Cooperative control method and system for realizing self-organizing protection of multi-unmanned aerial vehicle cluster

Abstract

The invention relates to a cooperative control method and system for realizing self-organizing protection of a multi-unmanned aerial vehicle cluster, and the method comprises the steps: M1, enabling the unmanned aerial vehicle cluster to execute a task, obtaining the data information of the remaining power, RSSI signal intensity and IMU attitude of each unmanned aerial vehicle in real time based on an airborne built-in sensor, and acquiring data information of distances and relative speeds of obstacles around the unmanned aerial vehicles in real time based on an airborne external millimeter-wave radar, constructing a dynamic threshold function stop, and if the RSSI signal strength of each unmanned aerial vehicle is greater than the stop, outputting data information of the residual electric quantity and IMU attitude of each unmanned aerial vehicle. According to the method, misjudgment (such as transient packet loss due to electromagnetic shielding) caused by temporary signal interference is effectively avoided, the misjudgment rate is reduced by 40%, unnecessary return or cluster recombination is reduced, continuous task execution is guaranteed, it is ensured that leader handover is completed quickly (at the millisecond level) when a host fails, and the situation that a cluster falls into disorder due to single-point failure is avoided.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a collaborative control method and system for realizing self-organized protection of multiple UAV swarms. Background Technology

[0002] The development of multi-rotor drone swarm technology stands in stark contrast to its shortcomings in protection. While this technology has evolved from its initial single-drone operation mode to its current collaborative intelligence stage, it has consistently lagged behind in terms of protection mechanisms. Early multi-rotor drones were mainly used in basic scenarios such as aerial photography and power line inspection, achieving limited autonomy through pre-programmed paths. With the introduction of swarm intelligence and distributed control theory, swarm technology has begun to transform towards collaboration and self-organization, demonstrating the advantages of multi-drone cooperation in scenarios such as disaster relief and environmental monitoring.

[0003] However, this technological evolution has exposed serious deficiencies in protection capabilities: traditional protection mechanisms have long relied on centralized architectures and preset rules, making them unable to dynamically adapt to sudden situations under high-pressure environments (such as communication interruptions or airflow disturbances). This results in drones being able to only perform rigid return-to-home maneuvers when signals are weak, leading to mission interruptions and data loss. More seriously, swarm systems lack effective failover and self-healing capabilities. When the host or critical node fails, the entire swarm may become paralyzed. This severe disconnect between technological development and protection capabilities greatly restricts the reliable application of multi-rotor drone swarms in complex environments.

[0004] Although existing protection technologies optimize network connectivity through self-organizing network modules, they still have inherent defects. They rely on a "preset action" mechanism, and when the network signal is below the threshold, the drone can only return to the take-off airport, resulting in the incomplete transmission of inspection data and repeated operations. At the same time, although its multi-objective optimization function considers factors such as power and distance, it does not realize dynamic election and number adjustment, making it difficult to cope with scenarios where the master or slave drones suddenly fail.

[0005] In the prior art, Chinese patent application (application number: 202411171223.4, publication number: CN118672286A) discloses an emergency control method and system for UAVs with multi-airport collaborative inspection. For any given UAV, a set of self-organizing network modules capable of network connectivity is detected in real time. If the network signal strength of the current network connection is lower than a set threshold, all self-organizing network module sets are traversed, and the self-organizing network module with the maximum network signal strength is selected for emergency self-organizing network connection, where the maximum network signal strength is greater than or equal to the set threshold. However, this emergency control scheme still primarily relies on "single-UAV-airport" or "single-UAV-control center" interactions, with self-organizing network switching and landing decisions depending on the scoring calculation of the central node (such as the airport or control center). When the host or critical communication node in the cluster fails, the system cannot dynamically trigger distributed election or role reconfiguration (such as the slave autonomously taking over after the host fails), potentially causing the entire cluster to become inoperable due to a single point of failure, making it difficult to achieve true self-organizing protection.

[0006] In the prior art, Chinese patent application (application number: 202511196352.3, publication number: CN120704404A) discloses a method, device, and medium for implementing a multi-UAV formation flight system. The ground station and multiple UAVs adopt a network mode, and the UAVs are grouped through a task management thread. Based on the grouping, tasks can be distributed to the UAVs in real time, and feedback on the UAV execution results is obtained in real time. Based on the feedback information, the task management thread is triggered to distribute tasks to the next group of UAVs in real time. However, if the ground station fails due to communication interruption, hardware failure, or external interference, the entire cluster will lose its command center and will be unable to autonomously trigger emergency mechanisms (such as dynamic role election or formation reorganization), leading to cluster paralysis. For example, the document mentions that the task management thread is responsible for screening faulty UAVs and generating faulty sub-tasks, but if the ground station itself loses connection, the cluster cannot achieve self-healing through local interaction, exposing the risk of a single point of failure. Summary of the Invention

[0007] In view of the shortcomings of the prior art, the present invention provides a collaborative control method and system for realizing self-organized protection of multiple UAV swarms. It not only effectively avoids misjudgment caused by temporary signal interference (such as brief packet loss due to electromagnetic blockage), reducing the misjudgment rate by 40%, reducing unnecessary return to base or swarm reorganization, and ensuring continuous mission execution, but also ensures rapid (millisecond-level) handover of leadership when the host fails, avoiding chaos in the swarm due to single point of failure.

[0008] To achieve the above and other related objectives, the present invention provides the following technical solution: A collaborative control method for achieving self-organized protection of multiple unmanned aerial vehicle (UAV) swarms, the method comprising: M1. The drone swarm performs tasks by acquiring real-time data on the remaining battery power, RSSI signal strength, and IMU attitude of each drone based on the onboard built-in sensors, and acquiring real-time data on the distance and relative speed of obstacles around the drone based on the onboard external millimeter-wave radar. M2. Based on the RSSI signal strength data of each UAV, a dynamic threshold function s is constructed. top If the RSSI signal strength of each drone is greater than s top Then it outputs the remaining battery power and IMU attitude data for each drone; M3. Based on the remaining battery power and IMU attitude data of each drone, combined with the distance and relative speed data of obstacles around the drone, and the historical mission completion rate data of each drone, an improved weighted election algorithm is used to select the host of the drone, and the selected host of the drone is obtained. M4. When the main unit of the drone changes, the spatial region is divided using a Voronoi diagram, the path is planned using a particle swarm optimization algorithm, the numbering and position of the remaining slave drones are adjusted, the formation shape is maintained, and obstacles are avoided.

[0009] Furthermore, in step M2, the construction of the dynamic threshold function s top for, , Where, x i (t) represents the RSSI signal strength data of the i-th UAV at time t, where n is the total number of UAVs, T is the sampling period, and α, β, and δ are weighting coefficients.

[0010] Furthermore, the weighting coefficient α is any constant parameter between 0 and 1, the weighting coefficient β is any constant parameter between 0 and 1, the weighting coefficient δ is any constant parameter between 0 and 1, and the sum of the squares of the weighting coefficients α, β and δ is 1.

[0011] Furthermore, in step M3, the selection of the UAV host using the improved weighted election algorithm includes: M31. Based on the data information of the remaining battery power, IMU attitude, distance and relative speed of surrounding obstacles and historical task completion rate of each UAV, normalization processing is performed to obtain the normalized data information of the remaining battery power, IMU attitude, distance and relative speed of surrounding obstacles and historical task completion rate of each UAV. M32. Based on the normalized data of each UAV's remaining battery power, IMU attitude, distance and relative speed to surrounding obstacles, and historical mission completion rate, construct the scoring function F for the j-th UAV.j , , Among them, a j (t) represents the normalized remaining battery power of the j-th drone at time t. j (t) represents the IMU attitude data of the j-th UAV at time t after normalization, c j (t) represents the normalized distance data of the surrounding obstacles of the j-th UAV at time t, v j (t) represents the normalized data on the relative velocities of the surrounding obstacles of the j-th UAV at time t, and d j (t) represents the normalized historical task completion rate data of the j-th drone at time t, where ω, σ, ρ and η are weighting factors. The score value of the j-th drone is calculated to obtain the score value data of the j-th drone. M33. Based on the score data of the j-th UAV, construct a decision function H=max{F1,F2,...F...} j ,...,F n}, where n is the total number of drones, and the drone with the highest score is selected as the host drone.

[0012] Furthermore, the weight factor ω takes the value of any constant parameter between 0 and 1, the weight factor σ takes the value of any constant parameter between 0 and 1, the weight factor ρ takes the value of any constant parameter between 0 and 1, and the weight factor η takes the value of any constant parameter between 0 and 1.

[0013] Furthermore, the method also includes: M5. State Synchronization and Recovery: After a new host is elected, the latest cluster status is broadcast globally. When a slave rejoins after a fault is repaired, an idle number is assigned and the current task is synchronized.

[0014] Furthermore, the method also includes: M6. Abnormal Vibration Detection and Alarm: Combining the IMU data of the UAV body, abnormal vibrations are identified, structural health alarms are triggered and uploaded to the ground station.

[0015] Furthermore, the method also includes: M7. Enhanced communication protocol: All transmitted data is encrypted with AES-256 and authenticated with HMAC.

[0016] To achieve the above and other related objectives, the present invention also provides a collaborative control system for realizing self-organized protection of multiple UAV swarms, used to implement the aforementioned collaborative control method for realizing self-organized protection of multiple UAV swarms, the system comprising: The multi-drone swarm data acquisition module is used to acquire data information such as the remaining battery power, RSSI signal strength and IMU attitude of each drone in real time based on the onboard built-in sensors, and to acquire data information on the distance and relative speed of obstacles around the drone in real time based on the onboard external millimeter-wave radar. The multi-UAV swarm adaptive heartbeat monitoring module is connected to the multi-UAV swarm data acquisition module and is used to construct a dynamic threshold function s. top If the RSSI signal strength of each drone is greater than s top Then it outputs the remaining battery power and IMU attitude data for each drone; A multi-drone cluster weighted election module, connected to the multi-drone cluster adaptive heartbeat monitoring module, is used to select the host of the drones using an improved weighted election algorithm to obtain the selected host of the drones. The multi-UAV swarm intelligent formation reconstruction module is connected to the multi-UAV swarm weighted election module. When the host UAV changes, it uses a Voronoi diagram to divide the spatial region, a particle swarm algorithm to plan the path, adjusts the number and position of the remaining slave UAVs, maintains the formation shape, and avoids obstacles.

[0017] Furthermore, the system also includes a multi-UAV swarm status synchronization and recovery module, a multi-UAV swarm abnormal vibration detection and alarm module, and a multi-UAV swarm enhanced communication protocol module. The multi-UAV swarm status synchronization and recovery module is connected to the multi-UAV swarm intelligent formation reconstruction module. It is used to globally broadcast the latest swarm status after a new host is selected, and to assign idle numbers and synchronize current tasks when a slave rejoins after fault repair. The multi-UAV swarm abnormal vibration detection and alarm module is connected to the multi-UAV swarm intelligent formation reconstruction module. It is used to identify abnormal vibrations by combining the UAV's fuselage IMU data, trigger structural health alarms, and upload them to the ground station. The multi-UAV swarm enhanced communication protocol module is connected to the ground station and is used to perform AES-256 encryption and HMAC authentication on all transmitted data.

[0018] The present invention has the following positive effects: 1. This invention dynamically adjusts the timeout threshold (baseline 3 times, shortened to 2 times when the signal is unstable, and extended to 7 times when it is stable) through an adaptive heartbeat monitoring module, and combines ACK confirmation and a maximum of 3 retransmission mechanism to effectively avoid misjudgment caused by temporary signal interference (such as temporary packet loss due to electromagnetic shielding), reduce the misjudgment rate by 40%, reduce unnecessary return trips or cluster reorganization, and ensure continuous task execution.

[0019] 2. This invention employs an improved weighted election algorithm. When slaves compete for a new master, they comprehensively evaluate power consumption (30%), signal strength (25%), location advantage (20%), and historical reliability (25%) to generate a score. High-value nodes are prioritized, and a 10-50ms random backoff is introduced to avoid multi-machine election conflicts. This ensures that the handover of leadership is completed quickly (millisecond level) when the master fails, and avoids the cluster falling into chaos due to single point of failure.

[0020] 3. This invention optimizes position adjustments based on Voronoi diagrams or particle swarm optimization algorithms when nodes fail or hosts change through an intelligent formation reconstruction module. It integrates geometric formation constraints (such as maintaining a diamond formation), millimeter-wave radar obstacle avoidance data, and a high-battery-priority movement strategy to reduce invalid movements (path shortening by 15%), ensuring that the formation can quickly restore its complete form and reducing task interruption time by 50%.

[0021] 4. This invention introduces a new status synchronization mechanism. After a new host is elected, it broadcasts the global node number, location, and task status. After a faulty node is repaired, it is re-added with an idle number after identity verification. This avoids number conflicts or information gaps, ensures that the cluster always maintains a unified understanding, and improves the efficiency of collaborative operations.

[0022] 5. This invention supports dynamic switching between Wi-Fi and self-organizing network dual links by enhancing the communication protocol, and uses AES-256 encryption and HMAC authentication to ensure the anti-interference and anti-tampering capabilities of heartbeat packets, command and status data transmission, making it suitable for scenarios with strong electromagnetic interference or malicious attacks.

[0023] 6. This invention can be deployed on mainstream multi-rotor UAV platforms without disruptive modifications to the flight control architecture. It only requires embedding algorithms such as adaptive heartbeat and weighted election at the software layer, and supports mixed formations of heterogeneous aircraft such as reconnaissance and transport types. It has good engineering promotion value. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a flowchart illustrating the improved weighted election algorithm of the present invention; Figure 3 This is a schematic diagram of the system framework of the present invention. Detailed Implementation

[0025] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0026] Example 1: As Figure 1 As shown, a collaborative control method for achieving self-organized protection of multiple unmanned aerial vehicle (UAV) swarms is disclosed, the method comprising: M1. The drone swarm performs tasks by acquiring real-time data on the remaining battery power, RSSI signal strength, and IMU attitude of each drone based on the onboard built-in sensors, and acquiring real-time data on the distance and relative speed of obstacles around the drone based on the onboard external millimeter-wave radar. M2. Based on the RSSI signal strength data of each UAV, a dynamic threshold function s is constructed. top If the RSSI signal strength of each drone is greater than s top Then it outputs the remaining battery power and IMU attitude data for each drone; M3. Based on the remaining battery power and IMU attitude data of each drone, combined with the distance and relative speed data of obstacles around the drone, and the historical mission completion rate data of each drone, an improved weighted election algorithm is used to select the host of the drone, and the selected host of the drone is obtained. M4. When the main unit of the drone changes, the spatial region is divided using a Voronoi diagram, the path is planned using a particle swarm optimization algorithm, the numbering and position of the remaining slave drones are adjusted, the formation shape is maintained, and obstacles are avoided.

[0027] In this embodiment, in step M2, the dynamic threshold function s is constructed. top for, , Where, x i (t) represents the RSSI signal strength data of the i-th UAV at time t, where n is the total number of UAVs, T is the sampling period, and α, β, and δ are weighting coefficients.

[0028] In this embodiment, the weighting coefficient α is any constant parameter between 0 and 1, the weighting coefficient β is any constant parameter between 0 and 1, the weighting coefficient δ is any constant parameter between 0 and 1, and the sum of the squares of the weighting coefficients α, β and δ is 1.

[0029] In this embodiment, as Figure 2 As shown, in step M3, the selection of the UAV host using the improved weighted election algorithm includes: M31. Based on the data information of the remaining battery power, IMU attitude, distance and relative speed of surrounding obstacles and historical task completion rate of each UAV, normalization processing is performed to obtain the normalized data information of the remaining battery power, IMU attitude, distance and relative speed of surrounding obstacles and historical task completion rate of each UAV. M32. Based on the normalized data of each UAV's remaining battery power, IMU attitude, distance and relative speed to surrounding obstacles, and historical mission completion rate, construct the scoring function F for the j-th UAV. j , , Among them, a j (t) represents the normalized remaining battery power of the j-th drone at time t. j (t) represents the IMU attitude data of the j-th UAV at time t after normalization, c j (t) represents the normalized distance data of the surrounding obstacles of the j-th UAV at time t, v j (t) represents the normalized data on the relative velocities of the surrounding obstacles of the j-th UAV at time t, and d j (t) represents the normalized historical task completion rate data of the j-th drone at time t, where ω, σ, ρ and η are weighting factors. The score value of the j-th drone is calculated to obtain the score value data of the j-th drone. M33. Based on the score data of the j-th UAV, construct a decision function H=max{F1,F2,...F...} j ,...,F n}, where n is the total number of drones, and the drone with the highest score is selected as the host drone.

[0030] In this embodiment, the weight factor ω is any constant parameter between 0 and 1, the weight factor σ is any constant parameter between 0 and 1, the weight factor ρ is any constant parameter between 0 and 1, and the weight factor η is any constant parameter between 0 and 1.

[0031] In this embodiment, the method further includes: M5. State Synchronization and Recovery: After a new host is elected, the latest cluster status is broadcast globally. When a slave rejoins after a fault is repaired, an idle number is assigned and the current task is synchronized.

[0032] In this embodiment, the method further includes: M6. Abnormal Vibration Detection and Alarm: Combining the IMU data of the UAV body, abnormal vibrations are identified, structural health alarms are triggered and uploaded to the ground station.

[0033] In this embodiment, the method further includes: M7. Enhanced communication protocol: All transmitted data is encrypted with AES-256 and authenticated with HMAC.

[0034] In this embodiment, a cluster of 12 DJI M300 RTK drones is used to perform power line inspection in mountainous areas as an example: Initial deployment. The ground station issues the mission area (a polygon defined by latitude and longitude), safe altitude (120m), and inspection resolution (2cm / pixel). Host A (ID=001) generates 12 flight paths based on the flight trajectory analysis unit and assigns them to each aircraft.

[0035] Dynamic monitoring. At the 17th minute of operation, host A entered the tunnel, and the RSSI plummeted to -92dBm, below the current level. top =-78.3dBm, automatically masked. The remaining 11 machines initiate election: Drone 005: a j =82%, b j =1.8°, d j =96.7%, c j =14.2m, v j =25m / s→F5=0.741; Drone 008: a j =65%, b j =1.6°, d j =80.6%, c j =13.2m, v j =22m / s→F8=0.722; Drone 012: a j =79%, b j =1.5°, d j =93.7%, c j =8.2m, v j =20m / s→F 12 =0.738.

[0036] Drone 006: a j =55%, b j =1.2°, d j =82.6%, c j =12.2m, v j=21m / s→F8=0.721.

[0037] Drone 007: a j =45%, b j =1.6°, d j =85.6%, c j =11.2m, v j =23m / s→F8=0.712.

[0038] Drone 009: a j =75%, b j =1.3°, d j =89.6%, c j =13.5m, v j =26m / s→F8=0.70.

[0039] Drone 010: a j =85%, b j =1.2°, d j =81.5%, c j =10.5m, v j =23m / s→F8=0.723.

[0040] Drone 011: a j =75%, b j =1.7°, d j =82.3%, c j =19.6m, v j =25m / s→F8=0.735.

[0041] Drone 001: a j =55%, b j =1.9°, d j =83.5%, c j =14.6m, v j =24m / s→F8=0.715.

[0042] Drone 002: a j =85%, b j =1.1°, d j =81.7%, c j =17.5m, v j =23m / s→F8=0.733.

[0043] Drone 003: a j =35%, b j =1.5°, d j =82.9%, c j=19.2m, v j =20m / s→F8=0.707.

[0044] Drone 005 won with the highest score and completed the host switch within 50ms.

[0045] Formation reconfiguration. The new host 005 calls the Voronoi module to re-divide the formation into 12 sub-regions based on the real-time battery levels of each unit. For unit 008, the original target point deviates from the new Voronoi centroid by 37.5m, triggering PSO replanning: after 25 iterations, a smooth obstacle avoidance path is generated (maximum curvature 0.025m). -1 (meeting rotorcraft dynamics constraints), with no hovering or waiting throughout the entire flight.

[0046] In this embodiment, firstly, a dynamic threshold function s for RSSI is proposed. top (t), for the first time, link quality assessment is coupled with cluster motion state and channel statistical characteristics, solving the common problem of poor adaptability of fixed thresholds in dynamic scenarios; Second, a four-dimensional weighted election model of "battery power - attitude - history - environment" is constructed, and the weight coefficients support the adaptive configuration of mission mode, so that the host selection has the advantages of endurance guarantee, flight stability, mission reliability and safety redundancy. Third, a joint reconstruction mechanism of weighted Voronoi and obstacle perception PSO is designed to unify spatial division and path planning under the goal of maximizing safety margin, breaking through the limitation of separate optimization of the two in traditional methods. Fourth, establish a hierarchical response strategy: a three-level linkage of global host reconstruction, local neighborhood fine-tuning, and single-machine autonomous obstacle avoidance, taking into account both system robustness and real-time response.

[0047] Example 2: Based on the collaborative control method for realizing self-organized protection of multiple UAV clusters in Example 1, the present invention will be further explained and described below.

[0048] like Figure 3 As shown, the present invention also provides a collaborative control system for realizing self-organized protection of multiple UAV swarms, used to implement the aforementioned collaborative control method for realizing self-organized protection of multiple UAV swarms, the system comprising: The multi-drone swarm data acquisition module is used to acquire data information such as the remaining battery power, RSSI signal strength and IMU attitude of each drone in real time based on the onboard built-in sensors, and to acquire data information on the distance and relative speed of obstacles around the drone in real time based on the onboard external millimeter-wave radar. The multi-UAV swarm adaptive heartbeat monitoring module is connected to the multi-UAV swarm data acquisition module and is used to construct a dynamic threshold function s. top If the RSSI signal strength of each drone is greater than stop Then it outputs the remaining battery power and IMU attitude data for each drone; A multi-drone cluster weighted election module, connected to the multi-drone cluster adaptive heartbeat monitoring module, is used to select the host of the drones using an improved weighted election algorithm to obtain the selected host of the drones. The multi-UAV swarm intelligent formation reconstruction module is connected to the multi-UAV swarm weighted election module. When the host UAV changes, it uses a Voronoi diagram to divide the spatial region, a particle swarm algorithm to plan the path, adjusts the number and position of the remaining slave UAVs, maintains the formation shape, and avoids obstacles.

[0049] In this embodiment, the system further includes a multi-UAV swarm status synchronization and recovery module, a multi-UAV swarm abnormal vibration detection and alarm module, and a multi-UAV swarm enhanced communication protocol module. The multi-UAV swarm status synchronization and recovery module is connected to the multi-UAV swarm intelligent formation reconstruction module. It is used to globally broadcast the latest swarm status after a new host is selected, and to assign idle numbers and synchronize current tasks when a slave rejoins after fault repair. The multi-UAV swarm abnormal vibration detection and alarm module is connected to the multi-UAV swarm intelligent formation reconstruction module. It is used to identify abnormal vibrations by combining the UAV's fuselage IMU data, trigger structural health alarms, and upload them to the ground station. The multi-UAV swarm enhanced communication protocol module is connected and used to perform AES-256 encryption and HMAC authentication on all transmitted data.

[0050] This system employs a layered mesh architecture, specifically designed for self-organizing protection of drone swarms under dynamic high-pressure environments. Its core function is to achieve fault-tolerant collaboration within the swarm through a closed-loop "perception-decision-execution" process. The system is divided into five layers from top to bottom: dynamic high-pressure environment, drone swarm (master / slave), perception layer (data acquisition), decision-making layer (core control), and execution layer (action response). Each layer has independent functions and a clear linkage, as detailed below: 1. Top layer: Dynamic high-pressure environment As the external background for system operation, it provides complex scenario constraints (such as electromagnetic interference, signal obstruction, and task timeliness pressure), driving the cluster to have the protection capabilities of "adaptability, anti-interference, and high reliability". Dynamic factors in the environment (such as signal attenuation and obstacle movement) will affect the decision-making logic of the lower-level cluster in real time.

[0051] 2. Drone Swarm Layer: Master / Slave Collaborative Entity The cluster consists of one host and multiple slave devices, and is the core of the system's execution. The host: By default, it assumes the role of "commander" and is responsible for global task scheduling, heartbeat monitoring, and status synchronization; Slave device: Performs specific tasks (such as reconnaissance and transportation) and reports its status (heartbeat, location, battery level) to the master device. The master and slave devices interact through dual-link communication (Wi-Fi + self-organizing network) to ensure reliable transmission of commands and status information.

[0052] 3. Perception Layer: Data Acquisition and State Awareness Core function: Collects cluster status and environmental data to provide "first-hand input" for decision-making. Includes the following sensors and data flow: Basic status sensors: Deployed on each drone to collect battery power (remaining flight time), RSSI signal strength (communication link quality), and IMU attitude data (acceleration, angular velocity, used for flight control); Environmental perception sensor: Omnidirectional millimeter-wave radar, which detects the distance and relative speed of surrounding obstacles and other drones to avoid collisions; Data output: All sensing data is uploaded to the decision-making layer through dual links to support subsequent heartbeat judgment, formation adjustment and communication decisions.

[0053] 4. Decision-making level: Core control and intelligent decision-making Core Function: Processes data from the perception layer, generates "action commands" for the cluster, and serves as the system's "brain." It integrates the following six modules: Adaptive heartbeat monitoring: Receives heartbeat packets and RSSI data from the master and slave devices, and dynamically adjusts the heartbeat timeout threshold (e.g., set to 7 times for failure when the signal is good, and shorten to 2 times when the signal is poor); combined with the ACK confirmation mechanism (the master device requests the slave device to reply with confirmation after sending a heartbeat packet), it reduces misjudgments caused by temporary packet loss.

[0054] Improved weighted election algorithm: When the master fails (slave loses heartbeat), slaves compete for the new master through multi-factor scoring: battery power (30% weight, priority to high battery life), signal strength (25%, priority to good communication), location advantage (20%, priority to the center of the formation), historical task completion rate (25%, priority to reliable nodes); a 10-50ms random backoff is introduced during the competition to avoid multiple machines vying for the master at the same time.

[0055] Intelligent formation reconfiguration: When a node fails or the host changes, the spatial region is divided using a Voronoi diagram, the path is planned using a particle swarm algorithm, the numbering of the remaining slaves (e.g., after slave A fails, the number of slave B is increased by 1) and their positions are adjusted to maintain the formation shape (e.g., diamond / linear) and avoid obstacles and save power.

[0056] State synchronization and recovery: After a new host is elected, the latest cluster status (node ​​number, location, task progress) is broadcast globally; when a slave that has recovered from a fault rejoins, an idle number is assigned and the current task is synchronized to ensure cluster consistency.

[0057] Abnormal vibration detection and alarm: Combined with the IMU data of the fuselage, abnormal vibrations (such as motor imbalance, loose arm) are identified, structural health alarms are triggered and uploaded to the ground station to prompt maintenance.

[0058] Enhanced communication protocol: All transmitted data (heartbeat packets, commands, status) are encrypted with AES-256 (to prevent eavesdropping) and authenticated with HMAC (to prevent tampering) to ensure communication security.

[0059] 5. Execution layer: Action response and command execution Core function: Receiving instructions from the decision-making level and completing specific actions; it is the "limbs" of the system.

[0060] Master / Slave Control Unit: Host: Performs global scheduling, such as assigning tasks, triggering elections, and synchronizing states; Slave: Executes commands (such as "return home" or "adjust position") and reports the execution results back to the host; Dual-link transmission: Supports dynamic switching between Wi-Fi (high-speed short-range) and self-organizing network (interference-resistant and centerless) to ensure that commands are delivered (if the host command cannot be sent through Wi-Fi, it will automatically switch to self-organizing network).

[0061] This system, through its layered design and modular functionality, enables UAV swarms to achieve "self-sensing, self-decision-making, and self-execution" under dynamic high-pressure environments. It solves the problems of "single point failure, delayed response, and chaotic coordination" in traditional swarms and is suitable for high-risk scenarios such as rescue and patrol.

[0062] In this embodiment, the present invention provides a computer-readable storage medium storing a computer program programmed or configured to execute the aforementioned cooperative control method for implementing self-organized protection of multi-UAV swarms.

[0063] Any references to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0064] In summary, this invention not only effectively avoids misjudgments caused by temporary signal interference (such as brief packet loss due to electromagnetic shielding), reducing the misjudgment rate by 40%, reducing unnecessary return trips or cluster reorganizations, and ensuring continuous task execution, but also ensures rapid (millisecond-level) leadership handover when the host fails, preventing the cluster from falling into chaos due to single point of failure.

[0065] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for cooperative control of multi-UAV swarm self-organization protection, characterized in that, The method comprises: M1. The UAV group performs a task, real-time acquisition of data information of the remaining power, RSSI signal strength and IMU attitude of each UAV based on the on-board built-in sensor, and real-time acquisition of data information of the distance and relative speed of the obstacles around the UAV based on the on-board external millimeter wave radar; M2. Construct a dynamic threshold function s based on the data information of the RSSI signal strength of each UAV top , if the RSSI signal strength of each UAV is greater than s top , output the data information of the remaining power and the IMU attitude of each UAV; M3. Based on the data information of the remaining power and IMU attitude of each UAV, combined with the data information of the distance and relative speed of the obstacles around the UAV, and the data information of the historical task completion rate of each UAV, an improved weighted election algorithm is used to select the host of the UAV, and the selected host of the UAV is obtained; M4. When the host of the UAV changes, the space area is divided by a Voronoi diagram, the path is planned by a particle swarm algorithm, the number and position of the remaining slaves are adjusted, the formation shape is maintained, and the obstacles are avoided.

2. The method of claim 1, wherein, In step M2, the construction dynamic threshold function s top is ， wherein x i (t) is the data information of the RSSI signal strength of the ith UAV at time t, n is the total number of UAVs, T is the sampling period, and a, β, and δ are weight coefficients.

3. The method of claim 2, wherein: The weight coefficient a is any constant parameter between 0 and 1, the weight coefficient β is any constant parameter between 0 and 1, and the weight coefficient δ is any constant parameter between 0 and 1, and the sum of the squares of the weight coefficients a, β and δ is 1.

4. The method of claim 1, wherein, In step M3, the improved weighted election algorithm for selecting the host of the UAV comprises: M31. Based on the data information of the remaining power, IMU attitude, distance and relative speed of the obstacles around each UAV and historical task completion rate, normalization processing is performed to obtain the normalized data information of the remaining power, IMU attitude, distance and relative speed of the obstacles around each UAV and historical task completion rate; M32. Based on the remaining power, IMU attitude, distance and relative speed of surrounding obstacles, and historical task completion rate of each normalized drone, a weighted scoring function F of the jth drone is constructed j , ， wherein, a j (t) is the data information of the remaining power of the jth UAV at time t after normalization processing, b j (t) is the data information of the IMU attitude of the jth UAV at time t after normalization processing, c j (t) is the data information of the distance of the surrounding obstacles of the jth UAV at time t after normalization processing, v j (t) is the data information of the relative speed of the surrounding obstacles of the jth UAV at time t after normalization processing, d j (t) is the data information of the historical task completion rate of the jth UAV at time t after normalization processing, ω, σ, ρ and η are weight factors, and the score value of the jth UAV is calculated to obtain the data information of the score value of the jth UAV; M33. Based on the data information of the score value of the jth unmanned aerial vehicle, a decision function H = max{F1, F2,... F j ,...,F n} is constructed, n is the total number of unmanned aerial vehicles, and the unmanned aerial vehicle with the maximum score value is obtained, that is, the host of the selected unmanned aerial vehicle.

5. The method of claim 4, wherein, It is characterized in that: The value range of the weight factor ω is any constant parameter between 0 and 1, the value range of the weight factor σ is any constant parameter between 0 and 1, the value range of the weight factor ρ is any constant parameter between 0 and 1, and the value range of the weight factor η is any constant parameter between 0 and 1.

6. The method of claim 1, wherein, The method further comprises: M5. State synchronization and recovery: after the new host is elected, the latest cluster state is globally broadcasted, and when the slave after fault repair rejoins, an idle number is allocated and the current task is synchronized.

7. The method of claim 1, wherein, The method further comprises: M6. Abnormal vibration detection and alarm: combined with the fuselage IMU data of the UAV, abnormal vibration is identified, a structure health alarm is triggered and uploaded to the ground station.

8. The method of claim 1, wherein, The method further comprises: M7. Enhanced communication protocol: all transmission data is encrypted by AES-256 and authenticated by HMAC.

9. A cooperative control system for realizing multi-UAV swarm self-organization defense, characterized in that, The system for implementing the cooperative control method for realizing multi-UAV cluster self-organization protection according to any one of claims 1-8, the system comprises: A multi-UAV cluster data acquisition module for real-time acquisition of data information of the remaining power, RSSI signal strength and IMU attitude of each UAV based on the on-board built-in sensor, and real-time acquisition of data information of the distance and relative speed of the obstacles around the UAV based on the on-board external millimeter wave radar; The multi-unmanned aerial vehicle cluster adaptive heartbeat monitoring module is connected with the multi-unmanned aerial vehicle cluster data acquisition module, and is used for constructing a dynamic threshold function s top If the RSSI signal strength of each unmanned aerial vehicle is greater than s top , the data information of the residual power and the IMU attitude of each unmanned aerial vehicle is output. The multi-unmanned aerial vehicle cluster weighted election module is connected with the multi-unmanned aerial vehicle cluster adaptive heartbeat monitoring module, and is configured to select a host of the unmanned aerial vehicle by using an improved weighted election algorithm, and obtain the selected host of the unmanned aerial vehicle. The multi-unmanned aerial vehicle cluster intelligent formation reconstruction module is connected with the multi-unmanned aerial vehicle cluster weighted election module, and is configured to, when the host of the unmanned aerial vehicle is changed, divide a space region by using a Voronoi graph, plan a path by using a particle swarm algorithm, adjust numbers and positions of remaining slaves, maintain a formation shape, and avoid obstacles.

10. The cooperative control system for enabling multi-UAV swarm self-organization defense according to claim 9, wherein, The system further comprises a multi-unmanned aerial vehicle cluster state synchronization and recovery module, a multi-unmanned aerial vehicle cluster abnormal vibration detection and alarm module, and a multi-unmanned aerial vehicle cluster enhanced communication protocol module. The multi-unmanned aerial vehicle cluster state synchronization and recovery module is connected with the multi-unmanned aerial vehicle cluster intelligent formation reconstruction module, and is configured to, after a new host is elected, globally broadcast a latest cluster state, and when a slave is repaired, the slave is re-joined, an idle number is allocated, and a current task is synchronized. The multi-unmanned aerial vehicle cluster abnormal vibration detection and alarm module is connected with the multi-unmanned aerial vehicle cluster intelligent formation reconstruction module, and is configured to, in combination with IMU data of a fuselage of the unmanned aerial vehicle, identify abnormal vibration, trigger a structure health alarm, and upload the structure health alarm to a ground station. The multi-unmanned aerial vehicle cluster enhanced communication protocol module is connected, and is configured to perform AES-256 encryption and HMAC authentication on all transmission data.

Images