Multi-UAV dynamic task allocation method based on improved contract network algorithm

Abstract

The present invention discloses a method for dynamic task allocation of multiple UAVs based on an improved contract network algorithm. The method comprehensively considers the influence of UAV execution cost, payload cost and navigation cost on UAV performance to establish an objective function, analyzes the shortcomings of the traditional contract network algorithm in solving this problem, sets task priority, narrows the bidding scope, reduces the number of bids, and increases supervision and management to improve the contract network algorithm in four aspects, so as to reduce the communication volume during UAV task allocation and improve the system performance; thereby effectively solving the problem of dynamic task allocation of multiple UAVs.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) task allocation, and in particular to a multi-UAV dynamic task allocation method based on an improved contract network algorithm. Background Art

[0002] Drones (UAVs) have garnered widespread attention in the military due to their low cost, small size, high flexibility, excellent concealment, and strong adaptability. Current UAV research focuses primarily on collaborative combat capabilities, with UAV task allocation technology being a hot topic in recent years. Compared to static task allocation, dynamic UAV task allocation is more adaptable to ever-changing conditions, making research on this issue particularly relevant. The key challenges in dynamic UAV task allocation are reducing system communication while improving system performance and enabling better response to emergencies.

[0003] Currently, UAV task allocation includes single-UAV and multi-UAV collaborative task allocation. However, single UAVs are restricted in terms of type, communication range, resource operator, combat efficiency, etc., and the types of tasks they can perform are also limited. Therefore, UAVs can communicate and cooperate directly with other UAVs through ground base stations or other UAVs. Multiple UAVs can be dispatched to carry out collaborative operations according to different combat environments and mission requirements. Therefore, research on UAV collaborative task allocation technology will receive more attention. Summary of the Invention

[0004] Aiming at the problems that the completion rate of urgent tasks is poor when assigning tasks to multiple UAVs, and as the amount of tasks increases, the system communication volume increases sharply, resulting in reduced efficiency, the present invention provides a multi-UAV dynamic task allocation method based on an improved contract network algorithm.

[0005] To achieve the above objectives, this application proposes a multi-UAV dynamic task allocation method based on an improved contract network algorithm, including:

[0006] According to the influence of UAV execution cost, payload cost and navigation cost on UAV performance, a multi-UAV dynamic task allocation objective function is established;

[0007] The contract network algorithm was improved by setting task priorities, narrowing the bidding scope, reducing the number of bids, and increasing supervision and management to achieve the optimal task allocation method. Compared with the traditional contract network algorithm and the consistency bundle algorithm, the improved contract network algorithm has higher system efficiency and less communication traffic, and it also has certain advantages as the number of tasks increases.

[0008] Furthermore, let the set of drones be V and the set of tasks to be assigned be T; where V = {1, 2, ..., N}, N represents the number of drones; T = {1, 2, ..., M}, M represents the number of tasks.

[0009] Furthermore, when a drone performs a mission, it needs to pay an execution cost, as shown in the following formula:

[0010]

[0011] Among them, S i is the mission execution sequence of UAV i, is the execution cost of UAV i executing task j, ω1 is the execution cost constant coefficient and ω1>0; is the execution requirement of task j, A i The capability of drone i is obtained as follows:

[0012]

[0013] Where A i1 For the UAV mission completion capability, A i2 The remaining working capacity of the drone, which can be obtained by:

[0014]

[0015] Where V i1 is the number of completed tasks of UAV i, V i2 is the number of winning bids for drone i;

[0016]

[0017] Where V i3 is the number of tasks in the current task sequence of UAV i, V i4 is the maximum number of tasks that drone i can perform.

[0018] Furthermore, when the UAV is performing a mission, the fuselage load is the load cost of UAV i performing mission j. for:

[0019]

[0020] Among them, ω2 is the load cost coefficient and ω2>0, β is the attenuation factor, β∈(0,1), μ1 is a constant coefficient and μ1∈(0,1), t jstart is the task execution start time, t j (S i )-t jstart Waiting time for the drone, is the required load of task j, Lo i is the total payload of the UAV; j * is the position of task j in the current execution sequence, t j (S i) represents the current task sequence S of task j in UAV i i The actual execution time in the , which is composed of the UAV’s flight time between tasks and the task execution time, is as follows:

[0021]

[0022] in, For task sequence S i The position of the kth task in is the distance between the two tasks, is the execution time required for the kth task; v i is the flight speed of drone i.

[0023] Furthermore, when a drone performs a mission, the longer the flight distance, the greater the flight cost. The flight cost of drone i performing mission j is for:

[0024]

[0025] Among them, ω3 is the navigation cost coefficient and ω3>0, Le i The longest flight range of UAV i.

[0026] Furthermore, the task execution cost C of drone i performing task j is ij (S i )for:

[0027]

[0028] The profit R of drone i performing mission j ij (S i )for:

[0029]

[0030] Among them, α is the attenuation factor and α∈(0,1), μ2 is a constant coefficient and μ2∈(0,1), Val j is the execution value of task j, t jstart is the start execution time of task j;

[0031] Therefore, the objective function of multi-UAV dynamic task allocation is obtained as follows:

[0032]

[0033] Among them, x ij is the decision variable, x ij =1, it means that task j is assigned to drone i, otherwise, x ij =0 means each task is assigned to one UAV; This means that the mission execution requirements cannot exceed the UAV's execution capabilities; Indicates that the total payload requirement of the mission cannot exceed the total payload of the UAV; Indicates that the mission execution path distance cannot exceed the maximum flight range of the drone.

[0034] Furthermore, the contract network algorithm was proposed by Randall Davis and Reid G. Smith in 1980 from the economic field to solve the distributed task allocation problem in the multi-agent field. The agents in the contract network model have autonomous capabilities and can communicate and process information. In the contract network of the present invention, the agents are drones, which dynamically allocate tasks through four stages: bidding, winning the bid, and signing the contract. The specific process of the traditional contract network algorithm is as follows: Figure 1 shown.

[0035] Although the traditional contract network algorithm can assign tasks to drones, the drone task completion rate is low, the overall efficiency of the system is low, and the communication volume is large. The traditional contract network algorithm is improved from the following aspects to better solve the problem of dynamic task allocation of drones.

[0036] (1) Set task priority

[0037] In traditional contract network algorithms, there is no clear definition for the selection of bidding tasks. If urgent tasks cannot be completed as quickly as possible, the system's task completion level will be reduced.

[0038] For multiple tasks to be assigned, set priorities according to the urgency of the tasks, and select the task with the highest priority as the bidding task for this round.

[0039] (2) Narrowing the scope of bidding

[0040] In traditional contract network algorithms, bidders must send bidding information to all drones in the system. As the number of drones in the system continues to increase, the system communication volume will increase rapidly, increasing the communication burden and even causing system congestion and paralysis. Therefore, it is necessary to narrow the bidding scope and select drones with strong capabilities to send bidding information before sending bidding documents.

[0041] (3) Reduce the number of bids

[0042] In the traditional contract network algorithm, any drone that receives a tender document can bid. This will greatly increase the system communication volume and also increase the bid evaluation pressure on the bidder. The tender document includes the bidder's own performance in completing the task. The drone that receives the tender document calculates the performance of completing the task. If it is greater than the bidder's performance, it checks its own task time window. If there is a suitable time window in the drone sequence to add a new task, it will bid; otherwise, it will not bid. Through these two indicator constraints, unqualified drones can be screened out, thereby reducing the number of bidders. The schematic diagram of the drone task execution time window is as follows: Figure 2 shown.

[0043] (4) Increase supervision and management

[0044] In the traditional contract network algorithm, if the winning drone fails or is damaged, the tasks it has bid for and the tasks in its task sequence will not be executed, which will reduce the system task completion rate. Therefore, the bidding drone is used to supervise and manage the communication of the winning drone;

[0045] Assume that the communication time in the objective function is t a , the average information processing time of the drone is From the above, we can know that the task execution time is τ j , then at t a +τ j If the bidding drone does not receive feedback from the winning drone within the specified time, it will assume that the winning drone has failed or is damaged and will send a reminder message to the winning drone. If no reply is received within the time limit, the bidding drone will re-bid the task.

[0046] Dynamic task allocation flow chart based on improved contract network, such as Figure 3 shown.

[0047] Compared with the existing technology, the above technical solution adopted by the present invention has the following advantages: this method comprehensively considers the impact of drone execution cost, payload cost and navigation cost on drone performance to establish an objective function, analyzes the shortcomings of the traditional contract network algorithm in solving this problem, sets task priority, narrows the bidding scope, reduces the number of bids, and increases supervision and management to improve the contract network algorithm in four aspects, thereby reducing the communication volume during drone task allocation and improving system efficiency; thereby effectively solving the problem of dynamic task allocation for multiple drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a specific flow chart of the traditional contract network algorithm;

[0049] Figure 2 Schematic diagram of the UAV mission execution time window;

[0050] Figure 3 To improve the dynamic task allocation flow chart of the contract network;

[0051] Figure 4 This is a simulation diagram of the system traffic of the present invention;

[0052] Figure 5 This is a simulation diagram of the system performance of the present invention;

[0053] Figure 6 This is a simulation diagram of system communication volume with different task numbers according to the present invention;

[0054] Figure 7 This is a simulation diagram of the system performance of different task quantities of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of this application more clearly understood, this application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application. That is, the embodiments described are only part of the embodiments of this application, not all of them.

[0056] Example 1

[0057] This embodiment provides a multi-UAV dynamic task allocation method based on an improved contract network algorithm, specifically including: establishing a multi-UAV dynamic task allocation objective function based on the impact of UAV execution cost, payload cost and navigation cost on UAV performance; analyzing the shortcomings of the traditional contract network algorithm in solving this problem, and improving the contract network algorithm by setting task priorities, narrowing the bidding scope, reducing the number of bids, and increasing supervision and management. This reduces the communication volume during UAV task allocation and improves system performance.

[0058] The improved contract network algorithm is compared with the traditional contract network algorithm and CBBA algorithm. The improved contract network algorithm is more effective in solving the multi-UAV task allocation problem. The relevant parameter settings are shown in Table 1:

[0059] Table 1 Related parameter settings

[0060]

[0061]

[0062] 8 UAVs were randomly generated in the area, 25 tasks were carried out, and 10 simulation experiments were conducted to compare the system communication volume and system performance. The simulation results of system communication volume are as follows: Figure 4 The system performance simulation results are shown in Figure 5 The task assignment completion time is shown in Table 2.

[0063] In the 10 simulation experiments, the improved contract network algorithm and the CBBA algorithm successfully completed the dynamic task allocation, while the traditional contract network algorithm failed in the fifth experiment and did not complete all tasks, so its system efficiency was greatly reduced. Figure 4 It can be seen that the traditional contract network algorithm has the highest communication volume, followed by the CBBA algorithm, while the improved contract network algorithm has a significantly reduced number of communications; Figure 5 It can be seen that the improved contract network algorithm has higher system efficiency than the traditional contract network algorithm and CBBA algorithm.

[0064] Table 2 Task assignment completion time

[0065]

[0066] As shown in Table 2, the average time to complete the 10 experimental task assignments of the improved contract net algorithm is lower than that of the traditional contract net algorithm and the CBBA algorithm. It can be seen that the calculation speed of the improved contract net algorithm has been improved, and it has a certain timeliness.

[0067] In order to verify that the improved contract network algorithm is still effective as the number of tasks in the system increases, simulation comparisons are performed under different task numbers. The number of tasks to be assigned in the system is continuously increased, and the experiment is repeated 10 times for each task number, and the average value is taken. The system communication volume simulation results are shown in the figure below. Figure 6 The system performance simulation results are shown in Figure 7 shown.

[0068] Figure 6 and Figure 7 The following chart compares the system communication volume and system performance of the three task allocation algorithms for the cases where the number of tasks is 20, 25, 30, 35, 40, 45, 50, 55, and 60, respectively. The horizontal axis represents the number of tasks, and the vertical axis represents the communication volume and system performance achieved by all winning drones completing all tasks. As the number of tasks increases, the system communication volume and system performance of the three algorithms increase. The improved contract net algorithm uses fewer communications than the traditional contract net algorithm and the CBBA algorithm, achieving higher system performance. This advantage becomes more pronounced as the number of tasks increases. This verifies the effectiveness of the proposed method.

[0069] The foregoing descriptions of specific exemplary embodiments of the present invention are for purposes of illustration and description. These descriptions are not intended to limit the invention to the precise forms disclosed, and it is apparent that many variations and modifications are possible in light of the foregoing teachings. The exemplary embodiments have been selected and described for the purpose of explaining the specific principles of the invention and their practical application, thereby enabling those skilled in the art to realize and utilize a variety of exemplary embodiments of the invention and various options and modifications. The scope of the invention is intended to be defined by the claims and their equivalents.

Claims

1. A multi-UAV dynamic task allocation method based on an improved contract network algorithm is characterized by: include: According to the influence of UAV execution cost, payload cost and navigation cost on UAV performance, a multi-UAV dynamic task allocation objective function is established; The contract network algorithm is improved by setting task priorities, narrowing the bidding scope, reducing the number of bids, and increasing supervision and management to obtain the optimal task allocation method; Let the set of drones be V and the set of tasks to be assigned be T; where V = {1, 2, ..., N}, N is the number of drones; T = {1, 2, ..., M}, M is the number of tasks; When a drone performs a mission, it needs to pay an execution cost, as shown in the following formula: Among them, S i is the mission execution sequence of UAV i, is the execution cost of UAV i executing task j, ω1 is the execution cost constant coefficient and ω1>0; is the execution requirement of task j, A i The capability of drone i is obtained as follows: Where A i1 For the UAV mission completion capability, A i2 The remaining working capacity of the drone, which can be obtained by: Where V i1 is the number of completed tasks of UAV i, V i2 is the number of winning bids for drone i; Where V i3 is the number of tasks in the current task sequence of UAV i, V i4 The maximum number of tasks that drone i can perform; The load of the drone during the mission, the load cost of drone i performing mission j for: Among them, ω2 is the load cost coefficient and ω2>0, β is the attenuation factor, β∈(0,1), μ1 is a constant coefficient and μ1∈(0,1), t jstart is the task execution start time, t j (S i )-t jstart Waiting time for the drone, is the required load of task j, Lo i is the total payload of the UAV; j * is the position of task j in the current execution sequence, t j (S i ) represents the current task sequence S of task j in UAV i i The actual execution time in the , which is composed of the UAV’s flight time between tasks and the task execution time, is as follows: in, For task sequence S i The position of the kth task in is the distance between the two tasks, is the execution time required for the kth task; v i is the flight speed of drone i; When a drone performs a mission, the longer the flight distance, the greater the flight cost. The flight cost of drone i performing mission j is for: Among them, ω3 is the navigation cost coefficient and ω3>0, Le i is the longest flight range of drone i; The execution cost C of drone i performing task j ij (S i )for: The profit R of drone i performing mission j ij (S i )for: Among them, α is the attenuation factor and α∈(0,1), μ2 is a constant coefficient and μ2∈(0,1), Val j is the execution value of task j, t jstart is the start execution time of task j; Therefore, the objective function of multi-UAV dynamic task allocation is obtained as follows: Among them, x ij is the decision variable, x ij =1, it means that task j is assigned to drone i, otherwise, x ij =0 means each task is assigned to one UAV; This means that the mission execution requirements cannot exceed the UAV's execution capabilities; Indicates that the total payload requirement of the mission cannot exceed the total payload of the UAV; Indicates that the mission execution path distance cannot exceed the maximum flight range of the drone; The intelligent agent in the contract network algorithm is a drone, which performs dynamic task allocation through four stages: bidding, winning the bid, and signing the contract; Assume that the communication time in the objective function is t a , the average information processing time of the drone is The task execution time is τ j , then at t a +τ j If the bidding drone does not receive feedback from the winning drone within the specified time, it will assume that the winning drone has failed or is damaged and will send a reminder message to the winning drone. If no reply is received within the time limit, the bidding drone will re-bid the task.

2. The multi-UAV dynamic task allocation method based on the improved contract network algorithm according to claim 1 is characterized in that: For multiple tasks to be assigned, priorities are set according to the urgency of the tasks, and the tasks with the highest priority are selected as the bidding tasks for this round; and before sending the bidding documents, drones with strong capabilities are screened to send bidding information.

3. The multi-UAV dynamic task allocation method based on the improved contract network algorithm according to claim 1 is characterized in that: The bidder's own efficiency in completing the task is included in the bidding document. The drone that receives the bidding document calculates the efficiency of completing the task. If it is greater than the bidder's efficiency, it checks its own task time window. If there is a suitable time window in the drone sequence to add the new task, it will bid; otherwise, it will not bid.

Images