Honeycomb type unmanned aerial vehicle air control platform system and control method thereof

Abstract

The invention belongs to the technical field of unmanned aerial vehicle control, particularly relates to a cellular unmanned aerial vehicle air control platform system and a control method thereof, and improves the control accuracy and flexibility of an air unmanned aerial vehicle group. The honeycomb type unmanned aerial vehicle air control platform system is designed, multiple unmanned aerial vehicles are carried on the honeycomb type unmanned aerial vehicle air control platform system, and compared with the conventional unmanned aerial vehicle control mode that only one unmanned aerial vehicle can be controlled, the honeycomb type unmanned aerial vehicle air control platform system can achieve multi-vehicle collaborative operation, and the task efficiency is greatly improved; according to the control method, the motor driving force numerical value and the action behavior instruction can be packaged and synthesized into the task instruction, unified scheduling of available resources is completed, the communication and control complexity of multi-machine cooperation is effectively reduced, and the control method has good application prospects in the task processes of reconnaissance, decision making, commanding, execution and feedback and is suitable for popularization and application. And an efficient carrier is provided for large-scale unmanned aerial vehicle application.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) control technology, specifically relating to a honeycomb-type UAV aerial control platform system and its control method. Background Technology

[0002] Currently, drone swarms are increasingly widely used in military and civilian fields. They are seen in various scenarios, including coordinated reconnaissance and attacks in the military, and logistics delivery, agricultural protection, and large-scale performances in the civilian sector. In logistics, companies are exploring the use of drone swarms for cargo delivery to improve efficiency. In agriculture, drone swarms can quickly and accurately spray pesticides, supporting agricultural production. At some large-scale events, drone swarm performances have also delivered stunning visual effects.

[0003] In terms of control, controlling multiple drones is far more difficult than controlling a single drone. Flight control systems are inherently a technical challenge for drones. The spatial and temporal control of group control and the design of flight control algorithms under complex tasks all contribute to a geometric increase in control complexity, which seriously affects decision-making speed and mission coordination efficiency.

[0004] At the level of artificial intelligence, although it is hoped that drone swarms will have a high degree of autonomy, it is difficult to achieve a high degree of autonomous intelligence at this stage. They still rely heavily on remote assistance from control stations, which not only leads to signal transmission delays, but also places extremely high demands on flight, formation control and communication. Moreover, the realization of functions such as information interaction and task allocation between drones requires a powerful artificial intelligence system, which in turn faces challenges in terms of airborne computer capabilities and energy consumption.

[0005] In terms of information and communication, the information exchange of drone swarms needs to be high-speed, interception-proof, and interference-resistant. However, under the current technology, swarm communication has weak anti-interference capabilities. Once the key nodes of the communication link are damaged, the collaborative cooperation will collapse.

[0006] Energy storage limits the mobility of drone swarms. Due to the requirements of low cost and lightweight design, drone platforms have poor endurance and load-bearing capacity. Ideal low-energy devices and high-capacity battery technologies are not yet mature.

[0007] In addition, due to cost constraints, the payload capacity of a single drone is limited, and its sensor, communication, and computing capabilities are low, affecting its destructive capabilities. It can only be compensated for by swarm saturation attacks, which also increases the complexity and uncertainty of the overall operation. Summary of the Invention

[0008] In view of this, embodiments of this application provide a cellular unmanned aerial vehicle (UAV) airborne control platform system and its control method, which improves the control accuracy and battlefield application flexibility of UAV swarms.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows:

[0010] A honeycomb-type unmanned aerial vehicle (UAV) air control platform system includes: a honeycomb-type mounting platform and N UAVs, where N is a set value;

[0011] The drone is mounted on a honeycomb platform, receives mission instructions from the drone and executes the mission, and transmits the collected information to the honeycomb platform.

[0012] The honeycomb-style mounting platform is used to carry drones, process information collected by itself and the drones, generate task instructions, and send them to the drones. The honeycomb-style mounting platform includes a power supply module, an early warning module, and an intelligent processing module. The power supply module includes solar panels and battery panels to provide power to the early warning module and the intelligent processing module. The early warning module automatically analyzes, judges, and predicts early warning targets that fly into the warning area of ​​the honeycomb-style mounting platform, and generates prediction-related task instructions to send to the drones. The intelligent processing module receives information collected by the drones, generates task execution instructions based on the collected information, and sends them to the drones.

[0013] The honeycomb-type unmanned aerial vehicle (UAV) air control platform system includes: a software layer, a communication layer, and a service layer.

[0014] The software layer performs information processing, intelligent decision-making, sending and receiving instructions, and automatic networking.

[0015] Among them, information processing involves processing information collected by the UAV and instructions sent by communication satellites; intelligent decision-making involves generating mission instructions using intelligent algorithms; command transmission involves receiving mission instructions sent by space communication satellites, other airborne control platforms, and itself, and completing command transmission through intelligent path planning and topology networking; and automatic networking involves completing intelligent dynamic networking based on wireless communication, positioning and navigation, and intelligent control algorithms.

[0016] The communication layer includes a complex communication link consisting of an airborne control and command platform, a maritime control and command platform, reconnaissance satellites, and communication satellites, used to complete the intelligent encrypted transmission, reception, and transmission of dynamic data between platforms and between UAVs.

[0017] The service layer is used to generate endurance tasks, intelligence reconnaissance tasks, intelligent decision-making tasks and intelligent allocation tasks, early warning and interception tasks, and handling and evaluation tasks.

[0018] The drones include reconnaissance and command drones and disposal drones;

[0019] The reconnaissance and command drones perform endurance missions, intelligence reconnaissance missions, intelligent decision-making missions, and intelligent allocation missions, while also directing and handling drones for early warning and interception missions.

[0020] Handling drones, carrying out early warning and interception tasks, and handling and assessment tasks.

[0021] The present invention also provides a control method for the honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system described in the present invention, wherein the generation of mission instructions includes the following steps:

[0022] Step 1: Obtain real-time motion values ​​of the drone using a honeycomb-style mounting platform. t :

[0023] a t =π θ (o t )

[0024] Where, π θ To define a function that follows a Gaussian distribution, o t To observe the information value, the calculation process is as follows:

[0025]

[0026] in, Indicates the spatial state of the real-time measurement direction. It is a tuple containing information about its neighbors. This represents the relative spatial position of the i-th and j-th adjacent UAVs at time t. Let K represent the relative velocity between the i-th and j-th adjacent UAVs at time t, and K≤N-1 means that the number of friendly neighboring units is less than or equal to the total number of UAVs minus 1.

[0027] Step 2: Based on the real-time action value a t Calculate the motor driving force f of the drone. t :

[0028] f t =clip(a t ,0,1);

[0029] Here, clip() is a setting function used to convert the real-time motion value to the range [0, 1], where 0 represents no power and 1 represents maximum power;

[0030] Step 3: Generate action commands μ based on the information collected by the UAV. a :

[0031] μ a =φ a (e j ,e η ,e o ,e att )

[0032] Where, φ a e is the setpoint function based on the neural network reinforcement algorithm.j Encode the current state of the j-th UAV, e o Encode the current obstacle for the j-th UAV, e η For the j-th UAV, embed information encoding of the current friendly neighboring units, e att Encode the current target of the j-th UAV;

[0033] Step 4, set the motor drive force value f t and action / behavior instructions μ a Package it into a task instruction.

[0034] Among them, the real-time action value a t Calculate the motor driving force f of the drone. t The honeycomb-style platform calculates the motor drive force value while simultaneously calculating the optimal reward value r. t :

[0035]

[0036] in, This represents the reward value for the shortest straight-line distance to the target position. This represents the penalty value after a drone collision. This represents the auxiliary reward function. This represents the function that rewards completion of a task.

[0037] Among them, the reward value of the shortest target position straight-line distance Drone collision penalty value Auxiliary reward function and task completion reward function The calculation process is as follows:

[0038]

[0039] Where, α pos Indicates the location reward factor. This represents the distance between the i-th drone and the target at time t in the "Honeycomb" response group;

[0040]

[0041] α col Represents the collision penalty factor, usually At time t, it indicates that the collision of the i-th drone was detected. The value is 1, α prox This indicates a penalty factor for being too close. d represents the distance between the i-th and j-th UAV mass points. prox This indicates twice the drone's arm span.

[0042]

[0043] α ω This represents the initial angular velocity penalty factor. Let α represent the angular velocity of the i-th drone in the "Hive" swarm at time t. f f represents the initial thrust penalty factor. i t In expression (6), the thrust of the i-th UAV in the "Hive" swarm at time t is α. rot R represents the horizontal plane flipping factor in the "hive" treatment group at time t. t This represents the flipped vector in the "honeycomb" processing group at time t;

[0044]

[0045] α att Indicates the reward factor for hitting, α atp The reward factor indicates that the target was too close. d represents the distance between the i-th and j-th UAV mass points. prox This indicates twice the length of a drone's arm span.

[0046] Wherein, the e j The current status code of the jth drone is e j The obstacle code for the j-th drone is e. o The embedded information code e of the current friendly unit of the j-th UAV η The target code of the j-th drone is currently being attacked. att The calculation process is as follows:

[0047]

[0048] in, It is a fully connected neural network used to convert observed features of neighboring units into embedded information, s i Indicates the current status of the drone. and Let e ​​represent the relative distance and relative speed between the j-th UAV and the i-th UAV, respectively. m This represents the sum of embedded information from all neighboring units;

[0049]

[0050] Among them, e η This indicates that neighboring units embed information. Represents an additional hidden layer in the neural network, α j According to e j and e m Weight parameters calculated using a fully connected neural network;

[0051]

[0052] Where, φ o Represents a multilayer perceptron neural network, e o Represents obstacle coding, This represents the obstacle information encountered by the i-th drone in the "Hive" swarm at time t, mainly including the obstacle radius. Location and the relative speed of the i-th drone

[0053]

[0054] in, This indicates the target information of the i-th drone in the "Hive" response group at time t, mainly including the target radius. Location and the relative speed of the i-th drone φ att This represents a multilayer perceptron neural network.

[0055] The UAV airborne control platform system, upon receiving a mission instruction, sequentially completes the mission flow of reconnaissance, decision-making, command, execution, and feedback; the reconnaissance, decision-making, command, execution, and feedback specifically include the following:

[0056] Reconnaissance involves receiving mission instructions, performing target detection and early warning, intelligently analyzing and judging the collected data, and completing target profiling and generating functional parameters.

[0057] The decision-making process is based on reconnaissance information, which involves intelligently allocating tasks according to the current distribution locations of the target and our drones. These tasks include handling, escaping, and interception.

[0058] Command involves selecting the drone in the optimal current position to execute the mission based on the intelligent task allocation plan generated by the decision-making process.

[0059] Execution involves the drone receiving a task and then carrying out the task in a formation of one main drone and two backup drones. If the main drone fails to complete the task, the two backup drones will take over until the task is completed. After receiving an escape mission, the drone will quickly leave the designated area. After receiving an interception mission, the drone will intercept and block the target in a formation of one main drone and two backup drones.

[0060] Feedback involves evaluating the execution process after it has begun, and then sending the evaluation results to the higher-level control platform via radio.

[0061] Beneficial effects:

[0062] 1. This invention proposes a honeycomb-type UAV aerial control platform system, which carries multiple UAVs. Compared with conventional UAV control systems that can only control a single UAV, this invention can achieve multi-UAV collaborative operation, greatly improving mission efficiency. The platform's integrated management and control system can uniformly schedule resources, reduce the communication and control complexity of multi-UAV collaboration, and provide an efficient carrier for large-scale UAV applications.

[0063] 2. In the system of this invention, the power supply unit of the honeycomb platform integrates solar panels and battery panels, which can provide continuous power for the drone during the patrol flight, solving the pain points of traditional drones such as "short flight time and reliance on manual resupply".

[0064] 3. The UAV control platform proposed in this invention has a communication layer consisting of a honeycomb UAV airborne control platform system, an airborne control and command platform, a maritime control and command platform, a reconnaissance satellite, and a communication satellite. It is used to complete the intelligent encrypted transmission, reception, and transmission of dynamic data between the platform and between UAVs. The complete and fixed communication layer can adapt to the dynamic adjustment of task allocation and adapt to complex battlefield environments.

[0065] 4. The control platform system proposed in this invention automatically generates the optimal handling path based on the target location and the distribution of UAVs, avoiding delays caused by manual intervention and achieving a rapid response capability of "detection and strike".

[0066] 5. When calculating the motor driving force value, the present invention simultaneously calculates the optimal reward value, which ensures that the system has a corresponding judgment basis in subsequent calculations, thereby ensuring the optimal calculation results.

[0067] 6. This invention features a fully automated design encompassing reconnaissance (R), decision-making (D), command (D), execution (E), and feedback (F), eliminating the need for excessive human intervention throughout the mission. Taking the "one primary, two backup" execution mode as an example, if the primary UAV fails, the backup UAV can automatically take over, reducing the burden of human decision-making and improving mission reliability. Attached Figure Description

[0068] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0069] Figure 1 This is a schematic diagram of multi-area communication for a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system.

[0070] Figure 2 This is a schematic diagram illustrating the decision-making process for long-range, medium-range, and short-range surveillance areas of a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system.

[0071] Figure 3 This is a schematic diagram of the overall system of a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform.

[0072] Figure 4 RDEF flowchart for a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system. Detailed Implementation

[0073] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0074] The core of this invention's honeycomb-style unmanned aerial vehicle (UAV) airborne control platform system consists of a honeycomb-style platform and a UAV swarm, where the number of UAVs N is a preset value (typically N≥10). The honeycomb-style platform, serving as the central control system, comprises three core components: a power supply module, an early warning module, and an intelligent processing module. The power supply module utilizes high-efficiency solar panels and lithium-ion battery packs working in tandem, achieving a conversion efficiency of no less than 22%, supporting continuous offline operation for 72 hours. The early warning module integrates millimeter-wave radar and infrared thermal imaging sensors, enabling automatic detection, trajectory analysis, and threat assessment of targets within a 5-kilometer radius. The intelligent processing module is equipped with a high-performance edge computing unit (such as NVIDIA Jetson AGX Orin), responsible for processing multi-source information and generating decision commands.

[0075] The drone swarm comprises two functional units: reconnaissance and command drones and response drones. The reconnaissance and command drones are equipped with high-definition electro-optical turrets and multispectral sensors, performing intelligence gathering, environmental monitoring, and formation command tasks. The response drones are equipped with physical interception devices and non-lethal weapon systems, responsible for target interception and damage assessment. The system architecture adopts a three-layer design: the software layer implements intelligent decision-making and dynamic networking functions, generating control commands based on the near-end policy optimization (PPO) algorithm and achieving millisecond-level low-latency networking through a TDMA-Mesh hybrid protocol; the communication layer establishes an integrated air-space-sea data link, utilizing Ka-band satellite communication for encrypted data transmission; and the service layer integrates a mission planning engine, supporting the dynamic generation of six types of tasks, including endurance management, intelligence reconnaissance, and intelligent interception.

[0076] The task instruction generation process follows a standardized four-step procedure. First, the cellular platform calculates the real-time action values ​​of the UAVs based on environmental observation data. This calculation integrates spatial orientation and motion parameters of neighboring UAVs, introducing a random exploration factor through a Gaussian distribution function. Taking a swarm of 12 UAVs as an example, the observation value of each UAV includes the relative position vector and velocity vector of the other 11 friendly UAVs. Second, the system converts the action values ​​into motor driving force values ​​using a Sigmoid function, which are mapped to the power output percentage in the [0,1] interval. During this process, a comprehensive reward value is calculated simultaneously, consisting of four parts: target distance reward (inversely proportional to the UAV-target distance), collision penalty (triggered when contact with the target is detected), attitude stability reward (suppressing abnormal angular velocities), and mission completion reward (activated upon successful interception).

[0077] Next, the system generates behavioral commands through multi-source information fusion. A neural network architecture is used to process environmental features in parallel across four dimensions: a fully connected network encodes the system's own state vector (position, velocity, and attitude); an attention mechanism aggregates neighbor interaction information; a multilayer perceptron analyzes obstacle features (position, size, and relative velocity); and a convolutional network extracts target attributes. Finally, the motor drive force values ​​and behavioral commands are encapsulated into structured data packets and transmitted to the drone terminal via a LoRa-5G dual-channel transmission.

[0078] The mission execution follows a standardized workflow. During the reconnaissance phase, the UAV swarm performs a three-dimensional scan of the target area, using deep learning algorithms to complete target identification and 3D modeling with a positioning accuracy of ±0.5 meters. The decision engine generates three types of mission instructions based on the real-time situation: initiating a response mission when the target threat level is ≥7; assigning an interception mission to targets with speeds >200km / h; and triggering an evacuation mission when the system battery level is below 20%. The command system adopts a spatial proximity principle, prioritizing UAVs within 300 meters of the target as mission execution units. The execution process employs a "one primary, two backup" redundancy strategy: if the primary UAV fails during the response mission, the backup unit automatically takes over within 500 milliseconds; the interception mission uses a triangular formation for coordinated interception. After the mission is completed, damage assessment data is transmitted back to the command center via satellite link, forming a closed-loop decision-making process. Figure 1 This is a schematic diagram of multi-area communication for a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system. Figure 2 This is a schematic diagram illustrating the decision-making process for long-range, medium-range, and short-range surveillance areas of a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system. Figure 3 This is a schematic diagram of the overall system of a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform. Figure 4 RDEF flowchart for a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system.

[0079] A typical application scenario for this system is border security patrol. When the early warning module detects an illegally crossing aircraft, the intelligent processing module generates an interception plan within 2 seconds. Three response drones form a formation to approach the target and launch net-like projectiles for physical interception while maintaining a safe distance of 50 meters. The entire mission is monitored in real time by a reconnaissance and command drone, and the evidence data is transmitted to the higher-level command platform via an encrypted channel.

[0080] The present invention also provides a control method for the honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system described in the present invention, wherein the generation of mission instructions includes the following steps:

[0081] Step 1: Obtain real-time motion values ​​of the drone using a honeycomb-style mounting platform. t :

[0082] a t =π θ (o t )

[0083] Where, π θ To define a function that follows a Gaussian distribution, o t To observe the information value, the calculation process is as follows:

[0084]

[0085] in, Indicates the spatial state of the real-time measurement direction. It is a tuple containing information about its neighbors. This represents the relative spatial position of the i-th and j-th adjacent UAVs at time t. Let K represent the relative velocity between the i-th and j-th adjacent UAVs at time t, and K≤N-1 means that the number of friendly neighboring units is less than or equal to the total number of UAVs minus 1.

[0086] Step 2: Based on the real-time action value a t Calculate the motor driving force f of the drone. t :

[0087] f t =clip(a t ,0,1);

[0088] Here, clip() is a setting function used to convert the real-time motion value to the range [0, 1], where 0 represents no power and 1 represents maximum power;

[0089] Step 3: Generate action commands μ based on the information collected by the UAV. a :

[0090] μ a =φ a (ej ,e η ,e o ,e att )

[0091] Where, φ a e is the setpoint function based on the neural network reinforcement algorithm. j Encode the current state of the j-th UAV, e o Encode the current obstacle for the j-th UAV, e η For the j-th UAV, embed information encoding of the current friendly neighboring units, e att Encode the current target of the j-th UAV;

[0092] Step 4: Package the motor drive force values ​​and action commands into a task command.

[0093] Among them, the real-time action value a t Calculate the motor driving force f of the drone. t The honeycomb-style platform calculates the motor drive force value while simultaneously calculating the optimal reward value r. t :

[0094]

[0095] in, This represents the reward value for the shortest straight-line distance to the target position. This represents the penalty value after a drone collision. This represents the auxiliary reward function. This represents the function that rewards completion of a task.

[0096] Among them, the reward value of the shortest target position straight-line distance Drone collision penalty value Auxiliary reward function and task completion reward function The calculation process is as follows:

[0097]

[0098] Where, α pos Indicates the location reward factor. This represents the distance between the i-th drone and the target at time t in the "Honeycomb" response group;

[0099]

[0100] α col Represents the collision penalty factor, usually At time t, it indicates that the collision of the i-th drone was detected. The value is 1, α proxThis indicates a penalty factor for being too close. d represents the distance between the i-th and j-th UAV mass points. prox This indicates twice the drone's arm span.

[0101]

[0102] α ω This represents the initial angular velocity penalty factor. Let α represent the angular velocity of the i-th drone in the "Hive" swarm at time t. f Indicates the initial thrust penalty factor. In expression (6), the thrust of the i-th UAV in the "Hive" swarm at time t is α. rot R represents the horizontal plane flipping factor in the "hive" treatment group at time t. t This represents the flipped vector in the "honeycomb" processing group at time t;

[0103]

[0104] α att Indicates the reward factor for hitting, α atp The reward factor indicates that the target was too close. d represents the distance between the i-th and j-th UAV mass points. prox This indicates twice the length of a drone's arm span.

[0105] Wherein, the e j The current status code of the jth drone is e j The obstacle code for the j-th drone is e. o The embedded information code e of the current friendly unit of the j-th UAV η The target code of the j-th drone is currently being attacked. att The calculation process is as follows:

[0106]

[0107] in, It is a fully connected neural network used to convert observed features of neighboring units into embedded information, s i Indicates the current status of the drone. and Let e ​​represent the relative distance and relative speed between the j-th UAV and the i-th UAV, respectively. m This represents the sum of embedded information from all neighboring units;

[0108]

[0109] Among them, e η This indicates that neighboring units embed information. Represents an additional hidden layer in the neural network, α j According to e j and e m Weight parameters calculated using a fully connected neural network;

[0110]

[0111] Where, φ o Represents a multilayer perceptron neural network, e o Represents obstacle coding, This represents the obstacle information encountered by the i-th drone in the "Hive" swarm at time t, mainly including the obstacle radius. Location and the relative speed of the i-th drone

[0112]

[0113] in, This indicates the target information of the i-th drone in the "Hive" response group at time t, mainly including the target radius. Location and the relative speed of the i-th drone φ att This represents a multilayer perceptron neural network.

[0114] The UAV airborne control platform system, upon receiving a mission instruction, sequentially completes the mission flow of reconnaissance, decision-making, command, execution, and feedback; the reconnaissance, decision-making, command, execution, and feedback specifically include the following:

[0115] Reconnaissance involves receiving mission instructions, performing target detection and early warning, intelligently analyzing and judging the collected data, and completing target profiling and generating functional parameters.

[0116] The decision-making process is based on reconnaissance information, which involves intelligently allocating tasks according to the current distribution locations of the target and our drones. These tasks include handling, escaping, and interception.

[0117] Command involves selecting the drone in the optimal current position to execute the mission based on the intelligent task allocation plan generated by the decision-making process.

[0118] Execution involves the drone receiving a task and then carrying out the task in a formation of one main drone and two backup drones. If the main drone fails to complete the task, the two backup drones will take over until the task is completed. After receiving an escape mission, the drone will quickly leave the designated area. After receiving an interception mission, the drone will intercept and block the target in a formation of one main drone and two backup drones.

[0119] Feedback involves evaluating the execution process after it has begun, and then sending the evaluation results to the higher-level control platform via radio.

[0120] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system, characterized in that, include: A honeycomb-style platform and N drones, where N is a set value; The drone is mounted on a honeycomb platform, receives mission instructions from the drone and executes the mission, and transmits the collected information to the honeycomb platform. The honeycomb-type mounting platform is used to carry drones, process information collected by itself and drones, generate task instructions and send them to drones; The honeycomb-style platform includes: a power supply module, an early warning module, and an intelligent processing module; wherein the power supply module includes solar panels and battery panels to provide power to the early warning module and the intelligent processing module; the early warning module automatically analyzes, judges, and predicts early warning targets that fly into the warning area of ​​the honeycomb-style platform, and generates prediction-related task instructions to send to the drone; the intelligent processing module receives information collected by the drone, and generates execution instructions based on the collected information and sends them to the drone.

2. The system as described in claim 1, characterized in that, The honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system includes: a software layer, a communication layer, and a service layer. The software layer performs information processing, intelligent decision-making, sending and receiving instructions, and automatic networking. Among them, information processing involves processing information collected by the UAV and instructions sent by communication satellites; intelligent decision-making involves generating mission instructions using intelligent algorithms; command transmission involves receiving mission instructions sent by space communication satellites, other airborne control platforms, and itself, and completing command transmission through intelligent path planning and topology networking; and automatic networking involves completing intelligent dynamic networking based on wireless communication, positioning and navigation, and intelligent control algorithms. The communication layer includes a complex communication link consisting of an airborne control and command platform, a maritime control and command platform, reconnaissance satellites, and communication satellites, used to complete the intelligent encrypted transmission, reception, and transmission of dynamic data between platforms and between UAVs. The service layer is used to generate endurance tasks, intelligence reconnaissance tasks, intelligent decision-making tasks and intelligent allocation tasks, early warning and interception tasks, and handling and evaluation tasks.

3. The system as described in claim 1 or 2, characterized in that, The drones include reconnaissance and command drones and disposal drones; The reconnaissance and command drones perform endurance missions, intelligence reconnaissance missions, intelligent decision-making missions, and intelligent allocation missions, while also directing and handling drones for early warning and interception missions. Handling drones, carrying out early warning and interception tasks, and handling and assessment tasks.

4. A control method for a honeycomb-type unmanned aerial vehicle (UAV) airborne control platform system as described in any one of claims 1-3, characterized in that, The task generation instruction includes the following steps: Step 1: Obtain real-time motion values ​​of the drone using a honeycomb-style mounting platform. t : to t =π θ (the t ) Where, π θ To define a function that follows a Gaussian distribution, o t To observe the information value, the calculation process is as follows: in, Indicates the spatial state of the real-time measurement direction. It is a tuple containing information about its neighbors. This represents the relative spatial position of the i-th and j-th adjacent UAVs at time t. Let K represent the relative velocity between the i-th and j-th adjacent UAVs at time t, and K≤N-1 means that the number of friendly neighboring units is less than or equal to the total number of UAVs minus 1. Step 2: Based on the real-time action value a t Calculate the motor driving force f of the drone t : f t =clip(a t ,0,1); Among them, clip() is a setting function used to convert the real-time motion value to the range [0, 1], where 0 represents no power and 1 represents maximum power; Step 3: Generate action commands μ based on the information collected by the UAV. a : μ a =φ a (And j ,And η ,And o ,And att ) Where, φ a e is the setpoint function based on the neural network reinforcement algorithm. j Encode the current state of the j-th UAV, e o Encode the current obstacle for the j-th UAV, e η For the j-th UAV, embed information encoding of the current friendly neighboring units, e att Encode the current target of the j-th UAV; Step 4, set the motor drive force value f t and action / behavior instructions μ a Package it into a task instruction.

5. The method as described in claim 4, characterized in that, The real-time action value a t Calculate the motor driving force f of the drone t The honeycomb-style platform calculates the motor drive force value while simultaneously calculating the optimal reward value r. t : in, This represents the reward value for the shortest straight-line distance to the target position. This represents the penalty value after a drone collision. This represents the auxiliary reward function. This represents the function that rewards completion of a task.

6. The method as described in claim 5, characterized in that, The shortest target position straight-line distance reward value Drone collision penalty value Auxiliary reward function and task completion reward function The calculation process is as follows: Where, α pos Indicates the location reward factor. This represents the distance between the i-th drone and the target at time t in the "Honeycomb" response group; α col Represents the collision penalty factor, usually At time t, it indicates that the collision of the i-th drone was detected. The value is 1, α prox This indicates a penalty factor for being too close. d represents the distance between the i-th and j-th UAV mass points. prox This indicates twice the drone's arm span. α ω This represents the initial angular velocity penalty factor. Let α represent the angular velocity of the i-th drone in the "Hive" swarm at time t. f This represents the initial thrust penalty factor. In expression (6), the thrust of the i-th UAV in the "Hive" swarm at time t is α. rot R represents the horizontal plane flipping factor in the "hive" treatment group at time t. t This represents the flipped vector in the "hive" processing group at time t; α att Indicates the reward factor for hitting, α atp The reward factor indicates that the target was too close. d represents the distance between the i-th and j-th UAV mass points. prox This indicates twice the length of a drone's arm span.

7. The method as described in claim 5 or 6, characterized in that, The e j The current status code of the jth drone is e j The obstacle code for the j-th drone is e. o The embedded information code e of the current friendly unit of the j-th UAV η The target code of the j-th drone is currently being attacked. att The calculation process is as follows: in, It is a fully connected neural network used to convert observed features of neighboring units into embedded information, s i Indicates the current status of the drone. and Let e ​​represent the relative distance and relative speed between the j-th UAV and the i-th UAV, respectively. m This represents the sum of embedded information from all neighboring units; Among them, e η This indicates that neighboring units embed information. Represents an additional hidden layer in the neural network, α j According to e j and e m Weight parameters calculated using a fully connected neural network; Where, φ o This represents a multilayer perceptron neural network, e o Represents obstacle coding, This represents the obstacle information encountered by the i-th drone in the "Hive" swarm at time t, mainly including the obstacle radius. Location and the relative speed of the i-th drone in, This indicates the target information of the i-th drone in the "Hive" response group at time t, mainly including the target radius. Location and the relative speed of the i-th drone φ att This represents a multilayer perceptron neural network.

8. The method as described in claim 5 or 6, characterized in that, Upon receiving a mission instruction, the UAV airborne control platform system sequentially completes the mission flow of reconnaissance, decision-making, command, execution, and feedback; the reconnaissance, decision-making, command, execution, and feedback specifically include the following: Reconnaissance involves receiving mission instructions, performing target detection and early warning, intelligently analyzing and judging the collected data, and completing target profiling and generating functional parameters. The decision-making process is based on reconnaissance information, which involves intelligently allocating tasks according to the current distribution locations of the target and our drones. These tasks include handling, escaping, and interception. Command involves selecting the drone in the optimal current position to execute the mission based on the intelligent task allocation plan generated by the decision-making process. Execution is the process where, after receiving a task, the drones carry out the task in a formation of one main drone and two backup drones. If the main drone fails to complete the task during execution, the two backup drones will take over the task until it is completed. After receiving the escape mission, the drone quickly escapes the designated area; after receiving the interception mission, the drone intercepts and blocks the target in a formation of one main drone and two backup drones. Feedback involves evaluating the execution process after it has begun, and then sending the evaluation results to the higher-level control platform via radio.

Images