A multi-uav flight task allocation optimization method and system

Abstract

The application discloses a multi-unmanned aerial vehicle flight task allocation optimization method and system, and belongs to the technical field of unmanned aerial vehicle intelligent control. In order to solve the problem of reasonable and efficient allocation of multi-unmanned aerial vehicle flight tasks, the application constructs a feasibility constraint discrimination method of unmanned aerial vehicles and unmanned aerial vehicle tasks, filters unmanned aerial vehicles and unmanned aerial vehicle task pairs meeting the feasibility constraint; a priority dynamic updating mechanism of the unmanned aerial vehicle task is established, the initial priority is updated, and the task priority is obtained; a rule scoring function model for the unmanned aerial vehicle to execute the task is constructed; a scoring function model for the unmanned aerial vehicle to execute the task based on a deep neural network is constructed, a time limit sensitive weighting factor is introduced into a fusion scoring of the unmanned aerial vehicle to execute the task, and the final scoring of the unmanned aerial vehicle to execute the task is obtained; the multi-unmanned aerial vehicle flight task allocation problem is converted into solving the maximum total scoring value of all unmanned aerial vehicle task pairs, and the multi-unmanned aerial vehicle flight task allocation optimization scheme is obtained. The application improves the overall task completion rate.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of intelligent control of unmanned aerial vehicles, and particularly relates to a multi-unmanned aerial vehicle flight task allocation optimization method and system. BACKGROUND

[0002] With the rapid progress of artificial intelligence, navigation control, communication network and new energy technology, unmanned aerial vehicle technology is evolving towards intelligence, clustering and autonomy. In the future, unmanned aerial vehicles will play an increasingly important role in urban logistics, disaster relief, environmental monitoring, agricultural operations, aerial inspection and other fields. In particular, in complex scenarios such as low-altitude economy, urban air traffic and disaster emergency, unmanned aerial vehicle operations require higher system collaboration capabilities, autonomous decision-making capabilities and task execution efficiency. Multi-unmanned aerial vehicle cooperative task execution will become the mainstream application mode and technology development trend.

[0003] In practical applications, a single unmanned aerial vehicle often cannot meet the complex operation requirements of large task range, large task volume or tight task time limit. At this time, multiple unmanned aerial vehicles can be used to fly cooperatively to significantly improve system efficiency and completion rate through task decomposition and parallel operation. For example, multiple unmanned aerial vehicles can simultaneously cover different areas in urban inspection; multiple unmanned aerial vehicles can quickly respond to multiple disaster points in emergency rescue; and multiple unmanned aerial vehicles can simultaneously complete multi-point distribution in warehouse logistics. Task coordination and allocation of multi-unmanned aerial vehicle systems has become an important support for improving operation capability and system intelligence level.

[0004] Multi-unmanned aerial vehicle task allocation not only involves the timing and priority of tasks, but also considers various state attributes such as power, flight speed, load capacity and path risk of unmanned aerial vehicles. If not properly optimized, it may lead to task conflicts, resource waste, task delays, and even task failures. Traditional allocation methods based on distance or static weights cannot adapt to such complex scenarios with multiple factors and high dynamics. Current multi-unmanned aerial vehicle task allocation faces the following main problems: first, there is a lack of unified modeling of task attributes and unmanned aerial vehicle states, making it difficult to achieve targeted scheduling optimization; second, the scoring method is single and cannot consider multiple factors such as task timeliness, path energy consumption and flight risk; third, traditional rule models have limited expression ability and are difficult to cope with nonlinearities and uncertainties in the task environment.

[0005] Therefore, there is an urgent need for a task allocation optimization method that can consider multiple constraint conditions and performance indicators to ensure the rationality, feasibility and overall system operation efficiency of the task allocation results. SUMMARY

[0006] The problem to be solved by the present application is to achieve rational and efficient allocation of multi-unmanned aerial vehicle flight tasks, and to propose a multi-unmanned aerial vehicle flight task allocation optimization method and system.

[0007] To achieve the above object, the present application is realized by the following technical solutions:

[0008] A multi-UAV flight task allocation optimization method, comprising the following steps:

[0009] S1. Collecting UAV key attribute data to construct a UAV set, and collecting UAV task data to construct a UAV task set;

[0010] S2. Based on the data collected in step S1, calculating the total cost of UAV task execution considering the flight time, energy consumption and danger level on the path, and then constructing a UAV and UAV task feasibility constraint discrimination method to screen UAVs and UAV task pairs that meet the feasibility constraint;

[0011] S3. Establishing a dynamic update mechanism for UAV task priority, updating the initial priority to obtain the task priority;

[0012] S4. Considering the energy utilization efficiency under unit execution cost, the matching degree between the current load capacity of the UAV and the required load of the task, the task priority, and the energy and speed adaptation efficiency in unit task time, constructing a rule scoring function model for UAV task execution for the UAV and UAV task pairs that meet the feasibility constraint obtained in step S2;

[0013] S5. Constructing a deep neural network-based UAV task execution scoring function model, learning historical samples to extract deep-level correlations between UAV states and task attributes, and then fusing the rule scoring function model for UAV task execution obtained in step S3 to obtain a fusion score for UAV task execution;

[0014] S6. Introducing a time limit sensitive weighting factor to the fusion score for UAV task execution obtained in step S5 to obtain the final score for UAV task execution;

[0015] S7. Based on the final score for UAV task execution obtained in step S6, converting the multi-UAV flight task allocation problem into a problem of maximizing the total score of all UAV task pairs to obtain a multi-UAV flight task allocation optimization scheme.

[0016] Further, the collection of UAV key attribute data in step S1 includes the current position , the remaining power , the flight speed , the maximum load capacity , and the collection of UAV task data includes the target position , the initial priority , and the required time , required load .

[0017] Further, the specific implementation method of step S2 includes the following steps:

[0018] S2.1. Integrate the flight time, energy consumption and path risk of the unmanned aerial vehicle to comprehensively evaluate the total cost of the unmanned aerial vehicle to perform the task, and obtain the expression:

[0019]

[0020] wherein, is the total cost of the i-th unmanned aerial vehicle to perform the j-th task; is the required time for the i-th unmanned aerial vehicle to perform the j-th task; is the path risk value of the i-th unmanned aerial vehicle to perform the j-th task, is the differential item of time; , , are respectively the flight time weight coefficient, the energy consumption weight coefficient and the path risk weight coefficient, , , , determined by expert experience; ;

[0021] S2.2. From the three constraint conditions, the feasibility constraint discrimination method of the unmanned aerial vehicle and the unmanned aerial vehicle task is constructed, the first constraint condition is whether the remaining power of the unmanned aerial vehicle is sufficient to complete the path cost, the second constraint condition is whether the maximum load capacity of the unmanned aerial vehicle meets the task requirement, and the third constraint condition is whether the unmanned aerial vehicle can complete the execution within the required time range of the task at the current time, and the unmanned aerial vehicle and the unmanned aerial vehicle task pair that meet the three constraint conditions at the same time are screened.

[0022] Further, the dynamic updating mechanism of the priority of the unmanned aerial vehicle task of step S3 is to update the initial task priority dynamically by combining the delay degree and the urgency of the task, and the expression is:

[0023]

[0024] wherein, is the task priority of the j-th task, , are respectively the delay time coefficient and the urgency index coefficient, which are determined by expert experience; is the delay time of the j-th task, which is obtained by comparing the task requirement deadline time with the current time; is the urgency index, .

[0025] Further, step S4 constructs an explicit scoring function model of the UAV executing the task, introduces a nonlinear function and a combined feature factor to enhance the comprehensive expression ability of the energy consumption, load capacity, priority and urgency of the UAV, and obtains an expression of the rule score of the UAV executing the task as follows:

[0026]

[0027] wherein, is a rule score value of the ith UAV executing the jth task; is a task priority of the jth task; is an energy utilization efficiency coefficient, is a load matching coefficient, is a priority coefficient, is a comprehensive adaptation coefficient, is a constant, ~ and determined by expert experience; is a comprehensive work efficiency index; .

[0028] Further, the specific implementation method of step S5 includes the following steps:

[0029] S5.1. Defining a UAV-task feature vector as follows:

[0030]

[0031] wherein, is a UAV-task feature vector of the ith UAV executing the jth task;

[0032] S5.2. After the parameters in the UAV-task feature vector are normalized by using a numerical normalization method, the normalized parameters are input into the constructed scoring function model of the UAV executing the task based on the deep neural network, and an output score of the UAV executing the task based on the deep neural network is output, and the expression is as follows:

[0033]

[0034] wherein, is an output score value of the scoring function model of the UAV executing the task based on the deep neural network; is a multilayer perceptron neural network.

[0035] S5.3. Calculating a fusion score of the UAV executing the task, and the expression is as follows:

[0036]

[0037] wherein, a fusion score of the task performed by the UAV; is a fusion ratio, determined by expert experience.

[0038] Further, the time limit sensitive weighting factor introduced in step S6 is:

[0039]

[0040] wherein, is a time limit sensitive weighting factor; is a time sensitive coefficient, determined by expert experience.

[0041] The expression of the final score of the task performed by the UAV is:

[0042]

[0043] wherein, is the final score of the task performed by the UAV.

[0044] Further, the task assignment of the multi-UAV flight in step S7 is modeled as a constrained optimization problem, the objective of which is to maximize the total score of all assigned task pairs; by constructing binary assignment variables and imposing resource and uniqueness constraints, a greedy matching, linear programming or Hungarian algorithm is used to solve, and an optimal solution is obtained.

[0045] The optimization objective is as follows:

[0046]

[0047] The constraint conditions are as follows:

[0048]

[0049] wherein, represents whether the task j is assigned to the UAV i, if yes, the value is 1, if not, the value is 0; represents the maximum task capacity limit of the i-th UAV, determined by expert experience.

[0050] A system of a multi-UAV flight task assignment optimization method, comprising a processor, a memory and a computer program stored in the memory and executable on the processor, the computer program realizes the steps of the multi-UAV flight task assignment optimization method when executed.

[0051] The beneficial effects of the present application are:

[0052] The multi-unmanned aerial vehicle flight task allocation optimization method constructs a cost evaluation method integrating energy consumption, path risk and task allocation timeliness based on the state modeling of unmanned aerial vehicles and tasks, and filters feasible task combinations combined with resource and time constraints. On this basis, an explicit scoring model and an implicit scoring model based on deep representation learning are designed, and the comprehensive score is determined by weighted fusion. At the same time, the method introduces a time limit sensitive weighting function and a dynamic priority adjustment method to improve the response ability to urgent tasks. Finally, the task allocation solution is generated by a constraint optimization algorithm, and local redistribution update is performed when the task or flight state changes.

[0053] The multi-unmanned aerial vehicle flight task allocation optimization method constructs a scoring model integrating energy consumption, risk and time limit factors, and introduces explicit and implicit scoring expressions combining deep neural networks. The scoring mechanism has both interpretability and learning ability, significantly improving the accuracy and adaptability of task allocation evaluation; it realizes intelligent task allocation under complex scenarios with multiple targets and multiple constraint conditions, making the allocation result more scientific and reasonable, and improving the overall task completion rate and system scheduling efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 The flowchart of the multi-unmanned aerial vehicle flight task allocation optimization method of the present application;

[0055] Figure 2 The multi-unmanned aerial vehicle flight task allocation optimization method of the present application; DETAILED DESCRIPTION

[0056] In order to make the purpose, technical scheme and advantages of the present application clearer and more understandable, the present application will be further described in detail below in combination with the drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application, i.e. the described specific embodiments are only a part of the embodiments of the present application, but not all the specific embodiments. The components of the specific embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations, and the present application can also have other embodiments.

[0057] Therefore, the detailed description of the specific embodiments of the present application provided below in the drawings is not intended to limit the scope of the claimed present application, but only represents selected specific embodiments of the present application. Based on the specific embodiments of the present application, all other specific embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0058] In order to further understand the invention content, characteristics and effects of the present application, the following specific embodiments are exemplified, and the drawings are combinedFigure 1 - attached Figure 2 The detailed description is as follows:

[0059] Embodiment 1:

[0060] A multi-UAV flight task allocation optimization method, comprising the following steps:

[0061] S1. Collecting UAV key attribute data to construct a UAV set, and collecting UAV task data to construct a UAV task set;

[0062] Further, the step S1 of collecting UAV key attribute data includes the current position of the i-th UAV , the remaining power , the flight speed , the maximum load capacity , and the task data of the UAV includes the target position of the j-th task , the initial priority , the required time , and the required load .

[0063] Further, each UAV is represented by a state vector, which includes spatial position information, power, load capacity, and speed key attributes; each task includes target position, time window, load demand, and priority key parameters. This modeling serves as the basis for information input of the entire task allocation.

[0064] The UAV set is denoted as: ;

[0065] wherein is the key attribute set of the i-th UAV, ; the current position, the remaining power, the flight speed, and the total number of UAVs can be obtained from the flight control system; the maximum load capacity is pre-set; the target position, the initial priority, the required time for the task, and the required load for the task are directly given by the task system;

[0066] S2. Based on the data collected in step S1, the total cost of the UAV executing the task is calculated by considering the flight time required by the UAV, the energy consumption, and the danger degree on the path, and then a feasibility constraint discrimination method for the UAV and the UAV task is constructed to screen the UAV and the UAV task pair satisfying the feasibility constraint;

[0067] Further, the specific implementation method of step S2 includes the following steps:

[0068] S2.1. The total cost of the UAV executing the task is comprehensively evaluated by integrating the flight time required by the UAV, the energy consumption, and the danger degree on the path, and the expression is:

[0069]

[0070] wherein, is the total cost of the ith UAV performing the jth task; is the required time of the ith UAV performing the jth task; is the path risk value of the ith UAV performing the jth task, is the differential item of time; , , are respectively the flight time weight coefficient, the energy consumption weight coefficient, and the path risk weight coefficient, , , , are determined by expert experience; ;

[0071] The flight time reflects efficiency, the speed square term embodies energy consumption, and the risk value represents safety, and the three are flexibly adjusted through the weight coefficients to finally obtain a comprehensive cost evaluation value;

[0072] S2.2. A feasibility constraint discrimination method for UAVs and UAV tasks is constructed from three constraint conditions. The first constraint condition is whether the remaining power of the UAV is sufficient to complete the path cost, the second constraint condition is whether the maximum load capacity of the UAV meets the task requirement, and the third constraint condition is whether the UAV can complete the execution within the required time range of the task at the current time. The UAV and the UAV task pair that simultaneously satisfy the three constraint conditions are screened.

[0073] Further, only the task pair that simultaneously satisfies the above three types of constraints enters the next stage of scoring and allocation process to ensure the physical feasibility and time accessibility of the task allocation result. At the same time, the method is optimized by introducing factors such as path risk discount, power utilization efficiency, and task time window smoothing penalty, so that the judgment process is more reasonable and continuous, avoids task rejection or incorrect allocation due to boundary condition fluctuations, and significantly improves the robustness and fault tolerance of task scheduling.

[0074] The feasibility discrimination formula is as follows:

[0075]

[0076] wherein, is the remaining power of the ith UAV, obtained by the flight control system; is the path risk value of the ith UAV performing the jth task; is the maximum value of the path risk value; is the power safety coefficient, determined by expert experience; total cost of the ith UAV to perform the jth task; maximum load capacity of the ith UAV; task required load of the jth task; time required for the jth task; current position of the ith UAV. target position of the jth task. flight speed of the ith UAV. a micro constant, determined by expert experience; a carrying capacity threshold, determined by expert experience; a time matching threshold, determined by expert experience;

[0077] Further, is a judgment on the available power, which needs to meet the task execution cost, and also considers the path risk and the urgency of the task time. is a judgment on the carrying capacity, which is proportional to the task required load, which can improve the fault tolerance and flexibility of the model and avoid boundary misjudgment. is a judgment on the time matching, which judges whether the UAV can complete the task within the required time.

[0078] S3. Establish a dynamic priority updating mechanism for UAV tasks to update the initial priority and obtain the task priority;

[0079] Further, the dynamic priority updating mechanism of the UAV task in step S3 is to dynamically update the initial task priority by combining the task delay degree and the urgency, and the expression is:

[0080]

[0081] wherein, the task priority of the jth task, , respectively, are the delay time coefficient and the urgency index coefficient, determined by expert experience; the delay time of the jth task, obtained by comparing the task required deadline with the current time; the urgency index, .

[0082] Further, the task priority can evolve dynamically over time in the task allocation period, especially for tasks that have been delayed or are about to expire, the urgency should be appropriately increased.

[0083] S4. Constructing a rule scoring function model of the UAV executing the task by considering the energy utilization efficiency under the unit execution cost, the matching degree between the current load capacity of the UAV and the required load of the task, the task priority, and the energy and speed adaptation efficiency within the unit task time;

[0084] Further, step S4 constructs an explicit scoring function model of the UAV executing the task, introduces a nonlinear function and a combination feature factor to enhance the comprehensive expression ability of the energy consumption, load capacity, priority and urgency of the UAV, and obtains an expression of the rule score of the UAV executing the task as follows:

[0085]

[0086] wherein, is the rule score value of the ith UAV executing the jth task; is the task priority of the jth task; is the energy utilization efficiency coefficient, is the load matching coefficient, is the priority coefficient, is the comprehensive adaptation coefficient, is a constant, ~ and are determined by expert experience; is a comprehensive work efficiency index; .

[0087] Further, the explicit scoring function model of the UAV executing the task not only considers the energy efficiency ratio under the unit cost, but also introduces the logarithmic weighting of the electric quantity on the task cost and the nonlinear mapping of the load matching. Meanwhile, in order to better reflect the task adaptation and execution efficiency, a complex feature exponential function is designed to fuse the speed, electric quantity and demand time. The model has more distinguishability and robustness in structure, and can adapt to the task allocation decision demand in complex dynamic environment. In the above formula, is used to represent the energy utilization efficiency under the unit execution cost; is used to represent the matching degree between the current load capacity of the UAV and the required load of the task; is used to represent the task priority; comprehensively reflects the energy-speed adaptation efficiency within the unit task time, and embodies the comprehensive execution ability of the UAV for the current task.

[0088] S5. Constructing a score function model of the UAV performing the task based on a deep neural network, extracting deep-level correlations between the UAV state and the task attributes by learning historical samples, and then fusing the rule score function model of the UAV performing the task obtained in step S3 to obtain a fusion score of the UAV performing the task;

[0089] Further, the specific implementation method of step S5 includes the following steps:

[0090] S5.1. Defining the UAV-task feature vector as:

[0091]

[0092] wherein, is the UAV-task feature vector of the ith UAV performing the jth task;

[0093] S5.2. After normalizing the parameters in the UAV-task feature vector by using a numerical normalization method, inputting the normalized parameters into the constructed score function model of the UAV performing the task based on a deep neural network, and outputting the score of the UAV performing the task based on the deep neural network, the expression is:

[0094]

[0095] wherein, is the output score value of the score function model of the UAV performing the task based on a deep neural network; is a multi-layer perceptron neural network.

[0096] Further, considering that the traditional explicit function is difficult to depict the nonlinear task allocation strategy in a complex environment, a score prediction model based on a deep neural network is introduced. The model extracts deep-level correlations between the UAV state and the task attributes by learning historical samples, thereby estimating the potential benefits of task completion. The model can adapt to different scene strategy preferences and improve the adaptability of the task allocation method to high-dimensional and variable inputs.

[0097] For the convenience of understanding, the specific implementation process of the model is described as follows:

[0098] (1) Feature engineering and data preprocessing.

[0099] To ensure the effect and generalization ability of model training, systematic preprocessing of the original feature data is needed. This step mainly solves the problem of data imbalance and provides high-quality training samples for the subsequent deep learning model.

[0100] The input feature vector is ;

[0101] The numerical normalization method is used to scale the score to the interval [0, 1].

[0102] (2) Neural network structure design

[0103] The architecture of the neural network directly affects the model's expressive power and learning effect. Through the structure of the multilayer perceptron, the deep correlation between features is extracted layer by layer, and finally the normalized score value is output. Each layer of the network is designed to balance the model complexity and computational efficiency.

[0104] The architecture of the multilayer perceptron neural network is as follows:

[0105] 1. Input layer: 10 neurons, corresponding to .

[0106] 2. Hidden layer 1: 32 neurons, ReLU activation function;

[0107] 3. Hidden layer 2: 16 neurons, ReLU activation function;

[0108] 4. Output layer: 1 neuron, corresponding to , Sigmoid activation function.

[0109] The interlayer connection adopts a full connection mode.

[0110] (3) Model training process

[0111] Model training is the core step of the entire learning process, which requires reasonable setting of training parameters and establishment of effective monitoring mechanism. Through the designed training strategy, it is ensured that the model can fully learn the rules in the historical data, while avoiding the problem of overfitting.

[0112] The training parameters of the model are set as follows:

[0113] 1. Batch size: 64;

[0114] 2. Learning rate: 0.001, using Adam optimizer;

[0115] 3. Training rounds: 100, using early stopping strategy;

[0116] 4. Loss function: binary cross-entropy;

[0117] After the model training is completed, the parameters are input into the multilayer perceptron neural network, and the output is obtained.

[0118] S5.3. Calculate the fusion score of the unmanned aerial vehicle performing the task, the expression is:

[0119] ​

[0120] wherein, is the fusion score of the UAV performing the task; is the fusion ratio, determined by expert experience.

[0121] S6. Introducing a time limit sensitive weighting factor to the fusion score of the UAV performing the task obtained in step S5, to obtain the final score of the UAV performing the task;

[0122] Further, the time limit sensitive weighting factor introduced in step S6 is:

[0123]

[0124] wherein, is the time limit sensitive weighting factor; is the time sensitive coefficient, determined by expert experience;

[0125] Further, in order to make the multi-UAV task allocation method more sensitive to respond to tasks with time limit approaching, an exponential weighting function with remaining time as input is introduced. This function significantly suppresses or boosts the score value when the remaining time is short, forcing high time efficiency tasks to be allocated preferentially within limited time. This method can adapt to the time limit urgency of different tasks and improve the overall task completion rate.

[0126] The expression of the final score of the UAV performing the task is:

[0127]

[0128] wherein, is the final score of the UAV performing the task.

[0129] S7. Based on the final score of the UAV performing the task obtained in step S6, the multi-UAV flight task allocation problem is converted into solving the maximum total score value of all UAVs performing task pairs, to obtain the multi-UAV flight task allocation optimization scheme.

[0130] Further, in step S7, the multi-UAV flight task allocation is modeled as an optimization problem with constraints, the goal is to maximize the total score value of all allocated task pairs; by constructing binary allocation variables and imposing resource and uniqueness constraints, greedy matching, linear programming or Hungarian algorithm is used for solving to obtain the optimal solution;

[0131] The optimization goal is as follows:

[0132]

[0133] The constraint conditions are as follows:

[0134]

[0135] wherein, represents whether the task j is assigned to the unmanned aerial vehicle i, and if yes, the value is 1, and if not, the value is 0; represents the maximum task capacity limit of the i-th unmanned aerial vehicle, which is determined by expert experience.

[0136] The implementation scheme of the present embodiment is as follows:

[0137] A typical multi-unmanned aerial vehicle cooperative task scenario is considered, in which there are 4 unmanned aerial vehicles and 8 tasks to be executed. Each task has a fixed target position, initial priority, task required time, and task required load; each unmanned aerial vehicle has a current position, remaining power, flight speed, maximum load capacity, and other attributes. The specific information is as follows, Table 1 is the unmanned aerial vehicle information, and Table 2 is the task information;

[0138] Table 1

[0139]

[0140] Table 2

[0141]

[0142] Based on the above information, the following is calculated: Then, the feasibility is judged, and the fusion score value is obtained through the calculation of the rule score value and the learning model output value. Considering the time weighting factor, the final score is finally calculated. Part of the results are shown in Table 3;

[0143] Table 3

[0144]

[0145] The optimal allocation result of the task is shown in Table 4 and Figure 2

[0146] Table 4

[0147]

[0148] Embodiment 2:

[0149] A system of a multi-unmanned aerial vehicle flight task allocation optimization method, comprising a processor, a memory and a computer program stored in the memory and executable on the processor, the computer program implements the steps of a multi-unmanned aerial vehicle flight task allocation optimization method as described in embodiment 1 when running.

[0150] ​It has to be noted that the terms "first", "second", and the like in connection with an entity or action refer to this entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without further constraints, exclude the presence of additional elements of the process, method, article, or apparatus.

[0151] While the application has been described with reference to specific implementations thereof, it should be understood that various modifications and substitutions can be made by those skilled in the art without departing from the scope of the present application. In particular, any one of the features of the present application disclosed above can be utilized independently of any other and the scope of the application should not be limited by the specific embodiments disclosed herein, but should be given the widest coverage possible in its true scope.

Claims

1. A method for optimizing task allocation during multi-UAV flight, characterized in that, Includes the following steps: S1. Collect key attribute data of drones to construct a drone set, and collect task data of drones to construct a task set of drones; S2. Based on the data collected in step S1, the total cost of the UAV mission is calculated, taking into account the flight time, energy consumption, and danger level along the path. Then, a feasibility constraint judgment method for UAVs and UAV missions is constructed to screen UAVs and UAV mission pairs that meet the feasibility constraints. S3. Establish a dynamic priority update mechanism for UAV missions to update the initial priority and obtain the mission priority; S4. For the UAVs and UAV mission pairs that satisfy the feasibility constraints obtained in step S2, construct a rule-based scoring function model for UAV mission execution, considering the energy utilization efficiency per unit execution cost, the degree of matching between the current payload capacity of the UAV and the payload required by the mission, mission priority, and energy and speed adaptation efficiency per unit mission time. S5. Construct a scoring function model for UAV task execution based on deep neural networks. By learning from historical samples, extract the deep-level correlation between UAV state and task attributes. Then, fuse it with the rule-based scoring function model for UAV task execution obtained in step S3 to obtain the fused score for UAV task execution. S6. The fusion score of the UAV mission obtained in step S5 is used to introduce a time-sensitive weighting factor to obtain the final score of the UAV mission. S7. Based on the final score of the UAV mission obtained in step S6, the task allocation problem of multi-UAV flight is transformed into solving the problem of maximizing the total score of all UAV mission pairs, thus obtaining the task allocation optimization scheme for multi-UAV flight.

2. The task allocation optimization method for multi-UAV flight according to claim 1, characterized in that, Step S1 involves collecting key attribute data for the UAV, including the current position of the i-th UAV. Remaining battery power Flight speed Maximum load capacity The mission data collected from the UAV includes the target location of the j-th mission. Initial priority Time required Required load capacity .

3. A system for optimizing task allocation during multi-UAV flight, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed, implements the steps of a task allocation optimization method for multi-UAV flight as described in any one of claims 1-2.

Images