An unmanned aerial vehicle swarm intelligent cooperative control method and system

Abstract

This invention relates to the field of unmanned aerial vehicle (UAV) control technology and discloses an intelligent collaborative control method and system for UAV swarms. The method includes acquiring task urgency, remaining battery percentage, task execution time limit, battery consumption rate, real-time location coordinates, and channel identifiers; ranking UAVs in a competitive manner to obtain a priority sequence; assigning risk weights to obtain a conflict risk distribution map; classifying UAVs based on the conflict risk distribution map to obtain low-battery and high-battery passage sequences; adjusting passage time and occupancy period to obtain an optimized passage sequence; filtering channel conflicts based on the optimized passage sequence to obtain a channel priority sequence; evaluating collaborative consistency to obtain a synchronization coordination index; and correcting the timing based on the synchronization coordination index to obtain a corrected passage sequence; iteratively adjusting the scheme to obtain the final allocation scheme. This method can achieve intelligent solutions to resource competition and path conflicts.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to an intelligent collaborative control method and system for UAV swarms. Background Technology

[0002] Currently, drone swarms face an urgent need for real-time multi-drone collaboration in complex and dynamic scenarios such as emergency rescue and environmental monitoring. When multiple drones simultaneously request to use the same communication channel or fly over the same airspace node, the system lacks a mechanism to quickly and fairly resolve resource competition and path conflicts based on mission urgency and flight status under decentralized command conditions, leading to mission delays and even safety hazards. Therefore, there is an urgent need for a control technology capable of coordinating drone swarm resource allocation and passage order in real time, incorporating such conflict resolution mechanisms into fault prediction and health management to ensure the overall order of swarm operation and mission success rate.

[0003] In one existing technology, the system pre-sets fixed task priorities for each UAV and uses a centralized scheduling method to handle multi-UAV collaboration. When multiple UAVs simultaneously request to occupy the same communication channel or plan to fly over the same airspace node, the system allocates resources sequentially according to the preset priority order; if priorities are the same, resources are selected randomly. The system first obtains the task type and predetermined path of each UAV, identifies overlapping areas of paths through a spatiotemporal grid, and adjusts UAVs with lower priority in the overlapping areas to wait or detour. Regarding communication resource allocation, the system authorizes channel occupancy sequentially based on the communication needs transmitted back by the UAVs; if channel conflicts occur, allocation is based on the order of application time. This processing method may cause UAVs with low battery or urgent tasks to be unable to complete their tasks due to battery depletion or timeouts while waiting. Existing technology identifies conflicts through fixed priorities and predetermined paths, setting low-priority or later-applying UAVs to wait or detour, leading to delays in urgent tasks or timeouts for low-battery UAVs.

[0004] Therefore, existing technologies cannot achieve intelligent solutions to resource competition and path conflicts. Summary of the Invention

[0005] This invention provides an intelligent collaborative control method and system for unmanned aerial vehicle (UAV) swarms to achieve intelligent solutions to resource competition and path conflicts.

[0006] Firstly, in order to solve the above-mentioned technical problems, the present invention provides an intelligent cooperative control method for unmanned aerial vehicle (UAV) swarms, comprising: Acquire drone status information and perform drone competition ranking to obtain a drone priority sequence; Based on the drone priority sequence, the risk probability is calculated to obtain the collision probability. Based on the collision probability, the risk weight is assigned to the preset task area to obtain the conflict risk distribution map. Based on the conflict risk distribution map, a weighted correction calculation is performed to obtain a correction priority score. Based on the correction priority score, the drones are classified according to a preset battery alarm threshold to obtain a low battery passage sequence and a high battery passage sequence. Based on the low-power passage sequence and the high-power passage sequence, the passage time and occupancy period are adjusted to obtain an optimized passage sequence; Based on the optimized passage sequence, channel conflict filtering is performed to obtain the channel priority sequence; Based on the channel priority sequence, the distribution pattern is deduced to obtain the group distribution pattern, and based on the group distribution pattern, the coordination consistency is evaluated to obtain the synchronization coordination index. Based on the aforementioned synchronization and coordination indicators, high-risk screening is performed to obtain high-risk drone sequences, and time-series correction is performed based on the high-risk drone sequences to obtain corrected passage sequences. Based on the modified passage sequence and the channel priority sequence, the scheme is iteratively adjusted to obtain the final allocation scheme.

[0007] Secondly, the present invention provides an intelligent collaborative control system for unmanned aerial vehicle (UAV) swarms, comprising: The data preprocessing module is used to acquire UAV status information and perform competitive sorting of UAVs to obtain a priority sequence of UAVs; The weight assignment module is used to calculate the risk probability based on the priority sequence of the UAVs, obtain the collision probability, and assign risk weights to the preset task area based on the collision probability to obtain a conflict risk distribution map. The power classification module is used to perform weighted correction calculations based on the conflict risk distribution map to obtain a correction priority score, and to classify drones based on the correction priority score and a preset power alarm threshold to obtain a low power passage sequence and a high power passage sequence. The cycle adjustment module is used to adjust the passage time and occupancy cycle according to the low power passage sequence and the high power passage sequence to obtain an optimized passage sequence; The conflict filtering module is used to perform channel conflict filtering based on the optimized passage sequence to obtain a channel priority sequence; The collaborative evaluation module is used to deduce the distribution pattern based on the channel priority sequence to obtain the group distribution pattern, and to evaluate the collaborative consistency based on the group distribution pattern to obtain the synchronization coordination index. The timing correction module is used to perform high-risk screening based on the synchronization coordination index to obtain a high-risk UAV sequence, and to perform timing correction based on the high-risk UAV sequence to obtain a corrected passage sequence. The output module is used to perform iterative adjustments to the scheme based on the corrected passage sequence and the channel priority sequence to obtain the final allocation scheme.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention converts the remaining power ratio and power consumption rate into redundant time, and weights the reciprocal of the redundant time with the urgency of the task to obtain the initial priority score. Since the redundant time reflects the power buffering capacity of the UAV within the task time limit, the priority score after integrating the power urgency and the task importance more realistically reflects the urgency when multiple UAVs request resources. It solves the problem that the existing technology relies on fixed priority, which causes low-power UAVs to crash due to waiting. It achieves both fairness and urgency in resource competition.

[0009] (2) This invention generates a conflict risk distribution map by identifying overlapping areas of the path and assigning risk weight coefficients. It adjusts the passage time and compresses the occupancy period according to the low-battery passage sequence and the high-battery passage sequence. Since the latest passage time of the low-battery drone is constrained by the remaining endurance and the safety time margin, the high-battery drone shortens the occupancy period by increasing its speed to make room for the timing window. This solves the problem of setting low-priority drones to wait or detour, which leads to delays in high-urgent tasks, and realizes the dual collaborative optimization of physical space and communication resources.

[0010] (3) This invention obtains the synchronization coordination index by deriving the distribution pattern through the group evolution model, and recalculates the timing correction amount according to the risk distribution map when the index is lower than the threshold. It verifies the non-conflictability of the allocation scheme through zero-space projection and iteratively adjusts it until the modulus is zero. Since the synchronization coordination index quantifies the synchronization tightness of the formation operation, the projection modulus being zero mathematically proves that all conflict points have been eliminated, solving the problem that the existing static scheme cannot adapt to dynamic changes, and realizing the closed-loop iteration and adaptive optimization of UAV swarm collaborative control. Attached Figure Description

[0011] Figure 1 This is a schematic flowchart of the intelligent collaborative control method for unmanned aerial vehicle swarms provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the intelligent collaborative control system for unmanned aerial vehicle swarms provided in the second embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] Reference Figure 1 The first embodiment of the present invention provides an intelligent collaborative control method for unmanned aerial vehicle (UAV) swarms, comprising the following steps: S11: Obtain UAV status information and perform UAV competition sorting to obtain UAV priority sequence; S12, calculate the risk probability according to the priority sequence of the UAVs to obtain the collision probability, and assign risk weights to the preset task area according to the collision probability to obtain a conflict risk distribution map. S13. Based on the conflict risk distribution map, perform weighted correction calculation to obtain a correction priority score, and based on the correction priority score, classify drones according to a preset power alarm threshold to obtain a low power passage sequence and a high power passage sequence. S14, based on the low-power passage sequence and the high-power passage sequence, adjust the passage time and occupancy period to obtain an optimized passage sequence; S15, based on the optimized passage sequence, perform channel conflict filtering to obtain a channel priority sequence; S16. Based on the channel priority sequence, perform distribution pattern deduction to obtain the group distribution pattern, and based on the group distribution pattern, perform coordination consistency evaluation to obtain synchronization coordination index. S17. Based on the synchronization and coordination indicators, high-risk screening is performed to obtain a high-risk drone sequence, and the timing is corrected based on the high-risk drone sequence to obtain a corrected passage sequence. S18, based on the modified passage sequence and the channel priority sequence, the scheme is iteratively adjusted to obtain the final allocation scheme.

[0014] In step S11, the drone status information is obtained, and the drone competition ranking is performed to obtain the drone priority sequence, including: Obtain the drone's mission urgency, remaining battery percentage, mission execution time limit, battery consumption rate, real-time location coordinates, and channel identifier; Divide the remaining battery percentage by the battery consumption rate to obtain the theoretical battery life, and subtract the task execution time limit from the theoretical battery life to obtain the redundancy time; The initial priority score is obtained by weighting and summing the reciprocal of the redundant time with the urgency of the task. Based on the real-time position coordinates, the Euclidean distance between each pair of drones is calculated to obtain the spatial distance; By filtering out drones whose spatial distance is less than a preset safe distance threshold and whose channel identifiers are the same, a drone competition sequence is obtained; Based on the initial priority score, the drone competition sequence is sorted in descending order to obtain the drone priority sequence.

[0015] Specifically, the system receives real-time status data packets from the onboard sensors of each UAV via a distributed network communication protocol at a fixed sampling period. The data packets contain six data items: mission urgency, remaining battery percentage, mission execution time limit, battery consumption rate, real-time location coordinates, and channel identifier. The mission urgency is pre-set by the ground mission planning system based on mission attributes and sent to the UAV; its value is an integer from 1 to 10, with higher values ​​indicating more urgent tasks. The remaining battery percentage is the percentage of the current battery's remaining charge relative to its full charge. The mission execution time limit is the maximum remaining time allowed for the UAV to complete the current mission, in minutes. The battery consumption rate is the percentage of battery power consumed per minute under the current flight conditions. The real-time location coordinates are obtained through the onboard GPS and include longitude, latitude, and altitude. The channel identifier is the communication channel number planned for use by the UAV, pre-assigned by the communication resource management system according to mission requirements.

[0016] During data acquisition, the system employs a timestamp alignment mechanism to ensure that the status data reported by each UAV within the same sampling period belongs to the same time segment. For missing data due to network latency or packet loss, the system performs linear extrapolation to complete the data using data from the previous sampling period and marks it as an estimated value for subsequent processing. All raw data undergoes validity verification upon receipt, and outliers exceeding the physical range are removed before being stored in the system's operational status database in a structured format for use in subsequent steps.

[0017] The above data forms the foundation for the collaborative control of UAV swarms. Task urgency and remaining battery percentage jointly determine the priority of UAVs in resource competition; task execution time and battery consumption rate are used to assess the remaining endurance of UAVs; real-time position coordinates are used for spatial conflict detection; and channel identifiers are used to identify communication resource contention. By acquiring this multi-dimensional state information, the system integrates physical battery constraints, task importance constraints, spatial location constraints, and communication resource constraints into a unified decision-making framework, providing data support for subsequent priority scoring, conflict identification, and resource allocation.

[0018] Specifically, the remaining battery percentage, battery consumption rate, and mission execution time limit are extracted from the status data of each drone. Dividing the remaining battery percentage by the battery consumption rate yields the theoretical flight time of the drone at the current battery consumption level. This theoretical flight time represents the total duration the drone can fly continuously while maintaining its current power consumption. Then, the mission execution time limit is subtracted from the theoretical flight time to obtain the redundancy time. This redundancy time reflects the remaining time buffer after the drone completes its assigned task. A shorter redundancy time indicates a higher urgency for the drone to complete the task; a negative redundancy time indicates that the current battery power is insufficient to support the completion of the entire task.

[0019] The system obtains the mission urgency level of each drone and takes the reciprocal of its redundancy time. The reciprocal of the redundancy time is inversely proportional to the redundancy time; the shorter the redundancy time, the larger its reciprocal, thus giving drones with higher urgency a greater weight in subsequent weighted calculations. The system weights and sums the reciprocal of the redundancy time and the mission urgency level, where the sum of the weight coefficients for mission urgency and redundancy time is 1. This weight coefficient is determined through statistical analysis of historical mission data. Multiple sets of flight records containing both successful and failed missions are collected. With mission success rate as the optimization objective, the system iterates through the mission urgency weight values ​​within a preset range with a fixed step size, selecting the weight combination that maximizes the mission success rate as the preset weight. For example, if the mission urgency weight values ​​are iterated within the range of 0.1 to 0.9 with a step size of 0.1, and the highest mission success rate is found when the mission urgency weight is 0.6 and the redundancy time reciprocal weight is 0.4, then the mission urgency weight is preset to 0.6 and the redundancy time reciprocal weight is preset to 0.4. The weighted summation method involves multiplying the task urgency by its weight, multiplying the reciprocal of the redundancy time by its weight, and then summing the two products to obtain the initial priority score. This score integrates task importance and battery urgency into a comprehensive indicator; a higher score indicates that the drone should receive higher priority in resource competition.

[0020] It should be noted that when the redundancy time is positive and greater than the preset minimum redundancy threshold, its reciprocal is directly used for weighted summation; when the redundancy time is less than or equal to the minimum redundancy threshold, the redundancy time is fixed to the minimum redundancy threshold and then its reciprocal is taken; when the redundancy time is negative, the absolute value of the redundancy time is added to the minimum redundancy threshold and then its reciprocal is taken, and the upper limit of the priority score is set to the maximum value of the task urgency to prevent score divergence; wherein, the minimum redundancy threshold is 0.1 minutes, and by setting the redundancy time to vary from 0.01 minutes to 1 minute in the simulation environment, the fluctuation range of the initial priority score is observed, and the critical value that makes the score change smooth and does not have abnormal sudden changes is selected as the minimum redundancy threshold.

[0021] Based on the real-time position coordinates of each UAV, the spatial distance between each UAV and all other UAVs is calculated. For any two UAVs, their three-dimensional coordinates are extracted, the sum of squares of the coordinate differences on the three coordinate axes is calculated, and the square root of the sum of squares is taken to obtain the Euclidean distance between the two UAVs. The system iterates through all UAV pairs to complete the Euclidean distance calculation between every two UAVs, thus obtaining the spatial distance.

[0022] The system presets a safe distance threshold, which is determined by collecting measured data on the relationship between communication signal strength and distance during UAV flight. The signal-to-noise ratio (SNR) of the communication signal is recorded at different distances during the measurements. When the SNR falls below the minimum required communication quality, the critical distance is recorded. Critical distance values ​​are collected from multiple flight missions, and the maximum value is taken as the safe distance threshold. The system compares the calculated spatial distance with the preset safe distance threshold, and simultaneously compares the channel identifiers acquired by each UAV. UAVs that simultaneously meet the conditions of having a spatial distance less than the safe distance threshold and the same channel identifier are selected. These UAV pairs are considered to have a communication interference risk and constitute a competitive relationship. The system extracts all UAVs involved in these pairs, removes duplicates, and obtains the UAV competition sequence.

[0023] Using the initial priority score of each drone in the drone competition sequence as the ranking basis, a descending sorting method is adopted, placing the drone with the highest score first in the sequence, the second highest score second, and so on, ultimately obtaining the drone priority sequence. This sequence reflects the processing order of drones in a spatially proximate drone group with communication resource contention, from high to low overall urgency, providing a decision-making basis for resource allocation and path conflict analysis in subsequent steps.

[0024] In step S12, based on the UAV priority sequence, a risk probability calculation is performed to obtain the collision probability. Then, based on the collision probability, a risk weight is assigned to a preset task area to obtain a conflict risk distribution map, including: Based on the preset obstacle distribution map, the preset task area is divided into three-dimensional grids according to the preset grid side lengths to obtain a discretized spatial grid. Based on the discretized spatial grid, the potential path trajectory is obtained by calculating the uniform motion trajectory of the UAV priority sequence using the uniform linear motion formula. Identify path segments in the potential path trajectories that pass through the same discretized spatial grid within the same time window to obtain the path overlap region; Extract the relative speed and approach angle of each potential path trajectory within the path overlap area, and calculate the collision probability using a pre-constructed risk mapping function; Based on the urgency of the task, the collision probabilities are weighted and summed to obtain risk weight coefficients, which are then assigned to the corresponding discretized spatial grid to obtain a conflict risk distribution map.

[0025] Specifically, based on the drone priority sequence, a preset 3D map of the mission area and a preset obstacle distribution map are loaded. The preset mission area is the 3D spatial range in which the drone performs the mission, defined by latitude, longitude, and altitude. The preset obstacle distribution map contains labeled data of impassable areas such as buildings and mountains within this spatial range. The mission area is discretized into a 3D grid according to a preset grid side length. The preset grid side length is determined through simulation analysis. In the simulation environment, the mission area is divided with different grid side lengths, and a path conflict identification algorithm is run for each. The optimal grid side length is selected with the goal of minimizing the sum of the false alarm rate and the missed alarm rate. For example, after simulation comparison, a grid side length of 10 meters results in the lowest sum of the false alarm rate and the missed alarm rate, so the preset grid side length is 10 meters. During the grid division process, each obstacle in the preset obstacle distribution map is traversed, and the grid cells occupied by the obstacle are marked as impassable areas. After marking, a discretized spatial grid is obtained.

[0026] Based on the real-time position coordinates and flight velocity vectors of each drone in the drone priority sequence, the trajectory of each drone within a future time window is calculated using the formula for uniform linear motion. Specifically, for each drone, its current 3D spatial coordinates are extracted as the starting point, and its flight velocity vector is extracted as the direction and speed of motion. The time window is divided into several discrete time points. Starting from the current moment, the spatial position at each time point is calculated sequentially. The spatial position at each time point is equal to the starting point coordinates plus the product of the velocity vector and the time interval. The calculated spatial position at each time point is mapped onto a discretized spatial grid, and the grid cell occupied by the drone at that time point is recorded. This process is repeated for all time points to obtain the potential path trajectory of the drone. The above calculation is repeated for each drone in the drone priority sequence to obtain the potential path trajectories of all drones.

[0027] The algorithm iterates through all potential UAV paths. For each discretized spatial grid cell, it checks whether two or more UAVs have potential paths passing through that grid cell within the same time window. Specifically, for each grid cell, it extracts the time points when all UAVs appear in that grid cell. If the time difference between different UAVs is less than the preset time window width, it determines that the paths of these UAVs overlap in that grid cell. All grid cells that meet the above conditions, along with their corresponding UAV identifiers and time window information, are recorded to obtain the path overlap area. The preset time window width is 2 seconds, determined by statistically analyzing the grid passage time of UAV swarms at typical flight speeds. Using a grid side length of 10 meters and a common UAV speed of 10 meters per second, the passage time for a single grid is calculated to be 1 second. This time window width is doubled to cover the overlap of adjacent time points, and simulations verify that the collision false negative rate is minimized.

[0028] The overlapping UAV pairs are extracted from the overlapping path region. The flight velocity vector of each UAV is obtained, and the relative velocity vector between the two UAVs is calculated. The relative velocity vector equals the velocity vector of one UAV minus the velocity vector of the other UAV. The magnitude of this vector is then calculated to obtain the relative velocity magnitude. The approach angle is determined by calculating the angle between the velocity vectors of the two UAVs and taking the cosine of the angle. The relative velocity magnitude and approach angle are input into a pre-constructed risk mapping function, which is built using historical flight data and collision simulation data. During the construction process, multiple sets of flight records containing relative velocity, approach angle, and whether a collision occurred are collected. Relative velocity and approach angle are used as input features, and collision occurrence is used as the output label. Logistic regression is used to fit the function parameters, and the function outputs a collision probability value between 0 and 1. The function is called, and the relative speed and approach angle of the current drone pair are input to obtain the collision probability. Specifically, the risk mapping function adopts the logistic function, and the output value of the function is equal to a negative exponent raised to the power of a natural constant, where the exponent is a linear combination of the relative speed magnitude and the cosine of the approach angle. The linear combination coefficient and intercept are obtained by fitting historical collision event data using the maximum likelihood estimation method.

[0029] The task urgency levels of drones participating in path overlap are obtained, and the collision probability and task urgency are weighted and summed. The sum of the weights for collision probability and task urgency is 1, and the weight coefficients are determined through statistical analysis of historical task data. Multiple sets of flight mission records containing different combinations of collision probability and task urgency are collected. With the mission success rate as the optimization objective, the collision probability weights are iterated through within a preset range, and the weight combination that maximizes the mission success rate is selected as the preset weight. For example, if statistics show that a collision probability weight of 0.5 and a task urgency weight of 0.5 result in the highest mission success rate, then both weights are preset to 0.5. The weighted sum is calculated by multiplying the collision probability by its weight and the task urgency by its weight, then summing the two products to obtain the risk weight coefficient.

[0030] The calculated risk weight coefficients are assigned to the corresponding discretized spatial grid cells. Each grid cell is assigned a value based on the risk weight coefficients calculated within its path overlap region. If the same grid cell is involved in multiple pairs of overlapping UAV paths, the maximum value among all risk weight coefficients is taken as the final risk value for that grid cell. After assigning values ​​to all high-risk grid cells, a conflict risk distribution map is output. This distribution map is presented in a three-dimensional grid format, with each grid cell carrying a risk weight coefficient to characterize the degree of collision risk at that location in both time and space dimensions.

[0031] In step S13, a weighted correction calculation is performed based on the conflict risk distribution map to obtain a correction priority score. Then, based on the correction priority score and a preset battery alarm threshold, the drones are classified to obtain low-battery passage sequences and high-battery passage sequences, including: The UAVs whose risk weight coefficients in the conflict risk distribution map exceed a preset conflict risk threshold are selected to obtain a set of conflict UAVs; Based on a preset task relationship matrix, the urgency of the tasks corresponding to the conflicting drone sets is weighted and calculated to obtain a corrected priority score; Based on the correction priority score, the conflicting drone set is sorted in descending order to obtain a correction priority sequence; The correction priority sequence is classified according to a preset power alarm threshold to obtain a high power passage sequence and a low power passage sequence.

[0032] Specifically, each discretized spatial grid cell in the conflict risk distribution map carries a risk weight coefficient. All grid cells in the conflict risk distribution map are traversed, and the risk weight coefficients are extracted and compared with a preset conflict risk threshold. The preset conflict risk threshold is determined through statistical analysis of historical flight accident data. Multiple sets of event records of collisions or close-range conflicts between UAVs during flight are collected, and the values ​​of the risk weight coefficients at the time of the events are extracted. The 95th percentile of all event risk weight coefficients is taken as the conflict risk threshold, ensuring that the risk weight coefficient in 95% of historical conflict events is higher than this threshold. Grid cells with risk weight coefficients exceeding this threshold are selected, and all UAV identifiers involved in these grid cells are extracted. After removing duplicates, a set of conflicting UAVs is obtained.

[0033] Obtain the preset task relationship matrix. The construction process of this task relationship matrix is ​​as follows: First, enumerate all drone task types that may appear in the same task scenario and assign a unique identifier to each task type. For each pair of task types, determine the dependency weight based on their dependency relationship in actual task execution. If task A must depend on task B to execute, then set the dependency weight of task B on task A to 1.0; if task A assists task B in execution but does not constitute a necessary dependency, then set the dependency weight of task B on task A to 0.5; if there is no dependency relationship between the two tasks, then set the dependency weight to 0. For the case of bidirectional dependency, set the dependency weights in both directions respectively. Arrange the dependency weights of all task types in order of task identifier to form a square matrix. The rows and columns of this square matrix correspond to task types, and the element in the i-th row and j-th column of the square matrix represents the dependency weight of task j on task i. After construction, the task relationship matrix is ​​stored in the system configuration file in the form of a two-dimensional array for loading when called in step S13.

[0034] Step S11 retrieves the mission urgency level of each drone. The mission type identifier for each drone is extracted from the conflicting drone set. Based on the mission type identifier, the dependency weight between that drone and other conflicting drones is retrieved from a pre-defined mission relationship matrix. For each drone in the conflicting drone set, all other drones in the set are iterated through. The mission urgency level of each other drone is multiplied by its dependency weight on the current drone to obtain the dependency contribution value of that other drone to the current drone. The dependency contribution values ​​of all other drones are summed to obtain the dependency correction value. The dependency correction value is added to the original mission urgency level of the drone to obtain the corrected priority score.

[0035] Using the correction priority score of each drone in the conflict drone set as the sorting basis, a descending sorting method is adopted, placing the drone with the highest score at the first position in the sequence, the drone with the second highest score at the second position, and so on, to obtain the correction priority sequence after sorting.

[0036] A preset battery warning threshold is obtained, which is determined by the battery discharge characteristic curve and the flight safety boundary. A constant current discharge test is performed on the battery model used, and the voltage change curve over discharge time is recorded. When the voltage drops to the minimum operating voltage required by the UAV flight control system, the remaining battery percentage at that moment is read. The test is repeated under different ambient temperatures, and the maximum value among all test results is taken as the battery warning threshold. This ensures that the threshold does not exceed the actual crash risk battery value under any operating condition. For example, if the test results show that the remaining battery corresponding to the minimum operating voltage is between 15% and 25%, the maximum value of 25% is taken as the battery warning threshold. Using this threshold as a boundary, each UAV in the priority sequence is traversed, and the remaining battery percentage obtained in step S11 is extracted. If the remaining battery percentage is lower than the battery warning threshold, the UAV is classified into the low battery passage sequence; otherwise, it is classified into the high battery passage sequence.

[0037] In step S14, the passage time and occupancy period are adjusted according to the low-battery passage sequence and the high-battery passage sequence to obtain an optimized passage sequence, including: Divide the remaining battery percentage in the low battery passage sequence by the battery consumption rate to obtain the remaining flight time, and subtract a preset safety time margin from the remaining flight time to obtain the available flight time; Subtract the available flight time from the task execution time limit to obtain the latest passage time. Based on the low battery passage sequence and the real-time position coordinates, calculate the arrival and departure times of each UAV in the overlapping path area using the uniform linear motion formula to obtain the estimated arrival and departure times. Calculate the difference between the estimated arrival time and the latest departure time to obtain the time offset, and add the time offset to the estimated arrival time and the estimated departure time respectively to obtain the low battery update sequence; When the low battery update sequence and the high battery passage sequence do not overlap in time, the low battery update sequence and the high battery passage sequence are merged and sorted according to the correction priority score to obtain an optimized passage sequence. When the low battery update sequence and the high battery passage sequence overlap in time, the flight speed of the UAV in the high battery passage sequence is increased until the low battery update sequence and the high battery passage sequence no longer overlap in time, thus obtaining an optimized passage sequence.

[0038] Specifically, the low-battery passage sequence and high-battery passage sequence output in step S13 are obtained. From step S11, the remaining battery percentage, battery consumption rate, task execution time limit, and real-time location coordinates of each drone are obtained. From step S12, the coordinate range of the overlapping path area is obtained. For each drone in the low-battery passage sequence, the remaining battery percentage is divided by the battery consumption rate to obtain the remaining flight time. This remaining flight time represents the total time the drone can continue flying under the current battery consumption level. The remaining flight time is subtracted from the preset safety time margin to obtain the available flight time. The preset safety time margin is determined through flight test statistics. Multiple sets of extra flight time data for drones encountering sudden gusts or needing temporary detours are collected, and the maximum value among all test data is taken as the safety time margin. For example, if the test results show that the maximum extra flight time under emergency conditions is 150 seconds, then the preset safety time margin is 150 seconds. This margin is used to ensure that the drone still has battery support when dealing with emergencies.

[0039] Subtracting the available flight time from the mission execution time limit yields the latest passage time. This latest passage time indicates the time the drone must pass through the overlapping path area; otherwise, it will be unable to complete the remaining flight mission within the mission execution time limit. Obtain the real-time position coordinates and flight speed of each drone in the low-battery passage sequence, and calculate the arrival time of each drone in the overlapping path area using the uniform linear motion formula. Specifically, extract the entrance and exit coordinates of the overlapping path area, calculate the spatial straight-line distance between the drone's real-time position coordinates and the entrance coordinates, and divide this distance by the flight speed to obtain the arrival time. Calculate the spatial straight-line distance between the entrance and exit coordinates, divide this distance by the flight speed to obtain the passage duration. Add the arrival time and passage duration to obtain the departure time. After completing the calculations for all low-battery drones, obtain the estimated arrival time and estimated departure time.

[0040] Calculate the difference between the estimated arrival time and the latest departure time. A positive difference indicates the drone's arrival time is later than the latest departure time, requiring it to pass earlier; a negative difference indicates the drone's arrival time is earlier than the latest departure time, requiring no adjustment. Use this difference as a time offset, and add it to both the estimated arrival and departure times to obtain the adjusted arrival and departure times. After adjustment, the arrival time of all low-battery drones will not be later than their latest departure time. Combine the adjusted arrival and departure times with the corresponding drone identifier to obtain the low-battery update sequence.

[0041] The low-battery update sequence and the high-battery passage sequence are merged, and the occupancy periods of all drones after merging are checked for time overlap. The occupancy period is defined as the time interval from arrival time to departure time. For any two drones, if the arrival time of one drone is less than the departure time of the other drone and the departure time of the first drone is greater than the arrival time of the second drone, then the occupancy periods of the two drones are determined to overlap. If there is no time overlap, then according to the corrected priority score calculated in step S13, all drones in the low-battery update sequence and the high-battery passage sequence are merged and rearranged in descending order of corrected priority score to obtain the optimized passage sequence.

[0042] If time overlap exists, the flight speed of drones in the high-battery passage sequence is increased. The increase in flight speed is determined iteratively. Each time, the flight speed of all drones in the high-battery passage sequence is increased by a preset speed step size, for example, 5% of the original speed. The arrival and departure times of each drone in the high-battery passage sequence are recalculated, where arrival and departure times are inversely proportional to flight speed; as the speed increases, arrival and departure times decrease accordingly. The recalculated high-battery passage sequence is merged with the low-battery update sequence, and time overlap is checked again. The above steps of increasing speed and recalculating are repeated until the merged sequence has no time overlap. When the overlap is eliminated, the flight speed of each drone in the current high-battery passage sequence is recorded as the adjusted speed. Based on the adjusted arrival and departure times, combined with the corrected priority score, all drones are merged and sorted in descending order to obtain the optimized passage sequence.

[0043] It should be noted that the preset speed step size is five percent of the original flight speed. This value was determined through speed adjustment sensitivity testing. In the simulation environment, the speed of the high-power drone was increased with different step sizes. If the step size is too small, the number of iterations will be too many. If the step size is too large, it may cause new conflicts. The test results showed that a step size of five percent achieved the optimal balance between convergence speed and safety.

[0044] In step S15, channel conflict filtering is performed based on the optimized passage sequence to obtain a channel priority sequence, including: By filtering out drones with the same channel identifier in the optimized passage sequence, a channel conflict set is obtained; Based on the corrected priority score, the channel conflict set is sorted in descending order to obtain the channel priority sequence.

[0045] Specifically, the optimized passage sequence output in step S14 is obtained, where each drone carries the channel identifier obtained in step S11. All drones in the optimized passage sequence are traversed, and the channel identifier of each drone is extracted. Drones with the same channel identifier are grouped together. For each channel identifier group, if the number of drones in the group is greater than or equal to two, there is a communication channel conflict between these drones. The system extracts all drones in that group, merges all conflicting drone groups, removes duplicate drone identifiers, and obtains a channel conflict set.

[0046] The corrected priority score calculated in step S13 is obtained, which integrates the task urgency and task dependency of the UAVs. Using the corrected priority scores of each UAV in the channel conflict set as the sorting criterion, a descending sorting method is adopted, placing the UAV with the highest score first in the sequence, the second highest score second, and so on, to obtain the channel priority sequence. This sequence reflects the priority of each UAV in acquiring communication resources in descending order of overall urgency when communication channel competition exists. Through this sorting, the system can prioritize allocating communication channels to UAVs with high corrected priority scores in subsequent steps, ensuring that high-urgency tasks receive priority in communication resource contention, while preventing UAVs with low battery or high task dependency from being delayed due to waiting for communication resources. The channel priority sequence and the optimized passage sequence output in step S14 together constitute a dual collaborative scheme for physical space passage and communication resource allocation.

[0047] In step S16, based on the channel priority sequence, a distribution pattern is deduced to obtain the group distribution pattern, and based on the group distribution pattern, a coordination consistency assessment is performed to obtain synchronization coordination indicators, including: The expected arrival coordinates are obtained by multiplying the flight speed of each UAV in the channel priority sequence by the preset command distribution delay and then adding the result to the real-time position coordinates. Movement commands are sent to the drones in the channel priority sequence to obtain a feedback drone sequence, and the real-time position coordinates of the feedback drone sequence are extracted to obtain the actual arrival coordinates; The difference between the actual arrival coordinates and the expected arrival coordinates is calculated to obtain the coordinate deviation vector; and based on the coordinate deviation vector, the distribution pattern of the feedback UAV sequence is deduced through a pre-constructed swarm evolution model to obtain the swarm distribution pattern. Calculate the modulus of the Euclidean distance and velocity difference between each UAV in the group distribution pattern, and perform a weighted summation based on the calculation results to obtain the collaborative consistency matrix; The variance of all off-diagonal elements in the coordination consistency matrix is ​​calculated to obtain the synchronization coordination index.

[0048] Specifically, the channel priority sequence output in step S15 is obtained, and the real-time position coordinates and flight speed of each UAV are obtained from step S11. The preset command distribution delay is determined by statistically analyzing the measured delay of the communication link between the ground control station and the UAV. Multiple sets of time difference data from the issuance of commands to their reception by the UAV are collected, and the maximum value among all test data is taken as the command distribution delay. For example, if the test results show that the maximum delay is 0.5 seconds, then the preset command distribution delay is 0.5 seconds. For each UAV in the channel priority sequence, its flight speed is multiplied by the preset command distribution delay to obtain the displacement vector of the UAV during that delay time. This displacement vector is superimposed on the real-time position coordinates of the UAV to obtain the expected arrival coordinates of the UAV at the actual time the command takes effect.

[0049] Movement commands, containing target position or heading adjustment information, are issued to drones in the channel priority sequence. After executing the commands, the drones transmit execution feedback data packets via their onboard communication modules. These packets contain the drones' real-time position coordinates after executing the commands. The execution feedback from all drones is received, and the real-time position coordinates are extracted to obtain the actual arrival coordinates. The actual arrival coordinates reflect the actual position the drones reached after command execution. Due to factors such as flight environment disturbances and command execution accuracy, there is a deviation between the actual arrival coordinates and the expected arrival coordinates.

[0050] Calculate the difference between the actual arrival coordinates and the expected arrival coordinates. For each UAV, calculate the difference between each component of the three-dimensional coordinates to obtain the coordinate deviation vector, which contains the magnitude of the deviation in three directions.

[0051] A pre-constructed swarm evolution model is obtained. This model is built using historical flight data and dynamic simulation. The data preprocessing process is as follows: Complete flight records of multiple UAVs performing cooperative flight missions are collected. Each record contains real-time position coordinates, velocity vectors, command issuance time, and command execution feedback time in a continuous time series. Each record is resampled at fixed time intervals, and missing data between sampling points is filled using linear interpolation. Noise filtering is applied to the resampled data. For the position coordinate sequence, a sliding window averaging method is used, taking the average of the coordinates of five consecutive sampling points within the window as the filtered coordinates for that point, with a window step size of one sampling point. For the velocity vector sequence, it is calculated based on the difference results of the position coordinates and smoothed using the same window. The data segment between the command issuance time and the command execution feedback time in each record is extracted, and the coordinate deviation vector at each sampling point is calculated, i.e., the difference between the actual coordinate and the coordinate expected to be reached according to the command. The coordinate deviation vector, current spatial position, and current velocity vector at each sampling point are used as input features, and the coordinate deviation vector at the next sampling point is used as the output label to construct a training dataset.

[0052] The swarm evolution model employs a nonlinear state-space model structure. The model is a nonlinear function of the deviation vector, spatial position, and velocity vector, where the rate of change of the deviation vector is equal to the deviation vector, spatial position, and velocity vector. Model parameters are determined by fitting the training dataset using the least squares method. Specifically, the fitting process involves substituting the input features and output labels into the nonlinear function expression to construct a residual sum of squares function. Gradient descent is then used to iteratively solve for the parameter values ​​that minimize the residual sum of squares. The iteration terminates when the change in the residual sum of squares is less than a preset threshold. After model construction, the current coordinate deviation vector, the current spatial position of each drone in the feedback drone sequence, and the current velocity vector are input into the model. Based on the current state, the model infers the spatial position changes of the drone swarm within a preset time window and outputs the predicted coordinates of each drone at various future time points. The set of these predicted coordinates constitutes the swarm distribution pattern.

[0053] It is worth noting that the population evolution model uses a radial basis function network to fit the nonlinear function. The network inputs are the coordinate deviation vector, spatial position, and velocity vector at the current moment, and the output is the rate of change of the deviation vector. The radial basis function centers are selected from historical flight data through a clustering algorithm, and the network weights are determined using the least squares method.

[0054] The predicted coordinates and velocities of each UAV at the same moment are extracted from the group distribution pattern. For each pair of UAVs in the group distribution pattern, their Euclidean distance is calculated, which is the square root of the sum of the squares of the differences in the three-dimensional coordinates of the two UAVs. Simultaneously, the magnitude of the velocity difference is calculated, which is the square root of the sum of the squares of the differences in the components of the velocity vectors of the two UAVs. The Euclidean distance and the magnitude of the velocity difference are then weighted and summed, where the sum of the weights of the Euclidean distance and the magnitude of the velocity difference is 1. The weight coefficients are determined through statistical analysis of formation flight experimental data. Multiple sets of formation flight records containing different combinations of distance and velocity differences are collected. With the stability of the formation as the optimization objective, the values ​​of the Euclidean distance weights are iterated, and the weight combination that optimizes the formation stability index is selected. For example, statistical analysis shows that the formation is most stable when the Euclidean distance weight is 0.6 and the magnitude of the velocity difference weight is 0.4, so the preset weights are 0.6 and 0.4 respectively. The weighted summation is calculated by multiplying the Euclidean distance by its weight, multiplying the magnitude of the velocity difference by its weight, and then adding the two products to obtain the cooperative consistency coefficient of the UAV pair. The cooperative consistency coefficients of all UAV pairs are arranged in order of UAV identification to construct a cooperative consistency matrix. The rows and columns of this matrix correspond to the UAV identification, and the diagonal elements are set to 0.

[0055] Calculate the variance of all off-diagonal elements in the coordination consistency matrix. First, calculate the mean of all off-diagonal elements, which is the sum of all off-diagonal elements divided by the number of off-diagonal elements. Then, calculate the square of the difference between each off-diagonal element and the mean. Sum all the squares of the differences and divide by the number of off-diagonal elements to obtain the variance. This variance value is output as a synchronization coordination indicator. The smaller the variance, the higher the coordination consistency among UAVs and the more synchronized the formation operation; the larger the variance, the greater the differences between UAVs and the worse the formation coordination.

[0056] In step S17, high-risk screening is performed based on the synchronization and coordination indicators to obtain high-risk drone sequences, and timing correction is performed based on the high-risk drone sequences to obtain corrected passage sequences, including: When the synchronization and coordination index is not lower than the preset synchronization and coordination threshold, the current state is maintained for drone control. When the synchronization and coordination index is lower than the preset synchronization and coordination threshold, the flight trajectory is deduced by using the uniform linear motion formula based on the real-time position coordinates in the feedback UAV sequence to obtain the deduced risk distribution map. By filtering out the regions and corresponding drones whose risk weight coefficients in the simulated risk distribution map exceed the conflict risk threshold, high-risk regions and high-risk drone sequences are obtained. Calculate the time offset required for each drone in the high-risk drone sequence to avoid the high-risk area, and obtain the timing correction amount; The timing correction is superimposed on the planned passage time of each drone in the high-risk drone sequence to generate an initial instruction set; Based on the initial instruction set, the UAVs corresponding to each instruction in the initial instruction set are prioritized according to the pre-built instruction priority rules to obtain the corrected passage sequence.

[0057] Specifically, the synchronization coordination index output in step S16 is obtained, and a preset synchronization coordination threshold is loaded. The preset synchronization coordination threshold is determined through statistical analysis of formation flight experiment data. Multiple sets of UAV formation synchronization coordination index values ​​are collected under normal operating conditions, and the 75th percentile of all index values ​​is calculated. This 75th percentile is used as the synchronization coordination threshold, ensuring that the synchronization coordination index is higher than this threshold in 75% of normal operating conditions. The synchronization coordination index is compared with the preset synchronization coordination threshold. If the synchronization coordination index is not lower than the threshold, it indicates that the current formation is operating with good synchronization, and no adjustment is needed. The current passage sequence and command state are maintained, and coordinated control continues.

[0058] When the synchronization and coordination index is lower than the preset synchronization and coordination threshold, the feedback UAV sequence output in step S16 is obtained. This sequence contains the actual arrival coordinates and current flight speed of each UAV after executing the command. The real-time position coordinates of each UAV are extracted from the feedback UAV sequence. Using the same uniform linear motion formula as in step S12, the flight trajectory within a preset future time window is extrapolated, with the current real-time position coordinates as the starting point and the current flight speed as the direction and rate. The spatial position of each UAV at each future time point is calculated and mapped to a discretized spatial grid to obtain the extrapolated risk distribution map. This distribution map has the same grid structure as the conflict risk distribution map in step S12, and each grid cell carries a risk weight coefficient calculated based on the trajectory extrapolation results.

[0059] From the simulated risk distribution map, select grid cells whose risk weight coefficient exceeds the preset conflict risk threshold in step S12, extract the spatial area covered by these grid cells as high-risk areas, and extract all UAV identifiers involved in these grid cells. After removing duplicates, obtain the high-risk UAV sequence.

[0060] For each drone in the high-risk drone sequence, calculate the time offset required to avoid the high-risk area. Specifically, obtain the planned passage time of the drone through the high-risk area, which is derived from the arrival time allocated to the drone in the optimized passage sequence output in step S14. Obtain the shortest path distance between the drone's current real-time position coordinates and the entrance boundary of the high-risk area, and divide this distance by the drone's current flight speed to obtain the shortest arrival time required for avoidance. Compare the planned passage time with the shortest arrival time required for avoidance. If the planned passage time is earlier than the shortest arrival time required for avoidance, the time offset is zero; if the planned passage time is later than the shortest arrival time required for avoidance, subtract the shortest arrival time required for avoidance from the planned passage time to obtain the amount of time needed to be advanced as a time adjustment.

[0061] The timing correction is added to the planned transit time of each drone in the high-risk drone sequence; that is, the timing correction is subtracted from the original planned transit time to obtain the adjusted transit time. Based on the adjusted transit time, combined with the drone's current flight speed and path length, the arrival and departure times of the drones on the path are recalculated to form an initial command set. This initial command set contains new commands for each high-risk drone, including adjusted flight speed and heading, to ensure passage through the high-risk area within the adjusted time window.

[0062] Obtain pre-built command priority rules, which are categorized by task type and command attribute. The construction process of command priority rules is as follows: enumerate all possible command types, including avoidance maneuver commands, path adjustment commands, speed adjustment commands, communication commands, and status reporting commands. Assign priority values ​​to each command type based on its impact on mission success and flight safety. For example, the avoidance maneuver command has the greatest impact on safety and is assigned the highest priority value of 5; the path adjustment and speed adjustment commands are assigned priority values ​​of 4; the communication command is assigned priority value of 3; and the status reporting command is assigned priority value of 2. Store the mapping relationship between command types and priority values ​​as a priority rule table. In step S17, for each command in the initial command set, find the corresponding priority value from the priority rule table according to its command type. Sort all commands according to their priority values ​​from high to low. Commands with the same priority value are arranged according to the corrected priority score calculated by the UAV in step S13 from high to low. After sorting, use the UAV order in the command sequence as the new passage order and output the corrected passage sequence.

[0063] In step S18, the scheme is iteratively adjusted according to the corrected passage sequence and the channel priority sequence to obtain the final allocation scheme, including: A spatiotemporal state matrix is ​​constructed based on the corrected passage sequence, the channel priority sequence, and the preset instruction synchronization state; Based on the spatiotemporal state matrix, a smoothed state trajectory is obtained by using a cubic spline interpolation algorithm. Based on the smooth state trajectory, an initial allocation scheme is generated, and the conflict check vector of the initial allocation scheme is projected onto a preset conflict vector space to obtain the projection result; When the modulus of the projection result is zero, the initial allocation scheme is directly used as the final allocation scheme. When the modulus of the projection result is not zero, the timing correction amount is recalculated based on the projection result until the modulus of the projection result is zero, thus obtaining the final allocation scheme.

[0064] Specifically, the system acquires the corrected passage sequence output in step S17, the channel priority sequence output in step S15, and the preset command synchronization status. The preset command synchronization status is command execution status data maintained by the system in real time, including the current execution stage of each command. The status of each command is divided into three types: pending reception, verification in progress, and execution completed. This status is dynamically updated by the feedback information after the command is issued. The physical space passage order in the corrected passage sequence, the communication resource allocation order in the channel priority sequence, and the execution stage information in the command synchronization status are fused. The fusion method is to construct a multi-dimensional spatiotemporal state matrix. The rows of the matrix correspond to time sampling points, and the columns of the matrix correspond to the three-dimensional spatial coordinates of each UAV, the communication channel occupancy identifier, and the command execution stage identifier, respectively. For each time sampling point, the spatial position of each UAV at that time is extracted from the corrected passage sequence, the communication channel that each UAV should occupy at that time is extracted from the channel priority sequence, and the execution stage of each command at that time is extracted from the command synchronization status. This information is filled into the corresponding positions in the matrix to complete the construction of the spatiotemporal state matrix.

[0065] The constructed spatiotemporal state matrix is ​​smoothed using a cubic spline interpolation algorithm. The specific implementation of the cubic spline interpolation algorithm is as follows: For each column of the data sequence in the matrix, the time point of the sequence is used as the independent variable, and the sequence value as the dependent variable. A cubic polynomial is constructed between every two adjacent time points, such that the function value of the polynomial at each time point equals the original data value, and the first and second derivatives of the polynomial are continuous at each time point, thus forming a smooth curve. Cubic spline interpolation is performed on all columns to obtain the smooth curve corresponding to each column of data. The set of all smooth curves constitutes the smooth state trajectory.

[0066] An initial allocation scheme is generated based on the smoothed state trajectory. This initial allocation scheme includes the physical spatial coordinates of each UAV at each moment, its communication channel occupancy plan, and the command execution sequence. The spatial coordinates of each UAV at each time point within a preset time window are extracted from the smoothed state trajectory to form a flight path plan; the communication channel occupancy identifiers of each UAV at each time point are extracted to form a communication resource occupancy plan; and the command execution phases of each UAV at each time point are extracted to form a command execution plan. These three plans are then aligned by time and combined to form the initial allocation scheme.

[0067] The conflict check vector is extracted from the initial allocation scheme. The conflict check vector consists of conflict indices for all potentially conflicting UAV pairs in terms of time, space, and resources. Specifically, for each UAV pair, it checks whether they occupy the same spatial grid cell at the same time; if so, the spatial conflict index is recorded as 1, otherwise as 0. It also checks whether they occupy the same communication channel at the same time; if so, the resource conflict index is recorded as 1, otherwise as 0. Finally, it checks whether command execution times overlap and cause resource contention; if so, the temporal conflict index is recorded as 1, otherwise as 0. The spatial conflict index, resource conflict index, and temporal conflict index of all UAV pairs are arranged in order of UAV identification, forming a one-dimensional vector, which is the conflict check vector.

[0068] The conflict check vectors are projected onto a pre-defined conflict vector space. This space is pre-constructed by enumerating all possible conflict types, where each basis vector represents a conflict type, and the dimensions of the basis vectors are the same as those of the conflict check vectors. The projection process involves calculating the inner product of the conflict check vectors with each basis vector to obtain the projection coefficients on each basis vector. The square root of the sum of the squares of all projection coefficients yields the magnitude of the projection result. The pre-defined conflict vector space is constructed using historical conflict event data, collecting at least one hundred sets of historical conflict event records. For each record, a three-dimensional conflict check vector containing spatial conflict indicators, resource conflict indicators, and temporal conflict indicators is extracted to form a sample matrix. After centering the sample matrix, the covariance matrix is ​​calculated, and the eigenvalues ​​and eigenvectors of the covariance matrix are solved. The eigenvalues ​​are sorted from largest to smallest, and the top few eigenvectors with a cumulative variance contribution rate exceeding 95% are selected as basis vectors, forming the conflict vector space. When the magnitude of the projection result is zero, it indicates that all conflict points have been resolved; therefore, the initial allocation scheme is directly used as the final allocation scheme.

[0069] When the magnitude of the projection result is not zero, the timing correction is recalculated based on the projection result. Specifically, the projection coefficients on each basis vector in the projection result are analyzed, and the basis vector with the largest absolute value of the projection coefficient is identified. The conflict type corresponding to this basis vector is the most severe conflict type at present. Based on this conflict type, the process returns to the sub-step of calculating the timing correction in step S17, recalculates the timing correction for the UAV pair that caused the conflict, and updates the corrected passage sequence. The updated corrected passage sequence then re-enters step S18 to reconstruct the spatiotemporal state matrix, generate smooth state trajectories, form an initial allocation scheme, and perform projection verification. This iterative process is repeated until the magnitude of the projection result is reduced to zero; the resulting allocation scheme is the final allocation scheme. This iterative mechanism ensures that the final output allocation scheme is conflict-free in the three dimensions of time, space, and communication resources, achieving closed-loop convergence of UAV swarm cooperative control.

[0070] It should be noted that during the iterative adjustment process, the maximum number of iterations is preset to twenty. When the number of iterations reaches the maximum number of iterations and the projection result modulus has not dropped to zero, the initial allocation scheme with the smallest modulus is selected as the final allocation scheme, and conflict warning information is output. The projection modulus is recorded in each iteration. If the modulus does not decrease for three consecutive iterations, the iteration is terminated in advance, and the scheme with the smallest current modulus is adopted.

[0071] Reference Figure 2 The second embodiment of the present invention provides an intelligent collaborative control system for unmanned aerial vehicle (UAV) swarms, comprising: The data preprocessing module is used to acquire UAV status information and perform competitive sorting of UAVs to obtain a priority sequence of UAVs; The weight assignment module is used to calculate the risk probability based on the priority sequence of the UAVs, obtain the collision probability, and assign risk weights to the preset task area based on the collision probability to obtain a conflict risk distribution map. The power classification module is used to perform weighted correction calculations based on the conflict risk distribution map to obtain a correction priority score, and to classify drones based on the correction priority score and a preset power alarm threshold to obtain a low power passage sequence and a high power passage sequence. The cycle adjustment module is used to adjust the passage time and occupancy cycle according to the low power passage sequence and the high power passage sequence to obtain an optimized passage sequence; The conflict filtering module is used to perform channel conflict filtering based on the optimized passage sequence to obtain a channel priority sequence; The collaborative evaluation module is used to deduce the distribution pattern based on the channel priority sequence to obtain the group distribution pattern, and to evaluate the collaborative consistency based on the group distribution pattern to obtain the synchronization coordination index. The timing correction module is used to perform high-risk screening based on the synchronization coordination index to obtain a high-risk UAV sequence, and to perform timing correction based on the high-risk UAV sequence to obtain a corrected passage sequence. The output module is used to perform iterative adjustments to the scheme based on the corrected passage sequence and the channel priority sequence to obtain the final allocation scheme.

[0072] It should be noted that the intelligent collaborative control system for unmanned aerial vehicle (UAV) swarms provided in this embodiment of the invention is used to execute all the process steps of the intelligent collaborative control method for UAV swarms described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0073] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0074] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for intelligent collaborative control of unmanned aerial vehicle (UAV) swarms, characterized in that, include: Acquire drone status information and perform drone competition ranking to obtain a drone priority sequence; Based on the drone priority sequence, the risk probability is calculated to obtain the collision probability. Based on the collision probability, the risk weight is assigned to the preset task area to obtain the conflict risk distribution map. Based on the conflict risk distribution map, a weighted correction calculation is performed to obtain a correction priority score. Based on the correction priority score, the drones are classified according to a preset battery alarm threshold to obtain a low battery passage sequence and a high battery passage sequence. Based on the low-battery passage sequence and the high-battery passage sequence, the passage time and occupancy period are adjusted to obtain an optimized passage sequence; Based on the optimized passage sequence, channel conflict filtering is performed to obtain the channel priority sequence; Based on the channel priority sequence, the distribution pattern is deduced to obtain the group distribution pattern, and based on the group distribution pattern, the coordination consistency is evaluated to obtain the synchronization coordination index. Based on the aforementioned synchronization and coordination indicators, high-risk screening is performed to obtain high-risk drone sequences, and time-series correction is performed based on the high-risk drone sequences to obtain corrected passage sequences. Based on the modified passage sequence and the channel priority sequence, the scheme is iteratively adjusted to obtain the final allocation scheme.

2. The intelligent collaborative control method for unmanned aerial vehicle swarms according to claim 1, characterized in that, The process of acquiring UAV status information and performing UAV competition ranking to obtain a UAV priority sequence includes: Obtain the drone's mission urgency, remaining battery percentage, mission execution time limit, battery consumption rate, real-time location coordinates, and channel identifier; Divide the remaining battery percentage by the battery consumption rate to obtain the theoretical battery life, and subtract the task execution time limit from the theoretical battery life to obtain the redundancy time; The initial priority score is obtained by weighting and summing the reciprocal of the redundant time with the urgency of the task. Based on the real-time position coordinates, the Euclidean distance between each pair of drones is calculated to obtain the spatial distance; By filtering out drones whose spatial distance is less than a preset safe distance threshold and whose channel identifiers are the same, a drone competition sequence is obtained; Based on the initial priority score, the drone competition sequence is sorted in descending order to obtain the drone priority sequence.

3. The intelligent collaborative control method for unmanned aerial vehicle (UAV) swarms according to claim 2, characterized in that, The step of calculating the risk probability based on the drone priority sequence to obtain the collision probability, and assigning risk weights to a preset task area based on the collision probability to obtain a conflict risk distribution map, includes: Based on the preset obstacle distribution map, the preset task area is divided into three-dimensional grids according to the preset grid side lengths to obtain a discretized spatial grid. Based on the discretized spatial grid, the potential path trajectory is obtained by calculating the uniform motion trajectory of the UAV priority sequence using the uniform linear motion formula. Identify path segments in the potential path trajectories that pass through the same discretized spatial grid within the same time window to obtain the path overlap region; Extract the relative speed and approach angle of each potential path trajectory within the path overlap area, and calculate the collision probability using a pre-constructed risk mapping function; Based on the urgency of the task, the collision probabilities are weighted and summed to obtain risk weight coefficients, which are then assigned to the corresponding discretized spatial grid to obtain a conflict risk distribution map.

4. The intelligent collaborative control method for unmanned aerial vehicle swarms according to claim 3, characterized in that, The step involves performing a weighted correction calculation based on the conflict risk distribution map to obtain a correction priority score. Then, based on the correction priority score and a preset battery alarm threshold, drones are classified to obtain low-battery passage sequences and high-battery passage sequences, including: The UAVs whose risk weight coefficients in the conflict risk distribution map exceed a preset conflict risk threshold are selected to obtain a set of conflict UAVs; Based on a preset task relationship matrix, the urgency of the tasks corresponding to the conflicting drone sets is weighted and calculated to obtain a corrected priority score; Based on the correction priority score, the conflicting drone set is sorted in descending order to obtain a correction priority sequence; The correction priority sequence is classified according to a preset power alarm threshold to obtain a high power passage sequence and a low power passage sequence.

5. The intelligent collaborative control method for unmanned aerial vehicle swarms according to claim 3, characterized in that, The step of adjusting the passage time and occupancy period based on the low-power passage sequence and the high-power passage sequence to obtain an optimized passage sequence includes: Divide the remaining battery percentage in the low battery passage sequence by the battery consumption rate to obtain the remaining flight time, and subtract a preset safety time margin from the remaining flight time to obtain the available flight time; Subtract the available flight time from the task execution time limit to obtain the latest passage time. Based on the low battery passage sequence and the real-time position coordinates, calculate the arrival and departure times of each UAV in the overlapping path area using the uniform linear motion formula to obtain the estimated arrival and departure times. Calculate the difference between the estimated arrival time and the latest departure time to obtain the time offset, and add the time offset to the estimated arrival time and the estimated departure time respectively to obtain the low battery update sequence; When the low battery update sequence and the high battery passage sequence do not overlap in time, the low battery update sequence and the high battery passage sequence are merged and sorted according to the correction priority score to obtain an optimized passage sequence. When the low battery update sequence and the high battery passage sequence overlap in time, the flight speed of the UAV in the high battery passage sequence is increased until the low battery update sequence and the high battery passage sequence no longer overlap in time, thus obtaining an optimized passage sequence.

6. The intelligent cooperative control method for unmanned aerial vehicle swarms according to claim 2, characterized in that, The step of performing channel conflict filtering based on the optimized passage sequence to obtain a channel priority sequence includes: By filtering out drones with the same channel identifier in the optimized passage sequence, a channel conflict set is obtained; Based on the corrected priority score, the channel conflict set is sorted in descending order to obtain the channel priority sequence.

7. The intelligent collaborative control method for unmanned aerial vehicle swarms according to claim 4, characterized in that, The process involves performing distribution pattern deduction based on the channel priority sequence to obtain the group distribution pattern, and then conducting a coordination consistency assessment based on the group distribution pattern to obtain synchronization coordination indicators, including: The expected arrival coordinates are obtained by multiplying the flight speed of each UAV in the channel priority sequence by the preset command distribution delay and then adding the result to the real-time position coordinates. Movement commands are sent to the drones in the channel priority sequence to obtain a feedback drone sequence, and the real-time position coordinates of the feedback drone sequence are extracted to obtain the actual arrival coordinates; The difference between the actual arrival coordinates and the expected arrival coordinates is calculated to obtain the coordinate deviation vector; and based on the coordinate deviation vector, the distribution pattern of the feedback UAV sequence is deduced through a pre-constructed swarm evolution model to obtain the swarm distribution pattern. Calculate the modulus of the Euclidean distance and velocity difference between each UAV in the group distribution pattern, and perform a weighted summation based on the calculation results to obtain the collaborative consistency matrix; The variance of all off-diagonal elements in the coordination consistency matrix is ​​calculated to obtain the synchronization coordination index.

8. The intelligent cooperative control method for unmanned aerial vehicle swarms according to claim 7, characterized in that, The process involves high-risk screening based on the synchronization and coordination indicators to obtain high-risk drone sequences, followed by timing correction based on these high-risk drone sequences to obtain corrected passage sequences, including: When the synchronization and coordination index is not lower than the preset synchronization and coordination threshold, the current state is maintained for drone control. When the synchronization and coordination index is lower than the preset synchronization and coordination threshold, the flight trajectory is deduced by using the uniform linear motion formula based on the real-time position coordinates in the feedback UAV sequence to obtain the deduced risk distribution map. By filtering out the regions and corresponding drones whose risk weight coefficients in the simulated risk distribution map exceed the conflict risk threshold, high-risk regions and high-risk drone sequences are obtained. Calculate the time offset required for each drone in the high-risk drone sequence to avoid the high-risk area, and obtain the timing correction amount; The timing correction is superimposed on the planned passage time of each drone in the high-risk drone sequence to generate an initial instruction set; Based on the initial instruction set, the UAVs corresponding to each instruction in the initial instruction set are prioritized according to the pre-built instruction priority rules to obtain the corrected passage sequence.

9. The intelligent collaborative control method for unmanned aerial vehicle swarms according to claim 8, characterized in that, The step of iteratively adjusting the scheme based on the modified passage sequence and the channel priority sequence to obtain the final allocation scheme includes: A spatiotemporal state matrix is ​​constructed based on the corrected passage sequence, the channel priority sequence, and the preset instruction synchronization state; Based on the spatiotemporal state matrix, a smoothed state trajectory is obtained by using a cubic spline interpolation algorithm. Based on the smooth state trajectory, an initial allocation scheme is generated, and the conflict check vector of the initial allocation scheme is projected onto a preset conflict vector space to obtain the projection result; When the modulus of the projection result is zero, the initial allocation scheme is directly used as the final allocation scheme. When the modulus of the projection result is not zero, the timing correction amount is recalculated based on the projection result until the modulus of the projection result is zero, thus obtaining the final allocation scheme.

10. An intelligent collaborative control system for unmanned aerial vehicle (UAV) swarms, characterized in that, include: The data preprocessing module is used to acquire UAV status information and perform competitive sorting of UAVs to obtain a priority sequence of UAVs; The weight assignment module is used to calculate the risk probability based on the priority sequence of the UAVs, obtain the collision probability, and assign risk weights to the preset task area based on the collision probability to obtain a conflict risk distribution map. The power classification module is used to perform weighted correction calculations based on the conflict risk distribution map to obtain a correction priority score, and to classify drones based on the correction priority score and a preset power alarm threshold to obtain a low power passage sequence and a high power passage sequence. The cycle adjustment module is used to adjust the passage time and occupancy cycle according to the low power passage sequence and the high power passage sequence to obtain an optimized passage sequence; The conflict filtering module is used to perform channel conflict filtering based on the optimized passage sequence to obtain a channel priority sequence; The collaborative evaluation module is used to deduce the distribution pattern based on the channel priority sequence to obtain the group distribution pattern, and to evaluate the collaborative consistency based on the group distribution pattern to obtain the synchronization coordination index. The timing correction module is used to perform high-risk screening based on the synchronization coordination index to obtain a high-risk UAV sequence, and to perform timing correction based on the high-risk UAV sequence to obtain a corrected passage sequence. The output module is used to perform iterative adjustments to the scheme based on the corrected passage sequence and the channel priority sequence to obtain the final allocation scheme.

Images