, etc.); 3) Mission spatial geographic information, mission type information and mission capability requirement information.

[0081] 2) Action Space: There are two behaviors in the task allocation problem. The first is that drone a is assigned a task m, then the drone executes the task and completes the task after a period of time. The second behavior is that drone a is not assigned any task during the time slice, it will be idle for a time slice and enter the next round of task allocation.

[0082] 3) The reward function expression is as follows:

[0083]

[0084] Where γ is the discount factor, taking a decimal value between 0 and 1; T is the discretized time series T = {t1,...,t} all}, where t∈T is any time point in the time series, and is a discrete quantity. Δt is the time discretization threshold; R evaluate It is a numerical evaluation of a certain UAV's completion of a certain task in the allocation scheme (including task execution time, resource utilization, energy consumption, damage effect, and task priority), T execute The mission execution time includes the time it takes for the drone to travel from its current location to the mission location, and the time it takes to complete the mission.

[0085] 4) State transition: State transition occurs along with behavior. After UAV a executes task m, the UAV's own position state changes and the platform's capabilities change.

[0086] 5) Solve for the dynamic value function V πd (s) is represented using a multi-layer neural network, and the function parameters are updated using a TD (Temporal-Dierence) based update algorithm, with the update formula as follows:

[0087]

[0088] Where s′ is the next state, T execute The task execution time includes the drone's travel time from its current location to the task location, and the time required to complete the task. For unassigned actions, T... execute =1, and R γ =0.

[0089] For example, the construction and solution of the matching value matrix includes the following:

[0090] This paper models a drone task allocation scenario. The task allocation problem assumes an existing drone swarm A that needs to complete a set of tasks M and W. am Let represent the fitness edge weights of drone a for task m. Task allocation aims to determine the allocation relationship between drones and tasks, maximizing the overall fitness of the task allocation results in each round.

[0091] Figure 1 A schematic diagram for constructing a matching value matrix.

[0092] The connection between drones and tasks represents the appropriate value for different drones performing different tasks. By calculating static and dynamic value gains, a matching value matrix can be obtained.

[0093]

[0094] Corrected edge reference weights and output values ​​w am =R am +β(V πd (S am )); where R am Here, β is the static return function, and β is the weighting factor.

[0095] The above description is merely a further embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope disclosed in the present invention, based on the technical solution and concept of the present invention, shall fall within the scope of protection of the present invention.

Claims

1. A method for UAV task allocation with adaptive matching mode switching, characterized in that, include: Upon receiving the task set M, determine whether there is historical task execution data of UAV swarm A performing the same type of task; If so, then the state value function is trained using reinforcement learning based on historical task execution data. The dynamic reward matrix of the task set M executed by the UAV swarm A is obtained based on the trained state value function. Based on the dynamic reward matrix, the task set M is assigned to the UAV swarm A using a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus coalition auction algorithm. If not, calculate the static revenue function of the drone swarm A executing task set M. Based on the task type of each task in task set M and the sufficiency of drone resources, use the bipartite graph allocation algorithm, consensus auction algorithm, or consensus coalition auction algorithm to obtain the allocation scheme with the maximum value of the static revenue function as the drone task allocation scheme. Task types include: point tasks and area tasks; Static return function ; In the formula, R am For drones a Complete the task m The benefits, ; v importance Assigning importance to the task; To fulfill the required capabilities of unmanned aerial vehicles (UAVs), at least the following capabilities must be included: reconnaissance capability, attack capability, endurance capability, production capability, electronic jamming capability, and maneuverability. λ j The capability weighting coefficient satisfies ; a j Indicates drone a The numerical representation of capability j; m j This represents the capability required to complete task m. j The numerical representation of θ; θ is the capability verification factor, for a certain capability j If the requirements are met Only drone a can perform task m, then when When θ=1, θ=0 otherwise; X am Indicates drone a Has the task been completed? m， For decision variables, it is represented as: 。 2. The UAV task allocation method with adaptive matching mode switching according to claim 1, characterized in that, Based on the task type of each task in the task set M and the sufficiency of UAV resources, a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus federation auction algorithm is used to obtain the allocation scheme that maximizes the static reward function as the UAV task allocation scheme, including: For task set M, when there are only point tasks in task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme. For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme. For a task set M, when there are surface tasks in the task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme.

3. The UAV task allocation method with adaptive matching mode switching according to claim 1, characterized in that, The state-value function is trained using reinforcement learning based on historical task execution data, including: Based on historical mission execution data, obtain the mission completion status of a certain drone performing a certain mission in a time series, as well as the evaluation of the completion status, and establish a reinforcement learning problem model. Based on the reinforcement learning problem model, obtain the state-value function; The state-value function is trained using reinforcement learning methods; Once the state-value function is trained, the matching value of assigning a task m to a drone a can be calculated. ,in S am The state space represents the attributes of drone a and task m. By traversing the task set M and drone swarm A, the dynamic reward matrix of drone swarm A executing task set M can be obtained. The reinforcement learning problem model includes: state space, action space, state transition, and immediate reward; the state space includes: current time segment information, UAV position attribute information, payload attribute information, weapon attribute information and platform capability information, mission spatial geographic information, mission type information, and mission capability requirement information; the platform capability information includes: reconnaissance capability, attack capability, endurance capability, generation capability, electronic jamming capability, and maneuverability capability; The action space includes: when a drone is assigned a task and when a drone is not assigned a task in the current time slice; Instant rewards ; γ This is the discount factor, which takes the value of a decimal between 0 and 1. T For the entire time series after discretization , Let be any point in the time series, and be a discrete quantity. , R evaluate It is a numerical evaluation of a certain drone's completion of a certain task in the allocation scheme. T execute The mission execution time includes the time it takes for the UAV to travel from its current location to the mission location, as well as the time it takes to complete the mission. Numerical evaluation includes: mission execution time, resource utilization, energy consumption, damage effect, and mission priority.

4. The UAV task allocation method with adaptive matching mode switching according to claim 3, characterized in that, Based on the dynamic payoff matrix, the task set M is assigned to the drone swarm A using a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus federation auction algorithm, including: Based on the dynamic revenue matrix, for a task set M, when there are only point tasks in the task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme. For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme. For task set M, when there are surface tasks in task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme. The reference weights of edges in bipartite graph assignment methods and the value of drone bids in consensus-based consortium auction algorithms. Where α is the weighting factor.

5. A UAV task allocation device with adaptive matching mode switching, characterized in that, include: The judgment module is used to determine, upon receiving the task set M, whether there is historical task execution data of the UAV swarm A performing the same type of task; The first allocation module is used to train a state value function based on historical task execution data of the same type of task performed by UAV swarm A, and to obtain the dynamic reward matrix of task set M performed by UAV swarm A based on the trained state value function. Based on the dynamic reward matrix, the task set M is allocated to UAV swarm A using a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus coalition auction algorithm. The second allocation module is used to calculate the static revenue function of the task set M executed by UAV swarm A when there is no historical task execution data of the same type of task executed by UAV swarm A. Based on the task type of each task in task set M and the sufficiency of UAV resources, the module uses a bipartite graph allocation algorithm, a consensus auction algorithm, or a consensus coalition auction algorithm to obtain the allocation scheme with the maximum value of the static revenue function as the UAV task allocation scheme. Task types include: point tasks and area tasks; Static return function ; R am For drones a Complete the task m The benefits, ; v importance Assigning importance to the task; To fulfill the required capabilities of unmanned aerial vehicles (UAVs), at least the following capabilities must be included: reconnaissance capability, attack capability, endurance capability, production capability, electronic jamming capability, and maneuverability. λ j The capability weighting coefficient satisfies ; a j Indicates drone a The numerical representation of capability j; m j This represents the capability required to complete task m. j The numerical representation of θ; θ is the capability verification factor, for a certain capability j If the requirements are met Only drone a can perform task m, then when When θ=1, θ=0 otherwise; X am Indicates drone a Has the task been completed? m， For decision variables, it is represented as: 。 6. The UAV task allocation device with adaptive matching mode switching according to claim 5, characterized in that, The second allocation module is specifically used for: For task set M, when there are only point tasks in task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme. For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme. For a task set M, when there are surface tasks in the task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme.

7. The UAV task allocation device with adaptive matching mode switching according to claim 5, characterized in that, The first allocation module is specifically used for: Based on historical mission execution data, obtain the mission completion status of a certain drone performing a certain mission in a time series, as well as the evaluation of the completion status, and establish a reinforcement learning problem model. Based on the reinforcement learning problem model, obtain the state-value function; The state-value function is trained using reinforcement learning methods; Once the state-value function is trained, the matching value of assigning a task m to a drone a can be calculated. ,in S am The state space represents the attributes of drone a and task m. By traversing the task set M and drone swarm A, the dynamic reward matrix of drone swarm A executing task set M can be obtained. The reinforcement learning problem model includes: state space, action space, state transition, and immediate reward; the state space includes: current time segment information, UAV position attribute information, payload attribute information, weapon attribute information and platform capability information, mission spatial geographic information, mission type information, and mission capability requirement information; the platform capability information includes: reconnaissance capability, attack capability, endurance capability, generation capability, electronic jamming capability, and maneuverability capability; The action space includes: when a drone is assigned a task and when a drone is not assigned a task in the current time slice; Instant rewards ; γ This is the discount factor, which takes the value of a decimal between 0 and 1. T For the entire time series after discretization , Let be any point in the time series, and be a discrete quantity. , R evaluate It is a numerical evaluation of a certain drone's completion of a certain task in the allocation scheme. T execute The mission execution time includes the time it takes for the UAV to travel from its current location to the mission location, as well as the time it takes to complete the mission. Numerical evaluation includes: mission execution time, resource utilization, energy consumption, damage effect, and mission priority.

8. The UAV task allocation device with adaptive matching mode switching according to claim 7, characterized in that, The first allocation module is specifically used for: Based on the dynamic revenue matrix, for a task set M, when there are only point tasks in the task set M and the number of drones is greater than the number of tasks, the bipartite graph allocation algorithm is called to solve the task allocation scheme. For task set M, when there are only point tasks in task set M and the number of drones is less than the number of tasks, the consensus auction algorithm is called to solve the task allocation scheme. For task set M, when there are surface tasks in task set M, the consensus consortium auction algorithm is called to solve the task allocation scheme. The reference weights of edges in bipartite graph assignment methods and the value of drone bids in consensus-based consortium auction algorithms. ;in, β This is the weighting factor.

Images