Method for predicting number of aerial drones based on ground vehicle trajectory data

Abstract

The application discloses a method for predicting the number of air unmanned aerial vehicles (UAVs) based on ground vehicle trajectory data, and belongs to the technical field of UAV control. In order to solve the problem of predicting the number of air unmanned aerial vehicles based on ground traffic state, the application comprises the following steps: constructing a vehicle trajectory set; constructing a traffic density thermal field; constructing an average speed trend vector field; constructing a direction consistency index; taking the traffic density thermal field as a basic load, the modulus of the average speed trend vector field representing fluidity, and the direction consistency index representing traffic order, and performing weighted combination to construct a demand intensity of air resources converted from a distribution pressure intensity; considering a spatial vector from a vehicle position to a nearest UAV take-off point and a terrain complexity factor, and constructing a preliminary UAV demand estimation field; introducing a task urgency weighting function, a current task backlog and a weather flight risk to construct a final UAV deployment suggestion function, and completing the prediction of the number of air unmanned aerial vehicles based on ground vehicle trajectory data.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) control technology, specifically relating to a method for predicting the number of aerial UAVs based on ground vehicle trajectory data. Background Technology

[0002] In recent years, with the accelerating pace of urbanization and residents' increasing demands for timely logistics services, urban logistics and distribution are facing unprecedented pressure. Especially during peak hours and in special scenarios (such as emergencies, severe weather, or concentrated deliveries during holidays), traditional delivery methods relying on ground vehicles are severely impacted by road congestion and poor traffic management, significantly affecting logistics efficiency and service quality. Drones, with their advantages of convenient flight, rapid response, and ability to avoid ground congestion, are becoming an important supplement to the urban delivery system, widely used in emergency medical care, express delivery, and urban management.

[0003] However, drone resources are limited, flight airspace is restricted, and take-off and landing point layouts are limited. Furthermore, flight safety is affected by various factors such as weather and terrain, making global deployment impossible. Therefore, scientific allocation based on real-time needs by region and time period is essential. Determining the number of drones needed for each region at different times has become a key technical challenge in realizing an air-ground collaborative scheduling system. Currently, methods for predicting drone deployment numbers mainly rely on historical delivery data, order density, population distribution, or simple geographic information, employing static or semi-static allocation strategies. These methods have several prominent problems: first, they lack real-time perception of the dynamic state of ground traffic, failing to reflect the impact of traffic congestion and flow trends on ground delivery efficiency; second, they do not consider the complexity and feasibility of the flight environment, such as take-off and landing point distances, building density, and no-fly zones; and third, they ignore the differentiated impact of task urgency and backlog, leading to insufficient or wasted resources. Moreover, very few methods organically integrate these factors to establish a unified, dynamically updated predictive model. Summary of the Invention

[0004] This invention aims to solve the problem of scientifically predicting the number of aerial drones required in different areas at different times based on ground traffic conditions, and proposes a method for predicting the number of aerial drones based on ground vehicle trajectory data.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for predicting the number of aerial drones based on ground vehicle trajectory data includes the following steps:

[0007] S1. Collect ground vehicle trajectory data and construct a vehicle trajectory set;

[0008] S2. Using vehicle speed, acceleration, and historical congestion frequency as weighting factors, a traffic density thermodynamic field is constructed using a direction-aware kernel function;

[0009] S3. Combine the speed and direction angle of each vehicle to form a speed vector, and then perform a weighted average in space using the direction sensing kernel function to construct an average speed trend vector field;

[0010] S4. For all vehicles in the spatial neighborhood, calculate the cosine similarity between the vehicle's heading angle and the average heading angle, and construct a heading consistency index;

[0011] S5. Using the traffic density thermal field as the basic load, the magnitude of the average velocity trend vector field represents the flow rate, and the direction consistency index represents the traffic order, a weighted combination is used to construct the delivery pressure intensity.

[0012] S6. Transform the delivery pressure intensity into the demand intensity of aerial resources, and construct a preliminary drone demand estimation field by considering the spatial vector from the vehicle location to the nearest drone take-off and landing point and the terrain complexity factor.

[0013] S7. The initial UAV demand estimate is revised by introducing a task urgency weighting function, current task backlog, and weather flight risk, and the final UAV deployment suggestion function is constructed to complete the prediction of the number of aerial UAVs based on ground vehicle trajectory data.

[0014] Furthermore, the expression for the set of vehicle trajectories in step S1 is:

[0015]

[0016] in, , Let x and y be the x and y coordinates of the position of the k-th vehicle at time t, respectively. , , These are the velocity, acceleration, and heading angle of the k-th vehicle at time t, respectively, obtained and calculated from the vehicle's GPS data; Let N be the set of vehicle trajectories; N is the total number of vehicles.

[0017] Furthermore, the expression for constructing the traffic density thermodynamic field in step S2 is as follows:

[0018]

[0019] in, The space density thermodynamic field; x is the abscissa of the spatial position, and y is the ordinate of the spatial position. The normalized historical congestion frequency of the k-th vehicle. The normalized speed of the kth vehicle. Let the normalized acceleration of the k-th vehicle be... This is the direction-aware kernel function.

[0020] Furthermore, the expression for the average velocity trend vector field constructed in step S3 is:

[0021]

[0022] in, It is a unit vector in direction; The weighted sum of the direction-aware kernel functions; It represents the average velocity trend vector field.

[0023] Furthermore, the expression for the directional consistency index constructed in step S4 is as follows:

[0024]

[0025]

[0026] in, As an indicator of directional consistency; The set of vehicles within radius r is obtained from vehicle trajectory data; is the average direction angle; i is the imaginary unit; arg is the phase angle function of the complex number; Let r be the set of vehicles within a radius r.

[0027] Furthermore, the expression for the delivery pressure intensity in step S5 is as follows:

[0028]

[0029] in, To assess the intensity of delivery pressure; The pressure intensity coefficient is determined by expert experience and is also used to adjust the dimensions. The magnitude of the velocity trend vector; The maximum road speed is determined based on the design value; The consistency penalty coefficient is set empirically.

[0030] Furthermore, the expression for the drone demand estimation field constructed in step S6 is:

[0031]

[0032] in, Preliminary demand estimation for drones; The safety redundancy factor is determined by expert experience; The maximum number of tasks a single drone can perform per unit of time is determined from the design specifications. The spatial vector from the vehicle to the nearest take-off and landing point is calculated from a GIS map. The reference space vector is determined by expert experience; This is the terrain complexity factor, obtained by experts based on GIS map analysis.

[0033] Furthermore, the specific implementation method of step S7 includes the following steps:

[0034] S7.1. Construct a task urgency weighting function, with the following expression:

[0035]

[0036] in, Weight the urgency of the task; The remaining task response time is obtained from the real-time scheduling system; The expected response time is determined by expert experience or actual requirements. The historical failure rate is obtained from historical data. The weighting of the current regional task type is determined by expert experience or actual requirements. As a time urgency factor, it is determined by expert experience and is also used to adjust the dimensions; As a risk-sensitive factor, These are task priority factors, all determined by expert experience;

[0037] S7.2. Construct the final drone deployment suggestion function, with the expression:

[0038]

[0039] in, Recommended number of drones for final deployment; Weighted by urgency; The backlog of tasks in the region is calculated by the scheduling system. The maximum task backlog threshold is determined by expert experience or actual requirements. Wind speed risk factors are determined by experts based on meteorological data.

[0040] The beneficial effects of this invention are:

[0041] This invention discloses a method for predicting the number of aerial drones based on ground vehicle trajectory data. First, by introducing multi-dimensional features such as traffic density, speed trends, and directional consistency, it achieves high-precision modeling of ground traffic conditions, enhancing the real-time sensitivity of drone deployment. Second, by integrating factors such as mission urgency, take-off and landing point accessibility, and terrain complexity, the prediction results better reflect actual flight conditions, avoiding situations where drones are unreachable or resources are wasted. Third, the system also incorporates mission backlog and weather risks to output dynamically adjustable deployment suggestions, improving the intelligence level of the scheduling system. Overall, this method possesses advantages such as strong dynamism, high adaptability, precise deployment, and good security, making it suitable for air-ground collaborative logistics scheduling in complex urban scenarios.

[0042] This invention presents a method for predicting the number of aerial drones (UAVs) needed based on ground vehicle trajectory data. It addresses the problem of scientifically predicting the required number of UAVs for different regions at different times based on ground traffic conditions. Traditional scheduling methods lack awareness of ground traffic dynamics, neglecting the impact of congestion, poor mobility, and directional confusion on ground delivery efficiency, and failing to accurately identify the necessity and spatiotemporal distribution of UAV intervention. Furthermore, existing methods fail to comprehensively consider key parameters such as flight environment complexity, takeoff and landing point accessibility, mission urgency, and flight risks, leading to unreasonable configurations and impacting system efficiency and safety. This invention, by constructing a multi-dimensional traffic scenario modeling and delivery pressure assessment mechanism, achieves a dynamic mapping from ground trajectory data to aerial resource demand, filling a core technological gap in air-ground collaborative scheduling. Attached Figure Description

[0043] Figure 1 This is a flowchart of a method for predicting the configuration number of aerial drones based on ground vehicle trajectory data, as described in this invention.

[0044] Figure 2 This is a prediction result diagram of the aerial unmanned aerial vehicle configuration quantity prediction method based on ground vehicle trajectory data according to the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0046] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0047] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 -Appendix Figure 2 Detailed explanation is as follows:

[0048] Example 1:

[0049] A method for predicting the number of aerial drones based on ground vehicle trajectory data includes the following steps:

[0050] S1. Collect ground vehicle trajectory data and construct a vehicle trajectory set;

[0051] Furthermore, the expression for the set of vehicle trajectories in step S1 is:

[0052]

[0053] in, , Let x and y be the x and y coordinates of the position of the k-th vehicle at time t, respectively. , , These are the velocity, acceleration, and heading angle of the k-th vehicle at time t, respectively, obtained and calculated from the vehicle's GPS data; Let N be the set of vehicle trajectories; N is the total number of vehicles.

[0054] Furthermore, at the very beginning of the model building process, high-frequency ground vehicle trajectory data needs to be collected. To support subsequent spatial modeling and UAV scheduling predictions, the raw data must be complete, continuous, and dynamic. Therefore, in addition to the usual timestamps and location coordinates, this step also extracts behavioral features of each vehicle at each time point, such as speed, acceleration, and heading angle. This data not only reflects the movement of vehicles in the urban road network but also provides the basic input for subsequently constructing the density field, velocity trend field, and orientation consistency field.

[0055] S2. Using vehicle speed, acceleration, and historical congestion frequency as weighting factors, a traffic density thermodynamic field is constructed using a direction-aware kernel function;

[0056] After obtaining the vehicle trajectory dataset, the next step is to map the discrete trajectory points into a continuous spatial density field, thereby revealing the degree of traffic congestion in different areas of the city at different times. Traffic density is not merely a simple sum of vehicle numbers, but should also consider the dynamic behavioral characteristics of vehicles. Therefore, multiple weighting factors were introduced when constructing the density thermodynamic field, including vehicle speed, acceleration, and historical congestion frequency. These factors collectively influence the instantaneous traffic state of the area. Simultaneously, to more accurately reflect the influence range of vehicles in different directions, a direction-aware high-order kernel function was employed, enabling the density distribution to reflect road orientation characteristics. The construction of this density field provides crucial support for subsequent assessments of ground delivery efficiency and the necessity of drone intervention.

[0057] Furthermore, the expression for constructing the traffic density thermodynamic field in step S2 is as follows:

[0058]

[0059] in, The space density thermodynamic field; x is the abscissa of the spatial position, and y is the ordinate of the spatial position. The normalized historical congestion frequency of the k-th vehicle. The normalized speed of the kth vehicle. Let the normalized acceleration of the k-th vehicle be... This is the direction-aware kernel function.

[0060] Furthermore, the normalization formula and the calculation formula for the direction-aware kernel function are as follows:

[0061] , ,

[0062]

[0063] in, Obtained from map data; The historical congestion frequency of the area where the k-th vehicle is located is obtained from historical data; This is the kernel width parameter, set by expert experience or references; This is the offset of the current calculation point in the x-direction, obtained from map data; This is the offset of the current calculation point in the y-direction, obtained from map data; This is a speed reference value, determined by expert experience in conjunction with actual conditions; The acceleration reference value is determined by expert experience in combination with actual conditions; This is a reference value for congestion frequency, determined by experts based on their experience and actual conditions.

[0064] S3. Combine the speed and direction angle of each vehicle to form a speed vector, and then perform a weighted average in space using the direction sensing kernel function to construct an average speed trend vector field;

[0065] After understanding the traffic density distribution in different urban areas, density alone is insufficient to fully depict the traffic situation; further understanding of vehicle movement trends and flow directions is also necessary. Therefore, this step constructs a spatially continuous velocity trend vector field based on existing trajectory information to describe the average movement direction and velocity of vehicles in different areas. The velocity of each vehicle is combined with its azimuth angle to form a velocity vector, which is then spatially weighted and averaged using a direction-sensitive kernel function to obtain the vector value of the trend field. This trend field not only reveals the dominant patterns of traffic flow within a region but also reflects traffic mobility, providing essential information for assessing the difficulty of delivery tasks and predicting potential bottleneck areas, further refining the modeling of ground traffic conditions.

[0066] Furthermore, the expression for the average velocity trend vector field constructed in step S3 is:

[0067]

[0068]

[0069]

[0070] in, It is a unit vector in direction; The weighted sum of the direction-aware kernel functions; It represents the average velocity trend vector field.

[0071] S4. For all vehicles in the spatial neighborhood, calculate the cosine similarity between the vehicle's heading angle and the average heading angle, and construct a heading consistency index;

[0072] After modeling traffic density and speed trends, we further introduce directional consistency as a traffic order indicator to more comprehensively characterize the complexity of ground traffic. Directional consistency reflects the degree of convergence of vehicle travel directions within a certain area. It is closely related to traffic safety and flight predictability, and is of great significance in air-ground cooperative systems. To achieve this goal, we calculate the cosine similarity between the azimuth angles of all vehicles in the spatial neighborhood and the main direction, thus obtaining the directional consistency index. High consistency usually indicates good traffic order and low flight risk; while low consistency may represent complex scenarios such as intersections and U-turn areas. This index will participate in the modeling along with density and speed trends when constructing the delivery pressure function, forming a three-dimensional assessment of delivery difficulty.

[0073] Furthermore, the expression for the directional consistency index constructed in step S4 is as follows:

[0074]

[0075]

[0076] in, As an indicator of directional consistency; The set of vehicles within radius r is obtained from vehicle trajectory data; is the average direction angle; i is the imaginary unit; arg is the phase angle function of the complex number; Let r be the set of vehicles within a radius r.

[0077] S5. Using the traffic density thermal field as the basic load, the magnitude of the average velocity trend vector field represents the flow rate, and the direction consistency index represents the traffic order, a weighted combination is used to construct the delivery pressure intensity.

[0078] Having established multi-dimensional traffic state descriptions including density, speed trend, and directional consistency fields, this information needs to be integrated to form a quantifiable delivery pressure index. Delivery pressure reflects the degree to which ground traffic hinders conventional delivery methods and is a core basis for assessing the need to introduce aerial drones. The density field is considered the basic load, the magnitude of the speed trend represents mobility, and directional consistency reflects traffic order. These three are weighted and combined to construct a function that comprehensively reflects traffic difficulty. If a region experiences high density, low speed, and chaotic direction at a given moment, the delivery pressure in that region will significantly increase. This pressure field is not only used for spatial analysis but also provides a quantitative basis for subsequent drone demand estimation, connecting ground conditions with aerial resource allocation.

[0079] Furthermore, the expression for the delivery pressure intensity in step S5 is as follows:

[0080]

[0081] in, To assess the intensity of delivery pressure; The pressure intensity coefficient is determined by expert experience and is also used to adjust the dimensions. The magnitude of the velocity trend vector; The maximum road speed is determined based on the design value; The consistency penalty coefficient is set empirically.

[0082] S6. Transform the delivery pressure intensity into the demand intensity of aerial resources, and construct a preliminary drone demand estimation field by considering the spatial vector from the vehicle location to the nearest drone take-off and landing point and the terrain complexity factor.

[0083] After calculating the delivery pressure, this pressure value needs to be further transformed into the demand intensity of aerial resources to construct a spatially continuous UAV demand estimation field. Areas with higher delivery pressure have more limited ground delivery capabilities and, theoretically, a stronger dependence on UAVs. Based on this, the basic regional demand is calculated using the delivery pressure field and the unit capacity of UAVs. To improve the model's realism, two key correction factors are introduced: first, the spatial vector from the current location to the nearest UAV take-off and landing point; a larger spatial vector indicates a higher response cost, and the demand should be appropriately increased; second, the terrain complexity factor, used to reflect the accessibility and operational difficulty of the flight environment in the area. The resulting demand estimation field constitutes a spatial reference map for the initial deployment quantity of UAVs.

[0084] Furthermore, the expression for the drone demand estimation field constructed in step S6 is:

[0085]

[0086] in, Preliminary demand estimation for drones; The safety redundancy factor is determined by expert experience; The maximum number of tasks a single drone can perform per unit of time is determined from the design specifications. The spatial vector from the vehicle to the nearest take-off and landing point is calculated from a GIS map. The reference space vector is determined by expert experience; This is the terrain complexity factor, obtained by experts based on GIS map analysis.

[0087] S7. The initial UAV demand estimate is revised by introducing a task urgency weighting function, current task backlog, and weather flight risk, and the final UAV deployment suggestion function is constructed to complete the prediction of the number of aerial UAVs based on ground vehicle trajectory data.

[0088] Furthermore, the specific implementation method of step S7 includes the following steps:

[0089] S7.1. Construct a task urgency weighting function, with the following expression:

[0090]

[0091] in, Weight the urgency of the task; The remaining task response time is obtained from the real-time scheduling system; The expected response time is determined by expert experience or actual requirements. The historical failure rate is obtained from historical data. The weighting of the current regional task type is determined by expert experience or actual requirements. As a time urgency factor, it is determined by expert experience and is also used to adjust the dimensions; As a risk-sensitive factor, These are task priority factors, all determined by expert experience;

[0092] Although the demand estimation field provides the basic demand intensity for drones in each region, the scheduling system still needs to prioritize tasks based on their urgency when resources are limited. Therefore, this step introduces a task urgency weighting function to dynamically adjust the basic demand value. This function characterizes the urgency of tasks from three dimensions: first, the difference between the remaining response time and the expected response time of the current task; second, the historical task failure rate, reflecting the reliability of tasks in the region; and third, the task type weight, such as medical and emergency tasks should receive higher priority. This weighting mechanism allows high-priority regions to receive more drone resources, improving overall scheduling efficiency and system responsiveness.

[0093] S7.2. Construct the final drone deployment suggestion function, with the expression:

[0094]

[0095] in, Recommended number of drones for final deployment; Weighted by urgency; The backlog of tasks in the region is calculated by the scheduling system. The maximum task backlog threshold is determined by expert experience or actual requirements. Wind speed risk factors are determined by experts based on meteorological data.

[0096] After constructing the initial demand estimation field and urgency weighting function, two dynamic factors in actual operation are further considered: the current task backlog and weather flight risks. These two factors directly affect the actual deployment capability and safety. The task backlog reflects the current system load; a larger backlog indicates insufficient resources, and the number of deployments should be appropriately increased. Weather risks, especially wind speed, directly affect the flight stability and safety of UAVs; higher risks should lead to a reduction in deployment values. Finally, the demand estimate, urgency weight, backlog ratio, and weather risk are integrated to construct the final UAV deployment recommendation function. This function is the core output of the entire system, directly used for scheduling to achieve intelligent response integrating air and ground.

[0097] With the rapid growth of urban logistics and delivery demands, traditional ground transportation systems often face congestion bottlenecks during peak hours, leading to decreased delivery efficiency and delayed responses. To alleviate ground delivery pressure, more and more cities are exploring the auxiliary scheduling role of drones in urban logistics. However, how to scientifically and efficiently allocate the number of aerial drones has become a key challenge restricting the application of air-ground collaborative scheduling systems. This embodiment proposes a method for predicting the number of aerial drones based on ground vehicle trajectory data. By comprehensively analyzing the urban traffic operation situation, it constructs a delivery pressure field based on multi-dimensional parameters such as traffic density, speed trends, and directional consistency. Furthermore, it combines factors such as task urgency, geographical complexity, and flight risks to form a final drone deployment recommendation scheme. This method realizes intelligent linkage between ground conditions and aerial resource allocation, providing a scientific basis and technical support for air-ground collaborative delivery in smart cities.

[0098] The following is an example illustrating its application:

[0099] In a certain urban area A (320, 420) (area 1.5 km²), the dispatch system collected ground vehicle trajectory data during the morning rush hour (7:30-8:00). A total of 2000 vehicle trajectories were collected, with each vehicle's position, speed, acceleration, and heading angle recorded every 5 seconds. Specifically, the k-th vehicle at time t has x-coordinates of 312.6 and y-coordinates of 415.2, a speed of 8.2 m / s, an acceleration of 1.1 m / s², and a heading angle of 50 degrees.

[0100] Using the method in this embodiment, the final recommended number of drones is calculated to be 12. Similarly, calculations were performed for region B (800, 800) (area 2.0 km²), region C (1200, 1500) (area 2.5 km²), region D (2000, 2300) (area 1.8 km²), and region E (280, 400) (area 3.0 km²), and the final recommended number of drones were 10, 7, 8, and 9 respectively.

[0101] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0102] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for predicting the number of aerial unmanned aerial vehicles (UAVs) based on ground vehicle trajectory data, characterized in that, Includes the following steps: S1. Collect ground vehicle trajectory data and construct a vehicle trajectory set; S2. Using vehicle speed, acceleration, and historical congestion frequency as weighting factors, a traffic density thermodynamic field is constructed using a direction-aware kernel function; S3. Combine the speed and direction angle of each vehicle to form a speed vector, and then perform a weighted average in space using the direction sensing kernel function to construct an average speed trend vector field; The expression for constructing the traffic density thermodynamic field in step S2 is: ； in, The space density thermodynamic field; x is the abscissa of the spatial position, and y is the ordinate of the spatial position. The normalized historical congestion frequency of the k-th vehicle. The normalized speed of the kth vehicle. Let the normalized acceleration of the k-th vehicle be... For the direction-aware kernel function, , Let x and y be the x and y coordinates of the position of the k-th vehicle at time t, respectively. Let be the direction angle of the k-th vehicle at time t; S4. For all vehicles in the spatial neighborhood, calculate the cosine similarity between the vehicle's heading angle and the average heading angle, and construct a heading consistency index; S5. Using the traffic density thermal field as the basic load, the magnitude of the average velocity trend vector field represents the flow rate, and the direction consistency index represents the traffic order, a weighted combination is used to construct the delivery pressure intensity. The expression for the delivery pressure intensity in step S5 is as follows: ； in, To assess the intensity of delivery pressure; The pressure intensity coefficient is determined by expert experience and is also used to adjust the dimensions. The magnitude of the velocity trend vector; The maximum road speed is determined based on the design value; The consistency penalty coefficient is set empirically. As an indicator of directional consistency; S6. Transform the delivery pressure intensity into the demand intensity of aerial resources, and construct a preliminary drone demand estimation field by considering the spatial vector from the vehicle location to the nearest drone take-off and landing point and the terrain complexity factor. S7. The initial UAV demand estimate is revised by introducing a task urgency weighting function, current task backlog, and weather flight risk, and the final UAV deployment suggestion function is constructed to complete the prediction of the number of aerial UAVs based on ground vehicle trajectory data.

2. The method for predicting the number of aerial unmanned aerial vehicles (UAVs) based on ground vehicle trajectory data according to claim 1, characterized in that, The expression for the set of vehicle trajectories in step S1 is: ； in, , The velocity and acceleration of the k-th vehicle at time t are respectively obtained and calculated from the vehicle's GPS data; Let N be the set of vehicle trajectories; N is the total number of vehicles.

3. The method for predicting the number of aerial unmanned aerial vehicles (UAVs) based on ground vehicle trajectory data according to claim 2, characterized in that, The expression for the average velocity trend vector field constructed in step S3 is: ； in, It is a unit vector in direction; The weighted sum of the direction-aware kernel functions; It represents the average velocity trend vector field.

4. The method for predicting the number of aerial unmanned aerial vehicles (UAVs) based on ground vehicle trajectory data according to claim 3, characterized in that, The expression for the directional consistency index constructed in step S4 is: ； ； in, As an indicator of directional consistency; The set of vehicles within radius r is obtained from vehicle trajectory data; is the average direction angle; i is the imaginary unit; arg is the phase angle function of the complex number. Let be the direction angle of the j-th vehicle at time t, where j is... Any one of them.

5. The method for predicting the number of aerial unmanned aerial vehicles (UAVs) based on ground vehicle trajectory data according to claim 4, characterized in that, The expression for the UAV demand estimation field constructed in step S6 is: ； in, Preliminary demand estimates for drones; The safety redundancy factor is determined by expert experience; The maximum number of tasks a single drone can perform per unit of time is determined from the design specifications. The spatial vector from the vehicle to the nearest take-off and landing point is calculated from a GIS map. The reference space vector is determined by expert experience; This is the terrain complexity factor, obtained by experts based on GIS map analysis.

6. The method for predicting the number of aerial unmanned aerial vehicles (UAVs) based on ground vehicle trajectory data according to claim 5, characterized in that, The specific implementation method of step S7 includes the following steps: S7.

1. Construct a task urgency weighting function, with the following expression: ； in, Weight the urgency of the task; The remaining task response time is obtained from the real-time scheduling system; The expected response time is determined by expert experience or actual requirements; The historical failure rate is obtained from historical data. The weight of the current regional task type is determined by expert experience or actual requirements; As a time urgency factor, it is determined by expert experience and is also used to adjust the dimensions; As a risk-sensitive factor, These are task priority factors, all determined by expert experience; S7.

2. Construct the final drone deployment suggestion function, with the expression: ； in, Recommended number of drones for final deployment; Weighted by urgency; The backlog of tasks in the region is calculated by the scheduling system. The maximum task backlog threshold is determined by expert experience or actual requirements. Wind speed risk factors are determined by experts based on meteorological data.

Images