Reliability constraint fault-tolerant task scheduling method for unmanned aerial vehicle cluster system

Abstract

The invention discloses a reliability constraint fault-tolerant task scheduling method for an unmanned aerial vehicle cluster system, and belongs to the technical field of unmanned aerial vehicle cluster cooperative control, and the method comprises the steps: building a directed acyclic graph task, and obtaining an initial scheduling scheme based on the directed acyclic graph task; calculating the overall task reliability according to the initial scheduling scheme; whether the overall task reliability meets a preset reliability constraint target or not is judged, and if yes, the initial scheduling scheme is executed; and if the overall task reliability does not meet a preset reliability constraint target, according to the comprehensive priority coefficient, selecting a sub-task for copying to obtain a sub-task copy, placing the sub-task copy to the optimal unmanned aerial vehicle node, and updating the initial scheduling scheme and the overall task reliability until the overall task reliability meets the preset reliability constraint target. According to the method, the effect of minimizing the total task completion time on the premise of meeting the preset task reliability target can be achieved.

Description

Technical Field

[0001] This invention relates to the field of collaborative control technology for unmanned aerial vehicle (UAV) swarms, and in particular to a reliability-constrained fault-tolerant task scheduling method for UAV swarm systems. Background Technology

[0002] With the maturation of UAV (Unmanned Aerial Vehicle) technology and the development of artificial intelligence, swarm systems composed of multiple UAVs are playing an increasingly important role in fields such as military reconnaissance, disaster relief, smart logistics, and agricultural surveying. Compared to single-unit operations, UAV swarms can complete more complex and larger-scale tasks through collaborative cooperation. In these applications, a complex macroscopic task is often modeled as a Directed Acyclic Graph (DAG) task. A DAG consists of nodes and edges. A node represents a subtask that must be performed by a particular UAV (such as "flying to a designated area," "capturing high-resolution images," or "analyzing image data"), while edges represent the dependencies and data flows between subtasks.

[0003] The core challenge in leveraging the advantages of drone swarms lies in efficiently and reliably scheduling these subtasks to different drones within the swarm. This process faces the dual challenges of environmental uncertainty and scheduling complexity: First, drone swarms are typically heterogeneous, with significant differences in computing power (such as onboard processor performance), load status, and remaining battery life among the drone nodes, which may fail at any time due to power depletion, signal loss, or hardware malfunction. Second, wireless communication links between drones may experience bandwidth fluctuations or even interruptions due to distance, obstacles, or electromagnetic interference, threatening the smooth handover of data-dependent tasks.

[0004] Since task scheduling in a directed acyclic graph (DAG) is inherently an NP-hard problem, in such an unreliable environment, the scheduler not only needs to optimize traditional performance metrics, such as minimizing the total task completion time (makespan), but also must simultaneously satisfy the reliability constraints of the entire cluster. Existing research largely focuses on DAG task scheduling in general distributed environments. Classical heuristic algorithms, such as the HEFT algorithm, are high-performance, low-complexity task scheduling algorithms for heterogeneous computing environments. They primarily focus on optimizing the total task completion time (makespan) under reliable conditions, but neglect the impact of drone node and wireless link failures, limiting their applicability in dynamic, high-risk drone operation environments. While some fault-tolerant scheduling algorithms introduce reliability mechanisms, they often employ coarse-grained full replication strategies—that is, having multiple drones perform the same task as backups. Although such methods can improve the reliability of drone swarm systems, they lead to excessive consumption of valuable onboard resources (such as computing power and battery power) and may actually prolong the total task time due to communication congestion. Summary of the Invention

[0005] The purpose of this invention is to provide a reliability-constrained fault-tolerant task scheduling method for unmanned aerial vehicle (UAV) swarm systems. This method determines whether the calculated overall task reliability meets a preset task reliability target; if not, it iterates until the target is met. This invention achieves the effect of minimizing the total task completion time while satisfying the preset task reliability target. This invention is implemented through the following technical solutions.

[0006] This invention provides a reliability-constrained fault-tolerant task scheduling method for unmanned aerial vehicle (UAV) swarm systems, comprising the following:

[0007] Based on the dependencies between tasks to be executed, a directed acyclic graph task consisting of multiple subtasks and multiple data edges is constructed; each subtask needs to be scheduled to a drone node to execute the subtask, and each data edge represents the data dependency and communication relationship between subtasks.

[0008] Determine the execution priority of subtasks, and allocate subtasks to UAV nodes in sequence according to their execution priorities to form a scheduling mapping table of subtasks and UAV nodes, thus obtaining the initial scheduling scheme.

[0009] Calculate the overall task reliability based on the initial scheduling scheme;

[0010] Determine whether the overall task reliability meets the preset reliability constraint target. If it does, execute the initial scheduling scheme. If it does not, select a subtask to copy according to the comprehensive priority coefficient to obtain a subtask copy. Place the subtask copy in the optimal UAV node and update the initial scheduling scheme and the overall task reliability until the overall task reliability meets the preset reliability constraint target.

[0011] Optionally, the directed acyclic graph task expression is as follows:

[0012] ,

[0013] In the formula, For directed acyclic graph tasks, Let the set of subtasks be the set of tasks in the directed acyclic graph task. This is the set of data edges for the subtasks;

[0014] The set of subtasks The expression is as follows:

[0015] ,

[0016] In the formula, and These are the first two tasks in the directed acyclic graph task. The and the first Sub-tasks and Both represent subtask numbers, with values ​​ranging from 1 to k. This represents the total number of subtasks.

[0017] Optionally, the execution priority of a subtask is determined by a ranking value, which is calculated using the following formula:

[0018] ,

[0019] In the formula, and These are the first two tasks in the directed acyclic graph task. The and the first Sub-tasks For subtasks The ranking value For subtasks Average execution time across all drone nodes Subtask The set of subsequent tasks, For subtasks The ranking value

[0020] For subtasks To the successor task Average communication time in a drone swarm, where subtasks To the successor task Communication time It is obtained by calculation using the following formula:

[0021] ,

[0022] In the formula, For data edge The amount of data transmitted For subtasks Scheduled drone nodes to subtasks Link bandwidth between scheduled drone nodes, data edge Subtasks Waiting for subtasks Once the calculation results are completed and passed to the end, execution will begin. yes The precursor mission. yes The subsequent task; a subtask without a successor task is called an exit task. This marks the end of the entire directed acyclic graph task.

[0023] Optional, if subtask For export mission Then subtask The execution priority is calculated using the following formula:

[0024] ,

[0025] In the formula, For export mission Average execution time across all drone nodes.

[0026] Optionally, the allocation of drone nodes is performed using the following formula:

[0027] ,

[0028] In the formula, For the directed acyclic graph task, the first Sub-tasks To assign to subtasks The earliest drone node to complete this sub-task This refers to the drone node number. For the first One drone node, For a collection of drone nodes, For subtasks In the One drone node The earliest start time on, For subtasks In the One drone node The estimated execution time is as follows. For subtasks In the One drone node The earliest completion time.

[0029] Optionally, the overall task reliability is obtained using the following formula:

[0030] ,

[0031] In the formula, For the overall reliability of the mission, For the directed acyclic graph task, the first There are several subtasks, where i is the subtask number, ranging from 1 to k. The total number of subtasks.

[0032] The reliability of subtask combination is calculated using the following formula:

[0033] ,

[0034] In the formula, For subtasks The reliability of subtask dungeon combinations, subtasks For subtasks Any of the precursor missions. For subtasks The set of precursor missions, For the pioneer mission To subtask Combined communication reliability.

[0035] Optionally, the precursor task To subtask Combined communication reliability It is obtained by calculation using the following formula:

[0036] ,

[0037] In the formula, subtask For subtasks Any of the precursor missions. For the pioneer mission The replica number, with a value ranging from 1 to , For the pioneer mission The total number of copies, For the pioneer mission The The drone node where each replica is located. For subtasks The drone node it is located at For link transmission reliability;

[0038] The subtask Sub-task dungeon combination reliability It is obtained by calculation using the following formula:

[0039] ,

[0040] In the formula, This is the drone node number, ranging from 1 to z, where z is the subtask. The total number of copies generated across multiple drone nodes, For the first One drone node, For the first The reliability of each drone node.

[0041] Optionally, the link transmission reliability It is obtained by calculation using the following formula:

[0042] ,

[0043] In the formula, For subtasks To sub-task Crossing drone nodes and drone nodes Data transmission time, For drone nodes To drone node Link failure rate;

[0044] The first Reliability of each drone node It is obtained by calculation using the following formula:

[0045] ,

[0046] In the formula, For the first One drone node Link failure rate, For subtasks In the One drone node The estimated execution time.

[0047] Optionally, the step of selecting subtasks for replication based on the comprehensive priority coefficient to obtain subtask copies includes selecting the subtask with the highest comprehensive priority coefficient value to generate a subtask copy.

[0048] The overall priority coefficient is calculated using the following formula:

[0049] ,

[0050] In the formula, Let be the overall priority coefficient of the i-th subtask. For subtasks Global scheduling impact factors Represented as subtask The reliability gain of adding a replica to the drone swarm system A very small positive number set to prevent the denominator from being zero.

[0051] Optionally, the calculation of the optimal drone node includes calculating the comprehensive score of each candidate drone node, and the drone node with the highest comprehensive score is the optimal drone node. The comprehensive score is calculated using the following formula:

[0052] ,

[0053] In the formula, For the first The overall score of each drone node. As a scoring factor, For the first Normalized reliability score of each drone node For the first Normalized performance score of each drone node;

[0054] No. Normalized reliability score of each drone node It is obtained by calculation using the following formula:

[0055] ,

[0056] In the formula, For the sub-task instance in the 1st Execution reliability on each candidate drone node and These represent the minimum and maximum execution reliability among all candidate drone nodes, respectively.

[0057] The first Normalized performance score of each drone node It is obtained by calculation using the following formula:

[0058] ,

[0059] For the sub-task instance in the 1st The earliest completion time on each candidate drone node. and These are the minimum and maximum earliest completion times for the subtask replica among all candidate nodes, respectively.

[0060] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0061] This invention establishes a directed acyclic graph task, enabling multiple subtasks to be scheduled to multiple UAV nodes for execution. By combining ranking values ​​to determine the priority of each subtask, it helps to prioritize subtasks that have a greater impact on the overall task completion time during the scheduling process, thereby improving scheduling efficiency. This invention also effectively shortens the total task completion time while ensuring the reliability of the UAV swarm system by selecting the subtask with the highest comprehensive priority coefficient for replication. Attached Figure Description

[0062] Figure 1 The diagram shown is a schematic flowchart of a reliability-constrained fault-tolerant task scheduling method for a drone swarm system according to an embodiment of the present invention.

[0063] Figure 2 The diagram shown is a schematic representation of the iterative process in one embodiment of the present invention. Detailed Implementation

[0064] The following description, in conjunction with the accompanying drawings and specific embodiments, provides further details. In this description, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature.

[0065] Example 1 This embodiment provides a reliability-constrained fault-tolerant task scheduling method for unmanned aerial vehicle (UAV) swarm systems, including the following steps:

[0066] Based on the dependencies between tasks to be executed, a directed acyclic graph task consisting of multiple subtasks and multiple data edges is constructed; each subtask needs to be scheduled to a drone node to execute the subtask, and each data edge represents the data dependency and communication relationship between subtasks.

[0067] Determine the execution priority of subtasks, and allocate subtasks to UAV nodes in sequence according to their execution priorities to form a scheduling mapping table of subtasks and UAV nodes, thus obtaining the initial scheduling scheme.

[0068] Calculate the overall task reliability based on the initial scheduling scheme;

[0069] Determine whether the overall task reliability meets the preset reliability constraint target. If it does, execute the initial scheduling scheme. If it does not, select a subtask to copy according to the comprehensive priority coefficient to obtain a subtask copy. Place the subtask copy in the optimal UAV node and update the initial scheduling scheme and the overall task reliability until the overall task reliability meets the preset reliability constraint target.

[0070] Example 2 Based on Example 1, this example describes the specific implementation process of a reliability-constrained fault-tolerant task scheduling method for UAV swarm systems, such as... Figure 1 As shown, it specifically includes the following:

[0071] I. Constructing a Distributed Computing Network Environment and a Directed Acyclic Graph Task

[0072] In one specific embodiment of the present invention, the distributed computing network environment consisting of multiple UAV computing nodes and wireless communication links is composed of a set of UAV nodes. It consists of wireless communication links between nodes, among which, This represents the drone node number, ranging from 1 to N, where N is the total number of drone nodes in the drone swarm system. The expression for each drone node is: , Indicates the first The computing power of each drone node Failure rate per unit time node; UAV node arrive The expression for the wireless communication link is: Wireless communication link With available bandwidth matrix Failure rate per unit time node Among them, drone nodes and Let m and n represent the m-th and n-th drone nodes in the drone swarm system, where m and n both range from 1 to N. For drone nodes Failure rate, For drone nodes arrive The failure rate.

[0073] In distributed computing networks, tasks are often represented as a directed acyclic graph (DAG) task consisting of a set of subtasks with sequential dependencies and multiple data edges. The expression for a DAG task is as follows:

[0074] ,

[0075] In the formula, For directed acyclic graph tasks, Let the set of subtasks be the set of tasks in the directed acyclic graph task. The set of data edges for subtasks; the set of subtasks The expression is as follows:

[0076] ,

[0077] In the formula, and These are the first two tasks in the directed acyclic graph task. The and the first Sub-tasks and Both represent subtask numbers, with values ​​ranging from 1 to k. This represents the total number of subtasks.

[0078] II. Determine the priority of subtasks and allocate drone nodes to subtasks

[0079] 2.1 Determine the execution priority of subtasks

[0080] In one specific embodiment of the present invention, the execution priority of a subtask is determined by an upper ranking value, which is calculated using the following formula:

[0081] ,

[0082] In the formula, and These are the first two tasks in the directed acyclic graph task. The and the first Sub-tasks For subtasks The ranking value For subtasks Average execution time across all drone nodes Subtask The set of subsequent tasks, For subtasks The ranking value

[0083] For subtasks To the successor task Average communication time in a drone swarm, where subtasks To the successor task Communication time It is obtained by calculation using the following formula:

[0084] ,

[0085] In the formula, For data edge The amount of data transmitted For subtasks Scheduled drone nodes to subtasks Link bandwidth between scheduled drone nodes, data edge Subtasks Waiting for subtasks Once the calculation results are completed and passed to the end, execution will begin. yes The precursor mission. yes The subsequent task; a subtask without a successor task is called an exit task. This marks the end of the entire directed acyclic graph task.

[0086] If subtask For export mission Then subtask The execution priority is calculated using the following formula:

[0087] ,

[0088] In the formula, For export mission Average execution time across all drone nodes.

[0089] 2.2 Assigning UAV nodes to subtasks

[0090] Step 2.1 calculates the ranking value for each subtask, and drone nodes are assigned to each subtask according to the ranking value. The specific drone node assigned to each subtask is determined using the following formula:

[0091] ,

[0092] In the formula, For the directed acyclic graph task, the first Sub-tasks To assign to subtasks The earliest drone node to complete this sub-task This refers to the drone node number. For the first One drone node, For a collection of drone nodes, For subtasks In the One drone node The earliest start time on, For subtasks In the One drone node The estimated execution time is as follows. For subtasks In the One drone node The earliest completion time.

[0093] 2.3 Initial Scheduling Scheme

[0094] After the allocation is completed in steps 2.1 and 2.2, a scheduling mapping table of subtasks and UAV nodes is formed, resulting in an initial scheduling scheme. The initial scheduling scheme is generated to minimize the total task completion time (makespan).

[0095] III. Calculate the overall task reliability

[0096] 3.1 Calculate the execution reliability of UAV nodes

[0097] In one specific embodiment of the present invention, the first Reliability of each drone node It is obtained by calculation using the following formula:

[0098] ,

[0099] In the formula, For the first One drone node Link failure rate, For subtasks In the One drone node The estimated execution time is given by the formula, which represents the probability that the subtask will complete without failure on the specified drone node.

[0100] 3.2 Calculate link transmission reliability

[0101] In one specific embodiment of the present invention, for subtasks To sub-task Data transmission, if the transmission crosses drone nodes and Then the link transmission reliability It is obtained by calculation using the following formula:

[0102] ,

[0103] In the formula, For subtasks To sub-task Crossing drone nodes and drone nodes Data transmission time, For drone nodes To drone node Link failure rate;

[0104] 3.3 Calculate the reliability of subtask replica combinations

[0105] In one specific embodiment of the present invention, for any subtask If it generates on multiple drone nodes Each copy, then the subtask The probability of successful execution is the probability that at least one of the copies will successfully complete, denoted as a subtask. Sub-task dungeon combination reliability It is obtained by calculation using the following formula:

[0106] ,

[0107] In the formula, This is the drone node number, ranging from 1 to z, where z is the subtask. The total number of copies generated across multiple drone nodes, For the first One drone node, For the first The reliability of each drone node.

[0108] 3.4 Calculate the reliability of communication between the precursor missions.

[0109] In one specific embodiment of the present invention, for subtasks Any of the precursor missions If the precursor mission If it contains one or more copies, then it is related to the task. Combined communication reliability is defined as at least one copy to the task The probability of successful data transmission, the precursor mission To subtask Combined communication reliability It is obtained by calculation using the following formula:

[0110] ,

[0111] In the formula, subtask For subtasks Any of the precursor missions. For the pioneer mission The replica number, with a value ranging from 1 to , For the pioneer mission The total number of copies, For the pioneer mission The The drone node where each replica is located. For subtasks The drone node it is located at For link transmission reliability;

[0112] 3.5 Calculate the reliability of subtask combination

[0113] In one specific embodiment of the present invention, for subtasks The combined reliability is defined as the product of the combined reliability of the task itself and the combined communication reliability of all its predecessor tasks. For subtasks The combined reliability is calculated using the following formula:

[0114] ,

[0115] In the formula, For subtasks The reliability of subtask dungeon combinations, subtasks For subtasks Any of the precursor missions. For subtasks The set of precursor missions, For the pioneer mission To subtask Combined communication reliability.

[0116] 3.6 Calculate the overall task reliability

[0117] In one specific embodiment of the present invention, the overall task reliability is obtained by the following formula:

[0118] ,

[0119] In the formula, For the overall reliability of the mission, For the directed acyclic graph task, the first There are several subtasks, where i is the subtask number, ranging from 1 to k. The total number of subtasks.

[0120] IV. Determine whether the overall task reliability meets the preset reliability constraints.

[0121] 4.1 Assessment of overall mission reliability

[0122] In one specific embodiment of the present invention, the overall task reliability is calculated through the above steps. After that, it is necessary to... Make a judgment:

[0123] Judgment Condition 1: Overall Task Reliability Meets or exceeds preset reliability constraints;

[0124] Judgment condition 2: . The preset minimum threshold, and These represent the overall mission reliability of the current round and the previous round, respectively. This indicates the improvement in overall mission reliability.

[0125] If any of the above judgment conditions are not met, the following iterative process will be performed.

[0126] 4.2 Iterative Process

[0127] In one specific embodiment of the present invention, the iterative process is illustrated as follows: Figure 2 As shown, Figure 2 middle, The preset reliability constraints are as follows:

[0128] 4.2.1 Calculate the overall priority coefficient

[0129] In one specific embodiment of the present invention, before calculating the overall priority coefficient, a heuristic screening is first performed to generate a subset containing a finite number of candidate tasks. All subtasks are evaluated and ranked according to preset key indicators, and the top M subtasks are selected as the candidate set based on their ranking. The key indicators include at least the combinatorial reliability of the subtasks. The overall priority coefficient for each subtask in the candidate set is calculated using the following formula:

[0130] ,

[0131] In the formula, Let be the overall priority coefficient of the i-th subtask. For subtasks Global scheduling impact factors Represented as subtask The reliability gain of adding a replica to the drone swarm system A very small positive number set to prevent the denominator from being zero.

[0132] 4.2.2 Copying and Placing Subtask Instances

[0133] Step 4.2.1 yields the overall priority coefficient value for each subtask. .choose The subtask with the largest value is replicated to obtain a subtask replica, and the optimal drone node is selected for this replica. Placing the subtask replica on the optimal drone node minimizes the total task completion time (makespan) and improves the performance of the drone swarm system.

[0134] The calculation of the optimal drone node involves calculating the comprehensive score of each candidate drone node. The drone node with the highest comprehensive score is the optimal drone node. The comprehensive score is calculated using the following formula:

[0135] ,

[0136] In the formula, For the first The overall score of each drone node. As a scoring factor, For the first Normalized reliability score of each drone node For the first Normalized performance score of each drone node;

[0137] No. Normalized reliability score of each drone node It is obtained by calculation using the following formula:

[0138] ,

[0139] In the formula, For the sub-task instance in the 1st Execution reliability on each candidate drone node and These represent the minimum and maximum execution reliability among all candidate drone nodes, respectively.

[0140] No. Normalized performance score of each drone node It is obtained by calculation using the following formula:

[0141] ,

[0142] For the sub-task instance in the 1st The earliest completion time on each candidate drone node. and These are the minimum and maximum earliest completion times for the subtask replica among all candidate nodes, respectively.

[0143] 4.2.3 Update overall task reliability

[0144] Update the initial scheduling scheme obtained in step 2.3 and the overall task reliability obtained in step 3.6 based on the subtask replicas and optimal drone nodes obtained in step 4.2.2. And determine the overall reliability of the updated task. Check if the conditions in step 4.1 are met (iteration stops if any one condition is met). If met, execute the updated initial scheduling scheme; otherwise, repeat the iterative process in step 4.2 until the overall task reliability is achieved. The iteration stops once any one of the judgment conditions in step 4.1 is met. The updated scheduling scheme is based on the initial scheduling scheme, with the addition of the mapping between the subtask replicas and the optimal UAV node obtained in step 4.2.2.

[0145] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A reliability-constrained fault-tolerant task scheduling method for unmanned aerial vehicle (UAV) swarm systems, characterized in that, include: Based on the dependencies between tasks to be executed, a directed acyclic graph task consisting of multiple subtasks and multiple data edges is constructed; each subtask needs to be scheduled to a drone node to execute the subtask, and each data edge represents the data dependency and communication relationship between subtasks. Determine the execution priority of subtasks, and allocate subtasks to UAV nodes in sequence according to their execution priorities to form a scheduling mapping table of subtasks and UAV nodes, thus obtaining the initial scheduling scheme. Calculate the overall task reliability based on the initial scheduling scheme; Determine whether the overall task reliability meets the preset reliability constraint target. If it does, execute the initial scheduling scheme. If it does not, select a subtask to copy according to the comprehensive priority coefficient to obtain a subtask copy. Place the subtask copy in the optimal UAV node and update the initial scheduling scheme and the overall task reliability until the overall task reliability meets the preset reliability constraint target.

2. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 1, characterized in that, The expression for the directed acyclic graph task is as follows: ， In the formula, For directed acyclic graph tasks, Let the set of subtasks be the set of tasks in the directed acyclic graph task. This is the set of data edges for the subtasks; The set of subtasks The expression is as follows: ， In the formula, and These are the first two tasks in the directed acyclic graph task. The and the first Sub-tasks and Both represent subtask numbers, with values ​​ranging from 1 to k. This represents the total number of subtasks.

3. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 1, characterized in that, The execution priority of subtasks is determined by the ranking value, which is calculated using the following formula: ， In the formula, and These are the first two tasks in the directed acyclic graph task. The and the first Sub-tasks For subtasks The ranking value For subtasks Average execution time across all drone nodes Subtask The set of subsequent tasks, For subtasks The ranking value For subtasks To the successor task Average communication time in a drone swarm, where subtasks To the successor task Communication time It is obtained by calculation using the following formula: ， In the formula, For data edge The amount of data transmitted for Scheduled drone nodes to subtasks Link bandwidth between scheduled drone nodes, data edge Subtasks Waiting for subtasks Once the calculation results are completed and passed to the end, execution will begin. yes The precursor mission. yes The subsequent task; a subtask without a successor task is called an exit task. This marks the end of the entire directed acyclic graph task.

4. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 3, characterized in that, If subtask For export mission Then subtask The execution priority is calculated using the following formula: ， In the formula, For export mission Average execution time across all drone nodes.

5. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 1, characterized in that, The allocation of drone nodes is performed using the following formula: ， In the formula, For the directed acyclic graph task, the first Sub-tasks To assign to subtasks The earliest drone node to complete this sub-task This refers to the drone node number. For the first One drone node, For a collection of drone nodes, For subtasks In the One drone node The earliest start time on, For subtasks In the One drone node The estimated execution time is as follows. For subtasks In the One drone node The earliest completion time.

6. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 3, characterized in that, the... The overall mission reliability is obtained using the following formula: ， In the formula, For the overall reliability of the mission, For the directed acyclic graph task, the first There are several subtasks, where i is the subtask number, ranging from 1 to k. The total number of subtasks. The reliability of subtask combination is calculated using the following formula: ， In the formula, For subtasks The reliability of subtask dungeon combinations, subtasks For subtasks Any of the precursor missions. For subtasks The set of precursor missions, For the pioneer mission To subtask Combined communication reliability.

7. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 6, characterized in that, The precursor task To subtask Combined communication reliability It is obtained by calculation using the following formula: ， In the formula, subtask For subtasks Any of the precursor missions. For the pioneer mission The replica number, with a value ranging from 1 to , For the pioneer mission The total number of copies, For the pioneer mission The The drone node where each replica is located. For subtasks The drone node it is located at For link transmission reliability; The subtask Sub-task dungeon combination reliability It is obtained by calculation using the following formula: ， In the formula, This is the drone node number, ranging from 1 to z, where z is the subtask. The total number of copies generated across multiple drone nodes, For the first One drone node, For the first The reliability of each drone node.

8. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 7, characterized in that, The reliability of the link transmission It is obtained by calculation using the following formula: ， In the formula, For subtasks To sub-task Crossing drone nodes and drone nodes Data transmission time, For drone nodes To drone node Link failure rate; The first Reliability of each drone node It is obtained by calculation using the following formula: ， In the formula, For the first One drone node Link failure rate, For subtasks In the One drone node The estimated execution time.

9. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 1, characterized in that, The step of selecting subtasks for replication based on a comprehensive priority coefficient to obtain subtask copies includes selecting the subtask with the highest comprehensive priority coefficient value to generate a subtask copy. The overall priority coefficient is calculated using the following formula: ， In the formula, Let be the overall priority coefficient of the i-th subtask. For subtasks Global scheduling impact factors Represented as subtask The reliability gain of adding a replica to the drone swarm system A very small positive number set to prevent the denominator from being zero.

10. The reliability-constrained fault-tolerant task scheduling method for UAV swarm systems according to claim 1, characterized in that, The calculation of the optimal drone node includes calculating the comprehensive score of each candidate drone node, and the drone node with the highest comprehensive score is the optimal drone node. The comprehensive score is calculated using the following formula: ， In the formula, For the first The overall score of each drone node. As a scoring factor, For the first Normalized reliability score of each drone node For the first Normalized performance score of each drone node; No. Normalized reliability score of each drone node It is obtained by calculation using the following formula: ， In the formula, For the sub-task instance in the 1st Execution reliability on each candidate drone node and These represent the minimum and maximum execution reliability among all candidate drone nodes, respectively. The first Normalized performance score of each drone node It is obtained by calculation using the following formula: ， For the sub-task instance in the 1st The earliest completion time on each candidate drone node. and These are the minimum and maximum earliest completion times for the subtask replica among all candidate nodes, respectively.

Images