A dynamic hotspot unmanned aerial vehicle double-layer position adaptive hovering planning method and system

Abstract

This application relates to a dynamic hotspot UAV dual-layer position adaptive hovering planning method and system. The method includes: acquiring the status of each hotspot in the current time slot and the UAV's status in the previous time slot; constructing a multi-objective optimization objective function with weighted service revenue maximization as the core, superimposed with migration distance cost, handover action penalty, flight energy consumption, and hovering energy consumption as penalty terms; setting constraints; traversing the candidate hovering point set, calculating the optimal two-dimensional fine-tuning amount within the fine-tuning radius of each candidate hovering point, performing hierarchical capacity allocation, and calculating the comprehensive revenue value; verifying the minimum hovering duration constraint and selecting the candidate point with the largest comprehensive revenue as the primary hovering point; controlling the UAV to perform actions and update its status. Through multi-objective joint optimization and dual-layer position decision-making, a dynamic balance is achieved between communication service revenue, migration cost, handover penalty, and energy consumption, significantly improving communication stability, reducing energy consumption, adapting to dynamic hotspot drift, and demonstrating strong engineering feasibility.

Description

Technical Field

[0001] This application relates to the fields of unmanned aerial vehicle (UAV) communication and mobile edge computing technology, and in particular to a dynamic hotspot UAV dual-layer position adaptive hovering planning method and system. Background Technology

[0002] In mixed indoor and outdoor scenarios such as large-scale exhibitions and sporting events, the density of users and the demand for communication services in the venue will change dramatically with time and space. The capacity and coverage of traditional fixed 5G base stations are fixed and cannot effectively carry out sudden and time-varying hotspot communication traffic.

[0003] Currently, the industry generally uses drones as aerial base stations to supplement communication capacity, thereby relieving the pressure on ground base stations. Existing technical routes can be divided into three categories, all of which have obvious defects.

[0004] The first type is the greedy scheduling scheme. This type of scheme simply selects the area with the most people as the drone hovering point, which easily leads to frequent switching of hovering points, causing frequent reconnection of communication links, resulting in poor communication service continuity and increased switching losses; moreover, it completely ignores the drone's energy consumption and migration costs, resulting in extremely low overall operating efficiency.

[0005] The second category is static or cruise-type solutions. Some solutions use fixed static hovering points, which cannot adapt to the natural drift characteristics of hotspot areas; if the drone flies continuously for the entire time to follow the hotspot, it will greatly increase flight energy consumption, and at the same time cause wireless backhaul link jitter and decreased stability.

[0006] The third category is intelligent optimization solutions. Existing research mostly uses deep reinforcement learning and multi-agent algorithms to solve the deployment location of UAVs. Although it can achieve a certain degree of dynamic adaptation, it has three major engineering bottlenecks: First, model training depends on a large amount of scene data, and the generalization ability to new scenes is poor; second, the computational complexity is high, and airborne embedded devices cannot support real-time inference, requiring cloud deployment, which leads to decision delays; third, the optimization objective is relatively singular, and a joint trade-off mechanism for switching penalties, migration costs, and multiple types of energy consumption has not been established, making it difficult to balance service continuity and operational economy.

[0007] In addition, existing technical solutions generally lack supporting engineering mechanisms such as link redundancy backup, rapid adaptation to sudden hotspots, and refined capacity allocation, resulting in insufficient overall operational reliability and engineering practicality.

[0008] Therefore, there is an urgent need in related technologies to solve the problems of poor communication service stability, high energy consumption of drone operation, and inability to adapt to dynamic hotspot drift in existing drone hovering point planning schemes. Summary of the Invention

[0009] Therefore, it is necessary to provide a dynamic hotspot UAV dual-layer position adaptive hovering planning method and system to address the above-mentioned technical problems.

[0010] Firstly, this application provides a two-layer position-adaptive hovering planning method for dynamic hotspot unmanned aerial vehicles (UAVs). The method includes: Obtain the center location, real-time business demand, priority weight, and hovering position and continuous hovering duration of each hotspot in the current time slot; A multi-objective optimization objective function is constructed, which takes the maximization of weighted service revenue as the core and adds migration distance cost, switching action penalty, flight energy consumption and hovering energy consumption as penalty terms. Set constraints, including single master hovering point selection constraints, fine-tuning range constraints, migration speed constraints, minimum hovering duration constraints, backhaul link capacity constraints, and total UAV service capacity constraints. Traverse each candidate hovering point in the candidate hovering point set, calculate the optimal two-dimensional fine-tuning amount within the fine-tuning radius of each candidate hovering point, and substitute each candidate hovering point and its corresponding optimal two-dimensional fine-tuning amount into the multi-objective optimization objective function. Perform hierarchical capacity allocation under the current candidate hovering point, and calculate the comprehensive benefit value based on the capacity allocation result. The minimum hovering duration constraint is verified. If the continuous hovering duration of the current time slot is less than the minimum hovering threshold, the hovering point of the previous time slot is forcibly maintained as the main hovering point of the current time slot; otherwise, the candidate hovering point with the largest comprehensive benefit value is selected as the main hovering point of the current time slot. Control the drone to fly to the main hovering point and execute the corresponding two-dimensional fine-tuning, update the drone's real-time position and continuous hovering duration, and then enter the next time slot.

[0011] Optionally, in one embodiment of this application, the expression of the multi-objective optimization objective function is:

[0012] in, Hot Topics In the time slot Priority weight, Hot Topics The business demand, Hot Topics Corresponding ground foundation capacity, Drones are a hot topic Provided supplementary capacity, For the drone's migration distance, To switch the indicator variable for the hover point, , , , These are the penalty weighting coefficients for migration distance cost, switching action penalty, flight energy consumption, and hovering energy consumption, respectively. The current time slot flight energy consumption, This represents the energy consumption during hovering in the current time slot.

[0013] Optionally, in one embodiment of this application, the primary hovering point and the two-dimensional fine-tuning amount constitute a two-layer position decision mechanism, specifically including: The main hovering point serves as the primary hovering position for the drone, used to address large-scale positional shifts in hotspot areas. The two-dimensional fine-tuning amount serves as a local position adjustment around the main hovering point, used to adapt to the small-range natural drift of the hotspot area. The criteria for determining small-range natural drift is that the hotspot offset distance is less than or equal to the fine-tuning radius, and the criteria for determining large-range positional drift is that the hotspot offset distance is greater than the fine-tuning radius.

[0014] Optionally, in one embodiment of this application, the hierarchical capacity allocation rule is: Prioritize allocating capacity to the primary hotspot with the highest priority weight, with the allocation value being the smaller of the primary hotspot's business demand and the total effective capacity of the drones. If there is still remaining capacity after allocation, the remaining capacity is multiplied by a reduction factor and then allocated to the secondary hotspot with the second highest priority. The reduction factor decreases as the distance between the secondary hotspot and the actual location of the UAV increases. No additional drone capacity will be allocated to hotspots other than primary and secondary hotspots.

[0015] Optionally, in one embodiment of this application, the penalty weight coefficient adopts an adaptive tuning strategy, specifically: When the drone's remaining battery power is below a preset threshold, the flight energy consumption penalty coefficient is increased. and hovering energy consumption penalty coefficient At the same time, reduce the penalty coefficient for switching actions. ; When the drone's remaining battery power is higher than a preset threshold, the flight energy consumption penalty coefficient is reduced. and hovering energy consumption penalty coefficient ; Calculate the average offset distance of the hotspot center position within multiple consecutive time slots. When the average offset distance is less than a set proportion of the fine-tuning radius, increase... When the average offset distance continuously exceeds the fine-tuning radius, reduce... .

[0016] Optionally, in one embodiment of this application, the candidate hover point set supports dynamic updating, and the updating process includes: Real-time monitoring of business needs and location changes at various hotspots; When a new hotspot area is detected and all of the following triggering conditions are met, a new candidate hover point is automatically generated and added to the candidate hover point set: The distance from all existing candidate hovering points to the newly added hotspot area is greater than the effective service radius of the drone; The business demand in the newly added hotspot area exceeded the set percentage threshold of the ground infrastructure capacity.

[0017] Optionally, in one embodiment of this application, the backhaul link corresponding to the backhaul link capacity constraint adopts a dual-link architecture of a terahertz primary backhaul link and a microwave backup backhaul link. The signal-to-noise ratio (SNR) and packet loss rate of the terahertz primary backhaul link are monitored in real time. When the SNR is lower than a set threshold or the packet loss rate is higher than a set percentage threshold and continues for a preset number of sampling periods, the primary link is determined to be abnormal. The system immediately switches to the microwave backup backhaul link and updates the upper limit of the capacity in the backhaul link capacity constraint from the terahertz capacity upper limit to the microwave link rated bandwidth. Based on the updated backhaul link capacity constraint, the optimization solution for the current time slot is re-executed.

[0018] Optionally, in one embodiment of this application, the minimum hovering duration threshold adopts an adaptive adjustment scheme: When the remaining battery power of the drone is lower than a preset threshold, the minimum hovering time threshold is increased; when the remaining battery power of the drone is higher than the preset threshold, the minimum hovering time threshold is decreased. The frequency of hotspot location fluctuations is counted per unit time. The higher the fluctuation frequency, the lower the minimum hovering time threshold is; the lower the fluctuation frequency, the higher the minimum hovering time threshold is.

[0019] Optionally, in one embodiment of this application, the constraints include: The single primary hovering point selection constraint allows only one candidate hovering point to be selected as the primary hovering point for each time slot; The fine-tuning range constraint is that the sum of the squares of the lateral and longitudinal fine-tuning amounts around the main hovering point of the UAV is less than or equal to the square of the fine-tuning radius. The migration speed constraint is that the UAV migration distance is less than or equal to the product of the maximum migration speed and the time slot duration.

[0020] Secondly, this application also provides a dynamic hotspot UAV dual-layer position adaptive hovering planning system. The system includes: The scene modeling module is used to construct dynamic hotspot scenes, set up a set of candidate hovering points, and complete the initialization of system parameters; The status awareness module is used to collect the center location of each hotspot, real-time business demand, priority weight, and the hovering point and continuous hovering duration of the drone. The optimal choice module is used to construct multi-objective optimization objective functions and constraints, perform candidate hovering point traversal and optimal solution calculation, and output the optimal master hovering point and two-dimensional fine-tuning amount of the current time slot. The execution control module is used to receive decision commands and control the UAV to complete flight migration, hovering, and position fine-tuning actions; The communication module integrates a terahertz main backhaul unit and a microwave backup backhaul unit to achieve dual-link redundant data transmission and data interaction between the UAV and the ground receiving node.

[0021] The above-mentioned dynamic hotspot UAV dual-layer position adaptive hovering planning method and system has the following advantages compared with the prior art: First, the stability of communication services has been significantly improved. Through the combined hardware and software approach of switching penalty terms and minimum hovering duration constraints, frequent hovering point switching is effectively suppressed, the proportion of communication link interruption time is greatly reduced, and service continuity is significantly enhanced.

[0022] Second, the energy consumption of drones is optimized for high efficiency. The combined optimization of energy consumption during flight and hovering effectively reduces the overall operating energy consumption and extends the drone's endurance.

[0023] Third, it has strong dynamic scene adaptability. The two-layer location decision can adapt to the natural drift of hotspots; the dynamic candidate point generation mechanism can respond to sudden hotspots in a short time, and the adaptation speed is far superior to traditional solutions.

[0024] Fourth, it has outstanding engineering feasibility. The computational complexity of solving discrete candidate points is low, and it can be achieved in milliseconds by airborne embedded devices. It requires no training, has no generalization problems, and has an extremely low deployment threshold.

[0025] Fifth, higher capacity utilization. Layered capacity allocation and nested optimization of location decisions prioritize the service experience of high-priority core hotspots, resulting in a significant improvement in overall capacity utilization.

[0026] Sixth, the link reliability is strong. Seamless switching between terahertz and microwave dual links ensures no service interruption in fault scenarios, significantly improving system robustness.

[0027] Seventh, flexible scenario adaptation. Both the weighting coefficient and the reduction coefficient can be adaptively adjusted to adapt to various scenarios such as exhibitions, sports events, and emergency support. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the system architecture in one embodiment; Figure 2Here is a flowchart of a dynamic hotspot UAV dual-layer position adaptive hovering planning method in one embodiment; Figure 3 This is a schematic diagram of a two-layer location decision in one embodiment; Figure 4 This is a schematic diagram of nested optimization logic in one embodiment; Figure 5 Here is a scene model and hotspot distribution map of Shunyi Hall in one embodiment; Figure 6 This is a comparison diagram of the T1 time slot decision process in one embodiment; Figure 7 This is a timing diagram of the main hover point switching in one embodiment; Figure 8 This is a flowchart illustrating the collaborative process of two-layer decision-making and switching constraints in one embodiment; Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0030] In this embodiment, the present invention is divided into two main parts: a UAV hovering point planning method and a supporting planning system. It adopts a time-slot-by-time online optimization algorithm, and performs decision-making and control cyclically in discrete time slots. At the same time, it integrates supporting mechanisms such as two-layer position decision, hierarchical capacity allocation, dual-link redundancy, and dynamic candidate point generation.

[0031] Figure 1 A schematic diagram of the system architecture of the present invention is shown. For example... Figure 1 As shown, the system comprises a user terminal, a drone, five functional modules (scene modeling module, state perception module, optimal decision-making module, execution control module, and communication module), a terahertz backhaul link, a microwave backup link, a ground receiving node, and a 5G core network. The connection relationships and data flow between these components are illustrated in the figure. The user terminal connects to the drone's airborne base station or a ground-based 5G base station via wireless access. The drone communicates with the ground receiving node via the terahertz primary backhaul link or the microwave backup backhaul link. The ground receiving node is connected to the 5G core network. The system uses a time-slot-by-time online optimization algorithm as its core engine, cyclically executing decision-making and control in discrete time slots. At the beginning of each time slot, multi-source data is integrated for state acquisition. After the optimal hovering point and fine-tuning are solved by the optimal decision-making module, the execution control module drives the drone to complete the action execution. Simultaneously, the communication module ensures the reliability of the data backhaul link.

[0032] Specifically, this application provides a dynamic hotspot UAV dual-layer position adaptive hovering planning method, such as... Figure 2 As shown, its core process is as follows: First, obtain the center location, real-time service demand, priority weight, and hovering position and duration of the drone in the previous time slot for each hotspot in the current time slot. Specifically, optimize the time slots by dividing them into fixed durations, and in each time slot... In the initial stage, three types of data are integrated: ground 5G base station signaling, on-site IoT probes, and user terminal access reports. The central location, real-time service demand, and priority weight of each hotspot are statistically obtained. At the same time, the hovering position and continuous hovering duration of the UAV in the previous time slot are read.

[0033] Secondly, a multi-objective optimization objective function is constructed. This objective function focuses on maximizing weighted service revenue, incorporating migration distance cost, handover penalty, flight energy consumption, and hovering energy consumption as penalty terms. This objective function is not a simple aggregation of multiple indicators, but rather achieves dynamic balancing of multi-dimensional needs through weighting coefficients. The collaborative constraints between these indicators are as follows: The handover penalty and the minimum hovering time constraint form a "soft plus hard" dual handover inhibition: the handover penalty is a soft constraint, which reduces the willingness to handover under a small increase in benefits by reducing the benefits; the minimum hovering time is a hard constraint, which forcibly prevents frequent handovers in a short period of time. The two together ensure the continuity of communication services.

[0034] Migration distance cost and flight energy consumption form a dual energy consumption control mechanism: migration distance cost directly constrains long-distance cross-regional flights, while flight energy consumption is combined with real-time parameters such as wind speed and load to calculate energy consumption in a refined manner. The two control flight losses from the two dimensions of path length and physical energy consumption, respectively.

[0035] The hovering energy consumption and service revenue items form a hovering efficiency balance: avoiding prolonged ineffective hovering in pursuit of minimal revenue gains, and forcing capacity resources to be tilted towards high-value hotspots.

[0036] Simultaneously, constraints are set, including single master hovering point selection constraints, fine-tuning range constraints, migration speed constraints, minimum hovering duration constraints, backhaul link capacity constraints, and total UAV service capacity constraints, to ensure stable and compliant system operation.

[0037] Then, each candidate hovering point in the candidate hovering point set is traversed. The candidate hovering point set is preset during the scene construction phase, and candidate hovering points and hotspot areas support various deployment forms such as one-to-one, one-to-many, and many-to-many. The optimal two-dimensional fine-tuning amount is calculated within the fine-tuning radius of each candidate hovering point, and each candidate hovering point and its corresponding optimal two-dimensional fine-tuning amount are substituted into the multi-objective optimization objective function. Hierarchical capacity allocation is performed under the current candidate hovering point, and the comprehensive benefit value is calculated based on the capacity allocation result.

[0038] Next, the minimum hovering duration constraint is verified. If the continuous hovering duration of the current time slot is less than the minimum hovering threshold, the hovering point of the previous time slot is forcibly maintained as the main hovering point of the current time slot; otherwise, the candidate hovering point with the largest comprehensive benefit value is selected as the main hovering point of the current time slot.

[0039] Finally, the drone is controlled to fly to the main hovering point and the corresponding two-dimensional fine-tuning is performed. After updating the drone's real-time position and continuous hovering duration, it enters the next time slot.

[0040] Figure 2 The above single-time-slot decision execution process is fully demonstrated: First, the hotspot status and the UAV status of the previous time slot are input. Then, all candidate hovering points are traversed and the corresponding optimal fine-tuning amount and comprehensive target benefit value are calculated. After selecting the candidate point with the highest benefit, the minimum hovering time constraint is checked. If the constraint is met, the optimal decision is output. If not, the previous hovering point is forcibly maintained. Finally, the decision is executed and the status is updated, and the next time slot cycle begins.

[0041] In one embodiment of this application, the expression for the multi-objective optimization objective function is:

[0042] in, Hot Topics In the time slot Priority weight, Hot Topics The business demand, Hot Topics Corresponding ground foundation capacity, Drones are a hot topic Provided supplementary capacity, For the drone's migration distance, To switch the indicator variable for the hover point, , , , These are the penalty weighting coefficients for migration distance cost, switching action penalty, flight energy consumption, and hovering energy consumption, respectively. The current time slot flight energy consumption, This represents the energy consumption during hovering in the current time slot.

[0043] In the objective function, the first term The weighted service revenue item represents the total revenue obtained by drones after providing supplemental communication capacity to various hotspots; the second item... The third item is a migration distance penalty used to suppress long-distance flight; As a penalty for switching actions, a fixed penalty is applied when the main hover point changes; the fourth item For flight energy consumption penalty; Item 5 A hovering energy consumption penalty is imposed. Through a comprehensive consideration of the above five factors, a multi-dimensional dynamic balance is achieved between communication quality, operational energy consumption, and service stability.

[0044] In one embodiment of this application, the constraints include: The single master hover point selection constraint allows only one candidate hover point to be selected as the master hover point for each time slot. The mathematical expression is: , ,in Represents the selected number 1 candidate point This indicates that the candidate point is not selected.

[0045] The fine-tuning range constraint is that the sum of the squares of the lateral and longitudinal fine-tuning amounts around the main hovering point of the UAV is less than or equal to the square of the fine-tuning radius, i.e. .

[0046] The migration speed constraint is that the UAV's migration distance is less than or equal to the product of the maximum migration speed and the time slot duration, i.e. ,in This is the current hovering position in the time slot. This is the hovering position from the previous time slot. For maximum migration speed, This refers to the duration of the time slot.

[0047] The minimum hovering duration constraint is: if the number of consecutive hovering time slots of the drone in the previous time slot is less than the preset minimum hovering threshold. If so, switching the primary hover point is prohibited in the current time slot, i.e. ,like ,in This represents the number of consecutive hovering time slots for the drone in the previous time slot.

[0048] The backhaul link capacity constraint is: the total supplementary capacity of all hotspot-allocated UAVs shall not exceed the upper limit of the terahertz backhaul link capacity, i.e. When the terahertz link experiences obstruction or excessive loss, it automatically switches to the backup microwave link, using the rated bandwidth of the microwave link as the new capacity constraint. After switching to the microwave link, the constraint becomes... ,in This is the rated bandwidth of the microwave link.

[0049] The total service capacity constraint for drones is: the total supplementary capacity output by a drone shall not exceed its own nominal service capacity, i.e. .

[0050] In one embodiment of this application, the primary hovering point and the two-dimensional fine-tuning amount constitute a two-layer position decision mechanism. Figure 3 A schematic diagram of a two-layer location decision is shown. (For example...) Figure 3 As shown, taking time slot T1 as an example, the main hovering point P1 is the main hovering position of the UAV, and the hotspot areas H1 and H2 are located around the main hovering point. The UAV has a preset fine-tuning radius at the main hovering point. Two-dimensional fine-tuning is performed within the range, with the fine-tuning amount being... and .

[0051] Specifically, the main hovering point serves as the primary hovering position of the UAV, used to address large-scale positional shifts in hotspot areas; the two-dimensional fine-tuning amount serves as a local positional adjustment around the main hovering point, used to adapt to small-scale natural drift in hotspot areas. The criterion for determining small-scale natural drift is that the hotspot shift distance is less than or equal to the fine-tuning radius, and the criterion for determining large-scale positional shift is that the hotspot shift distance is greater than the fine-tuning radius.

[0052] The aforementioned two-layer position decision-making mechanism, combined with minimum hovering time constraints and switching penalties, enables tiered responses to different fluctuations in hotspots. The specific judgment and coordination logic is as follows: The determination is based on the straight-line distance between the center of the hotspot and the current main hovering point, combined with the preset fine-tuning radius. Divided into two levels: For small-range drift (hotspot offset distance less than or equal to the fine-tuning radius)... No need to switch the main hover point; it can adapt to hotspot displacement through only local fine-tuning. For large-scale offsets (hotspot offset distance greater than the fine-tuning radius)... It is necessary to evaluate the overall benefits of switching the main hovering point to determine whether to perform the main hovering point switch.

[0053] Based on this, the present invention also constructs a triple stable switching cooperative logic: The first layer is a hard constraint fallback. If the current consecutive hovering time slots are less than the minimum hovering threshold... It directly prohibits switching of the main hover point and only allows local fine-tuning, thus eliminating high-frequency switching at the source.

[0054] The second layer is a soft penalty screening. If the minimum hovering time constraint is met, the fixed penalty generated by the switching action is included in the objective function. Switching is only feasible when the incremental service benefits brought by the switching exceed the sum of the switching penalty and the migration cost.

[0055] The third layer is a fine-tuning buffer. When the main hover point remains unchanged, the service benefits are improved by adapting to the small-range natural drift of the hotspot through two-dimensional fine-tuning, without triggering a switch, and the frequency of switching is further reduced.

[0056] In one embodiment of this application, the hierarchical capacity allocation rule is as follows: capacity is allocated to the primary hotspot with the highest priority weight, and the allocation value is the smaller of the service demand of the primary hotspot and the total effective capacity of the drones; if there is still remaining capacity after allocation, the remaining capacity is multiplied by a reduction factor and allocated to the secondary hotspot with the second highest priority weight, and the reduction factor decreases as the distance between the secondary hotspot and the actual location of the drone increases; no additional capacity for drones is allocated to other hotspots other than the primary and secondary hotspots.

[0057] The hierarchical capacity allocation and candidate point traversal solution form a nested optimization logic: when traversing each candidate hovering point, the signal attenuation and serviceability of each hotspot under the current point are calculated based on the candidate point (including the corresponding fine-tuning amount), and then the hierarchical capacity allocation is performed. Finally, the service revenue obtained from the allocation is substituted into the objective function for comprehensive evaluation.

[0058] Compared to the traditional two-stage approach of "first determining the optimal location and then allocating capacity", this nested mechanism can incorporate capacity allocation efficiency into the evaluation during the location decision-making stage, avoiding the local optimal solution of "optimal location but low capacity utilization", and achieving joint optimal allocation of location resources and bandwidth resources. Figure 4 A schematic diagram of this nested optimization logic is shown.

[0059] The total effective capacity of a drone is determined by its nominal service capacity and the upper limit of its backhaul link capacity, calculated as follows:

[0060] in, For the nominal service capabilities of drones, This represents the upper limit of terahertz backhaul capacity.

[0061] In specific allocation, the capacity allocation for the main hotspot is the smaller value between the main hotspot service demand and the total effective capacity of drones:

[0062] If there is still remaining capacity after the primary hotspot is allocated, the remaining capacity will be multiplied by the reduction factor. It is assigned to the next highest priority secondary hotspot, with a reduction factor. The range of values ​​is The coefficient can be dynamically adjusted based on the straight-line distance between the secondary hotspot and the primary hovering point; the greater the distance, the smaller the coefficient value.

[0063]

[0064] in, Based on the reduction factor, The straight-line distance between the secondary hotspot center and the current actual location of the drone. This represents the maximum effective service distance for a single hop for a drone. The greater the distance, the smaller the reduction factor, ensuring that capacity allocation matches signal coverage capabilities.

[0065] Apart from primary and secondary hotspots, no additional capacity will be allocated to other hotspots, i.e. .

[0066] In one embodiment of this application, the penalty weight coefficient adopts an adaptive tuning strategy, specifically: When the drone's remaining battery power is below a preset threshold, the flight energy consumption penalty coefficient is increased. and hovering energy consumption penalty coefficient At the same time, reduce the penalty coefficient for switching actions. Prioritize ensuring the safety of drone battery life; when the remaining battery power of the drone is higher than a preset threshold, appropriately reduce the energy consumption weight and prioritize the quality of communication services.

[0067] Calculate the average offset distance of the hotspot center position within multiple consecutive time slots. When the average offset distance is less than a set proportion of the fine-tuning radius, the hotspot is determined to be in a stable state, and the handover penalty coefficient is increased. To suppress unnecessary switching; when the average offset distance continuously exceeds the fine-tuning radius, it is determined to be a large-scale hotspot drift, and the switching action penalty coefficient is reduced. Improve positional tracking capabilities.

[0068] In one embodiment of this application, the candidate hover point set supports dynamic updates. The system monitors the service demands and location changes of each hotspot in real time. When a new hotspot area is detected and all of the following triggering conditions are met, a new candidate hover point is automatically generated and added to the candidate hover point set: The distance from all existing candidate hovering points to the newly added hotspot area is greater than the effective service radius of the drone; The business demand in the newly added hotspot area exceeded the set percentage threshold of the ground infrastructure capacity.

[0069] The generation rule is as follows: based on the center coordinates of the newly added hotspot, combined with airspace no-fly restrictions and backhaul link obstruction, the optimal candidate hover point coordinates are generated and directly incorporated into the candidate point set. In the next time slot, they will automatically participate in the optimization traversal without interrupting the system operation process.

[0070] In one embodiment of this application, the backhaul link corresponding to the backhaul link capacity constraint adopts a dual-link architecture of a terahertz primary backhaul link and a microwave backup backhaul link.

[0071] The communication module monitors the signal-to-noise ratio (SNR) and packet loss rate of the terahertz main backhaul link in real time. When the SNR is lower than a set threshold or the packet loss rate is higher than a set percentage threshold and continues for a preset number of sampling periods, the main link is determined to be abnormal, triggering a seamless switching process. The first step is to immediately activate the microwave backup link and complete the link handshake and data synchronization; The second step is to synchronously update the backhaul capacity constraint parameters in the optimal selection module, and update the upper limit of the capacity in the backhaul link capacity constraint from the terahertz capacity upper limit to the microwave link rated bandwidth. The third step is to re-execute the fast optimization solution within the current time slot based on the new capacity constraints, adjust the capacity allocation scheme, and adapt to changes in link bandwidth.

[0072] Once the terahertz link returns to normal and remains stable for the preset duration, it will automatically switch back to the main link and restore the corresponding capacity constraints.

[0073] In one embodiment of this application, the minimum hovering duration threshold adopts an adaptive adjustment scheme: When the remaining battery power of the drone is lower than the preset threshold, the minimum hovering time threshold is increased to further suppress hovering point switching, reduce flight energy consumption, and prioritize flight safety; when the remaining battery power is sufficient, the threshold is appropriately reduced to improve position following flexibility.

[0074] The frequency of hotspot location fluctuations per unit time is statistically analyzed. The more frequent the fluctuations and the more drastic the drift, the smaller the minimum hovering time threshold should be to improve the system's response to rapid changes in hotspots. When the hotspots are stable, the threshold should be increased to enhance communication stability.

[0075] This application embodiment also provides a dynamic hotspot UAV dual-layer position adaptive hovering planning system, including: The scene modeling module is used to construct dynamic hotspot scenes, deploy candidate hovering point sets, and initialize system parameters. This module is responsible for building indoor and outdoor hybrid application scenarios containing multiple dynamic hotspots, pre-setting a candidate hovering point set that can fully cover all hotspot areas, uniformly initializing various operating parameters, including UAV parameters (working altitude, maximum migration speed, fine-tuning radius, minimum hovering duration), hotspot parameters (business requirements, location, priority weight), communication parameters (ground basic communication capacity, UAV nominal service capacity, terahertz backhaul capacity limit), and reserving an interface for generating dynamic candidate points.

[0076] The status awareness module is used to collect the center location of each hotspot, real-time service demand, priority weight, and hovering position and continuous hovering duration of drones. This module integrates multi-source data at the beginning of each time slot to obtain complete system operating status information.

[0077] The optimal decision-making module is used to construct the multi-objective optimization objective function and constraints, perform candidate hovering point traversal and optimal solution calculation, and output the optimal master hovering point and two-dimensional fine-tuning amount for the current time slot. This module is the core decision engine of the system. It adopts a deterministic solution scheme of discrete candidate point traversal plus local fine-tuning optimization, which has low computational complexity and can achieve millisecond-level real-time decision-making in airborne embedded devices.

[0078] The execution control module receives decision commands and controls the UAV to perform flight migration, hovering, and position fine-tuning maneuvers. This module receives decision commands from the optimal choice module, converts the coordinates of the optimal main hovering point and the two-dimensional fine-tuning parameters into flight control commands, and drives the UAV to perform the corresponding actions.

[0079] The communication module integrates a terahertz primary backhaul unit and a microwave backup backhaul unit to achieve dual-link redundant data transmission and data interaction between the UAV and the ground receiving node. This module implements dual-link redundant transmission, completing data interaction between the UAV, the ground receiving node, and the 5G core network.

[0080] It should be noted that the core mechanism of this invention supports multiple alternative implementation methods, which can be flexibly selected according to the application scenario, hardware computing power, and business needs, as detailed below: Regarding alternative tuning strategies for penalty weighting coefficients: In addition to the adaptive tuning strategy based on remaining power and hotspot fluctuation amplitude mentioned above, the following two implementation methods can also be adopted: One approach is a fuzzy logic tuning strategy. Using three parameters—remaining drone battery power, priority of hot-topic services, and ambient wind speed—as input, a fuzzy inference rule base is constructed. The strategy outputs the correction ratios for each penalty weight coefficient through three steps: fuzzification, rule inference, and defuzzification. This method does not require a precise mathematical model, is more robust to complex dynamic scenarios, and is suitable for outdoor sports events where environmental parameters fluctuate drastically.

[0081] The second approach is an online gradient descent tuning strategy. Using the comprehensive target return of historical time slots as feedback signals, stochastic gradient descent is employed to iteratively update the coefficients of each penalty weight online, with the iteration step size dynamically decreasing with the time slot number. This method can continuously converge towards the optimal weights during operation and is suitable for long-term, steady-state exhibition scenarios.

[0082] Regarding alternative adjustment methods for the secondary hotspot capacity reduction factor: in addition to the distance-based dynamic attenuation model, the following two adjustment rules can also be adopted: One approach is differentiated reduction based on service type. Differentiated base reduction coefficients are set according to the secondary hot service types, increasing the reduction coefficient for high-priority services such as emergency communications and voice calls, and decreasing the reduction coefficient for low-priority services such as ordinary file downloads and video streaming. This method allows capacity resources to be allocated to high-value services, adapting to complex scenarios with mixed services.

[0083] Secondly, there is proactive adjustment based on traffic prediction. By combining historical hotspot traffic data, a sliding window averaging method is used to predict the demand for hot services in the next time slot. If a secondary hotspot is predicted to become the primary hotspot, its reduction factor is increased in advance, and some capacity is pre-allocated to achieve a smooth transition. This method can reduce service fluctuations during hotspot switching and improve the service experience.

[0084] Example 1: Convention and Exhibition Center Scene In one embodiment of this application, the China International Exhibition Center (Shunyi Hall) is used as the simulation scene. The total area of ​​the venue is 231,400 square meters, which is divided into four dynamic hotspot areas H1, H2, H3 and H4, and four candidate hovering points P1(210,70), P2(305,165), P3(395,175) and P4(230,230) are set accordingly. Figure 5 The diagram shows the outline of the exhibition venue, the geographical distribution of four dynamic hotspots H1 to H4, and four candidate hovering points P1 to P4. The venue's operating cycle is divided into four time slots: opening and entry (T1), morning peak (T2), afternoon peak (T3), and closing and exit (T4).

[0085] Parameter configuration: Drone operating altitude 40m, maximum migration speed 12m / s, fine-tuning radius 20m, minimum hovering time slots 2; ground base capacity 12Gbps, drone nominal service capacity 10.0Gbps, terahertz backhaul capacity limit 8.0Gbps, secondary hotspot reduction factor. .

[0086] Figure 6 A comparison diagram of the decision-making process in time slot T1 is shown, comparing the comprehensive benefit values ​​of the objective function corresponding to the UAV hovering in place versus switching to other candidate points. The results are as follows: In time slot T1, the main hotspot is H1, and the UAV remains hovering at P1; in time slots T2 and T3, the main hotspot switches to H2, and the UAV switches to hovering at P2; in time slot T4, the main hotspot switches to H4, and the UAV switches to hovering at P4. Figure 7 The switching timing pattern of the main hover point as the hotspot changes is shown throughout the entire operation cycle.

[0087] Capacity allocation example (T1 time slot): The total effective capacity of the drone is 8.0Gbps, the main hotspot H1 service requires 4.5Gbps, and 4.5Gbps is allocated; remaining capacity... Gbps is allocated to the secondary hotspot H4, while no capacity is allocated to H2 and H3.

[0088] Example 2: Convention and Exhibition Center Scenario (Low Wind Speed ​​Parameters) In one embodiment of this application, parameters are adjusted based on the scenario of Embodiment 1: the maximum migration speed of the drone is changed to 8 m / s, the fine-tuning radius is changed to 15 m, and the remaining parameters, decision logic, and capacity allocation rules remain unchanged. This parameter configuration is suitable for exhibition scenarios with low wind speeds and small hotspot drift ranges, which can further reduce drone energy consumption and improve system operational stability.

[0089] Example 3: Outdoor large stadium scene In one embodiment of this application, a Class A stadium is used as the simulation scene. The total area of ​​the venue is 120,000 square meters. Six dynamic hotspot areas are set up, corresponding to six candidate hovering points. The maximum drift range of the hotspots can reach 35m, and the average wind speed in the environment is 5m / s.

[0090] Parameter configuration: Drone operating altitude 60m, maximum migration speed 10m / s, fine-tuning radius 25m, minimum hovering time slots 3; ground base capacity 10Gbps, drone nominal service capacity 12Gbps, terahertz backhaul capacity limit 9Gbps, secondary hotspot base reduction factor 0.75; energy consumption penalty factor dynamically increases by 15% with wind speed, switching penalty factor is appropriately reduced to adapt to a larger hotspot drift range.

[0091] Results: During the five stages of the event – ​​entry, first half, halftime, second half, and exit – the main hovering point switched only 3 times, far fewer than the 7 times of the greedy strategy. Throughout the event, by making minor adjustments to adapt to the small movements of the audience seating area, the duration of communication link interruptions was reduced by 62%, and the total energy consumption of the drones was reduced by 28%.

[0092] Example 4: Dynamic Adaptation Scenarios for Emerging Hot Topics In one embodiment of this application, based on the exhibition scenario of Embodiment 1, a sudden hotspot H5 (located in the temporary activity area on the east side of the venue) is added in the middle of the T2 time slot, with a service demand of 5Gbps, and the original candidate points cannot be effectively covered.

[0093] Figure 8 The flowchart illustrating the collaborative process of two-layer decision-making and switching constraints is shown. The operation is as follows: In the second sampling period of time slot T2, a sudden hotspot of H5 is detected, and the system automatically generates a candidate point P5 (450, 120) and adds it to the candidate set; P5 is included in the optimization traversal of time slot T3. After evaluation by the objective function, the comprehensive benefit of P5 is higher than that of the original P2. The UAV switches to P5 and hovers in time slot T3, allocating 4.2Gbps of supplementary capacity to H5; After H5 disappears at the end of the activity, the candidate point P5 is retained until the end of the operation cycle, or it can be manually removed.

[0094] Performance verification: It only takes 2 time slots from the emergence of the hotspot to the drone completing the hovering switch, and the response latency is 80% lower than that of the traditional manual scheduling solution, which can quickly handle sudden large traffic.

[0095] Example 5: Terahertz Link Failure Scenario In one embodiment of this application, during the T2 peak time slot of Embodiment 1, a terahertz backhaul link is simulated to experience an excessive attenuation fault due to high-altitude obstruction.

[0096] Operation process: After the system detects a link anomaly, it switches to the microwave backup link within 100ms. The microwave link has a rated bandwidth of 4Gbps. The capacity constraints are updated synchronously, and the optimization solution is re-executed: The main hotspot H2 service requires 6.5Gbps, but due to the limited microwave link capacity, 4Gbps is allocated; the remaining capacity is 0, and no further allocation is made to the secondary hotspot.

[0097] Performance verification: There was no service interruption during the link switching process, only a smooth decrease in speed, and no user disconnection occurred, with link reliability improved to 99.99%.

[0098] In this embodiment of the application, to verify the performance of the proposed solution, the exhibition scenario of Embodiment 1 is used as a benchmark, and it is compared with two baseline solutions: traditional greedy scheduling and DQN reinforcement learning. The core indicator results are shown in Table 1 below: Table 1

[0099] Based on the above data, the core beneficial effects of this invention are as follows: First, the stability of communication services is significantly improved. Through the combination of hardware and software mechanisms, including switching penalty terms and minimum hovering duration constraints, the number of hovering switches is reduced by 72.7% compared to the greedy strategy and by 57.1% compared to the reinforcement learning scheme; the proportion of communication link interruption time is greatly reduced, and service continuity is significantly enhanced.

[0100] Second, the energy consumption of drones is optimized efficiently. The energy consumption of flight and hovering is jointly optimized, and the overall operating energy consumption is reduced by 27.3% compared with the greedy strategy and by 15.2% compared with the reinforcement learning scheme, effectively extending the drone's endurance.

[0101] Third, it has strong dynamic scene adaptability. The two-layer location decision can adapt to the natural drift of hotspots within a range of 20 to 30 meters; the dynamic candidate point generation mechanism can respond to sudden hotspots within 2 time slots, and the adaptation speed is far superior to traditional solutions.

[0102] Fourth, it has strong practical applicability in engineering. The time complexity of solving the discrete candidate point traversal is O(log n). , With a limited number of candidate points, airborne embedded devices can perform millisecond-level calculations, with decision latency only one-fifteenth that of reinforcement learning schemes. It requires no training, has no generalization problems, and has an extremely low deployment threshold.

[0103] Fifth, higher capacity utilization. Layered capacity allocation and location-based decision-making are nested and optimized, resulting in a 22% increase in average throughput in hotspot areas compared to the average allocation method, prioritizing the service experience of high-priority core hotspots.

[0104] Sixth, the link reliability is strong. Seamless switching between terahertz and microwave dual links ensures no service interruption in fault scenarios, significantly improving system robustness.

[0105] Seventh, flexible scenario adaptation. Both the weighting coefficient and the reduction coefficient can be adaptively adjusted to adapt to various scenarios such as exhibitions, sports events, and emergency response.

[0106] It should be noted that the dynamic hotspot UAV dual-layer position adaptive hovering planning method and system provided by this invention has clear industrial applicability. This invention can be applied to scenarios requiring temporary supplementation of communication capacity, such as convention centers, stadiums, and emergency communication support, and is particularly suitable for the dynamic tracking and service of communication hotspots in large-scale indoor-outdoor mixed scenarios. This invention employs a time-slot-by-time online optimization algorithm, which has low computational complexity and can be directly deployed and run on existing UAV-borne embedded devices without relying on high-performance cloud computing power, resulting in low hardware implementation costs. The technical solution of this invention has been verified through multiple simulation examples, demonstrating mature technical feasibility and can be directly applied to the design and manufacturing of UAV communication systems.

[0107] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0108] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 9As shown, the computer device includes 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, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a dynamic hotspot UAV dual-layer position adaptive hovering planning method. 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.

[0109] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0110] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0111] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0112] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0113] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0114] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0115] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0116] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A dynamic hotspot UAV dual-layer position adaptive hovering planning method, characterized in that, The method includes: Obtain the center location, real-time business demand, priority weight, and hovering position and continuous hovering duration of each hotspot in the current time slot; A multi-objective optimization objective function is constructed, which takes the maximization of weighted service revenue as the core and adds migration distance cost, switching action penalty, flight energy consumption and hovering energy consumption as penalty terms. Set constraints, including single master hovering point selection constraints, fine-tuning range constraints, migration speed constraints, minimum hovering duration constraints, backhaul link capacity constraints, and total UAV service capacity constraints. Traverse each candidate hovering point in the candidate hovering point set, calculate the optimal two-dimensional fine-tuning amount within the fine-tuning radius of each candidate hovering point, and substitute each candidate hovering point and its corresponding optimal two-dimensional fine-tuning amount into the multi-objective optimization objective function. Perform hierarchical capacity allocation under the current candidate hovering point, and calculate the comprehensive benefit value based on the capacity allocation result. The minimum hovering duration constraint is verified. If the continuous hovering duration of the current time slot is less than the minimum hovering threshold, the hovering point of the previous time slot is forcibly maintained as the main hovering point of the current time slot; otherwise, the candidate hovering point with the largest comprehensive benefit value is selected as the main hovering point of the current time slot. Control the drone to fly to the main hovering point and execute the corresponding two-dimensional fine-tuning, update the drone's real-time position and continuous hovering duration, and then enter the next time slot.

2. The dynamic hotspot UAV dual-layer position adaptive hovering planning method according to claim 1, characterized in that, The expression for the multi-objective optimization objective function is: in, Hot Topics In the time slot Priority weight, Hot topic The business demand, Hot Topics Corresponding ground foundation capacity, Drones are a hot topic Provided supplementary capacity, For the drone's migration distance, To switch the indicator variable for the hover point, , , , These are the penalty weighting coefficients for migration distance cost, switching action penalty, flight energy consumption, and hovering energy consumption, respectively. The current time slot flight energy consumption, This represents the energy consumption during hovering in the current time slot.

3. The dynamic hotspot UAV dual-layer position adaptive hovering planning method according to claim 1, characterized in that, The primary hovering point and the two-dimensional fine-tuning amount constitute a two-layer position decision mechanism, specifically including: The main hovering point serves as the primary hovering position for the drone, used to address large-scale positional shifts in hotspot areas. The two-dimensional fine-tuning amount serves as a local position adjustment around the main hovering point, used to adapt to the small-range natural drift of the hotspot area. The criteria for determining small-range natural drift is that the hotspot offset distance is less than or equal to the fine-tuning radius, and the criteria for determining large-range positional drift is that the hotspot offset distance is greater than the fine-tuning radius.

4. The dynamic hotspot UAV dual-layer position adaptive hovering planning method according to claim 1, characterized in that, The hierarchical capacity allocation rule is as follows: Prioritize allocating capacity to the primary hotspot with the highest priority weight, with the allocation value being the smaller of the primary hotspot's business demand and the total effective capacity of the drones. If there is still remaining capacity after allocation, the remaining capacity is multiplied by a reduction factor and then allocated to the secondary hotspot with the second highest priority. The reduction factor decreases as the distance between the secondary hotspot and the actual location of the UAV increases. No additional drone capacity will be allocated to hotspots other than primary and secondary hotspots.

5. The dynamic hotspot UAV dual-layer position adaptive hovering planning method according to claim 2, characterized in that, The penalty weight coefficient adopts an adaptive tuning strategy, specifically: When the drone's remaining battery power is below a preset threshold, the flight energy consumption penalty coefficient is increased. and hovering energy consumption penalty coefficient At the same time, reduce the penalty coefficient for switching actions. ; When the drone's remaining battery power is higher than a preset threshold, reduce... and hovering energy consumption penalty coefficient ; Calculate the average offset distance of the hotspot center position within multiple consecutive time slots. When the average offset distance is less than a set proportion of the fine-tuning radius, increase... When the average offset distance continuously exceeds the fine-tuning radius, reduce... .

6. The dynamic hotspot UAV dual-layer position adaptive hovering planning method according to claim 1, characterized in that, The candidate hover point set supports dynamic updates, and the update process includes: Real-time monitoring of business needs and location changes at various hotspots; When a new hotspot area is detected and all of the following triggering conditions are met, a new candidate hover point is automatically generated and added to the candidate hover point set: The distance from all existing candidate hovering points to the newly added hotspot area is greater than the effective service radius of the drone; The business demand in the newly added hotspot area exceeded the set percentage threshold of the ground infrastructure capacity.

7. The dynamic hotspot UAV dual-layer position adaptive hovering planning method according to claim 1, characterized in that, The backhaul link corresponding to the capacity constraint adopts a dual-link architecture of terahertz primary backhaul link and microwave backup backhaul link. The signal-to-noise ratio (SNR) and packet loss rate of the terahertz primary backhaul link are monitored in real time. When the SNR is lower than a set threshold or the packet loss rate is higher than a set percentage threshold and continues for a preset number of sampling periods, the primary link is determined to be abnormal. The system immediately switches to the microwave backup backhaul link and updates the upper limit of the capacity in the backhaul link capacity constraint from the terahertz capacity upper limit to the microwave link rated bandwidth. Based on the updated backhaul link capacity constraint, the optimization solution for the current time slot is re-executed.

8. The dynamic hotspot UAV dual-layer position adaptive hovering planning method according to claim 1, characterized in that, The minimum hovering duration threshold adopts an adaptive adjustment scheme: When the remaining battery power of the drone is lower than a preset threshold, the minimum hovering time threshold is increased; when the remaining battery power of the drone is higher than the preset threshold, the minimum hovering time threshold is decreased. The frequency of hotspot location fluctuations within a unit of time is counted. The higher the fluctuation frequency, the lower the minimum hovering time threshold is; the lower the fluctuation frequency, the higher the minimum hovering time threshold is.

9. The dynamic hotspot UAV dual-layer position adaptive hovering planning method according to claim 1, characterized in that, Among the constraints: The single primary hovering point selection constraint allows only one candidate hovering point to be selected as the primary hovering point for each time slot; The fine-tuning range constraint is that the sum of the squares of the lateral and longitudinal fine-tuning amounts around the main hovering point of the UAV is less than or equal to the square of the fine-tuning radius. The migration speed constraint is that the UAV migration distance is less than or equal to the product of the maximum migration speed and the time slot duration.

10. A dynamic hotspot UAV dual-layer position adaptive hovering planning system, characterized in that, The system includes: The scene modeling module is used to construct dynamic hotspot scenes, set up candidate hover point sets, and complete system parameter initialization. The status awareness module is used to collect the center location of each hotspot, real-time business demand, priority weight, and the hovering point and continuous hovering duration of the drone. The optimal selection module is used to construct multi-objective optimization objective functions and constraints, perform candidate hovering point traversal and optimal solution calculation, and output the optimal master hovering point and two-dimensional fine-tuning amount of the current time slot. The execution control module is used to receive decision commands and control the UAV to complete flight migration, hovering, and position fine-tuning actions; The communication module integrates a terahertz main backhaul unit and a microwave backup backhaul unit to achieve dual-link redundant data transmission and data interaction between the UAV and the ground receiving node.

Images