Low-altitude economic unmanned aerial vehicle communication optimization method based on machine learning

Abstract

The invention discloses a low-altitude economic unmanned aerial vehicle communication optimization method based on machine learning, and relates to the related technical field of unmanned aerial vehicle communication optimization, and the method comprises the steps: obtaining a plurality of positions and a plurality of planning paths of a plurality of unmanned aerial vehicles; the method comprises the following steps: acquiring time demand coefficients of multiple unmanned aerial vehicles, sorting according to a minimum-to-maximum sequence, screening to obtain a target unmanned aerial vehicle, and performing cluster communication influence analysis on the target unmanned aerial vehicle according to a target position, a target planning path, other multiple positions and other multiple planning paths of the target unmanned aerial vehicle to obtain communication influence parameters. And based on machine learning, according to the target time demand coefficient and the communication influence parameter of the target unmanned aerial vehicle, performing path optimization to obtain an optimized target path. And performing flight control of the target unmanned aerial vehicle according to the optimized target path. A communication high-interference area is effectively avoided while the task time constraint is met, so that the communication stability of the flight of the unmanned aerial vehicle and the intelligence of path scheduling are improved.

Description

Technical Field

[0001] This application relates to the technical field of UAV communication optimization, specifically to a machine learning-based method for optimizing the communication of low-altitude economic UAVs. Background Technology

[0002] With the rapid development of the low-altitude economy, the application of drones in cluster operations in urban logistics, emergency delivery, and inspection and monitoring is becoming increasingly common. However, in the current technology, the frequent communication link congestion and signal interference during the collaborative operation of multiple drones seriously affect mission execution efficiency and flight safety. Traditional drone path planning methods usually focus on the shortest path for obstacle avoidance, ignoring the dynamic changes in the communication environment during flight, making it difficult to cope with real-time communication interference and timeliness conflicts in complex airspace.

[0003] Especially in areas with dense drone operations, the frequent spatial convergence of multiple drones causes a significant attenuation of communication link signals, which further leads to delays in mission data uploads, frequent path replanning, and even communication interruptions, affecting the stable operation of the entire drone system.

[0004] Therefore, in the existing technology, the communication stability and path scheduling intelligence of UAV flights are poor, which leads to the technical problem of declining operational stability of UAV scheduling systems. Summary of the Invention

[0005] This application provides a machine learning-based communication optimization method for low-altitude economical unmanned aerial vehicles (UAVs), addressing the technical problem of poor communication stability and path scheduling intelligence in existing UAV technologies, which leads to a decline in the operational stability of UAV scheduling systems. It effectively avoids high-interference communication areas while meeting mission time constraints, thereby improving the communication stability and path scheduling intelligence of UAVs.

[0006] This application provides a machine learning-based method for optimizing communication of low-altitude economic unmanned aerial vehicles (UAVs). The method includes: acquiring multiple locations and multiple planned paths of multiple UAVs, wherein the multiple UAVs form a UAV swarm; acquiring time demand coefficients of multiple UAVs, sorting them in ascending order, and filtering to obtain target UAVs; performing swarm communication impact analysis on the target UAVs based on their target locations, target planned paths, other locations, and other planned paths to obtain communication impact parameters; optimizing the paths based on machine learning, according to the target time demand coefficients and communication impact parameters of the target UAVs, to obtain optimized target paths, wherein optimization parameters are configured based on the target time demand coefficients and communication impact parameters for path optimization; and performing flight control of the target UAVs according to the optimized target paths.

[0007] In the implementation method, obtaining multiple locations and multiple planned paths of multiple drones includes: obtaining the real-time flight coordinates of multiple drones, wherein the multiple drones form a drone cluster; and obtaining multiple planned paths of multiple drones.

[0008] In the implementation method, multiple time requirement coefficients of drones are obtained, sorted in ascending order, and target drones are selected. This includes: obtaining the task time and task deadline of multiple drones, and obtaining the real-time time; calculating multiple remaining times based on the real-time time and multiple task deadlines; calculating multiple time requirement coefficients based on the multiple remaining times and multiple task times, wherein the size of the remaining time is positively correlated with the size of the time requirement coefficient; sorting the multiple time requirement coefficients in ascending order, and selecting the drone corresponding to the first time requirement coefficient as the target drone.

[0009] In the implementation method, based on the target UAV's target location, target planned path, and multiple other locations and multiple other planned paths, a cluster communication impact analysis of the target UAV is performed to obtain communication impact parameters. This includes: obtaining the target UAV's target location and target planned path, and filtering multiple associated locations and multiple associated planned paths of other UAVs; performing flight simulation before the task deadline based on the target location, target planned path, multiple associated locations, and multiple associated planned paths, obtaining flight coordinate points where the flight distance between the target UAV and other UAVs is less than or equal to a preset distance threshold, and obtaining a set of impact coordinate points; inputting the set of impact coordinate points into a communication impact analyzer to perform cluster communication impact analysis of the target UAV, and outputting communication impact parameters.

[0010] In the implementation, the training steps of the communication impact analyzer include: collecting multiple sets of sample impact coordinate points and the total impact parameters of UAV communication under different sets of sample impact coordinate points based on UAV swarm flight communication test data over a historical period, and labeling to obtain a set of sample communication impact parameters; constructing a communication impact analyzer based on a machine learning architecture; and using the multiple sets of sample impact coordinate points and the set of sample communication impact parameters to perform supervised training on the communication impact analyzer until the test accuracy converges.

[0011] In the implementation, based on machine learning, path optimization is performed according to the target time demand coefficient and communication impact parameters of the target UAV to obtain an optimized target path. This includes: processing and obtaining the optimization convergence count and optimization step size based on the target time demand coefficient and communication impact parameters, wherein the optimization step size includes the proportion of adjusting the target planned path; constructing a target path optimization space based on multiple other planned paths; randomly adjusting the target planned path within the target path optimization space according to the optimization step size to obtain a first target path; processing and obtaining a first communication fitness based on the first target path and multiple other planned paths; and continuing to adjust and optimize the target path until the optimization convergence count is reached to obtain the optimized target path with the highest communication fitness during the optimization process.

[0012] In the implementation method, based on the target time requirement coefficient and communication impact parameters, the optimization convergence count and optimization step size are obtained, including: obtaining a preset optimization convergence count and preset optimization step size; obtaining the average communication impact parameters during the UAV swarm operation process; calculating the ratio of the communication impact parameters to the average communication impact parameters; and adjusting the preset optimization convergence count and preset optimization step size in conjunction with the target time requirement coefficient to obtain the optimization convergence count and optimization step size.

[0013] In the implementation, based on the first target path and multiple other planned paths, a first communication fitness is obtained by processing, including: performing flight simulation before the task deadline based on the target location, the first target path, multiple associated locations of other UAVs, and multiple associated planned paths; obtaining flight coordinate points where the flight distance between the target UAV and other UAVs is less than or equal to a preset distance threshold; obtaining a first set of influencing coordinate points; inputting the first set of influencing coordinate points into a communication influence analyzer to perform cluster communication influence analysis of the target UAVs; outputting a first communication influence parameter; and calculating the ratio of the communication influence parameter to the first communication influence parameter to obtain the first communication fitness.

[0014] This application proposes a machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles (UAVs). The method involves acquiring multiple locations and planned paths for multiple UAVs. The time demand coefficients of these UAVs are obtained and sorted in ascending order to identify target UAVs. Based on the target UAV's target location, target planned path, and other locations and planned paths, a cluster communication impact analysis is performed to obtain communication impact parameters. Using machine learning, path optimization is conducted based on the target UAV's target time demand coefficient and communication impact parameters to obtain an optimized target path. Flight control of the target UAV is then performed according to the optimized target path. This method solves the technical problem of poor communication stability and intelligent path scheduling in existing UAV technologies, which leads to decreased operational stability of the UAV scheduling system. It effectively avoids high-interference communication areas while meeting task time constraints, thereby improving the communication stability and intelligent path scheduling of UAVs. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0016] Figure 1 A schematic flowchart of a machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles (UAVs) is provided for an embodiment of this application.

[0017] Figure 2 This is a schematic diagram illustrating the process of selecting target drones using a machine learning-based communication optimization method for low-altitude economic drones, as provided in an embodiment of this application. Detailed Implementation

[0018] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application.

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] In the following description, references to "some embodiments" describe a subset of all possible embodiments. However, it is understood that "some embodiments" may be the same or different subsets of all possible embodiments and may be combined with each other without conflict. The terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0021] This application provides a machine learning-based method for optimizing communication for low-altitude economical unmanned aerial vehicles (UAVs), such as... Figure 1 As shown, the method includes:

[0022] Multiple locations and multiple planned paths of multiple drones are obtained, wherein the multiple drones form a drone swarm; the time demand coefficients of multiple drones are obtained, sorted in ascending order, and target drones are selected; based on the target drone's target location, target planned path, multiple other locations, and multiple other planned paths, the swarm communication impact analysis of the target drone is performed to obtain communication impact parameters.

[0023] The scheduling platform acquires a drone swarm currently in a flight scheduling task. This swarm consists of multiple drones. The current positions and pre-planned flight paths of these drones are obtained. Subsequently, the time requirement coefficients of these drones are acquired and sorted in ascending order to filter and identify the target drone. The time requirement coefficient is the ratio of remaining task execution time to the total task time. The target drone is identified as the one with the most urgent time requirement. Further, based on the target drone's target position, target planned path, and multiple other positions and planned paths, a swarm communication impact analysis is performed to obtain communication impact parameters.

[0024] The method provided in this application embodiment further includes: obtaining the real-time flight coordinates of multiple drones, wherein the multiple drones form a drone cluster; and obtaining multiple planned paths of the multiple drones.

[0025] The system acquires multiple locations and planned paths for multiple drones, including: obtaining the real-time flight coordinates of multiple drones from a scheduling platform. These flight coordinates represent the spatial location information of the drones at the current moment, including longitude, latitude, and flight altitude data. The real-time flight coordinates of the drones are marked with timestamps. All drones are part of a drone swarm. The system also acquires the planned paths for each drone from the scheduling platform. These planned paths include the starting point, the ending point, the specific flight route, and altitude data.

[0026] like Figure 2 As shown in the embodiments of this application, the method further includes: obtaining the task time and task deadline of multiple drones, and obtaining the real-time time; calculating multiple remaining times based on the real-time time and multiple task deadlines; calculating multiple time demand coefficients based on the multiple remaining times and multiple task times, wherein the size of the remaining time is positively correlated with the size of the time demand coefficients; sorting the multiple time demand coefficients in ascending order, and selecting the drone corresponding to the first time demand coefficient as the target drone.

[0027] The process involves obtaining time requirement coefficients for multiple drones, sorting them in ascending order, and then filtering to select target drones. This includes: obtaining the task time and task deadline for each drone; the task time being the standard time required for the drone to complete its current task from start to finish; and the task deadline being the latest time the drone can complete its task. Real-time information is also obtained, representing the drone's current running time. Subsequently, based on the real-time information and multiple task deadlines, multiple remaining times are calculated. These remaining times represent the remaining time for each drone to execute its task, i.e., the difference between the task deadline and the real-time information. Based on these remaining times and task times, multiple time requirement coefficients are obtained by calculating the ratio of the remaining time to the task time. The magnitude of the remaining time is positively correlated with the magnitude of the time requirement coefficient. Finally, the multiple time requirement coefficients are sorted in ascending order, and the drone corresponding to the highest time requirement coefficient is selected as the target drone.

[0028] The method provided in this application embodiment further includes: obtaining the target location and target planned path of the target UAV; filtering multiple associated locations and multiple associated planned paths of other UAVs; performing flight simulation before the task deadline based on the target location, target planned path, multiple associated locations and multiple associated planned paths; obtaining flight coordinate points where the flight distance between the target UAV and other UAVs is less than or equal to a preset distance threshold; and obtaining a set of influencing coordinate points; inputting the set of influencing coordinate points into a communication influence analyzer to perform cluster communication influence analysis of the target UAVs, and outputting communication influence parameters.

[0029] Based on the target UAV's target location, target planned path, and multiple other locations and planned paths, a cluster communication impact analysis of the target UAV is performed to obtain communication impact parameters. This includes: acquiring the target UAV's target location and target planned path; filtering multiple associated locations and multiple associated planned paths of other UAVs; the associated locations being the distribution locations of UAVs in the UAV cluster whose distance from the target UAV at the target location is less than or equal to a preset distance threshold; and the multiple associated planned paths being the planned paths of UAVs that overlap with or are less than or equal to the preset distance threshold of the target UAV's target planned path. Based on the target location, target planned path, multiple associated locations, and multiple associated planned paths, a flight simulation is performed before the task deadline. The flight simulation is a uniform-speed flight simulation of each UAV's corresponding planned path at multiple associated locations and multiple associated planned paths, executed according to multiple remaining time intervals. Then, flight coordinates of the target UAV whose flight distance from other UAVs during the flight simulation is less than or equal to a preset distance threshold are obtained, resulting in an impact coordinate point set, which consists of coordinates within the target UAV's target planned path. Finally, the impact coordinate point set is input into a communication impact analyzer to perform cluster communication impact analysis of the target UAV, outputting the communication impact parameters.

[0030] The method provided in this application embodiment further includes: collecting multiple sets of sample influence coordinate points and the total influence parameters of UAV communication under different sets of sample influence coordinate points based on UAV swarm flight communication test data within a historical time period, and labeling to obtain a set of sample communication influence parameters; constructing a communication influence analyzer based on a machine learning architecture; and using the multiple sets of sample influence coordinate points and the set of sample communication influence parameters to perform supervised training on the communication influence analyzer until the test accuracy converges.

[0031] The training steps of the communication impact analyzer include: acquiring UAV swarm flight communication test data over a historical period; collecting multiple sample impact coordinate point sets from the communication test data; these sample impact coordinate point sets are the path coordinate points of the test UAV when the spatial flight distance between the test UAV and other UAVs is less than a preset distance threshold; and including the specific number and distance data of other UAVs in the samples. The total impact parameter of UAV communication impact is collected under different sample impact coordinate point sets, and the sample impact coordinate point sets are labeled with the corresponding total impact parameter to obtain a sample communication impact parameter set. The communication impact parameter is the UAV signal attenuation at each sample impact coordinate point under the corresponding impact condition; that is, the unit of measurement for the radio signal strength received by the UAV transmitter is usually dBm. The UAV signal attenuation is the signal attenuation under the influence of other UAVs compared to an interference-free environment. For example, if there are 5 UAVs interfering nearby, and the RSSI drops sharply from -55dBm to -75dBm, the corresponding total impact parameter is -20dBm. Furthermore, a communication impact analyzer based on a machine learning architecture is constructed, which is built based on a neural network. Finally, the multiple sets of sample influence coordinate points and sample communication influence parameters are input into an untrained communication influence analyzer for supervised training until the model converges and the test accuracy meets the requirements. The trained communication influence analyzer is then obtained. This analyzer is used to analyze the acquired set of influence coordinate points to obtain the corresponding communication influence parameters.

[0032] Based on machine learning, path optimization is performed according to the target time demand coefficient and communication impact parameters of the target UAV to obtain an optimized target path. Optimization parameters are configured based on the target time demand coefficient and communication impact parameters for path optimization. Flight control of the target UAV is then performed according to the optimized target path.

[0033] Based on machine learning, path optimization is performed according to the target time requirement coefficient and communication impact parameters of the target UAV to obtain an optimized target path. During path optimization, optimization parameters are configured based on the target time requirement coefficient and communication impact parameters. Specifically, the path optimization of the target UAV is performed based on the communication impact parameters and the target time requirement coefficient. The larger the communication impact parameter and the higher the target time requirement coefficient, the more rounds of path optimization can be performed to accurately obtain the best target path. Finally, the flight control of the target UAV is performed according to the optimized target path. This solves the technical problem in existing technologies where poor communication stability and path scheduling intelligence in UAV flight lead to a decline in the operational stability of the UAV scheduling system. It effectively avoids high-interference communication areas while meeting task time constraints, thereby improving the communication stability and intelligent path scheduling of UAV flight.

[0034] The method provided in this application embodiment further includes: processing and obtaining the number of optimization convergences and the optimization step size according to the target time demand coefficient and communication impact parameters, wherein the optimization step size includes the proportion of adjusting the target planning path; constructing a target path optimization space according to multiple other planning paths; randomly adjusting the target planning path within the target path optimization space according to the optimization step size to obtain a first target path; processing and obtaining a first communication fitness according to the first target path and multiple other planning paths; continuing to adjust and optimize the target path until the number of optimization convergences is reached to obtain the optimized target path with the largest communication fitness during the optimization process.

[0035] Based on machine learning, path optimization is performed according to the target UAV's target time requirement coefficient and communication impact parameters to obtain an optimized target path. This includes: to ensure a balance between mission timeliness and communication stability for the target UAV, multiplying the target time requirement coefficient by the normalized communication impact parameters to obtain a calculation result. Subsequently, the optimization convergence count and optimization step size corresponding to the calculation result are obtained from a demand mapping table. The demand mapping table stores different calculation results and their corresponding optimization convergence counts and optimization step sizes. Matching processing is performed based on the calculation results to obtain the optimization convergence count and optimization step size. The optimization step size includes the proportion of adjusting the target planned path, i.e., the maximum path adjustment percentage when randomly adjusting the target planned path. The optimization convergence count is the number of iterations for subsequent path optimization.

[0036] Furthermore, a target path optimization space is constructed based on multiple other planned paths. This target path optimization space is built by combining the original path with multiple other planned paths. The target path optimization space must ensure that the UAV does not exceed the designated flight area, avoids no-fly zones or altitude-restricted areas, and adopts an avoidance principle to reduce overlap with other UAV paths. Within the target path optimization space, the target planned path is randomly adjusted according to the optimization step size. Random adjustment methods include altitude adjustment and / or temporary path offset and recovery to obtain the first target path.

[0037] Based on the first target path and several other planned paths, a first communication fitness is obtained through communication fitness calculation. This first communication fitness is then compared with the fitness calculation results of the target planned paths, and the path with the highest fitness is selected as the current optimal target path. Using the current optimal target path as the optimization benchmark, the optimization target path continues to be adjusted until the optimization convergence count is reached, thus obtaining the optimized target path with the highest communication fitness during the optimization process.

[0038] The method provided in this application embodiment further includes: obtaining a preset number of optimization convergences and a preset optimization step size; obtaining an average communication impact parameter during the operation of a drone swarm; calculating the ratio of the communication impact parameter to the average communication impact parameter, and adjusting the preset number of optimization convergences and the preset optimization step size in conjunction with the target time requirement coefficient to obtain the optimized number of convergences and the optimization step size.

[0039] Based on the target time requirement coefficient and communication impact parameters, the optimization convergence count and optimization step size are obtained, including: acquiring a preset optimization convergence count and preset optimization step size, which are the maximum optimization convergence count and path optimization step size set in advance. Subsequently, the average communication impact parameter during the UAV swarm operation is acquired, which is the average of the communication impact parameters acquired by all UAVs during the operation. Further, the product of the ratio of the communication impact parameter to the average communication impact parameter and the target time requirement coefficient is calculated, and the product calculation result is obtained. The preset optimization convergence count is multiplied by the product calculation result, and the preset optimization step size is multiplied by the product calculation result. The result is rounded up to complete the adjustment calculation, yielding the optimization convergence count and optimization step size. For example, if the optimization step size is 10%, then 10% of the path is randomly adjusted each time, while ensuring path continuity, and the starting and ending points of the path are the target position and the task position, respectively.

[0040] The method provided in this application embodiment further includes: performing flight simulation before the task deadline based on the target location, the first target path, multiple associated locations of other UAVs, and multiple associated planned paths; obtaining flight coordinate points where the flight distance between the target UAV and other UAVs is less than or equal to a preset distance threshold; obtaining a first set of influencing coordinate points; inputting the first set of influencing coordinate points into a communication influence analyzer to perform cluster communication influence analysis of the target UAVs; outputting a first communication influence parameter; and calculating the ratio of the communication influence parameter to the first communication influence parameter to obtain a first communication fitness.

[0041] Based on the first target path and multiple other planned paths, a first communication fitness is obtained, including: based on the target location, the first target path, and multiple associated locations and associated planned paths of other UAVs, a flight simulation is performed before the task deadline to obtain flight coordinate points where the flight distance between the target UAV and other UAVs is less than or equal to a preset distance threshold, thus obtaining a first set of influencing coordinate points. Further, the first set of influencing coordinate points is input into a communication influence analyzer to perform cluster communication influence analysis of the target UAVs, outputting a first communication influence parameter. The first communication influence parameter is the communication influence parameter corresponding to the optimized first target path. The ratio of the communication influence parameter to the first communication influence parameter is calculated to obtain the first communication fitness, where a higher ratio corresponds to a better optimization effect, and vice versa.

[0042] The technical solution provided by this invention involves acquiring multiple locations and planned paths of multiple drones, where the drones form a drone swarm; acquiring time demand coefficients of the drones, sorting them in ascending order, and filtering to obtain target drones; performing swarm communication impact analysis on the target drones based on their target locations, target planned paths, other locations, and other planned paths to obtain communication impact parameters; optimizing the paths based on machine learning using the target drones' target time demand coefficients and communication impact parameters to obtain optimized target paths, wherein optimization parameters are configured based on the target time demand coefficients and communication impact parameters for path optimization; and performing flight control of the target drones according to the optimized target paths. This solves the technical problem in existing technologies where poor communication stability and intelligent path scheduling during drone flight lead to a decline in the operational stability of the drone scheduling system. It effectively avoids high-interference communication areas while meeting task time constraints, thereby improving the communication stability and intelligent path scheduling of drone flight.

[0043] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. A communication optimization method for low-altitude economic unmanned aerial vehicles (UAVs) based on machine learning, characterized in that, The method includes: The system acquires multiple locations and multiple planned paths for multiple drones, wherein the multiple drones form a drone swarm. The time requirement coefficients of multiple drones are obtained, sorted in ascending order, and the target drones are selected. Based on the target drone's target location, target planned path, multiple other locations, and multiple other planned paths, the cluster communication impact analysis of the target drones is performed to obtain communication impact parameters. Based on machine learning, the path is optimized according to the target time demand coefficient and communication impact parameters of the target UAV to obtain an optimized target path. The optimization parameters are configured according to the target time demand coefficient and communication impact parameters to perform path optimization. The target UAV is controlled to fly according to the optimized target path.

2. The machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles according to claim 1, characterized in that, Obtain multiple locations and multiple planned paths for multiple drones, including; The real-time flight coordinates of multiple drones are obtained, wherein the multiple drones form a drone swarm; Obtain multiple planned paths for multiple drones.

3. The machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles according to claim 1, characterized in that, Obtain the time requirement coefficients of multiple drones, sort them in ascending order, and filter to obtain the target drones, including: Obtain the mission times and deadlines of multiple drones, and retrieve the real-time time. Based on the real-time time and the deadlines for multiple tasks, multiple remaining times are calculated. Based on multiple remaining times and multiple task times, multiple time demand coefficients are calculated, where the magnitude of the remaining time and the magnitude of the time demand coefficient are positively correlated. The time demand coefficients are sorted in ascending order, and the drone corresponding to the first time demand coefficient is selected as the target drone.

4. The machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles according to claim 1, characterized in that, Based on the target UAV's target location, target planned path, multiple other locations, and multiple other planned paths, a cluster communication impact analysis of the target UAV is performed to obtain communication impact parameters, including: Obtain the target location and target planned path of the target UAV, and filter multiple associated locations and multiple associated planned paths of other UAVs; Based on the target location, target planned path, multiple associated locations and multiple associated planned paths, a flight simulation is performed before the task deadline to obtain the flight coordinates of the target UAV and other UAVs whose flight distance is less than or equal to a preset distance threshold, thus obtaining a set of influencing coordinates. The set of influencing coordinate points is input into the communication impact analyzer to perform cluster communication impact analysis on the target UAV and output the communication impact parameters.

5. The machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles according to claim 4, characterized in that, The training steps for the communication impact analyzer include: Based on historical data of drone swarm flight communication tests, multiple sets of sample influence coordinate points were collected, as well as the total influence parameters of drone communication under different sets of sample influence coordinate points, and the sample communication influence parameter set was obtained by labeling. Build a communication impact analyzer based on a machine learning architecture; The communication influence analyzer is trained under supervised supervision using the set of multiple sample influence coordinate points and the set of sample communication influence parameters until the test accuracy converges.

6. The machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles according to claim 1, characterized in that, Based on machine learning, and according to the target time demand coefficient and communication impact parameters of the target UAV, path optimization is performed to obtain an optimized target path, including: Based on the target time demand coefficient and communication impact parameters, the number of optimization convergences and the optimization step size are obtained, wherein the optimization step size includes the proportion of adjusting the target planning path; Construct the target path optimization space based on multiple other planning paths; Within the target path optimization space, the target planned path is randomly adjusted according to the optimization step size to obtain the first target path; Based on the first target path and multiple other planned paths, a first communication fitness is obtained; Continue to adjust and optimize the target path until the optimization convergence count is reached, and obtain the optimized target path with the highest communication fitness during the optimization process.

7. The machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles according to claim 6, characterized in that, Based on the target time requirement coefficient and communication impact parameters, the optimization convergence count and optimization step size are obtained, including: Obtain the preset number of convergence iterations and the preset optimization step size; Obtain the average communication impact parameters during drone swarm operations; Calculate the ratio of the communication impact parameter to the average communication impact parameter, and combine it with the target time requirement coefficient to adjust and calculate the preset number of optimization convergences and the preset optimization step size to obtain the optimized number of convergences and the optimization step size.

8. The machine learning-based communication optimization method for low-altitude economic unmanned aerial vehicles according to claim 6, characterized in that, Based on the first target path and multiple other planned paths, a first communication fitness is obtained, including: Based on the target location, the first target path, multiple associated locations of other UAVs, and multiple associated planned paths, a flight simulation is performed before the mission deadline to obtain the flight coordinates of the target UAV and other UAVs where the flight distance is less than or equal to a preset distance threshold, thus obtaining the first set of influencing coordinates. The first set of influencing coordinate points is input into the communication influence analyzer to perform cluster communication influence analysis of the target UAV and output the first communication influence parameter. Calculate the ratio of the communication impact parameter to the first communication impact parameter to obtain the first communication fitness.

Images