Multi-robot cooperative control method based on data link networking

Abstract

The present application relates to the technical field of unmanned aerial vehicle cooperative control, and discloses a method for unmanned aerial vehicle multi-aircraft cooperative control based on data link networking. The method constructs a fleet cooperative task data set containing flight path features, load state features and communication link features; extracts dynamic cooperative nodes therefrom, including fleet formation mode switching nodes and task target switching nodes; generates a flight path cooperative trigger signal based on the dynamic cooperative nodes, identifies local topology adjustment nodes triggered by fleet formation mode switching and global topology reconfiguration nodes triggered by task target switching; generates cooperative offset parameters containing flight path offset and communication delay offset according to the node classification; extracts fleet cooperative protocol parameters based on the cooperative offset parameters; and controls the unmanned aerial vehicle cluster to perform a cooperative flight task according to the protocol parameters. The method can improve the cooperative efficiency and task adaptability of the unmanned aerial vehicle fleet in a dynamic environment, and is suitable for various multi-aircraft cooperative operation scenarios.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle cooperative control, in particular to a multi-unmanned aerial vehicle cooperative control method based on data link networking. BACKGROUND

[0002] With the rapid development of unmanned aerial vehicle technology, multi-unmanned aerial vehicle cooperative task execution has become an important direction to improve operational efficiency and task adaptability. In military reconnaissance, disaster monitoring, logistics distribution and other scenarios, a single unmanned aerial vehicle is often limited by payload capacity, endurance time and operating range, and is difficult to meet complex task requirements, so a multi-unmanned aerial vehicle cooperative system is needed to form a complementary fleet system.

[0003] In current multi-unmanned aerial vehicle cooperative control methods, there is a common problem of insufficient adaptability to dynamic task environments. For example, when the formation of the fleet needs to be changed to avoid obstacles or adjust the task area, the traditional control method often relies on preset formation conversion rules and is difficult to respond to sudden conditions in complex environments in real time, resulting in flight path conflicts or decreased cooperative efficiency during fleet reconstruction.

[0004] In the task target switching scenario, such as from area search to point monitoring, the communication topology structure within the fleet needs to be adjusted to meet the information exchange needs of the new task. However, in existing methods, the adjustment of the communication topology is mostly in a global reconstruction mode, which triggers the reconfiguration of the entire network regardless of the size of the task change, which not only increases the consumption of communication resources, but also may affect the real-time performance of cooperative control due to data transmission delay during reconstruction.

[0005] Existing cooperative control protocols are usually designed for specific task scenarios and lack generality and reusability. When the fleet needs to perform multiple types of complex tasks, the switching process of the control protocol is complex and is prone to cooperative failure problems caused by protocol incompatibility. In terms of cooperative parameter calculation, existing methods do not adequately consider the correlation between flight path deviation and communication delay, and often set parameters separately, making it difficult to optimize flight path planning and communication quality in dynamic environments, affecting the overall operational efficiency of the fleet. SUMMARY

[0006] The purpose of the present application is to provide a multi-unmanned aerial vehicle cooperative control method based on data link networking to solve the problems raised in the background.

[0007] To achieve the above purpose, the present application provides a multi-unmanned aerial vehicle cooperative control method based on data link networking, which comprises:

[0008] Constructing a fleet cooperative task data set, the fleet cooperative task data set containing flight path characteristics, payload state characteristics and communication link characteristics;

[0009] extracting dynamic coordination nodes in the machine group coordination task data set, the dynamic coordination nodes including machine group formation mode switching nodes and task target switching nodes;

[0010] generating a flight path coordination trigger signal based on the dynamic coordination nodes, and identifying machine group communication topology change nodes according to the flight path coordination trigger signal, the machine group communication topology change nodes including local topology adjustment nodes triggered by formation mode switching and global topology reconstruction nodes triggered by task target switching;

[0011] generating coordination offset parameters according to the classification of the machine group communication topology change nodes, the coordination offset parameters including flight path offset amounts in the formation mode switching process and communication delay offset amounts in the task target switching process;

[0012] extracting machine group coordination protocol parameters based on the coordination offset parameters, the machine group coordination protocol parameters including a general flight protocol reusable across tasks and a load control protocol specific to a specific task, the specific task being a work scenario within a preset range that requires a special load configuration or a differentiated coordination strategy;

[0013] controlling the unmanned aerial vehicle cluster to perform a coordinated flight task according to the machine group coordination protocol parameters.

[0014] Preferably, the machine group coordination task data set includes:

[0015] extracting an initial formation parameter set and a corresponding flight time period based on flight path characteristics;

[0016] extracting a formation reconstruction parameter set and a corresponding switching time period based on the formation mode switching nodes;

[0017] processing the initial formation parameter set and the flight time period through a formation verification algorithm to generate a corrected formation reconstruction parameter set;

[0018] outputting the machine group coordination task data set by fusing the initial formation parameter set and the corrected formation reconstruction parameter set, wherein the formation verification algorithm introduces a flight environment disturbance factor when calculating a flight time deviation amount.

[0019] Preferably, the generation of the flight path coordination trigger signal includes:

[0020] extracting the formation reconstruction parameter set to generate a formation trigger signal;

[0021] extracting the corrected formation reconstruction parameter set to generate a task switching signal;

[0022] identifying a flight path coordination trend based on the formation trigger signal and the task switching signal;

[0023] dynamically adjusting a communication link feature sampling frequency according to the flight path coordination trend to generate the flight path coordination trigger signal.

[0024] Preferably, the identification of the fleet communication topology change node comprises:

[0025] Calculate the track similarity of the track cooperative trigger signal and the historical topology adjustment node;

[0026] When the track similarity exceeds the preset similarity threshold, mark it as a local topology adjustment node;

[0027] When the track similarity is lower than the preset similarity threshold, mark it as a global topology reconstruction node;

[0028] Wherein the track similarity is calculated by the track space distribution coincidence rate and the time synchronization deviation amount.

[0029] Preferably, the generation of the cooperative offset parameter comprises:

[0030] Extract the local topology adjustment node triggered by the formation mode switching, and mark it as a formation offset node;

[0031] Extract the global topology reconstruction node triggered by the task target switching, and mark it as a task offset node;

[0032] Calculate the track position offset based on the formation offset node, and mark it as a track cooperative offset parameter;

[0033] Calculate the communication delay increment based on the task offset node, and mark it as a communication cooperative offset parameter;

[0034] Fuse the track cooperative offset parameter and the communication cooperative offset parameter to generate the cooperative offset parameter.

[0035] Preferably, the extraction of the fleet cooperative protocol parameter comprises:

[0036] Analyze the cooperative offset parameter based on the flight path feature, and extract the track control protocol reusable across tasks as a general flight protocol;

[0037] Analyze the correlation between the load state feature and the cooperative offset parameter, and extract the sensor scheduling protocol dedicated to specific tasks as a load control protocol;

[0038] Wherein the general flight protocol is optimized by a track historical data reuse verification algorithm.

[0039] Preferably, the control of the unmanned aerial vehicle cluster to perform cooperative flight tasks comprises: taking the general flight protocol as a global track control benchmark, dynamically adjusting the fleet sensor working mode based on the load control protocol; and verifying the consistency of the protocol through the communication link feature, and triggering the protocol overload mechanism when detecting that the communication delay offset exceeds the limit.

[0040] Preferably, the method further comprises: establishing a formation coordination quality evaluation model, inputting the coordination offset parameter and the formation coordination protocol parameter; outputting the coordination quality evaluation value to a flight path coordination trigger signal generation module; and dynamically correcting the update frequency of the formation reconstruction parameter set based on the coordination quality evaluation value.

[0041] Preferably, the formation coordination quality evaluation model comprises: constructing a weighted evaluation function of the flight path position offset and the communication delay offset, introducing a flight environment disturbance factor as an evaluation function correction term, and training the evaluation function weight coefficient through historical task execution data.

[0042] Preferably, the identification of the formation communication topology change node further comprises:

[0043] When the communication link interruption or the data link quality is lower than the preset threshold, triggering a topology degradation mechanism;

[0044] Based on the flight path similarity and the real-time communication quality evaluation result, dynamically selecting the execution priority of the local topology adjustment node or the global topology reconstruction node;

[0045] The topology degradation mechanism maintains the basic formation communication capability through a preset minimum connected subgraph algorithm.

[0046] Compared with the prior art, the present application has the following beneficial effects:

[0047] By constructing a formation coordination task data set containing flight path, load state and communication link characteristics, comprehensive basic data support is provided for coordination control. The integration of multi-dimensional data enables the formation to comprehensively consider various influencing factors during task execution, avoiding decision bias caused by the lack of single data dimension.

[0048] In the dynamic coordination node extraction link, node identification is performed for two key scenarios of formation shape switching and task target switching, so that the coordination control can accurately focus on the core change point in the task process. This accurate identification mechanism helps to reduce the occupation of system resources by irrelevant data processing and improves the pertinence and efficiency of control response.

[0049] Based on the dynamic coordination node, a flight path coordination trigger signal is generated, and based on this, a communication topology change node is identified, further distinguishing the applicable scenarios of local topology adjustment and global topology reconstruction, and realizing the differentiated adjustment of the communication topology. In the local task change such as formation shape switching, local topology adjustment is adopted to reduce the fluctuation of the communication network and maintain the stable connection of most nodes; while in the global task change such as task target switching, global topology reconstruction is performed to ensure the overall optimization of the communication link under the new task. This flexible adjustment mode balances the communication efficiency and stability.

[0050] The cooperative offset parameter is generated according to the communication topology change node classification, the track offset amount and the communication delay offset amount are respectively calculated corresponding to different task scenes, and the cooperative optimization of track planning and communication quality is realized. In the formation mode switching process, the accurate calculation of the track offset amount can avoid the collision risk between unmanned aerial vehicles, and reduce the influence of track adjustment on the stability of the communication link; in the task target switching process, the consideration of the communication delay offset amount can reduce the interference of data transmission delay on the cooperative execution of the new task, so that the track planning and the communication state remain dynamically adaptive.

[0051] The cooperative offset parameter is generated according to the communication topology change node classification, the track offset amount and the communication delay offset amount are respectively calculated corresponding to different task scenes, and the cooperative optimization of track planning and communication quality is realized. In the formation mode switching process, the accurate calculation of the track offset amount can avoid the collision risk between unmanned aerial vehicles, and reduce the influence of track adjustment on the stability of the communication link; in the task target switching process, the consideration of the communication delay offset amount can reduce the interference of data transmission delay on the cooperative execution of the new task, so that the track planning and the communication state remain dynamically adaptive. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 The working principle diagram of the unmanned aerial vehicle multi-machine cooperative control method based on data link networking is described.

[0053] Figure 2 The flowchart for generating the track cooperative trigger signal is described.

[0054] Figure 3 The flowchart for identifying the communication topology change node of the swarm is described.

[0055] Figure 4 The flowchart for generating the cooperative offset parameter is described.

[0056] Figure 5 The flowchart for extracting the cooperative protocol parameter of the swarm is described.

[0057] Figure 6 The formation parameter and trigger signal analysis diagram is described.

[0058] Figure 7 The communication quality and topology adjustment analysis diagram is described.

[0059] Figure 8 The formation offset and cooperative parameter analysis is described. DETAILED DESCRIPTION

[0060] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments of the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work are within the protection scope of the present application.

[0061] Please refer to Figure 1 The present application provides a multi-vehicle cooperative control method for unmanned aerial vehicles based on data link networking, which comprises the following steps:

[0062] The flight path features record the latitude and longitude coordinates, height, speed vector and heading angle of each unmanned aerial vehicle. The load state features include sensor operating mode, energy consumption rate and remaining capacity of task load. The communication link features include data link signal strength, transmission delay and packet loss rate. The extraction of dynamic cooperative nodes adopts a sliding time window analysis method. The heading angle mutation point is identified as the formation mode switching node in the flight path features, and the operating mode jump point is detected as the task target switching node in the load state features. The generation of the flight path cooperative trigger signal is achieved by comparing the Euclidean distance change rate of the dynamic cooperative nodes in adjacent time windows. When the change rate exceeds the threshold, the signal generation is triggered. The identification of the swarm communication topology change node adopts a graph theory analysis method. The unmanned aerial vehicle communication connection graph model is established. According to the flight path cooperative trigger signal, the node degree distribution change rate in the graph is calculated. The nodes with a change rate lower than 20% are marked as local topology adjustment nodes, and the nodes with a change rate higher than 50% are marked as global topology reconstruction nodes. The calculation of the cooperative offset parameter adopts a differential positioning method. The flight path offset in the formation mode switching process is calculated by the position difference between the current flight path point and the preset flight path point. The communication delay offset in the task target switching process is determined by the sliding average value of the data link transmission delay. The extraction of the swarm cooperative protocol parameter adopts a protocol template matching algorithm. The general flight protocol matches the flight path control parameters with a similarity higher than 85% in the historical task database. The load control protocol calls the preset sensor scheduling scheme from the protocol library according to the current task type. The cooperative control of the unmanned aerial vehicle swarm is achieved by a distributed decision system. Each unmanned aerial vehicle node adjusts the flight state autonomously according to the received cooperative protocol parameters, and simultaneously exchanges the state information in real time through the data link network to maintain the formation consistency.

[0063] Embodiment 1: Please refer to Figure 2, the specific operation process of flight path feature processing, formation parameter generation and trigger signal generation mechanism. The airspace processing adopts three-dimensional grid division technology, and the task airspace is discretized into cubic unit grids with an edge length of 50 meters. Each grid unit has a unique spatial coordinate identifier. The crossing event of the unmanned aerial vehicle in the grid is recorded by the airborne positioning system, and the continuous coordinate points with millisecond-level time stamps constitute the core content of the initial formation parameter set. The flight time period automatic division mechanism analyzes the time interval characteristics of the continuous coordinate points. When the time interval between adjacent coordinate points exceeds two seconds, the system automatically generates a new flight time period boundary identifier, which includes the starting coordinate and heading vector data of the time period.

[0064] The formation parameter generation adopts a dynamic reference frame construction method. In the initial formation parameter set, the system analyzes the communication signal strength parameters of each unmanned aerial vehicle in real time, selects the node with the highest signal strength and a duration of more than five seconds as the virtual leader. Other unmanned aerial vehicle nodes establish a local coordinate system based on the virtual leader, forming a position offset matrix containing relative azimuth angle, pitch angle and Euclidean distance. The matrix data is updated at a frequency of 10 Hz in the data link, and the matrix dimension is consistent with the cluster size. Each element records the six-degree-of-freedom parameters of the relative pose of the two machines.

[0065] The formation parameter verification adopts a disturbance compensation model. The verification process first calculates the real-time standard deviation of all elements in the position offset matrix to establish a formation stability index. Flight environment disturbance factors are obtained through airborne meteorological sensors, including three-dimensional wind speed vector and atmospheric density parameters. The system establishes a three-layer height disturbance model: zero to one hundred meters height layer is given a zero point three weight coefficient, one hundred to three hundred meters layer is given a zero point five coefficient, and three hundred meters above layer is given a zero point seven coefficient. The Monte Carlo simulation module imports the current formation parameter set into the kinematic model with disturbance factors and performs five hundred random simulations. The simulation output results are compared with the original position offset matrix to generate a modified parameter set containing wind speed compensation values. These modified parameters include heading angle compensation, speed adjustment and height correction, which are stored separately according to the unmanned aerial vehicle nodes.

[0066] The trigger signal generation adopts matrix feature analysis technology. The formation trigger signal is derived from the real-time singular value decomposition result of the position offset matrix. The system continuously monitors the fluctuation of the maximum singular value of the matrix. When it is detected that the singular value increases by more than one point five times of the preset reference value in three consecutive sampling periods, a trigger signal containing the current formation disintegration risk level is automatically generated. The signal is broadcast to the entire cluster through the data link, and the signal payload includes the virtual leader identifier, the matrix instability degree coefficient and the recommended adjustment strategy code.

[0067] Task switching signal is generated by load mode state machine. Sensor operation mode is modeled as a twelve-state finite automaton, each state contains three attributes: operating frequency, power consumption level and data output rate. Hidden Markov Model is configured with sixteen hidden state nodes, and mode transition probability is calculated by Viterbi algorithm. Model input is a three-dimensional vector of load state features: energy consumption rate, number of sensor activations, data buffer remaining. When state transition probability is detected below one in ten thousand in consecutive ten samples, system generates task switching event code. The code contains original task identification, recommended switching mode code and sensor configuration scenario index.

[0068] Communication sampling adjustment adopts closed-loop feedback mechanism. Base sampling frequency is set to ten hertz period. Sampling frequency adjustment procedure is started when both track coordination trend factor and load state change factor are detected by system. Adjustment algorithm establishes linear mapping relationship from zero to thirty hertz with heading angle change rate as input variable. Specific execution process is: absolute value of current heading angle and previous sampled angle difference is calculated in real time, and frequency increment is converted by proportional coefficient. The increment is limited by three-second sliding window to prevent frequency oscillation caused by instantaneous mutation. Frequency switching process adopts smooth transition strategy, gradually adjusting to target frequency with five hertz gradient per second.

[0069] Coordinated task data integration adopts multi-source fusion strategy. Initial formation parameter set and modified formation parameter set are aligned by time axis in data storage layer, each flight hour period corresponds to independent parameter file. File structure contains sixteen data sections: three-dimensional grid coordinate mapping table is stored in header; original parameters and modified parameters are stored in main part by unmanned aerial vehicle node grouping; environmental disturbance model configuration summary and verification algorithm version identification are recorded in tail. Data set output performs integrity check before execution, check algorithm calculates file checksum and content hash value to ensure parameter consistency during data transmission process. The whole data processing flow is executed in real-time task priority in embedded system, and the calculation period is strictly controlled within fifty-millisecond time window.

[0070] Referring to Figure 6 , the heading angle change of the unmanned aerial vehicle formation, the working mode state, and the generation of the coordination trigger signal are shown. The upper chart shows the heading angle change curves of multiple unmanned aerial vehicles, and the vertical dashed lines indicate the key nodes of formation shape switching. These nodes usually appear at positions where the heading angle changes significantly, reflecting the adjustment of the overall movement direction of the formation. The middle chart shows the transition process of the unmanned aerial vehicle working mode, including initialization, standby, scanning, and locking four states. The dotted line marks the task target switching nodes, which correspond to the sudden changes of the working mode. The lower chart presents the track coordination trigger signal generated based on the dynamic coordination nodes. The square marks represent local topology adjustment nodes, and the triangle marks represent global topology reconstruction nodes, which are triggered when the communication quality decreases or the task demand changes.

[0071] Embodiment 2: refer to Figure 3 The dynamic time warping algorithm in track matching adopts a sliding window comparison technique, with a window length of twenty sampling points, each containing three-dimensional coordinates and corresponding timestamp information. The historical topology adjustment event database maintains track pattern records in the last fifty missions, with each record storing the complete track point sequence and topology change type identifier. The similarity calculation process first resamples the current track signal to align with the time resolution of the historical record, then calculates the cross-correlation coefficient of the waveform profile. The determination of the preset similarity threshold uses a dynamic statistical method, which periodically updates the distribution characteristics of the historical similarity data, and sets the threshold at the third quartile position, which is adaptively adjusted with the accumulation of mission execution experience.

[0072] The quantitative analysis of spatial distribution overlap rate is based on a three-dimensional voxel space occupancy model. The task airspace is divided into cubic units with a side length of one meter, and each unit is set with a binary occupancy flag. The voxel occupancy of the current track and the historical track is stored in bitmap form, and the system performs a bitwise AND operation to count the number of overlapping units. The time synchronization deviation measurement system deploys a precise clock synchronization network, with the master node equipped with an atomic clock source and the slave node using software phase-locked loop technology to track the master clock. The clock offset record contains a transmission delay compensation value, which is calculated by measuring the round-trip time of data packets. The synchronization accuracy is maintained at the microsecond level, and abnormal clock deviation events trigger an immediate correction protocol.

[0073] The communication link interruption detection uses a multi-level decision mechanism. The physical layer continuously monitors the carrier signal strength indicator, and when the received power is below the sensitivity threshold for one hundred milliseconds, the link state diagnosis program is started. The data link layer statistics the response timeout rate of the last ten transmission windows, and the application layer verifies the continuity of the heartbeat packet. The activation of the topology degradation mechanism requires meeting double conditions: physical layer interruption lasting more than five hundred milliseconds and application layer heartbeat loss count reaching three times. The minimum connected subgraph algorithm uses an improved Prim implementation, introducing a communication quality weight factor in the adjacency matrix construction stage, preferentially selecting links with high signal strength to join the spanning tree. The algorithm dynamically excludes unstable nodes during execution, maintaining the minimum support structure of the basic communication network.

[0074] The real-time communication quality evaluation system constructs a composite index system. The signal strength index is normalized using a logarithmic scale, mapping the received power to a linear interval of zero to one hundred. The transmission delay index calculates the percentile value within a sliding window, eliminating the effects of transient jitter. The packet loss rate statistics are based on sequence number continuity analysis, identifying duplicate and out-of-order packet phenomena. The three indices are fused to generate an overall evaluation value through weighted integration. The weight distribution is dynamically adjusted according to the task phase: the cruising phase focuses on signal strength, the task execution phase emphasizes the delay index, and the data return phase focuses on the packet loss rate parameter. The evaluation results are output in a five-level quantization form, corresponding to different topology adjustment strategies.

[0075] The execution priority decision system is implemented using a fuzzy logic controller. The input variables include track similarity and communication quality evaluation values, which are divided into five fuzzy sets. The boundaries of the similarity variable fuzzy sets use trapezoidal membership functions, and the communication quality variable uses triangular membership functions. The rule base contains twenty-five fuzzy inference rules, each defining the mapping relationship between input and output variables. The defuzzification process uses the center of gravity method to calculate the exact output value, quantizing the urgency level to a continuous interval of zero to one hundred. The output interface maps this value to a five-level response code, driving the corresponding topology adjustment program.

[0076] The processing flow of the local topology adjustment node includes two stages: neighbor discovery and link reconstruction. When marked as a local adjustment node, the system first broadcasts a neighbor probe request to collect node state information within a two-hop range. The link selection algorithm evaluates the stability index of candidate paths, which is calculated by combining historical connection duration and signal fluctuation variance. The new link needs to meet the minimum bandwidth requirement and verify bidirectional connectivity through a three-way handshake protocol. The adjustment process maintains the original data flow routing table, ensuring that the ongoing task data is not affected.

[0077] The global topology reconstruction process uses a phased execution strategy. The first stage collects node state information throughout the network, constructing a complete network topology snapshot. The second stage analyzes the task requirements and resource distribution, calculating the optimal communication connection scheme. The third stage implements gradual reconstruction, adjusting the connection relationship in order of node importance. During the reconstruction process, rollback checkpoints are set to automatically restore to the previous stable state when critical performance indicators deteriorate. Each reconstruction operation records detailed execution logs, which are used to optimize subsequent decision parameters.

[0078] The abnormality handling mechanism contains predefined recovery plans and adaptive adjustment strategies. For transient communication link interruption, the system initiates a fast reconnection protocol to rebuild the physical layer connection within 50 ms. Persistent interruption triggers a routing detour mechanism to maintain network connectivity through alternative relay nodes. When the detected data link quality continuously falls below the working threshold, the system automatically switches to a reduced communication mode, shutting down non-critical data service functions. Resource allocation in topology degradation state adopts strict priority scheduling to ensure transmission bandwidth for flight control commands and critical state information.

[0079] The cooperative analysis of track similarity and communication quality employs time alignment matching techniques. The system establishes a unified time reference axis to accurately align the sampling points of both indicators according to the timestamp. The joint analysis module detects the time sequence correlation of indicator changes to identify the causal relationship between communication quality degradation and track deviation. The analysis results are used to modify the topology adjustment strategy, prioritizing track correction over network reconstruction when it is confirmed that track abnormalities cause communication problems. The system maintains a dynamically adjusted experience database that stores the handling records of various abnormal scenarios to optimize subsequent decision-making processes.

[0080] The execution monitoring of the response mechanism employs a distributed observation architecture. Each network node independently assesses the local execution effect and exchanges monitoring information through the data link. The central coordination node integrates the local observation results to determine whether the global adjustment progress meets expectations. When it detects that the execution deviation exceeds the tolerance value, it issues supplementary adjustment instructions. The monitoring data is updated in real-time to the topology management database, forming a closed-loop feedback control system. The entire processing flow is designed in a fault-tolerant execution mode, and the failure of a single node does not affect the overall adjustment process. The system automatically reallocates uncompleted operation tasks.

[0081] Referring to Figure 7 , the communication quality evaluation and topology adjustment process of the UAV cluster are shown. The upper chart shows the communication signal strength variation curves of each UAV, reflecting the stability of the communication link. Signal strength fluctuations are influenced by flight environment, distance changes, and interference factors. The middle chart shows the variation of communication delay, which usually increases during formation shape switching or when environmental interference increases. Delay optimization is crucial for maintaining real-time cooperative control. The lower chart comprehensively evaluates the communication quality index and identifies the topology adjustment nodes. Local topology adjustment nodes trigger small-range communication optimization, and global topology reconstruction nodes (triangles) trigger communication structure adjustment of the entire formation.

[0082] Example 3: Referring to Figure 4The specific operation process of formation offset node identification, cooperative parameter calculation and fusion processing is shown. The detection of formation offset node adopts kinematic characteristic analysis method, and the key event node is identified by monitoring the change of the centroid motion state of the UAV group in real time. The system establishes a centroid position calculation model in the global coordinate system, collects three-dimensional coordinate data of all UAVs in the cluster every 200 milliseconds, and calculates the arithmetic mean value as the instantaneous centroid position. The generation of velocity vector adopts five-point differential method, and the position data of the current time and the previous two sampling points are used to calculate the instantaneous velocity. When the direction angle change of the velocity vector in the adjacent calculation period is more than 15 degrees, the system marks the time as a potential formation offset node candidate event. The candidate event needs to be verified for three consecutive sampling periods, and after confirming that the velocity change trend is persistent and not a transient disturbance, the effective formation offset node is finally confirmed, and the complete motion state parameters of the node are recorded.

[0083] The relative coordinate system conversion process establishes a local reference frame based on the formation centroid. The calculation of the conversion matrix considers the spatial distribution characteristics of the current formation shape, and determines the direction of the local coordinate axis through principal component analysis. The X-axis points to the direction of the formation motion, the Y-axis is perpendicular to the X-axis and parallel to the horizontal plane, and the Z-axis completes the right-handed coordinate system. The conversion from the global coordinate to the local coordinate adopts the following transformation relationship:

[0084]

[0085] wherein, represents the global coordinate of the ith UAV, is the formation centroid coordinate, is the rotation matrix considering the yaw angle , the pitch angle and the roll angle . is the local coordinate after conversion. The displacement difference calculation is carried out in the local coordinate system, which eliminates the influence of the overall motion of the formation and highlights the relative position change characteristics of the individuals.

[0086] The communication delay measurement adopts an adaptive heartbeat detection mechanism. The length of the probe data packet is dynamically adjusted according to the current network load, and is set to 64 bytes by default, and is compressed to 32 bytes when the channel is busy. The round-trip time delay is measured by recording the time difference between the sending request and the receiving response, and deducting the local processing delay compensation value. The calculation of the sliding window average value adopts the exponential weighting method, with the weight of the new measurement value being 0.3 and the weight of the historical value being 0.7, balancing the real-time performance and stability. The delay increment is obtained by comparing the relative change rate of the current average value and the reference value, and the reference value is determined by 30 seconds of continuous measurement at the beginning of the communication link establishment.

[0087] The normalization of the cooperative deviation parameter adopts the piecewise linear mapping method. The normalization reference of the track position deviation is the current maximum allowed distance of the formation, which is dynamically adjusted according to the task phase: 50 meters for the gathering phase, 30 meters for the cruising phase, and 20 meters for the task execution phase. The normalization reference of the communication delay deviation is the maximum allowed delay, which considers the transmission distance and hop number, and is calculated as where n is the hop number and d is the transmission distance (km). The normalized result is subjected to amplitude limiting to ensure that the output value is within the interval [0, 1].

[0088] The dynamic adjustment mechanism of the weighting coefficient is linked with the task type identification system. The task type classifier analyzes the current load state characteristics and working mode sequence, and outputs the task type probability distribution. In the reconnaissance task mode, the track parameter weight is initially set to 0.6, and when a continuous regional coverage behavior is detected, it is increased by 0.02 every 5 minutes, with a maximum of 0.75. In the strike task mode, the communication parameter weight is initially set to 0.7, and is increased to 0.8 in the weapon system activation state. The weight adjustment process is set to a 5-second transition period to avoid control oscillation caused by parameter mutation.

[0089] The identification of the task deviation node is based on the sensor state transition characteristics. The system defines six basic working states: initialization, standby, scanning, locking, data transmission, and maintenance. The state transition condition matrix records the allowed transition paths between states, and the irregular transition sequence refers to the transition combination not defined in the matrix. When a state jump or reverse transition is detected, the transition sequence analysis process is started. The confirmation of an effective task deviation node requires two conditions to be met: the state transition sequence contains at least two irregular transitions, and the transition interval time is less than 50% of the normal value. The node record contains the transition path diagram and time characteristic parameters.

[0090] The parameter fusion process adopts a hierarchical weighting architecture. The first layer processes the track position deviation, vectorially combines the displacement differences of each unmanned aerial vehicle, and calculates the formation overall deformation degree index. The second layer processes the communication delay deviation, distinguishes between inter-aircraft communication and ground station communication channels, and calculates the delay influence coefficient for each type. The third layer performs weighted fusion, selects the corresponding weight template according to the task type, and generates the integrated cooperative deviation parameter. The fusion result is subjected to Kalman filter smoothing to eliminate the fluctuations caused by measurement noise.

[0091] The association analysis of the formation deviation and the task deviation adopts event synchronization detection technology. The system maintains a time-aligned event log that records the occurrence time and duration of both types of deviation nodes. When two types of events occur successively within a 5-second time window, the association analysis program is started. The analysis process compares the similarity of the event characteristic parameters to establish a possible causal relationship model. The deviation parameters of strongly associated events are subjected to joint optimization processing, the weight distribution strategy is adjusted, and the overall consistency of cooperative control is improved.

[0092] The abnormal offset handling mechanism includes a three-level response strategy. The first-level response compensates for slight offsets through local trajectory fine-tuning; the second-level response handles moderate offsets by adjusting the formation shape and communication scheduling strategy; the third-level response deals with serious offsets by starting a task reorganization process. The determination of the response level is based on the duration and gradient of change of the offset parameters, using a fuzzy logic decision method. The offset trend is monitored in real time during the process, and the response level is dynamically adjusted, forming a closed-loop control structure.

[0093] The historical offset pattern learning system continuously updates the typical offset scenario library. After each task is completed, the system extracts the feature parameters of the offset event, and through clustering analysis, it is classified into existing scenarios or new scenarios. Scenario records include trigger conditions, handling strategies and effect evaluation data, which are used to optimize real-time decision-making in subsequent tasks. The learning process uses incremental updating to maintain the continuity and adaptability of historical experience.

[0094] Referring to Figure 8 , the motion characteristics of the UAV formation and the calculation process of the cooperative offset parameters are shown. The upper chart shows the motion trajectory of the formation center of mass, and the X mark indicates the detected formation offset nodes. These nodes usually appear at positions where the trajectory direction changes significantly, reflecting the adjustment of the overall shape of the formation. The middle chart shows the variation curve of the average speed of the formation, and the dashed line marks the key nodes where the speed direction changes abruptly. These nodes need special attention and may trigger the recalculation of cooperative control parameters. The lower chart compares the variation trends of the position offset and the communication delay offset of the two cooperative parameters. The position offset reflects the formation shape retention ability, and the communication delay offset reflects the information synchronization efficiency, both of which determine the quality of cooperative control.

[0095] Example 4: Referring to Figure 5 , the specific operation process of flight path control protocol extraction, sensor scheduling implementation and protocol execution monitoring. In a typical multi-vehicle cooperative reconnaissance task scenario, the system analyzes the flight path pattern characteristics in the historical task library to achieve optimized extraction and dynamic adjustment of the control protocol. The historical task library uses a hierarchical storage structure, with the top-level index table recording task type, environmental conditions and execution time, etc. The bottom layer data block stores detailed flight path control parameter sequences. When a new cooperative flight task starts, the feature matching engine first extracts the flight path feature descriptor of the current task, including the flight path point distribution density, turn radius statistical value and height variation pattern, etc. The matching process uses the nearest neighbor search algorithm to retrieve the ten most similar flight path pattern records from the historical library, and the selection of these records is based on the Euclidean distance calculation in the multi-dimensional feature space.

[0096] The generation process of sensor scheduling protocol is tightly coupled with specific task requirements. Take the area search task as an example, the system predefines five basic scanning modes in the payload control rules: spiral expansion scan, grid line scan, concentric circle scan, random Brownian motion, and key area rescan. Each scanning mode corresponds to a combination of sensor parameter configurations, including optical zoom, infrared sensitivity, scanning speed, and data sampling rate. The rule engine monitors the feature changes of the target area in real time during task execution, and automatically switches from spiral expansion mode to key area rescan mode when suspicious target signs are detected. The mode switching instruction contains a transition parameter sequence to ensure data continuity during the sensor state transition process.

[0097] The verification mechanism of protocol execution consistency operates in a distributed network environment. The verification node selection strategy is based on the communication topology, and usually selects three drones in a triangular distribution as the verification group. In each verification period, the master node broadcasts control instructions containing timestamps, and the verification nodes immediately reply with confirmation signals after receiving them. The voting mechanism counts the number and content of valid responses, and confirms the protocol execution to be valid when two or more verification nodes report the same execution state. The abnormal situation handling adopts a gradual response strategy, which triggers instruction retransmission when inconsistencies are detected for the first time, and starts the protocol reload process after three consecutive abnormalities.

[0098] The following table shows the key parameter configurations of the sensor scheduling protocol in a typical reconnaissance task.

[0099] Table 1: Sensor scheduling protocol parameter configuration for area search task

[0100] Scan mode Optical zoom factor Infrared sensitivity level Scan speed (m / s) Sampling rate (Hz) Applicable area type Spiral expanding scan 4x 3 15 10 Unknown open area Grid line scan 6x 4 10 15 Structured terrain Concentric circle scan 5x 5 12 12 Point target area Random Brownian motion 3x 2 8 8 Complex interference environment Key area re-scan 8x 6 5 20 Suspected target identification area

[0101] The optimization process of the general flight protocol is achieved through iterative learning. Each individual in the genetic algorithm population represents a set of flight path control parameters, including speed curve, height profile, and turn rate variables. The fitness evaluation function considers three dimensions: flight path tracking accuracy, energy consumption, and task completion time. The selection operation uses a tournament strategy to randomly select five individuals from the population and retain the best one. The crossover operation uniformly mixes in the variable space, and the mutation operation introduces a Gaussian disturbance term. The optimization process runs continuously, performing incremental learning during task execution gaps, gradually improving the adaptability of the protocol parameters.

[0102] The dynamic adjustment mechanism of the payload control protocol responds to environmental change signals. When the optical sensor detects a change in lighting conditions exceeding a threshold, the exposure parameters and filter settings are automatically adjusted. The working mode of the infrared sensor is dynamically adjusted according to the environmental temperature gradient, with the sensitivity level increased in high temperature difference areas. The adjustment of scanning speed is synchronized with the flight speed of the UAV, maintaining the stability of the ground sampling interval. The data sampling rate is flexibly set according to the remaining capacity of the storage system and the communication bandwidth status, and the lossy compression algorithm is started when resources are tight.

[0103] The following table shows the type judgment of the general flight protocol and the payload control protocol.

[0104] Table 2: General flight protocol and payload control protocol distinction criteria and judgment basis

[0105] Differentiation dimension General flight protocol Payload control protocol Task relevance Irrelevant to task type (e.g. all tasks need to maintain formation) Strongly relevant to task type (e.g. high sampling rate for reconnaissance, weapon lock parameters for strike) Multiplexing frequency Cross-task multiplexing rate ≥ 85% (only need to fine-tune environmental parameters) Cross-task multiplexing rate ≤ 30% (need to completely reconstruct core parameters) Adjustment trigger condition Only affected by flight environment disturbances (e.g. wind, temperature) Directly triggered by task target changes (e.g. from "area search" to "point monitoring")

[0106] Cross-task reuse mechanism and dedicated adaptation process:

[0107] Reuse process of general flight protocol: Extract "general protocol template library" (including 10 basic formation modes and 5 obstacle avoidance algorithms) from historical task database; input "environmental disturbance factors" (such as wind speed, air pressure) of current task, and adjust parameters through formula Fine tuning (weight range 0.1-0.3);

[0108] Verify the feasibility of reuse: If the adjusted parameters meet the requirements of track tracking error ≤0.5 meters and energy consumption rate ≤120% of the baseline value, directly reuse; otherwise, start local parameter optimization (such as adjusting the turning radius threshold through genetic algorithm).

[0109] Dedicated adaptation process of payload control protocol: According to the task type (such as "power inspection" and "agricultural plant protection"), call "dedicated protocol framework" from protocol library; input "communication delay offset" in cooperative offset parameters, dynamically adjust payload parameters (such as closing non-critical sensors to reduce data volume when delay >200ms); verify the adaptability through "task-payload matching degree algorithm" (formula: ) When the matching degree ≥80%, lock the protocol, otherwise trigger manual intervention process.

[0110] The cooperative working logic of the two protocols: the general flight protocol is the "base layer", which provides the global trajectory control reference (such as the formation center of mass trajectory, the minimum safe distance), and all tasks share the same set of core logic; the payload control protocol is the "application layer", which dynamically adjusts the payload working mode (such as the general protocol maintains formation flight, and the payload protocol controls a UAV to turn on the thermal imaging sensor) based on the general protocol according to the task requirements; the cooperative triggering mechanism is that when the payload control protocol needs to adjust the UAV position to optimize the sensor field of view, it sends a "trajectory fine-tuning request" (offset ≤ 2 meters) to the general protocol, and the general protocol realizes local position correction through PID algorithm to avoid overall formation reconstruction.

[0111] The specific process of protocol parameter extraction is as follows:

[0112] General flight protocol extraction: input cooperative offset parameter ΔP (current trajectory offset) and environmental disturbance factor W (such as wind speed weight coefficient 0.3-0.7), retrieve trajectory control protocol samples from historical task database that satisfy |ΔP_i - ΔP|×W ≤ 0.5 meters. Use dynamic programming algorithm to calculate the spatial distribution coincidence rate of sample trajectory and current trajectory (threshold ≥ 85%), select the optimal sample, and generate the general flight protocol through the formula corrected heading angle = sample heading angle + ΔP×0.02×W correction amount of heading angle compensation.

[0113] Payload control protocol extraction: input communication delay offset ΔT (sliding window average), when ΔT < 100ms, call "standard sensor scheduling template" in protocol library (such as sampling rate 20Hz, no compression); when ΔT ≥ 100ms, start delay adaptation rule: sampling frequency = 20Hz - (ΔT-100ms) / 20ms (minimum not less than 5Hz); enable LZW compression algorithm, compression rate = 1 + ΔT / 200ms (maximum not more than 3:1); at the same time, through the balance coefficient α×ΔT + β×sensor accuracy (α=0.6, β=0.4) to select the optimal sensor combination, generate the payload control protocol.

[0114] The trigger condition of protocol overload mechanism is set with multiple levels of judgment standards. The communication delay offset is monitored by sliding window statistics method, and the window size is ten sampling periods. When more than half of the sample values in the window exceed the threshold, it is marked as primary abnormality. The data link packet loss rate is calculated based on sequence number continuity analysis, and the rising trend in three consecutive statistical periods confirms the secondary abnormality. After the double abnormality condition is met, the system starts the protocol overload preparation program, including state snapshot saving and resource pre-release operation. The overload process adopts atomic operation design to ensure that the overall switching of control parameters does not occur in the intermediate state.

[0115] The implementation of the hierarchical control strategy relies on the state monitoring network. The normal level corresponds to the normal task execution state, and the sensor works in the balance mode. The enhanced level is activated when the key target is detected, which improves the sensor accuracy and sampling density. The emergency level is enabled when the system resources are tight or disturbed, and unnecessary functions are closed to ensure core tasks. The level switching decision is based on multi-dimensional feature analysis, including energy reserves, computing load and communication quality, etc. Each level defines clear functional boundaries and rollback conditions, forming a finite state automaton.

[0116] The coordination mechanism of the distributed decision system is realized through the information exchange protocol. Each UAV node periodically broadcasts state summary information, including current position, load status and protocol version, etc. After receiving the summary information, the neighbor node updates the local topology graph and detects the consistency state of the cluster. When local deviation in protocol execution is detected, the adjustment strategy is determined through negotiation mechanism, and the correction scheme with the smallest influence range is preferred. Global inconsistency triggers the re-synchronization process, and the reference protocol parameters are distributed by the leader node.

[0117] The historical data reuse verification algorithm constructs the parameter evolution trajectory. After each task is completed, the system extracts the actual executed flight path control parameters and performs difference analysis with the initial recommended parameters in the historical library. The significantly improved parameter combination is marked as the preferred case, and obtains higher weight in subsequent matching. The verification process considers the particularity of the task scene to avoid cross-scene parameter generalization errors. The propagation of the preferred case adopts a limited diffusion strategy, and only shares between UAV nodes of similar task types.

[0118] Embodiment 5: The specific implementation process of the construction of the coordination quality evaluation model and the dynamic correction mechanism. The input feature processing of the coordination quality evaluation model adopts multi-source data fusion technology, and the eight core feature parameters enter the neural network model through independent preprocessing channels. The mean and variance of the flight path position offset are calculated using the sliding window statistical method, and the window length is dynamically adjusted according to the task stage, set to 30 seconds in the cruise stage, and shortened to 15 seconds in the task execution stage. The statistical processing of communication delay offset distinguishes between uplink and downlink, and calculates the delay distribution characteristics of each link. The current value of the flight environment disturbance factor is collected in real time by the on-board meteorological sensor array, and the predicted value is generated based on the time series analysis model, considering the pressure gradient and temperature change trend. The success rate statistics of protocol execution consistency use event counting method, record the voting results in each verification period. The overload frequency monitoring records the number of protocol resets per unit time, eliminating the statistical deviation caused by transient fluctuations.

[0119] The structure design of the neural network model adopts a deep cross-feature fusion architecture. A feature cross layer is set after the input layer to perform nonlinear combination of the track position offset and the communication delay offset. The hidden layer adopts a stepwise width design, with 128 neurons in the first layer and a gradual decrease to 16 neurons in the output layer. The activation function selection considers the feature distribution characteristics, with the ReLU function used in the hidden layer to process non-negative features, and the Sigmoid function used in the output layer to constrain the evaluation value within a standard range. The model training process adopts the mini-batch gradient descent method, with each batch containing 32 samples, and the learning rate is initially set to 0.001, which is adjusted according to the cosine annealing strategy after each training round. The loss function is designed as a weighted binary cross-entropy, which applies asymmetric penalties to overestimation and underestimation of the evaluation value.

[0120] The construction of the weighted evaluation function adopts a hierarchical linear regression method. The first layer processes the track position offset, establishing a linear mapping relationship between the position error and the evaluation value. The second layer superimposes the influence of the communication delay offset, capturing the synergistic effect of the two types of offsets through interaction terms. The flight environment disturbance factor is used as a multiplication correction term to apply an environmental attenuation coefficient to the basic evaluation value. The selection process of historical task execution data uses time decay weighting, with recent task data obtaining a higher weight. The weight coefficient training uses elastic network regression to balance the influence of L1 and L2 regularization terms, preventing overfitting. Feature standardization processing uses moving window statistics to continuously update mean and variance estimates.

[0121] The adjustment of the reconfiguration parameter set update frequency adopts a closed-loop control strategy. The input variable of the PID controller is the short-term change rate of the cooperative quality evaluation value, and the output variable is the adjustment amount of the parameter update interval. The proportional term quickly responds to evaluation value fluctuations, the integral term eliminates steady-state error, and the differential term suppresses overshoot. The control parameters are preset according to the task type, with a slower update rhythm for reconnaissance tasks and a agile update mode for strike tasks. The frequency adjustment process sets safety boundaries, with a minimum of 1 Hz and a maximum of 10 Hz, to prevent system oscillation in extreme cases.

[0122] The online learning mechanism of the evaluation function weight coefficients adopts an incremental update method. The training data is input in real time, and the model performs parameter fine-tuning every five minutes. Feature importance analysis uses the permutation test method, randomly shuffling single-dimensional feature values to observe the evaluation value change amplitude. The weight update step of important features is increased accordingly, and secondary features use a conservative adjustment strategy. The weight coefficients are implemented with soft constraints, ensuring that each coefficient is within a reasonable range through the projection gradient method. Historical weight versions are retained in a rolling window to support quick rollback to a stable state.

[0123] The quantification of flight environment disturbance factors adopts multi-dimensional comprehensive evaluation. The three-dimensional wind speed data is decomposed into axial component strength by vector decomposition, and the atmospheric density parameter is converted into air dynamic influence coefficient. The disturbance level is divided into ten discrete scales, and each level corresponds to a control compensation strategy. The correlation model between environmental factors and evaluation value adopts piecewise linear approximation, and local linear relationship is established in different disturbance intervals. Real-time disturbance monitoring sets an outlier filter to eliminate false judgments caused by sensor transient noise.

[0124] The output interface design of collaborative quality evaluation value considers the downstream system requirements. After the original evaluation value is filtered by moving average, it is converted into a percentage score output. The score change trend is calculated by first-order difference, and three state labels of rising, stable and falling are added. The evaluation result distribution adopts the publish-subscribe mode, and the track planning module and the communication scheduling module register the evaluation dimensions of interest respectively. The key evaluation event triggers the alarm notification mechanism, which is broadcast to related nodes through data link.

[0125] The storage and retrieval of historical task data adopt a spatio-temporal index structure. Task records are indexed by execution date and geographical area, supporting multi-dimensional conditional queries. The data compression algorithm retains key feature trajectories and discards redundant detail information. The retrieval process uses approximate nearest neighbor search to speed up the matching speed of similar task patterns. Data updates maintain version control, preserving previous correction records for traceability analysis.

[0126] The adaptive adjustment mechanism of the model sets multiple protection strategies. When it is detected that the output value of five consecutive evaluation periods exceeds the reasonable range, the model diagnosis program is started. The diagnosis process checks whether the input feature distribution is shifted and verifies whether the model parameters are abnormal. In the case of serious abnormality, switch to the standby simplified model to ensure that the basic evaluation function does not interrupt. The model recovery process uses a gradual warm start to gradually reintroduce complex features. The entire evaluation system is designed as a fault-tolerant architecture, and a single module failure will not cause the loss of overall functionality.

[0127] It should be noted that, in this text, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0128] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A method for multi-robot cooperative control based on data link networking, characterized in that, The method comprises the following steps: constructing a swarm cooperative task data set, which contains flight path features, load state features and communication link features; extracting dynamic cooperative nodes in the swarm cooperative task data set, which contains swarm formation mode switching nodes and task target switching nodes; generating a flight path cooperative trigger signal based on the dynamic cooperative nodes, and identifying swarm communication topology change nodes according to the flight path cooperative trigger signal, which includes local topology adjustment nodes triggered by formation mode switching and global topology reconstruction nodes triggered by task target switching; generating cooperative offset parameters according to the classification of the swarm communication topology change nodes, which contains flight path offset in the formation mode switching process and communication delay offset in the task target switching process; extracting swarm cooperative protocol parameters based on the cooperative offset parameters, which contains a general flight protocol reusable across tasks and a load control protocol specific to a specific task, which is a preset range of operation scenarios that require special load configuration or differentiated cooperative strategies; controlling the UAV cluster to perform cooperative flight tasks according to the swarm cooperative protocol parameters.

2. The method of claim 1, wherein, The method of constructing a swarm cooperative task data set comprises: extracting an initial formation parameter set and a corresponding flight time period based on flight path features; extracting a formation reconstruction parameter set and a corresponding switching time period based on formation mode switching nodes; processing the initial formation parameter set and the flight time period through a formation verification algorithm to generate a corrected formation reconstruction parameter set; outputting the swarm cooperative task data set by fusing the initial formation parameter set and the corrected formation reconstruction parameter set, wherein the formation verification algorithm introduces a flight environment disturbance factor when calculating the flight time deviation.

3. The method of claim 2, wherein, The method of generating a flight path cooperative trigger signal comprises: extracting the formation reconstruction parameter set to generate a formation trigger signal; extracting the corrected formation reconstruction parameter set to generate a task switching signal; identifying a flight path cooperative trend based on the formation trigger signal and the task switching signal; dynamically adjusting the communication link feature sampling frequency according to the flight path cooperative trend to generate the flight path cooperative trigger signal.

4. The method of claim 3, wherein, The method of identifying swarm communication topology change nodes comprises: calculating the flight path similarity of the flight path cooperative trigger signal and the historical topology adjustment nodes; when the flight path similarity exceeds a preset similarity threshold, marking it as a local topology adjustment node; when the flight path similarity is lower than the preset similarity threshold, marking it as a global topology reconstruction node; wherein the flight path similarity is calculated by combining the flight path space distribution overlap rate and the time synchronization deviation.

5. The method of claim 1, wherein, The method of generating cooperative offset parameters comprises: extracting local topology adjustment nodes triggered by formation mode switching, marking them as formation offset nodes; extracting global topology reconstruction nodes triggered by task target switching, marking them as task offset nodes; calculating the flight path position offset based on the formation offset nodes, marking it as a flight path cooperative offset parameter; calculating the communication delay increment based on the task offset nodes, marking it as a communication cooperative offset parameter; fusing the flight path cooperative offset parameter and the communication cooperative offset parameter to generate the cooperative offset parameter.

6. The method of claim 5, wherein, The method of extracting swarm cooperative protocol parameters comprises: The flight path characteristics are taken as a reference to analyze the cooperative deviation parameters, and a cross-task reusable flight path control protocol is extracted as a general flight protocol; The correlation between the load state characteristics and the cooperative deviation parameters is analyzed, and a specific task dedicated sensor scheduling protocol is extracted as a load control protocol; The general flight protocol is optimized by a flight history data reuse verification algorithm.

7. The method of claim 6, wherein, The control unmanned aerial vehicle cluster to perform a cooperative flight task includes: taking the general flight protocol as a global flight path control reference, dynamically adjusting the sensor working mode of the cluster based on the load control protocol; and verifying the consistency of the protocol through the communication link characteristics, and triggering the protocol overload mechanism when detecting that the communication delay deviation exceeds the limit.

8. The method of claim 7, wherein, It also includes: establishing a cluster cooperation quality evaluation model, inputting the cooperative deviation parameters and cluster cooperation protocol parameters; outputting the cooperative quality evaluation value to the flight path cooperation trigger signal generation module; and dynamically correcting the update frequency of the formation reconstruction parameter set based on the cooperative quality evaluation value.

9. The method of claim 8, wherein, The cluster cooperation quality evaluation model includes: constructing a weighted evaluation function of the flight path position deviation and the communication delay deviation, introducing a flight environment disturbance factor as an evaluation function correction term, and training the evaluation function weight coefficient through historical task execution data.

10. The method of claim 4, wherein, The recognition cluster communication topology change node also includes: When detecting that the communication link is interrupted or the data link quality is lower than the preset threshold, a topology degradation mechanism is triggered; Based on the flight path similarity and the real-time communication quality evaluation result, the execution priority of the local topology adjustment node or the global topology reconstruction node is dynamically selected; The topology degradation mechanism maintains the basic formation communication capability through the preset minimum connected subgraph algorithm.

Images