Low-altitude airspace distribution method and device based on machine learning and medium

Abstract

The invention discloses a low-altitude airspace distribution method and device based on machine learning, and a medium, and relates to the technical field of intelligent airspace scheduling, and the method comprises the steps: extracting the demand characteristics of flight state data and meteorological data from a preliminary optimization data set, carrying out the multi-modal data fusion, obtaining an airspace resource demand characteristic, and carrying out the multi-modal data fusion; summarizing to form an airspace demand analysis report; according to the airspace demand analysis report, variance analysis is carried out on the flight state demand characteristics and the meteorological state demand characteristics to obtain a variance analysis F value, main factors influencing airspace distribution are screened according to the variance analysis F value and are sorted, and an airspace priority sorting report is generated; and according to the airspace priority ranking report, airspace distribution optimization is carried out to obtain an airspace distribution scheme. According to the method, main factors influencing airspace allocation are screened and sorted, so that the flexibility and optimization potential of airspace resource allocation are improved, and the decision-making capability driven by machine learning is reflected.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent airspace scheduling, in particular to a low-altitude airspace allocation method, device and medium based on machine learning. BACKGROUND

[0002] In the field of low-altitude airspace management, with the rapid growth of unmanned aerial vehicles and other aircraft, achieving efficient and safe dynamic allocation of airspace resources has become a key technical challenge. Traditional airspace allocation methods usually rely on pre-set fixed flight routes, rule-based scheduling strategies or relatively static priority models. These methods focus on processing macro-static information such as flight plans and airspace structures, and to some extent can meet the demand for regular and predictable airspace use. In recent years, with the development of sensing technology and aviation data collection capabilities, machine learning technology has begun to be explored for application in airspace situation awareness, conflict prediction and other aspects, aiming to improve the automation level of airspace management through data-driven methods. These attempts provide new technical ideas for handling more complex dynamic airspace environments.

[0003] However, existing data-driven airspace management methods have certain limitations in processing efficiency when faced with multi-source, heterogeneous and real-time changing dynamic data. On the one hand, in terms of feature extraction and fusion, existing methods often use relatively independent analysis paths for flight state, weather environment and other multi-modal data, failing to deeply quantify and effectively fuse the comprehensive influence weight of various factors on airspace demand, which may lead to inaccurate assessment of airspace emergency or potential demand. On the other hand, in terms of factor selection and optimization of allocation decisions, conventional methods lack an effective mechanism that can dynamically identify and sort out dominant factors from numerous influencing factors in a self-adaptive and quantitative manner, which makes the generation process of airspace allocation schemes likely to rely on experience or fixed weights, making it difficult to quickly respond to the dynamic changes of airspace demand characteristics, thereby affecting the overall optimization efficiency of airspace resource utilization. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a low-altitude airspace allocation method based on machine learning to solve the problems of difficulty in effectively fusing multi-modal airspace demand characteristics and inability to dynamically identify dominant factors in the prior art.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a low-altitude airspace allocation method based on machine learning, comprising, collecting multi-source dynamic data, cleaning and preprocessing the multi-source dynamic data to obtain a preliminary optimized data set; The flight state data and weather data of the preliminary optimization data set are required to extract the demand characteristics and perform multi-modal data fusion to obtain airspace resource demand characteristics, and are summarized as an airspace demand analysis report; According to the airspace demand analysis report, variance analysis is performed on the flight state demand characteristics and weather state demand characteristics to obtain a variance analysis F value, the main factors affecting the airspace allocation are screened and sorted according to the variance analysis F value, and an airspace priority sorting report is generated. According to the airspace priority sorting report, the airspace allocation optimization is performed to obtain an airspace allocation scheme.

[0007] As a preferred scheme of the low-altitude airspace allocation method based on machine learning, wherein: The multi-source dynamic data includes flight parameters and weather data of the unmanned aerial vehicle; The flight parameters of the unmanned aerial vehicle include the position of the unmanned aerial vehicle, the instantaneous speed of the unmanned aerial vehicle, the height of the unmanned aerial vehicle, and the flight direction of the unmanned aerial vehicle; and the weather data includes wind speed.

[0008] As a preferred scheme of the low-altitude airspace allocation method based on machine learning, wherein: Collecting multi-source dynamic data, recording the sampling time points of the multi-source dynamic data, extracting the earliest time point of the sampling time points as a time reference, and performing translation on the multi-source dynamic data to obtain aligned multi-source dynamic data; The aligned multi-source dynamic data is subjected to missing value detection and abnormal value detection, and data cleaning is completed to obtain cleaned multi-source dynamic data; The cleaned data is subjected to standardization processing and summarization to obtain a preliminary optimization data set.

[0009] As a preferred scheme of the low-altitude airspace allocation method based on machine learning, wherein: The flight state data and weather data of the preliminary optimization data set are required to extract the demand characteristics and perform multi-modal data fusion to obtain airspace resource demand characteristics, and are summarized as an airspace demand analysis report; The flight state demand characteristics and weather state demand characteristics are subjected to weighted summation to obtain airspace resource demand characteristics, and all airspace resource demand characteristics are summarized as an airspace demand analysis report.

[0010] As a preferred scheme of the low-altitude airspace allocation method based on machine learning, wherein: Reading each unmanned aerial vehicle windless flight position point, calculating the total length of the unmanned aerial vehicle windless flight path and the overall angle change; The total flight time of the drone after completing the total length of the windless flight path is calculated, and the average speed of the drone is obtained by calculating the ratio of the total length of the windless flight path to the total flight time. Extract the instantaneous flight speed of the UAV and calculate the standard deviation of the instantaneous flight speed as the speed fluctuation rate; Extract the drone's flight direction, calculate the difference between the drone's flight directions at adjacent time points, and calculate the ratio with the time difference to obtain the local direction change rate. Calculate the average of all local direction change rates to obtain the overall direction change rate. The low-altitude airspace is divided into high-low altitude, middle-low altitude, and low-low altitude layers. The altitude information of the UAV is extracted, and the frequency of the UAV in the high-low altitude, middle-low altitude, and low-low altitude layers is calculated to obtain the altitude distribution of the UAV. The altitude difference between adjacent UAVs is calculated to obtain the altitude change, and the standard deviation of the altitude change is calculated to obtain the altitude volatility. The flight state requirement characteristics are obtained by linearly weighting and summing the total length of the windless flight path, the overall angle change, the average speed of the UAV, the speed fluctuation rate, the overall direction change rate, the altitude fluctuation rate, and the altitude distribution of the UAV.

[0011] As a preferred embodiment of the low-altitude airspace allocation method based on machine learning described in this invention, the extraction of demand features from the meteorological data includes the following steps: Extract wind speed, calculate wind speed change, and calculate the average of all wind speed changes to obtain the overall wind speed change. Extract the location points of windy flight and calculate the total length of the windy flight path. The change in flight path length is calculated based on the total length of the windy flight path and the total length of the windless flight path. At the same time, the ratio of this change to the overall wind speed change is calculated to obtain the wind speed adjustment coefficient. The wind speed adjustment factor is then calculated based on the wind speed adjustment coefficient. The flight path affected by wind speed is calculated based on the wind speed adjustment factor and the flight path length, and the wind speed influence degree and wind speed weight are calculated. The meteorological state requirement characteristics are obtained by multiplying the wind speed weight and the wind speed influence.

[0012] As a preferred embodiment of the machine learning-based low-altitude airspace allocation method described in this invention, the steps for generating the airspace priority ranking report are as follows: Based on the airspace demand analysis report, the total length of windless flight paths is grouped, and the between-group variance and within-group variance of the total length of windless flight paths are calculated to obtain the F-value of the variance analysis of the total length of windless flight paths. The variance analysis values ​​for overall angle change, average speed of the UAV, speed fluctuation rate, overall direction change rate, and altitude fluctuation rate were calculated using the same method. Based on the wind speed adjustment factor, the meteorological state demand characteristics are grouped, and the variance between meteorological state groups and the variance within meteorological state groups are calculated to obtain the F-value of the analysis of variance of meteorological state. Among the factors of total length of windless flight path, overall angle change, average speed of UAV, speed fluctuation rate, overall direction change rate, altitude fluctuation rate, and wind speed, the one with the largest F-value in the analysis of variance is selected as the main factor affecting airspace allocation, and they are sorted in descending order of the F-value in the analysis of variance to obtain an airspace priority ranking report.

[0013] As a preferred embodiment of the machine learning-based low-altitude airspace allocation method described in this invention, the steps for obtaining the airspace allocation scheme are as follows: Allocate airspace resources to drones according to the airspace priority ranking report; Record the location and allocation method of each UAV airspace resource, and summarize the location and allocation method of each UAV airspace resource into an airspace allocation scheme.

[0014] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the low-altitude airspace allocation method based on machine learning as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the low-altitude airspace allocation method based on machine learning as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: by extracting demand features and fusing multimodal data, the comprehensive impact of flight status and meteorological factors on airspace demand is quantified and integrated, realizing the standardization and structuring of multi-source heterogeneous airspace demand information; by screening and ranking the main factors affecting airspace allocation, the flexibility and optimization potential of airspace resource allocation are improved, demonstrating the decision-making ability driven by machine learning. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Fig. 1 This is a flowchart of a machine learning-based low-altitude airspace allocation method.

[0019] Fig. 2 A flowchart for obtaining a preliminary optimized dataset.

[0020] Fig. 3 A flowchart for obtaining an airspace demand analysis report.

[0021] Fig. 4 A flowchart for generating an airspace priority sorting report. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figs. 1-4 This is one embodiment of the present invention, which provides a low-altitude airspace allocation method based on machine learning, including the following steps: S1. Collect multi-source dynamic data, clean and preprocess the multi-source dynamic data to obtain a preliminary optimized dataset.

[0026] Multi-source dynamic data refers to dynamic data collected from different data sources, including UAV flight parameters and meteorological data.

[0027] The drone's flight parameters include its position, instantaneous speed, altitude, and flight direction; meteorological data includes wind speed.

[0028] Collect multi-source dynamic data, record the sampling time points of the multi-source dynamic data, extract the earliest time point as the time reference, and translate the multi-source dynamic data to obtain aligned multi-source dynamic data.

[0029] Furthermore, time alignment is performed on dynamic data from different data sources. The sampling time points of multi-source dynamic data are recorded, the earliest time point among all sampling time points is extracted, and the hourly time closest to the earliest time point is used as the time reference. The difference between the first sampling time point of each multi-source dynamic data and the time reference is calculated to obtain the time difference. Each multi-source dynamic data is shifted according to the time difference to obtain aligned multi-source dynamic data. A time axis is established based on the smallest sampling time interval among all multi-source dynamic data.

[0030] The multi-source dynamic data is aligned and subjected to missing value detection and outlier detection to complete data cleaning, resulting in cleaned multi-source dynamic data.

[0031] Furthermore, the multi-source dynamic data is traversed and aligned to perform missing value detection. On the timeline, it is checked whether each sampling time point contains dynamic data from all data sources. If dynamic data from different data sources is missing at a sampling time point, the missing data is filled in by interpolation.

[0032] Outlier detection is performed on the aligned multi-source dynamic data that has completed missing value detection. The mean and standard deviation of the multi-source dynamic data are calculated. The sum of the mean and twice the standard deviation of the multi-source dynamic data is calculated to obtain the upper limit of outliers. The difference between the mean and twice the standard deviation of the multi-source dynamic data is calculated to obtain the lower limit of outliers. The range between the upper and lower limits of outliers is defined as the outlier interval. Multi-source dynamic data outside the outlier interval are removed as outliers. Data is filled in at the outlier positions using interpolation to complete the data cleaning and obtain the cleaned multi-source dynamic data.

[0033] It should be noted that defining an interval based on the sum of the mean and two standard deviations, and the difference between the mean and two standard deviations, can retain normal values ​​and eliminate outliers. This interval contains 95% of the data.

[0034] The cleaned data is standardized and summarized to obtain a preliminary optimized dataset.

[0035] Furthermore, the cleaned multi-source dynamic data is standardized using z-score standardization. The mean and standard deviation of the dynamic data from different data sources are extracted and standardized, as shown in the expression: ; in, To standardize dynamic data, For dynamic data, This is the average value of dynamic data. The standard deviation is the dynamic data.

[0036] The standardized dynamic data from different data sources are aggregated to obtain a preliminary optimized dataset.

[0037] S2. Extract the demand characteristics of flight status data and meteorological data from the preliminary optimized dataset and perform multimodal data fusion to obtain airspace resource demand characteristics, and summarize them into an airspace demand analysis report.

[0038] Demand features for flight status data and meteorological data were extracted from the preliminary optimized dataset.

[0039] Read the location of each drone in windless flight, and calculate the total length of the drone's windless flight path and the overall angle change.

[0040] Furthermore, the required features of flight status data are extracted from the preliminary optimized dataset. A three-dimensional coordinate system is established with the UAV's takeoff position as the origin. On the time axis, windless UAV position data is extracted according to time intervals as windless flight position points, and the windless three-dimensional coordinates of each windless flight position point are read. The distance between two adjacent windless flight position points is calculated based on the windless three-dimensional coordinates, and all distances are summed to obtain the total length of the windless flight path. The total length of the windless flight path for all UAVs is summed to obtain the total length of the windless flight path for all UAVs. The ratio of the total length of the windless flight path to the total length of the windless flight path for all UAVs is calculated to obtain the influence degree of the total length of the windless flight path. This influence degree is normalized, and the ratio of the influence degree of the total length of the windless flight path to the sum of the influence degrees of the total length of the windless flight path for all UAVs is calculated to obtain the weight of the total length of the windless flight path.

[0041] It should be noted that "no wind" refers to a wind force level of 0, where the wind speed measured at a height of 10 meters in a standard meteorological observation field is between 0.0 and 0.2 meters per second.

[0042] The three-dimensional coordinates of the first three windless flight positions are extracted sequentially. The first and second windless flight positions are used to form the first position vector, and the second and third windless flight positions are used to form the second position vector. The first local angle change is then calculated, expressed as follows: ; in, For the first local angle change, Let be the first position vector. This is the second position vector.

[0043] The three-dimensional coordinates of three adjacent windless flight positions are extracted sequentially, and the local angle changes are calculated sequentially. The average value of all local angle changes is calculated to obtain the overall angle change. The overall angle changes of all UAVs are summed to obtain the overall angle change of all UAVs. The contribution of the overall angle change is calculated by comparing the overall angle change contribution with the contribution of the overall angle changes of all UAVs. The weight of the overall angle change is then calculated by comparing the overall angle change contribution with the contribution of the overall angle changes of all UAVs.

[0044] The total flight time of the drone to complete the total length of the windless flight path is calculated, and the average speed of the drone is obtained by calculating the ratio of the total length of the windless flight path to the total flight time.

[0045] The instantaneous flight speed of the drone at each time point is extracted on the time axis according to the time interval, and the standard deviation of the instantaneous flight speed of the drone is calculated as the speed fluctuation rate.

[0046] Using the geographic North Pole as a reference, the aircraft orientation angle at each time point on the time axis is extracted as the UAV flight direction. The difference between the UAV flight directions at adjacent time points is calculated and the ratio is calculated with the time difference to obtain the local direction change rate. The overall direction change rate is obtained by averaging all local direction change rates.

[0047] The low-altitude airspace is divided into three equal parts: the uppermost layer is designated as the high-low altitude layer, the middle layer as the mid-low altitude layer, and the lowest layer as the low-low altitude layer. UAV altitude information is extracted based on time intervals on the time axis, and the frequency of UAVs in the high-low altitude layer, mid-low altitude layer, and low-low altitude layer is calculated to obtain the altitude distribution of UAVs.

[0048] It should be noted that low-altitude airspace refers to the airspace below 1000 meters above the ground.

[0049] The altitude difference between adjacent UAVs is calculated to obtain the altitude change, and the standard deviation of the altitude change is calculated to obtain the altitude fluctuation rate.

[0050] The weights of the UAV's average speed, speed fluctuation rate, overall direction change rate, and altitude fluctuation rate are calculated using the same method as those used to calculate the total length of the windless flight path and the overall angle change. At the same time, the weights of the UAV in the high and low altitude layers, the middle and low altitude layers, and the low and low altitude layers are calculated separately.

[0051] The flight state requirement characteristics are obtained by linearly weighting and summing the total length of the windless flight path, the overall angle change, the average speed of the UAV, the speed fluctuation rate, the overall direction change rate, the altitude fluctuation rate, and the altitude distribution of the UAV.

[0052] Extract wind speed, calculate wind speed change, and calculate the average of all wind speed changes to obtain the overall wind speed change. Extract the location points of windy flight and calculate the total length of the windy flight path.

[0053] Furthermore, meteorological data requirements features are extracted from the preliminary optimized dataset. Wind speed is extracted at each time point along the time axis, and the wind speed change between adjacent time points is calculated. The average of all adjacent wind speed changes is then calculated to obtain the overall wind speed change. Windy drone location data is extracted at each time point along the time axis as windy flight location points, and the windy 3D coordinates of each windy flight location point are read. The distance between two adjacent windy flight location points is calculated based on the windy 3D coordinates, and all distances are summed to obtain the total length of the windy flight path.

[0054] The change in flight path length is calculated based on the total length of the windy flight path and the total length of the windless flight path. This change is then compared with the overall wind speed change to obtain the wind speed adjustment coefficient. Finally, the wind speed adjustment factor is calculated based on the wind speed adjustment coefficient.

[0055] Furthermore, the total length of the windy flight paths of all drones is summed to obtain the total length of the windy flight paths of all drones. The difference between the total length of the windy flight paths of all drones and the total length of the windless flight paths of all drones is calculated to obtain the change in flight path length. The ratio of the change in flight path length to the overall change in wind speed is calculated to obtain the wind speed adjustment coefficient. The wind speed adjustment factor is then calculated, and the expression is as follows: ; in, For wind speed adjustment factor, This is the wind speed adjustment coefficient. This refers to wind speed.

[0056] The flight path affected by wind speed is calculated based on the wind speed adjustment factor and the flight path length, and the wind speed influence degree and wind speed weight are calculated.

[0057] Furthermore, the wind speed adjustment factor is multiplied by the total length of the windless flight path to obtain the flight path affected by wind speed. The wind-affected flight paths of all drones are summed to obtain the total number of wind-affected flight paths. The ratio of the wind-affected flight paths to the total number of wind-affected flight paths is calculated and normalized to obtain the wind speed influence degree. The total number of drones is counted, and the wind speed weight is calculated using the following expression: ; in, Assuming wind speed as the weighting, For the influence of wind speed, Index for the number of drones The total number of drones, For the first The impact of wind speed on the deployment of drones For the first Wind speed adjustment factor for drones For the first Wind speed adjustment coefficient for drones.

[0058] The meteorological state requirement characteristics are obtained by multiplying the wind speed weight and the wind speed influence.

[0059] The total contribution of flight status is calculated by summing the weights of the total length of the windless flight path, the overall angle change, the average speed of the UAV, the speed fluctuation rate, the overall direction change rate, the altitude fluctuation rate, and the weights of the UAV at high and low altitudes, mid and low altitudes. The wind speed weight is used as the total contribution of meteorological status, and the weights of flight status and meteorological status are calculated separately. The expression for calculating the flight status weight is as follows: ; in, For flight state weights, The total contribution of flight status. This represents the total contribution of meteorological conditions.

[0060] The expression for calculating the meteorological state weights is: ; in, The weights are based on weather conditions.

[0061] The flight status demand characteristics and meteorological status demand characteristics of each UAV are linearly weighted and summed to obtain the airspace resource demand characteristics. All airspace resource demand characteristics are then summarized into an airspace demand analysis report.

[0062] S3. Based on the airspace demand analysis report, perform variance analysis on the flight status demand characteristics and meteorological status demand characteristics to obtain the variance analysis F value. Based on the variance analysis F value, screen the main factors affecting airspace allocation and sort them to generate an airspace priority ranking report.

[0063] Based on the airspace demand analysis report, the total length of windless flight paths is grouped, and the between-group variance and within-group variance of the total length of windless flight paths are calculated to obtain the F-value of the variance analysis of the total length of windless flight paths.

[0064] Furthermore, the flight status requirements were grouped based on the total length of the windless flight path, overall angle change, average speed of the UAV, speed fluctuation rate, overall direction change rate, and altitude fluctuation rate. For the total length of the windless flight path, the maximum and minimum values ​​of the total length of the windless flight path for all UAVs were extracted to define the range of the total length of the windless flight path. This range was then divided into three equal parts: a low-impact group (Group 1), a medium-impact group (Group 2), and a high-impact group (Group 3). The total length of windless flight paths in each group was counted. The average length of the total windless flight paths in each group was calculated to obtain the average total length of windless flight paths in the low-impact group, medium-impact group, and high-impact group. The average of these averages was then calculated to obtain the overall average total length of windless flight paths. Finally, the variance of the total length of windless flight paths between groups was calculated, expressed as: ; in, The variance between groups represents the total length of the windless flight path. The number of groups representing the total length of windless flight paths. For the first The total length of the group's windless flight paths. This is the index of the total length of the windless flight path. For the first The average total length of the group's windless flight path. This represents the average total length of the entire windless flight path.

[0065] Calculate the sum of squares of the differences between the total length of each windless flight path and the mean of the total length of windless flight paths within the same group. Use this sum as the sum of squares for the total length of windless flight paths within the group. Sum the total number of windless flight paths in each group to obtain the total number of windless flight paths. Then calculate the variance of the total length of windless flight paths within the group, expressed as: ; in, The variance within the set of the total length of the windless flight path. This represents the total length and number of windless flight paths. It is the sum of squares within the total length of the windless flight path.

[0066] The F-value of the analysis of variance for the total length of the windless flight path is obtained by calculating the ratio between the variances of the total length of the windless flight path and within the variances of the total length of the windless flight path.

[0067] It should be noted that the F-value in ANOVA is a statistical measure used in ANOVA to measure the degree of influence of different factors on changes in airspace demand. The larger the F-value, the more significant the influence of the factor on airspace demand.

[0068] Analysis of variance was performed using the same method to calculate the overall angle change, average speed of the UAV, speed fluctuation rate, overall direction change rate, and altitude fluctuation rate. value.

[0069] Based on the wind speed adjustment factor, the meteorological condition demand characteristics are grouped, and the variances between and within the meteorological condition groups are calculated to obtain the F-value of the analysis of variance for the meteorological conditions.

[0070] Furthermore, wind speed adjustment factors for all drones are extracted. The maximum and minimum values ​​of the wind speed adjustment factors are extracted as the range of wind speed adjustment factors. This range is then divided into three equal parts: a low-impact wind speed group (Group 1), a medium-impact wind speed group (Group 2), and a high-impact wind speed group (Group 3). The number of meteorological state demand features for each group is counted, and the average values ​​of these features are calculated to obtain the mean values ​​for the low-impact, medium-impact, and high-impact wind speed groups. The mean values ​​for these three groups are then calculated to obtain the overall mean value for meteorological state demand features. Finally, the variance between the meteorological state groups is calculated, expressed as: ; Among them, among them, The variance between groups for weather conditions. The number of characteristic groups required for meteorological conditions. For the first The number of meteorological condition requirements for the group For the index of meteorological state groups, For the first The mean of the meteorological condition demand characteristics of the group. This represents the average value of the overall meteorological condition requirements.

[0071] Calculate the sum of squared differences between each meteorological state demand feature and the mean of the meteorological state demand features within the same group. Use this sum as the within-group sum of squared meteorological state demand features for each group. Summate the total number of meteorological state demand features for each group to obtain the total number of meteorological state demand features. Then calculate the within-group variance of the meteorological state groups. The expression is as follows: ; in, For the within-group variance of weather conditions, The quantity of characteristics required for the overall meteorological state. This represents the sum of squares within the meteorological state group.

[0072] The F-value of the analysis of variance for meteorological states is obtained by calculating the ratio between the variances of meteorological state groups to the variances within the meteorological state groups.

[0073] Among the factors of total length of windless flight path, overall angle change, average speed of UAV, speed fluctuation rate, overall direction change rate, altitude fluctuation rate, and wind speed, the one with the largest F-value in the analysis of variance is selected as the main factor affecting airspace allocation, and they are sorted in descending order of the F-value in the analysis of variance to obtain an airspace priority ranking report.

[0074] S4. Based on the airspace priority sorting report, optimize the airspace allocation to obtain the airspace allocation scheme.

[0075] Furthermore, airspace resources are allocated to UAVs according to the airspace priority ranking report. Based on machine learning, historical airspace resource allocation records are retrieved in the order of the airspace priority ranking report. Machine learning algorithms automatically learn the impact of various demand characteristics on airspace allocation from historical data. Each UAV's various demand characteristics are compared with those in the historical airspace resource allocation records, following the order of the airspace priority ranking report. If the UAV's various demand characteristics are exactly the same as those in the historical airspace resource allocation records, airspace resources are allocated to the UAV according to the method of the historical airspace resource allocation records. If the UAV's various demand characteristics are not exactly the same as those in the historical airspace resource allocation records, airspace resources are allocated based on the historical airspace resource allocation records, immediately adjacent to the airspace resource allocation of the previous UAV. Simultaneously, the various demand characteristics and allocation method of the current UAV are recorded to provide a reference for allocation optimization.

[0076] Record the location and allocation method of each UAV airspace resource, and summarize the location and allocation method of each UAV airspace resource into an airspace allocation scheme.

[0077] This embodiment also provides a computer device applicable to the low-altitude airspace allocation method based on machine learning, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the low-altitude airspace allocation method based on machine learning as proposed in the above embodiment.

[0078] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0079] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the low-altitude airspace allocation method based on machine learning as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0080] In summary, this invention achieves standardization and structuring of multi-source heterogeneous airspace demand information by: extracting demand features and fusing multimodal data to quantify and integrate the comprehensive impact of flight status and meteorological factors on airspace demand; and by screening and ranking the main factors affecting airspace allocation, it enhances the flexibility and optimization potential of airspace resource allocation, demonstrating machine learning-driven decision-making capabilities.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A low-altitude airspace allocation method based on machine learning, characterized in that: include, Collect multi-source dynamic data, clean and preprocess the multi-source dynamic data to obtain a preliminary optimized dataset; Demand characteristics of flight status data and meteorological data were extracted from the preliminary optimized dataset and multimodal data fusion was performed to obtain airspace resource demand characteristics, which were then summarized into an airspace demand analysis report. Based on the airspace demand analysis report, variance analysis is performed on the flight status demand characteristics and meteorological status demand characteristics to obtain the variance analysis F value. The main factors affecting airspace allocation are screened and ranked according to the variance analysis F value to generate an airspace priority ranking report. Based on the airspace priority ranking report, airspace allocation is optimized to obtain an airspace allocation scheme.

2. The low-altitude airspace allocation method based on machine learning as described in claim 1, characterized in that: The multi-source dynamic data includes the UAV's flight parameters and meteorological data; The flight parameters of the drone include the drone's position, instantaneous speed, altitude, and flight direction. The meteorological data includes wind speed.

3. The low-altitude airspace allocation method based on machine learning as described in claim 2, characterized in that: The preliminary optimized dataset is obtained through the following steps. Collect multi-source dynamic data, record the sampling time points of the multi-source dynamic data, extract the earliest time point of the sampling time points as the time reference, and translate the multi-source dynamic data to obtain aligned multi-source dynamic data; The multi-source dynamic data is aligned and missing value and outlier detection is performed to clean the data and obtain cleaned multi-source dynamic data. The cleaned data is standardized and summarized to obtain a preliminary optimized dataset.

4. The low-altitude airspace allocation method based on machine learning as described in claim 3, characterized in that: The steps to obtain the airspace demand analysis report are as follows: Demand features for flight status data and meteorological data were extracted from the preliminary optimized dataset. The airspace resource demand characteristics are obtained by weighted summation of the flight status demand characteristics and the meteorological status demand characteristics, and all airspace resource demand characteristics are summarized into an airspace demand analysis report.

5. The low-altitude airspace allocation method based on machine learning as described in claim 4, characterized in that: The steps for extracting the required features from the flight status data are as follows: Read the windless flight position of each drone and calculate the total length of the windless flight path and the overall angle change of the drone. The total flight time of the drone after completing the total length of the windless flight path is calculated, and the average speed of the drone is obtained by calculating the ratio of the total length of the windless flight path to the total flight time. Extract the instantaneous flight speed of the UAV and calculate the standard deviation of the instantaneous flight speed as the speed fluctuation rate; Extract the drone's flight direction, calculate the difference between the drone's flight directions at adjacent time points, and calculate the ratio with the time difference to obtain the local direction change rate. Calculate the average of all local direction change rates to obtain the overall direction change rate. The low-altitude airspace is divided into high-low altitude, middle-low altitude, and low-low altitude layers. The altitude information of the UAV is extracted, and the frequency of the UAV in the high-low altitude, middle-low altitude, and low-low altitude layers is calculated to obtain the altitude distribution of the UAV. The altitude difference between adjacent UAVs is calculated to obtain the altitude change, and the standard deviation of the altitude change is calculated to obtain the altitude volatility. The flight state requirement characteristics are obtained by linearly weighting and summing the total length of the windless flight path, the overall angle change, the average speed of the UAV, the speed fluctuation rate, the overall direction change rate, the altitude fluctuation rate, and the altitude distribution of the UAV.

6. The low-altitude airspace allocation method based on machine learning as described in claim 5, characterized in that: The steps for extracting the demand features from the meteorological data are as follows: Extract wind speed, calculate wind speed change, and calculate the average of all wind speed changes to obtain the overall wind speed change. Extract the location points of windy flight and calculate the total length of the windy flight path. The change in flight path length is calculated based on the total length of the windy flight path and the total length of the windless flight path. At the same time, the ratio of this change to the overall wind speed change is calculated to obtain the wind speed adjustment coefficient. The wind speed adjustment factor is then calculated based on the wind speed adjustment coefficient. The flight path affected by wind speed is calculated based on the wind speed adjustment factor and the flight path length, and the wind speed influence degree and wind speed weight are calculated. The meteorological state requirement characteristics are obtained by multiplying the wind speed weight and the wind speed influence.

7. The low-altitude airspace allocation method based on machine learning as described in claim 6, characterized in that: The steps for generating the spatial priority ranking report are as follows: Based on the airspace demand analysis report, the total length of windless flight paths is grouped, and the between-group variance and within-group variance of the total length of windless flight paths are calculated to obtain the F-value of the variance analysis of the total length of windless flight paths. The variance analysis values ​​for overall angle change, average speed of the UAV, speed fluctuation rate, overall direction change rate, and altitude fluctuation rate were calculated using the same method. Based on the wind speed adjustment factor, the meteorological state demand characteristics are grouped, and the variance between meteorological state groups and the variance within meteorological state groups are calculated to obtain the F-value of the analysis of variance of meteorological state. Among the factors of total length of windless flight path, overall angle change, average speed of UAV, speed fluctuation rate, overall direction change rate, altitude fluctuation rate, and wind speed, the one with the largest F-value in the analysis of variance is selected as the main factor affecting airspace allocation, and they are sorted in descending order of the F-value in the analysis of variance to obtain an airspace priority ranking report.

8. The low-altitude airspace allocation method based on machine learning as described in claim 7, characterized in that: The steps to obtain the airspace allocation scheme are as follows: Allocate airspace resources to drones according to the airspace priority ranking report; Record the location and allocation method of each UAV airspace resource, and summarize the location and allocation method of each UAV airspace resource into an airspace allocation scheme.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the low-altitude airspace allocation method based on machine learning as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the low-altitude airspace allocation method based on machine learning as described in any one of claims 1 to 8.

Images