A multi-target optimization-based unmanned aerial vehicle group linkage control method and system

Abstract

This invention discloses a multi-objective optimization-based method and system for coordinated control of UAV swarms, relating to the field of UAV control technology. By constructing a unidirectional hierarchical structure within the current control cycle and generating a periodic frozen reference surface corresponding to the current control cycle based on the swarm state information at the end of the previous control cycle, the UAV swarm no longer engages in real-time bidirectional mutual tracking within the current control cycle. Instead, control corrections are sequentially performed from upstream to downstream layers under the combined constraints of the unidirectional hierarchical structure and the periodic frozen reference surface. This ensures that the control effects caused by local disturbances can only unfold unidirectionally, making it difficult to form a closed propagation chain within the same control cycle. Further multi-objective optimization is then performed, effectively suppressing the cyclical propagation and continuous amplification of control effects within the swarm while preserving multi-objective coordination capabilities. This avoids swarm control loop resonance and improves the stability, coordination, and mission execution reliability of the coordinated control.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, and specifically to a method and system for coordinated control of UAV swarms based on multi-objective optimization. Background Technology

[0002] In existing multi-objective UAV swarm coordinated control methods, the first step is to acquire state data such as position, velocity, attitude, remaining energy, obstacle information, and communication link information for each UAV. This data is then combined with mission objectives, formation requirements, and environmental constraints to construct a swarm control model. Based on this, multiple objectives, including mission advancement, formation maintenance, obstacle avoidance, communication connectivity, and energy consumption control, are integrated into a unified optimization framework. Through centralized or distributed collaborative solving, corresponding velocity commands, acceleration commands, or heading adjustments are generated for each UAV. This allows each UAV to maintain overall coordination while executing its individual control actions, enabling the UAV swarm to collaboratively search, inspect, track, transport, or fly in formation over target areas. This UAV swarm control method can accommodate multiple performance requirements and avoid local optima problems that arise under a single control objective, thus becoming an important implementation method in intelligent collaborative control of UAV swarms.

[0003] However, in actual operation, the aforementioned UAV swarm coordinated control method typically assumes that each UAV can continuously reference the latest state of its neighbors and synchronously complete coordinated corrections within the control cycle. When a UAV experiences a slight control deviation due to local wind disturbances, obstacle avoidance maneuvers, communication delays, or state errors, this deviation is easily transmitted by neighboring UAVs as a new cooperative reference, gradually forming a closed-loop influence chain within the same control cycle among multiple UAVs. Especially when multiple objectives such as formation maintenance, speed consistency, and mission advancement are coupled simultaneously, such local disturbances are not only difficult to attenuate in time, but may also circulate and be continuously amplified within the swarm, thereby inducing a swarm control loop resonance phenomenon, causing formation trajectory swaying, repeated control corrections, abnormally increased energy consumption, and decreased cooperative control stability. Therefore, how to effectively suppress the closed-loop backflow of control influence within the same cycle while retaining the multi-target cooperative control capability has become an urgent technical problem to be solved. Summary of the Invention

[0004] The purpose of this invention is to solve the problem mentioned in the background art, where the loop propagation within the UAV swarm is continuously amplified under the simultaneous coupling of multiple targets, thereby inducing the swarm control loop resonance phenomenon. Therefore, this invention proposes a UAV swarm linkage control method and system based on multi-target optimization.

[0005] A first aspect of this invention provides a method for coordinated control of unmanned aerial vehicle (UAV) swarms based on multi-objective optimization, the method comprising:

[0006] S1: Obtain the position status, speed status, formation adjacency, communication adjacency, and current mission direction of each UAV in the UAV swarm, and construct a unidirectional hierarchical structure of the UAV swarm within the current control cycle based on the position status, formation adjacency, communication adjacency, and current mission direction of each UAV.

[0007] S2: Based on the unidirectional hierarchical structure, generate the periodic freezing reference surface corresponding to the current control cycle based on the full group state information at the end of the previous control cycle, so that the periodic freezing reference surface and the unidirectional hierarchical structure together serve as the basic constraints for the UAV swarm linkage control in the current control cycle.

[0008] S3: Based on the unidirectional hierarchical structure and the periodic frozen reference surface, the control correction relationship of each UAV layer is determined sequentially from the upstream layer to the downstream layer, so that the downstream UAV completes the control correction of the current layer according to the periodic frozen reference surface and the correction result transmitted by the upstream layer, and restricts the control correction result of the current layer from being transmitted back to the upstream layer in the current control cycle to obtain the correction result; so that the control influence in the current control cycle unfolds unidirectionally along the unidirectional hierarchical structure.

[0009] S4: Based on the control correction results, under the constraints of the unidirectional hierarchical structure and the periodically frozen reference surface, perform multi-objective optimization on each layer of UAVs to obtain the linkage control results of each UAV in the current control cycle. The multi-objective optimization solution includes at least the mission advancement objective, formation maintenance objective, obstacle avoidance safety objective, communication connectivity objective, and energy consumption control objective.

[0010] S5: Based on the linkage control results, control each UAV to complete the coordinated flight within the current control cycle, and update the overall state information of the UAV swarm after the current control cycle ends. Then, based on the updated overall state information, reconstruct the unidirectional hierarchical structure and cycle-freezing reference surface corresponding to the next control cycle, so as to enter the linkage control process of the next control cycle.

[0011] Optionally, the position vectors of each drone in the drone swarm can be obtained to form a set of drone states at the current moment;

[0012] Based on the location of the mission target point and the geometric center of the UAV swarm, the current mission propulsion direction vector is calculated and normalized to obtain the unit mission propulsion direction vector.

[0013] Project the position vector of each UAV onto the propulsion direction vector of the unit task to obtain the propulsion projection value of each UAV;

[0014] A comprehensive adjacency matrix is ​​constructed based on the formation adjacency matrix and the communication adjacency matrix. Based on the propulsion projection value of each UAV and the comprehensive adjacency matrix, the first layer of nodes with the propulsion front and the predetermined adjacency coverage capability is first determined. Then, the nodes are expanded layer by layer according to the adjacency relationship with the node set of the previous layer and the propulsion projection value constraint to obtain the hierarchical node set consisting of the first layer to the Mth layer.

[0015] Finally, the control effects between adjacent layers are limited to propagation only in the direction of increasing layer number, thus forming a unidirectional hierarchical structure within the current control cycle.

[0016] Optionally, the steps to make the periodically frozen reference surface and the unidirectional hierarchical structure serve as the basic constraints for the coordinated control of the UAV swarm within the current control cycle are as follows:

[0017] Obtain the position vector and velocity vector of each UAV in the UAV swarm at the end of the previous control cycle, and determine the position vector of each UAV at the end of the previous control cycle as the frozen reference position of the corresponding UAV in the current control cycle;

[0018] The velocity vector of each UAV at the end of the previous control cycle is determined as the frozen reference velocity of the corresponding UAV in the current control cycle;

[0019] The freezing center of the entire swarm is calculated based on the freezing reference position of each UAV, and then the freezing center position corresponding to the node set of each layer is calculated by combining the unidirectional hierarchical structure.

[0020] Calculate the frozen main propulsion direction vector for the current control cycle based on the mission target point position vector and the frozen group center, and normalize the frozen main propulsion direction vector to obtain the frozen unit propulsion direction vector;

[0021] The distance between adjacent frozen layers is calculated based on the freezing center position of adjacent layers, and the freezing reference velocity of each layer is calculated based on the freezing reference velocity of each UAV in each layer.

[0022] The frozen reference position of each UAV, the frozen reference velocity of each UAV, the frozen center of the entire swarm, the frozen center position of each layer, the frozen unit's advance direction vector, the frozen reference velocity of each layer, and the frozen interlayer distance between adjacent layers are collectively constituted as the periodic frozen reference surface for the current control cycle. The periodic frozen reference surface and the unidirectional hierarchical structure are jointly determined as the basic constraints for the UAV swarm linkage control within the current control cycle.

[0023] Optionally, the control correction results of the current layer are restricted from being propagated back to the upstream layer within the current control cycle. The steps to obtain the correction results are as follows:

[0024] For each UAV in the first layer, calculate the position deviation vector between its frozen reference position and current position, and the velocity deviation vector between its frozen reference velocity and current velocity. Then, add the position deviation vector, velocity deviation vector and frozen unit thrust direction vector and perform modulus normalization to obtain the control correction amount for each UAV in the first layer.

[0025] The layer correction center amount of the first layer is calculated based on the control correction amount of each UAV in the first layer; then, for each UAV in any downstream layer, the position deviation vector between the frozen reference position and the current position, the velocity deviation vector between the frozen reference speed and the current speed, and the actual distance between the current position of the UAV and the frozen center of the previous layer are calculated respectively, and the interlayer distance deviation is calculated based on the actual distance and the corresponding interlayer distance.

[0026] The position deviation vector, velocity deviation vector, layer correction center value of the previous layer, and inter-layer distance deviation correction term along the direction of frozen unit advance are added together and then normalized to obtain the control correction value of each UAV in the current downstream layer.

[0027] The control correction values ​​for all UAVs within the current control cycle are calculated layer by layer from the first layer to the last layer. Within the current control cycle, only the layer correction center values ​​of the previous layer are allowed to participate in the calculation of the control correction values ​​of the next layer, and the control correction values ​​of the next layer are not allowed to participate in the recalculation of the previous layer. The set of control correction values ​​is output as the correction result of the current control cycle.

[0028] Optionally, the steps for performing multi-objective optimization on each layer of UAVs to obtain the coordinated control results of each UAV within the current control cycle are as follows:

[0029] Based on the unidirectional hierarchical structure, the generated periodically frozen reference surface, and the obtained set of control correction results;

[0030] Based on the current control cycle duration, the predicted position and predicted velocity of each UAV at the end of the current control cycle are established;

[0031] The control inputs of each UAV are limited to control inputs along the direction of the corresponding control correction result; based on the predicted position and control inputs, mission advancement objectives, formation maintenance objectives, obstacle avoidance and safety objectives, communication connectivity objectives, and energy consumption control objectives are constructed respectively.

[0032] Among them, the mission advancement target is used to characterize the distance between the predicted position of each UAV and the mission target point; the formation maintenance target is used to characterize the deviation between the predicted distance and the expected formation distance between UAVs with formation adjacency; the obstacle avoidance safety target is used to characterize the distance between the predicted position of each UAV and each obstacle; the communication connectivity target is used to characterize the overrange amount when the predicted distance between UAVs with communication adjacency exceeds the effective communication distance threshold; and the energy consumption control target is used to characterize the modulus of the control input of each UAV.

[0033] The multi-objective optimization solution is performed sequentially according to the constraints, following the order of obstacle avoidance and safety objectives, communication connectivity objectives, formation maintenance objectives, mission advancement objectives, and finally energy consumption control objectives. After each objective is solved, the current optimal feasible solution set for that objective is used as the basis for solving the next objective.

[0034] During the solution process, the following constraints are satisfied: the control input of the (m+1)th layer is determined only based on the control correction result of the mth layer and is not reversed based on the result of the downstream layer. The control input of each UAV is solved with the corresponding freezing reference position, freezing reference speed, freezing main propulsion direction, freezing center of each layer and distance between freezing layers as reference benchmarks. The control input direction of each UAV is consistent with the direction of the corresponding control correction result and satisfies the preset maximum control input amplitude constraint.

[0035] The optimized solution yields the control input combinations for each UAV, which are then output as the coordinated control results for each UAV within the current control cycle.

[0036] A second aspect of this invention provides a multi-objective optimization-based unmanned aerial vehicle (UAV) swarm coordinated control system, the system comprising:

[0037] Structure building module: Obtains the position status, speed status, formation adjacency, communication adjacency, and current mission direction of each UAV in the UAV swarm, and constructs the unidirectional hierarchical structure of the UAV swarm within the current control cycle;

[0038] Basic constraint module: Based on the unidirectional hierarchical structure, the periodic freezing reference surface corresponding to the current control cycle is generated based on the full group state information at the end of the previous control cycle, so that the periodic freezing reference surface and the unidirectional hierarchical structure together serve as the basic constraints for the UAV swarm linkage control in the current control cycle.

[0039] Correction module: Based on a unidirectional hierarchical structure and a periodically frozen reference surface, the control correction relationship of each UAV layer is determined sequentially from the upstream layer to the downstream layer. This enables the downstream UAV to complete the control correction of the current layer according to the periodically frozen reference surface and the correction results transmitted from the upstream layer, and restricts the control correction results of the current layer from being transmitted back to the upstream layer within the current control cycle to obtain the correction result.

[0040] Target optimization module: Based on the control correction results, under the constraints of a unidirectional hierarchical structure and a periodically frozen reference surface, multi-objective optimization is performed on each layer of UAVs to obtain the linkage control results of each UAV in the current control cycle. The multi-objective optimization includes mission advancement objectives, formation maintenance objectives, obstacle avoidance and safety objectives, communication connectivity objectives, and energy consumption control objectives.

[0041] Control module: Based on the linkage control results, it controls each UAV to complete the coordinated flight within the current control cycle, and updates the overall state information of the UAV swarm after the current control cycle ends. Then, based on the updated overall state information, it reconstructs the unidirectional hierarchical structure and the cycle-freezing reference surface corresponding to the next control cycle, so as to enter the linkage control process of the next control cycle.

[0042] The beneficial effects of this invention are:

[0043] This invention proposes a method and system for coordinated control of UAV swarms based on multi-objective optimization. By constructing a unidirectional hierarchical structure within the current control cycle, and generating a periodically frozen reference surface corresponding to the current control cycle based on the swarm state information at the end of the previous control cycle, the coordinated reference relationship of the UAV swarm within the current control cycle no longer unfolds through real-time bidirectional mutual following between nodes. Instead, it uses the unidirectional hierarchical structure and the periodically frozen reference surface as a common constraint basis, performing control corrections sequentially from the upstream layer to the downstream layer. Furthermore, it restricts the reverse propagation of downstream layer correction results to the upstream layer within the current control cycle, thus ensuring that the control effects caused by local disturbances can only unfold unidirectionally along the unidirectional hierarchical path. It is difficult to form a closed propagation chain and return to the original node within the same control cycle. On this basis, multi-objective optimization is performed by combining the objectives of mission advancement, formation maintenance, obstacle avoidance and safety, communication connectivity and energy consumption control. This not only retains the ability of UAV swarm cooperative control to take into account multiple performance indicators, but also effectively suppresses the control effects caused by local wind disturbance, obstacle avoidance actions, communication delays or state deviations from circulating and continuously amplifying within the swarm. In this way, it can effectively avoid the swarm control loop resonance phenomenon, reduce formation trajectory sway and repeated control corrections, reduce abnormal energy consumption, and improve the overall stability, coordination and mission execution reliability in the UAV swarm linkage control process. Attached Figure Description

[0044] Figure 1 A flowchart of a multi-objective optimization-based unmanned aerial vehicle (UAV) swarm coordinated control method provided in an embodiment of the present invention. Detailed Implementation

[0045] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0046] This invention provides a method for coordinated control of unmanned aerial vehicle (UAV) swarms based on multi-objective optimization. See also... Figure 1 , Figure 1 A flowchart illustrating a multi-objective optimization-based unmanned aerial vehicle (UAV) swarm coordinated control method provided in this embodiment of the invention. The method includes the following steps:

[0047] S1: Obtain the position status, speed status, formation adjacency, communication adjacency, and current mission direction of each UAV in the UAV swarm, and construct a unidirectional hierarchical structure of the UAV swarm within the current control cycle based on the position status, formation adjacency, communication adjacency, and current mission direction of each UAV.

[0048] S2: Based on the unidirectional hierarchical structure, generate the periodic freezing reference surface corresponding to the current control cycle based on the full group state information at the end of the previous control cycle, so that the periodic freezing reference surface and the unidirectional hierarchical structure together serve as the basic constraints for the UAV swarm linkage control in the current control cycle.

[0049] S3: Based on the unidirectional hierarchical structure and the periodic frozen reference surface, the control correction relationship of each UAV layer is determined sequentially from the upstream layer to the downstream layer, so that the downstream UAV completes the control correction of the current layer according to the periodic frozen reference surface and the correction result transmitted by the upstream layer, and restricts the control correction result of the current layer from being transmitted back to the upstream layer in the current control cycle to obtain the correction result; so that the control influence in the current control cycle unfolds unidirectionally along the unidirectional hierarchical structure.

[0050] S4: Based on the control correction results, under the constraints of the unidirectional hierarchical structure and the periodically frozen reference surface, perform multi-objective optimization on each layer of UAVs to obtain the linkage control results of each UAV in the current control cycle. The multi-objective optimization solution includes at least the mission advancement objective, formation maintenance objective, obstacle avoidance safety objective, communication connectivity objective, and energy consumption control objective.

[0051] S5: Based on the linkage control results, control each UAV to complete the coordinated flight within the current control cycle, and update the overall state information of the UAV swarm after the current control cycle ends. Then, based on the updated overall state information, reconstruct the unidirectional hierarchical structure and cycle-freezing reference surface corresponding to the next control cycle, so as to enter the linkage control process of the next control cycle.

[0052] This invention provides a multi-objective optimization-based UAV swarm coordinated control method. By constructing a unidirectional hierarchical structure within the current control cycle and generating a periodic frozen reference surface corresponding to the current control cycle based on the swarm state information at the end of the previous control cycle, the coordinated reference relationship of the UAV swarm within the current control cycle no longer unfolds through real-time bidirectional mutual following between nodes. Instead, it uses the unidirectional hierarchical structure and the periodic frozen reference surface as a common constraint basis, sequentially completing control corrections from the upstream layer to the downstream layer. Furthermore, it restricts the downstream layer's correction results from being transmitted back to the upstream layer within the current control cycle, thus ensuring that the control effects caused by local disturbances can only unfold unidirectionally along the unidirectional hierarchical path. Opening up makes it difficult to form a closed propagation chain and return to the original node within the same control cycle. On this basis, multi-objective optimization is performed by combining mission advancement objectives, formation maintenance objectives, obstacle avoidance safety objectives, communication connectivity objectives, and energy consumption control objectives. This not only retains the ability of UAV swarm collaborative control to comprehensively consider multiple performance indicators, but also effectively suppresses the control effects caused by local wind disturbances, obstacle avoidance actions, communication delays, or state deviations from circulating and continuously amplifying within the swarm. In this way, it can effectively avoid the swarm control loop resonance phenomenon, reduce formation trajectory sway and repeated control corrections, reduce abnormal energy consumption, and improve the overall stability, coordination, and mission execution reliability in the UAV swarm linkage control process.

[0053] In one embodiment, S1: The steps of acquiring the position status, speed status, formation adjacency, communication adjacency, and current mission direction of each UAV in the UAV swarm, and constructing the unidirectional hierarchical structure of the UAV swarm within the current control cycle based on the position status, formation adjacency, communication adjacency, and current mission direction of each UAV are as follows:

[0054] Obtain the position vectors of each drone in the drone swarm to form the drone state set at the current moment;

[0055] Based on the location of the mission target point and the geometric center of the UAV swarm, the current mission propulsion direction vector is calculated and normalized to obtain the unit mission propulsion direction vector.

[0056] Project the position vector of each UAV onto the propulsion direction vector of the unit task to obtain the propulsion projection value of each UAV;

[0057] A comprehensive adjacency matrix is ​​constructed based on the formation adjacency matrix and the communication adjacency matrix. Based on the propulsion projection value of each UAV and the comprehensive adjacency matrix, the first layer of nodes with the propulsion front and the predetermined adjacency coverage capability is first determined. Then, the nodes are expanded layer by layer according to the adjacency relationship with the node set of the previous layer and the propulsion projection value constraint to obtain the hierarchical node set consisting of the first layer to the Mth layer.

[0058] Finally, the control effects between adjacent layers are limited to propagation only in the direction of increasing layer number, thus forming a unidirectional hierarchical structure within the current control cycle.

[0059] Specifically, in one implementation method, step S1 is implemented as follows:

[0060] Step 1: Obtain the current position and speed information of each drone in the drone swarm. Assume there are a total of [number missing] drones in the swarm. The first drone The status of the drone is recorded as follows: , Indicates the first The status of the drone; Indicates the first The position vector of the drone; Indicates the first The velocity vector of the drone; Indicates the drone's serial number. , This represents the total number of drones; in a three-dimensional coordinate system, the position vector can be represented as... , Indicates the first A drone in Position coordinates in the direction; Indicates the first A drone in Position coordinates in the direction; Indicates the first A drone in The position coordinates in the direction; the velocity vector can be represented as: , Indicates the first A drone in Velocity component in the direction; Indicates the first A drone in Velocity component in the direction; Indicates the first A drone in Velocity component in the direction;

[0061] The second step: After obtaining the basic status of the drones, it is necessary to further obtain the formation constraints and communication connections between the drones.

[0062] 1. Formation adjacency relationships; Let the formation adjacency matrix be: , Represents the adjacency matrix of the formation; This indicates whether there is a formation adjacency relationship between the nth and nth drones; if there is a direct formation maintenance relationship between the nth and nth drones, then... If there is no direct formation maintenance relationship, then ;

[0063] 2. Let the communication adjacency matrix be: , Represents the communication adjacency matrix. This represents the communication information between the nth and nth drones; if there is a direct communication connection between the nth and nth drones, then... If there is no direct communication connection between the nth drone and the nth drone, then .

[0064] 3. Determine the main propulsion direction for group coordination within the current control cycle. Let the position vector of the mission target point be: , This represents the position vector of the task target point. These represent the coordinates of the task target point in three directions, respectively.

[0065] First, calculate the geometric center position of the drone swarm. : Then, based on the mission objective point and the geometric center of the group, the current mission propulsion direction vector is calculated. : After normalization, the unit propulsion direction vector is obtained: ; This represents the normalized unit task propulsion direction vector; Representing vectors The modulus length;

[0066] After obtaining the mission propulsion direction, it is necessary to determine whether each UAV is positioned in front or behind that direction. Define the projection value of the nth UAV along the mission propulsion direction. for: , Indicates the first The projection value of the UAV along the mission's direction of advancement;

[0067] The formation adjacency relation and the communication adjacency relation are merged to form a comprehensive adjacency relation. Let the comprehensive adjacency matrix be: , Represents the composite adjacency matrix; This indicates whether there is a valid linkage relationship between the nth and nth drones; the elements of the comprehensive adjacency matrix are defined as follows: If two drones have either a formation adjacency relationship or a communication adjacency relationship, then they are considered to have an effective linkage relationship. and These represent formation adjacency and communication adjacency, respectively.

[0068] Step 3: After obtaining the projection value and the comprehensive adjacency relationship, first determine the first layer of nodes in the current control cycle, which is the starting layer of the hierarchical structure.

[0069] Specifically, the node with the largest projection value along the mission advancement direction is selected from all UAVs as... , This represents the maximum value among all UAV projection values; the set of candidate front nodes is defined as: , This represents the set of candidate frontier nodes in the first layer; Indicates the projection value tolerance threshold; when the first The difference between the projected value of the drone and the maximum projected value shall not exceed At that time, the first The drone was included in the set of frontier candidate nodes;

[0070] The advantage of doing this is that instead of mechanically selecting a single leading node, it allows a group of drones in the leading area to collectively form the first layer of candidate ranges.

[0071] Then, the overall adjacency degree of each node in the candidate node set is calculated: , This represents the overall adjacency of the nth UAV; This represents the elements in the composite adjacency matrix. This indicates the number of effective linkages between the nth drone and other drones;

[0072] The first layer can be further defined as: , Represents the set of nodes at the first level; Represents the set of frontier candidate nodes; This represents the overall adjacency of the nth UAV; This represents the adjacency threshold.

[0073] In other words, the first layer prioritizes drones that are both at the forefront of advancement and have strong adjacent coverage capabilities.

[0074] Step 4: After determining the first layer, it is necessary to expand layer by layer according to the comprehensive adjacency relationship to generate the second layer, the third layer, and so on until the last layer. Let the set of nodes in the determined m-th layer be denoted as . Then the candidate set of the (m+1)th layer can be defined as: , Indicates the first The set of candidate nodes for the layer; Indicates the first Layer node set; This represents the union of all nodes that have been layered from the first layer to the nth layer. This indicates the ID of a drone that has not yet been stratified. Indicates the first The drone belongs to the first layer; Indicates the first The drone has an effective linkage relationship with at least one node in the Nth layer; Indicates existence;

[0075] To maintain consistency in the hierarchical direction, it can be further required that the projection value of the candidate node is no greater than the average projection value of the upstream layer; first calculate the... Average projection value of the layer: ; Indicates the first The average projection value of the layer nodes, Indicates the first Number of layer nodes;

[0076] Then layer m+1 can be defined as: ; This represents the projection tolerance threshold when layering from layer m to layer m+1; in this way, we can expand layer by layer from front to back until all drones are included in a certain layer.

[0077] The final result is a set of hierarchical structures: ;in, This indicates the unidirectional hierarchical structure of the drone swarm within the current control cycle; They represent the first layer, the second layer, and so on up to the [number]th layer. The set of nodes in a layer; This indicates the total number of floors.

[0078] Step 5: After completing the layering, it is also necessary to clarify the direction of influence transmission between each layer.

[0079] Define the inter-layer directed relation matrix as follows: , Represents the directed relation matrix between layers; This indicates whether a control influence transfer relationship is allowed between layer 𝑚 and layer 𝑛;

[0080] To achieve unidirectional hierarchical transfer, the following can be defined: ;when When, it means that only the first layer is allowed to send data to the layer below it. Layer-transmitted effects; in other cases... This means that direct propagation of the current cycle between non-adjacent layers is not allowed, nor is reverse propagation from downstream to upstream layers permitted. Thus, the inter-layer propagation path is limited to: That is, the influence can only spread unidirectionally in the direction of increasing layer number.

[0081] Finally, the following result is taken as the output of S1: the unit task propulsion direction vector of the current control cycle. The set of projection values ​​of each UAV along the mission advancement direction. Comprehensive Adjacency Matrix ; Stratification results Interlayer unidirectional transfer relationship matrix This completes the construction of the unidirectional hierarchical structure within the current control cycle.

[0082] It should be noted that the benefits of step S1 are as follows: By reconstructing the potentially intersecting and bidirectional backflow relationships within the UAV swarm into an ordered control structure that unfolds layer by layer along the mission's direction of advancement, the objects referenced by each UAV in subsequent control, its preceding and following levels, and the direction of influence transmission are all predetermined. The advantages are twofold: First, by using the mission target point and the geometric center of the swarm to determine the main propulsion direction, and combining this with the projection values ​​of each UAV along that direction, it is possible to accurately distinguish which UAVs are at the forefront and which are in the follower positions, thus ensuring that the layering results are consistent with the current mission's advancement trend and avoiding a disconnect between the control layer and the actual flight direction. Second, by integrating formation adjacency relationships and communication adjacency relationships to construct a comprehensive adjacency matrix, and then expanding the node set layer by layer on this basis, layering can be completed without disrupting the original formation coordination relationships and communication reachability relationships, preventing the coordination chain from breaking due to mechanical layering. Simultaneously, by ultimately limiting inter-layer influence to unidirectional transmission only along the increasing layer number direction, the reverse backflow path within the same control cycle can be directly cut off from the control structure, preventing local disturbances from propagating back and forth among multiple UAVs and being continuously amplified.

[0083] In one embodiment, S2: Based on the unidirectional hierarchical structure, the step of generating a periodic freezing reference surface corresponding to the current control cycle based on the swarm state information at the end of the previous control cycle, and making the periodic freezing reference surface and the unidirectional hierarchical structure serve as the basic constraints for the UAV swarm linkage control in the current control cycle, is as follows:

[0084] Obtain the position vector and velocity vector of each UAV in the UAV swarm at the end of the previous control cycle, and determine the position vector of each UAV at the end of the previous control cycle as the frozen reference position of the corresponding UAV in the current control cycle;

[0085] The velocity vector of each UAV at the end of the previous control cycle is determined as the frozen reference velocity of the corresponding UAV in the current control cycle;

[0086] The freezing center of the entire swarm is calculated based on the freezing reference position of each UAV, and then the freezing center position corresponding to the node set of each layer is calculated by combining the unidirectional hierarchical structure.

[0087] Calculate the frozen main propulsion direction vector for the current control cycle based on the mission target point position vector and the frozen group center, and normalize the frozen main propulsion direction vector to obtain the frozen unit propulsion direction vector;

[0088] The distance between adjacent frozen layers is calculated based on the freezing center position of adjacent layers, and the freezing reference velocity of each layer is calculated based on the freezing reference velocity of each UAV in each layer.

[0089] The frozen reference position of each UAV, the frozen reference velocity of each UAV, the frozen center of the entire swarm, the frozen center position of each layer, the frozen unit's advance direction vector, the frozen reference velocity of each layer, and the frozen interlayer distance between adjacent layers are collectively constituted as the periodic frozen reference surface for the current control cycle. The periodic frozen reference surface and the unidirectional hierarchical structure are jointly determined as the basic constraints for the UAV swarm linkage control within the current control cycle.

[0090] Specifically, in one implementation method, step S2 is implemented as follows:

[0091] First, read the status information of each drone in the drone swarm at the end of the previous control cycle. Let the status of the nth drone at the end of the previous control cycle be denoted as: , Indicates the first The state of the drone at the end of the previous control cycle; Indicates the first The position vector of the UAV at the end of the previous control cycle. Indicates the first The velocity vector of the drone at the end of the previous control cycle. Number the drone. , Indicates the total number of drones;

[0092] Then, the position vector and velocity vector of each UAV at the end of the previous control cycle are used as the frozen reference position and frozen reference velocity for the current control cycle, respectively: , This indicates the frozen reference position of the nth UAV during the current control cycle; This indicates the frozen reference velocity of the nth UAV during the current control cycle;

[0093] Next, combining the unidirectional hierarchical structure already constructed in S1 First, calculate the frozen group center for the current control period. : ; This represents the vector of the frozen population center positions; further, the frozen center positions of each layer are calculated separately: , This represents the vector indicating the location of the freezing center of the nth layer; Represents the set of nodes at level n. This represents the number of drones in layer X;

[0094] Then, based on the target point position vector already determined in S1 and freeze the entire group center Calculate the frozen main propulsion direction vector for the current control cycle. : The vector of the frozen unit's propulsion direction is then normalized to obtain the vector of the frozen unit's propulsion direction. : In the formula, Representing vectors The modulus length;

[0095] Maintaining the interlayer spatial relationship corresponding to the unidirectional cascading structure in S1, continue to calculate the frozen interlayer distance between adjacent layers: ; Indicates the first Layer and First Freezing distance between layers; and They represent the first Layer and first Layer freeze center position vector;

[0096] Calculate the reference freezing rate for each layer: , Indicates the first The freezing reference velocity vector of the layer; Indicates the first Freeze reference speed of the drone;

[0097] Finally, freeze the set of reference locations. Frozen reference velocity set Freeze the entire group center Collection of freezing centers at each level Freeze the unit's propulsion direction vector Set of reference velocities for each layer of freezing and the set of distances between each frozen layer Together they form the periodic freeze reference surface of the current control cycle. , This causes the periodic freezing reference surface to be... The unidirectional layered structure obtained in S1 Together, they serve as the fundamental constraints for the coordinated control of the UAV swarm within the current control cycle, namely... ; This represents the set of basic constraints within the current control cycle.

[0098] It should be noted that the advantage of determining the basic constraints in the above manner is that, based on the stable state information of the entire swarm at the end of the previous control cycle, a unified swarm reference benchmark that remains unchanged throughout the current cycle is constructed. This ensures that when subsequent UAVs perform control corrections and multi-objective optimizations, they do not directly chase the constantly changing instantaneous states of other UAVs within the current cycle, but are instead constrained by the same set of frozen reference positions, frozen swarm centers, frozen centers of each layer, frozen main propulsion directions, frozen reference velocities of each layer, and frozen inter-layer distances. The advantage is that, on the one hand, the stable swarm structure at the end of the previous cycle can be completely continued into the current cycle, avoiding the influence of a single UAV's initial state in the current cycle. Local offsets, obstacle avoidance maneuvers, or velocity fluctuations cause the entire group's reference baseline to be deflected in real time, thereby improving the stability and consistency of the coordinated control within the current control cycle. On the other hand, combined with the unidirectional hierarchical structure already constructed in S1, each layer of UAVs has a clear transmission sequence and a fixed group reference base as a constraint within this cycle. This ensures that the inter-layer control corrections are carried out in an orderly manner along the predetermined direction, while maintaining the original spatial interval relationship, overall advancement trend, and hierarchical movement trend between each layer. It prevents local disturbances from accumulating and amplifying layer by layer during the inter-layer transmission process, providing a continuous, stable, and non-distortion-prone basic constraint for determining the subsequent control correction relationship and solving multi-objective optimization problems.

[0099] In one embodiment, S3: Based on the unidirectional hierarchical structure and the periodically frozen reference surface, the control correction relationship of each UAV layer is determined sequentially from the upstream layer to the downstream layer, so that the downstream UAV completes the control correction of the current layer according to the periodically frozen reference surface and the correction result transmitted by the upstream layer, and restricts the control correction result of the current layer from being transmitted back to the upstream layer in the current control cycle. The steps to obtain the correction result are as follows:

[0100] Based on the unidirectional layered structure and the generated periodically frozen reference surface;

[0101] For each UAV in the first layer, calculate the position deviation vector between its frozen reference position and current position, and the velocity deviation vector between its frozen reference velocity and current velocity. Then, add the position deviation vector, velocity deviation vector and frozen unit thrust direction vector and perform modulus normalization to obtain the control correction amount for each UAV in the first layer.

[0102] The layer correction center amount of the first layer is calculated based on the control correction amount of each UAV in the first layer; then, for each UAV in any downstream layer, the position deviation vector between the frozen reference position and the current position, the velocity deviation vector between the frozen reference speed and the current speed, and the actual distance between the current position of the UAV and the frozen center of the previous layer are calculated respectively, and the interlayer distance deviation is calculated based on the actual distance and the corresponding interlayer distance.

[0103] The position deviation vector, velocity deviation vector, layer correction center value of the previous layer, and inter-layer distance deviation correction term along the direction of frozen unit advance are added together and then normalized to obtain the control correction value of each UAV in the current downstream layer.

[0104] The control correction values ​​for all UAVs in the current control cycle are calculated layer by layer from the first layer to the last layer in the above manner. Only the layer correction center value of the previous layer is allowed to participate in the calculation of the control correction value of the next layer in the current control cycle, and the control correction value of the next layer is not allowed to participate in the recalculation of the previous layer. The set of control correction values ​​is output as the correction result of the current control cycle.

[0105] Specifically, in one implementation method, step S3 is implemented as follows:

[0106] First, read the unidirectional hierarchical structure obtained from S1. and the periodic freezing reference surface obtained by S2 Following the ascending order of layer number, the control correction relationships for each UAV layer are determined sequentially from the upstream layer to the downstream layer. For the first... The first in the layer First, calculate the position deviation vector of the drone's current position relative to the frozen reference position: ;in: Indicates the first The positional deviation vector of the drone; Indicates the first The actual position vector of the drone at the current moment;

[0107] Next, calculate the velocity deviation vector of its current velocity relative to the frozen reference velocity: In the formula, Indicates the first The velocity deviation vector of the drone; Indicates the first The actual velocity vector of the drone at the current moment;

[0108] For the first layer For drones in this context, since there is no upstream layer, the position deviation vector, velocity deviation vector, and frozen unit thrust direction vector are summed to obtain the basic correction vector for the first layer. : In the formula, Indicates the first layer The basic correction vector for the drone;

[0109] Then, normalize the magnitude of the basic correction vector to obtain the first layer. Control corrections for drones : , Representing vectors The modulus length;

[0110] Then, calculate the first... Layer correction center amount: , Indicates the first Layer correction center amount;

[0111] Following the above method, the control correction quantities for all UAVs are calculated layer by layer from the first layer to the nth layer: ;

[0112] To prevent the control correction results of the current layer from being propagated back to the upstream layer within the current control cycle, it is stipulated that the first... The control correction amount of the layer is only determined by the first layer. Layer correction center quantity Participating in the calculation, and the first The control correction amount of the layer will no longer be based on the first layer in this control cycle. The layer's control correction results are updated, that is: ; Indicates the first The first in the layer The control correction of the drone to the first The first in the layer The sensitivity of the control correction amount for the UAV is set. When its value is 0, it indicates that the correction results generated by the downstream layer in this cycle do not have a reverse effect on the upstream layer. Finally, the control correction amount set is... Output as the correction result within the current control cycle.

[0113] It should be noted that, based on the unidirectional hierarchical structure established by S1 and the periodic frozen reference surface formed by S2, the control correction process of each layer of UAVs within the current control cycle is explicitly organized into a correction chain that "has a fixed reference benchmark, is passed layer by layer, and does not allow reverse flow." This ensures that the control correction of each layer of UAVs can respond to its own deviation relative to the frozen reference position and frozen reference velocity, and can also orderly receive the overall correction trend already formed by the previous layer, while maintaining the original frozen interlayer spatial relationship with the previous layer. The advantage of this approach is that, on the one hand, the first layer first forms a correction result based on its own deviation and the frozen advancement direction, and subsequent downstream layers follow up layer by layer on this basis. This allows the control correction of the entire UAV swarm to unfold stably along the determined hierarchical direction, avoiding simultaneous disorderly corrections by each layer that could lead to control errors. The control chain is chaotic; on the other hand, the correction of the downstream layer is not simply blindly following the upstream layer, but is simultaneously constrained by the frozen reference position, frozen reference velocity, and the distance between frozen layers. Therefore, it can maintain the proper spatial position and motion trend of the current layer while receiving the upstream correction trend, preventing local offsets from being amplified by the entire layer. In particular, by limiting the layer correction center quantity of the previous layer to participate in the correction of the next layer within the current cycle, and not allowing the correction result of the next layer to participate in the recalculation of the previous layer, the backflow path within the same control cycle can be directly cut off from the correction propagation mechanism. This avoids local wind disturbances, obstacle avoidance actions, or state fluctuations from propagating back and forth between layers and being amplified, thereby providing correction results with clear direction, clear hierarchy, and less distortion for subsequent multi-objective optimization, improving the stability, continuity, and anti-loop resonance capability of the entire UAV swarm linkage control process.

[0114] In one embodiment, S4: Based on the control correction results, under the constraints of the unidirectional hierarchical structure and the periodically frozen reference surface, the steps of performing multi-objective optimization on each layer of UAVs to obtain the linkage control results of each UAV in the current control cycle are as follows:

[0115] Multi-objective optimization solutions include at least the mission advancement objective, formation maintenance objective, obstacle avoidance and safety objective, communication connectivity objective, and energy consumption control objective;

[0116] Based on the unidirectional hierarchical structure, the generated periodically frozen reference surface, and the obtained set of control correction results;

[0117] Based on the current control cycle duration, the predicted position and predicted velocity of each UAV at the end of the current control cycle are established;

[0118] The control inputs of each UAV are limited to control inputs along the direction of the corresponding control correction result; based on the predicted position and control inputs, mission advancement objectives, formation maintenance objectives, obstacle avoidance and safety objectives, communication connectivity objectives, and energy consumption control objectives are constructed respectively.

[0119] Among them, the mission advancement target is used to characterize the distance between the predicted position of each UAV and the mission target point; the formation maintenance target is used to characterize the deviation between the predicted distance and the expected formation distance between UAVs with formation adjacency; the obstacle avoidance safety target is used to characterize the distance between the predicted position of each UAV and each obstacle; the communication connectivity target is used to characterize the overrange amount when the predicted distance between UAVs with communication adjacency exceeds the effective communication distance threshold; and the energy consumption control target is used to characterize the modulus of the control input of each UAV.

[0120] Based on this, multi-objective optimization is performed sequentially in the order of obstacle avoidance safety objective, communication connectivity objective, formation maintenance objective, mission advancement objective, and finally energy consumption control objective. After each objective is solved, the current best feasible solution set of that objective is used as the basis for solving the next objective.

[0121] During the solution process, the following constraints are satisfied: the control input of the (m+1)th layer is determined only based on the control correction result of the mth layer and is not reversed based on the result of the downstream layer. The control input of each UAV is solved with the corresponding freezing reference position, freezing reference speed, freezing main propulsion direction, freezing center of each layer and distance between freezing layers as reference benchmarks. The control input direction of each UAV is consistent with the direction of the corresponding control correction result and satisfies the preset maximum control input amplitude constraint.

[0122] Finally, the optimized combination of control inputs for each UAV is output as the coordinated control result of each UAV within the current control cycle.

[0123] Specifically, in one implementation method, step S4 is implemented as follows:

[0124] Read the unidirectional stacked structure obtained from S1 The periodic freezing reference surface obtained by S2 and the set of control correction results obtained in S3 The control inputs for each UAV are defined as follows: ; This represents the set of control inputs for all UAVs within the current control cycle. Indicates the first The control input vector of the drone;

[0125] Furthermore, based on the current control cycle length Δt, the predicted position and predicted velocity of each UAV at the end of the current control cycle are established: , , Indicates the first The predicted position vector of the UAV at the end of the current control cycle. Indicates the first The predicted velocity vector of the UAV at the end of the current control cycle;

[0126] Next, using the control correction results obtained in S3, the control input of each UAV is limited to the corresponding correction direction, that is: In the formula, Indicates the first The control amplitude of the UAV along the control correction direction. Indicates the first The first in the layer Control and correction direction of the drone;

[0127] Then, under the control input constraints, the mission advancement objective, formation maintenance objective, obstacle avoidance and safety objective, communication connectivity objective, and energy consumption control objective are constructed respectively.

[0128] The mission objective is defined as follows: , This represents the objective function for advancing the task. Represents the position vector of the task target point, and represents the first... The drone is used to predict the distance between its location and the mission objective.

[0129] Formation maintenance objective is defined as: , This indicates that the formation maintains the objective function. This represents the elements of the adjacency matrix in S1. This represents the expected formation distance between the nth drone and the nth drone; The actual distance between the predicted positions of the nth drone and the twelfth drone;

[0130] The obstacle avoidance safety goal is defined as: ; This represents the obstacle avoidance safety objective function. Indicates the total number of obstacles. Indicates the first The position vectors of the obstacles This represents the distance between the predicted position of the nth drone and the nth obstacle;

[0131] The communication connectivity target is defined as: ; The objective function represents communication connectivity. This represents the communication adjacency matrix element in S1. Indicates the effective communication distance threshold. This represents the communication overrange amount generated when the predicted distance exceeds the effective communication distance threshold;

[0132] The energy consumption control target is defined as: , This represents the objective function for energy consumption control. Let represent the magnitude of the control input vector for the nth UAV;

[0133] After constructing the above objective function, multi-objective optimization is performed in a fixed order, specifically: first, the obstacle avoidance safety objective is solved. The minimum value is then found within the set of feasible solutions that minimize the obstacle avoidance safety objective, and the communication connectivity objective is then solved. The minimum value; within the set of feasible solutions that satisfy both the obstacle avoidance safety objective and the communication connectivity objective as currently optimal, then solve for the formation maintenance objective. The minimum value is then calculated; within the set of feasible solutions that satisfy the current optimal objectives of obstacle avoidance safety, communication connectivity, and formation maintenance, the mission advancement objective is then solved. The minimum value is then calculated; finally, within the set of feasible solutions where all the aforementioned objectives are currently optimal, the energy consumption control objective is solved. The minimum value.

[0134] During the solution process, the following constraints must be satisfied simultaneously:

[0135] First, it satisfies the hierarchical transmission constraint of the unidirectional hierarchical structure in S1, that is, the control input of the (m+1)th layer depends only on the correction result of the mth layer, and does not depend on the result of the downstream layer.

[0136] Secondly, the freezing constraint of the periodic freezing reference surface in S2 is satisfied, that is, the control input of each UAV in the current control cycle is solved based on the freezing reference position, freezing reference speed, freezing main propulsion direction, freezing center of each layer and distance between freezing layers.

[0137] Third, it satisfies the control correction direction constraint in S3, that is, the control input direction of each UAV is consistent with the direction of the corresponding control correction amount;

[0138] Fourth, it satisfies the current control input amplitude constraint: ; Indicates the preset maximum control input amplitude;

[0139] After completing the above multi-objective optimization solution, the set of coordinated control results for each UAV within the current control cycle is obtained: , This represents the set of optimal coordinated control results for all UAVs within the current control cycle; Let represent the optimal control input vector for the nth UAV;

[0140] Finally, the results of the linkage control will be collected. As the output of the coordinated control results of each UAV within the current control cycle, it is used by S5 to control each UAV to complete the coordinated flight within the current control cycle.

[0141] It should be noted that, based on the unidirectional hierarchical structure established in S1, the periodically frozen reference surface formed in S2, and the control correction results obtained in S3, the UAVs are no longer allowed to directly execute actions based on a single correction direction. Instead, the objectives that must be simultaneously satisfied in actual flight—mission advancement, formation maintenance, obstacle avoidance safety, communication connectivity, and energy consumption control—are further integrated into the same solution process. Furthermore, feasible solutions are selected step by step in the order of obstacle avoidance, communication maintenance, formation stabilization, advancement, and finally energy consumption control, ensuring that each subsequent objective is always based on the satisfaction of the preceding objective. The advantage of this approach is that it ensures that the control results of the UAV swarm first satisfy the most basic requirements. The system meets the requirements of safety and interoperability, avoiding situations where high propulsion efficiency results in high collision risk, communication failure, or formation instability. It also ensures that the subsequently obtained control inputs are always constrained by the unidirectional hierarchical structure, the periodically frozen reference surface, and the control correction direction. This guarantees that each UAV continues to operate in an orderly manner along the predetermined hierarchical direction, without introducing cross-layer backflow or deviating from the frozen reference structure during the optimization process. As a result, the final linkage control result not only conforms to the group correction trend within the current control cycle but also takes into account mission efficiency, structural stability, communication reliability, and reasonable energy consumption. This improves the overall stability, executability, and actual mission completion quality of the entire UAV swarm linkage control process.

[0142] In one embodiment, S5: Based on the linkage control result, control each UAV to complete the cooperative flight within the current control cycle, update the overall state information of the UAV swarm after the current control cycle ends, and then reconstruct the unidirectional hierarchical structure and periodic frozen reference surface corresponding to the next control cycle based on the updated overall state information, so as to enter the linkage control process of the next control cycle.

[0143] The coordinated control results of each UAV within the current control cycle are used as the execution control input for the corresponding UAV within the current control cycle, and each UAV is controlled to complete the cooperative flight within the current control cycle according to the corresponding execution control input; that is, steps S1-S4 are repeated.

[0144] At the end of the current control cycle, the current position vector and current velocity vector of each UAV are collected respectively, and the current position vector and current velocity vector of each UAV at the end of the current control cycle are jointly determined as the updated state information of the entire group;

[0145] Then, based on the updated state information of the entire swarm, the geometric center position of the UAV swarm is recalculated, and the mission advance direction vector for the next control cycle is determined according to the mission target point position vector and the geometric center position. After normalizing the mission advance direction vector, the unit mission advance direction vector for the next control cycle is obtained.

[0146] Then, the position vector of each UAV is projected onto the unit task propulsion direction vector to obtain the propulsion projection value corresponding to each UAV, and a comprehensive adjacency matrix is ​​constructed by combining the formation adjacency relationship and the communication adjacency relationship;

[0147] Based on the propulsion projection values ​​of each UAV and the comprehensive adjacency matrix, the first layer of node set for the next control cycle is first determined, and then the node set of the previous layer is extended layer by layer according to the adjacency relationship of the node set and the propulsion projection value constraints to obtain the unidirectional hierarchical structure corresponding to the next control cycle.

[0148] After reconstructing the unidirectional hierarchical structure for the next control cycle, the position vector of each UAV at the end of the current control cycle is determined as the frozen reference position of the corresponding UAV in the next control cycle, and the velocity vector of each UAV at the end of the current control cycle is determined as the frozen reference velocity of the corresponding UAV in the next control cycle. Based on the frozen reference positions of each UAV, the frozen center of the entire swarm is calculated. Then, combined with the unidirectional hierarchical structure of the next control cycle, the frozen center position of each layer, the frozen unit propulsion direction vector, the frozen reference velocity of each layer, and the frozen interlayer distance between adjacent layers are calculated respectively, which together constitute the periodic frozen reference surface for the next control cycle.

[0149] Finally, the reconstructed unidirectional hierarchical structure and the periodically frozen reference surface are jointly determined as the basic constraints for the coordinated control of the UAV swarm in the next control cycle, so as to enter the coordinated control process of the next control cycle.

[0150] It should be noted that the linkage control results obtained through steps S1 to S4 within the current control cycle are first actually sent to each UAV for execution, enabling each UAV to complete position adjustment, speed adjustment, and formation coordination according to the corresponding control input within this cycle. After the end of this cycle, the actual position and speed of each UAV are collected again, and these executed real states are used as the updated swarm state information. Subsequently, based on the updated swarm state, the swarm geometric center and the mission advancement direction of the next control cycle are recalculated, and the hierarchical relationship and effective linkage relationship of each UAV along the mission advancement direction are re-evaluated, thereby reconstructing the unidirectional hierarchical structure corresponding to the next control cycle. After the new hierarchical structure is determined, the actual position and speed at the end of this cycle are determined as the frozen reference position and frozen reference speed for the next control cycle, and the frozen swarm center, the frozen center of each layer, the frozen advancement direction, the frozen reference speed of each layer, and the frozen inter-layer distance are recalculated accordingly to form the periodic frozen reference surface for the next control cycle. Finally, the reconstructed unidirectional hierarchical structure and periodic frozen reference surface are used as the basis for executing steps S3 and S4 in the next control cycle. The fundamental constraint is that this approach ensures that the linkage correction within each control cycle only applies to that cycle. It prevents minor deviations caused by local wind disturbances, obstacle avoidance maneuvers, communication delays, or state errors within a particular UAV from being directly used as dynamic references for other UAVs to continuously follow within the same cycle. Instead, these deviations are transformed into the input basis for re-layering and refreezing the reference in the next cycle. This breaks down the closed-loop influence chain that would easily form within the same control cycle into an orderly update process across cycles. Furthermore, while retaining multi-target collaborative control capabilities, it effectively suppresses the closed-loop backflow of control influence within the same cycle, reduces the risk of local disturbances circulating and amplifying within the swarm, avoids inducing swarm control loop resonance, reduces formation trajectory swaying, repeated control corrections, and abnormal energy consumption, and improves the overall stability and continuity of UAV swarm linkage control.

[0151] Based on the same inventive concept, embodiments of the present invention also provide a multi-objective optimization-based unmanned aerial vehicle (UAV) swarm coordinated control system. This includes:

[0152] Structure building module: Obtains the position status, speed status, formation adjacency, communication adjacency, and current mission direction of each UAV in the UAV swarm, and constructs the unidirectional hierarchical structure of the UAV swarm within the current control cycle;

[0153] Basic constraint module: Based on the unidirectional hierarchical structure, the periodic freezing reference surface corresponding to the current control cycle is generated based on the full group state information at the end of the previous control cycle, so that the periodic freezing reference surface and the unidirectional hierarchical structure together serve as the basic constraints for the UAV swarm linkage control in the current control cycle.

[0154] Correction module: Based on a unidirectional hierarchical structure and a periodically frozen reference surface, the control correction relationship of each UAV layer is determined sequentially from the upstream layer to the downstream layer. This enables the downstream UAV to complete the control correction of the current layer according to the periodically frozen reference surface and the correction results transmitted from the upstream layer, and restricts the control correction results of the current layer from being transmitted back to the upstream layer within the current control cycle to obtain the correction result.

[0155] Target optimization module: Based on the control correction results, under the constraints of a unidirectional hierarchical structure and a periodically frozen reference surface, multi-objective optimization is performed on each layer of UAVs to obtain the linkage control results of each UAV in the current control cycle. The multi-objective optimization includes mission advancement objectives, formation maintenance objectives, obstacle avoidance and safety objectives, communication connectivity objectives, and energy consumption control objectives.

[0156] Control module: Based on the linkage control results, it controls each UAV to complete the coordinated flight within the current control cycle, and updates the overall state information of the UAV swarm after the current control cycle ends. Then, based on the updated overall state information, it reconstructs the unidirectional hierarchical structure and the cycle-freezing reference surface corresponding to the next control cycle, so as to enter the linkage control process of the next control cycle.

[0157] Based on the multi-objective optimization-based UAV swarm linkage control system provided by this invention, a unidirectional hierarchical structure is constructed within the current control cycle. Then, based on the swarm state information at the end of the previous control cycle, a periodic frozen reference surface corresponding to the current control cycle is generated. This prevents the linkage reference relationship of the UAV swarm within the current control cycle from unfolding in a real-time bidirectional mutual following manner between nodes. Instead, it uses the unidirectional hierarchical structure and the periodic frozen reference surface as a common constraint basis, sequentially completing control corrections from the upstream layer to the downstream layer. Furthermore, it restricts the downstream layer's correction results from being transmitted back to the upstream layer within the current control cycle, thus ensuring that the control effects caused by local disturbances can only unfold unidirectionally along the unidirectional hierarchical path. Opening up makes it difficult to form a closed propagation chain and return to the original node within the same control cycle. On this basis, multi-objective optimization is performed by combining mission advancement objectives, formation maintenance objectives, obstacle avoidance safety objectives, communication connectivity objectives, and energy consumption control objectives. This not only retains the ability of UAV swarm collaborative control to comprehensively consider multiple performance indicators, but also effectively suppresses the control effects caused by local wind disturbances, obstacle avoidance actions, communication delays, or state deviations from circulating and continuously amplifying within the swarm. In this way, it can effectively avoid the swarm control loop resonance phenomenon, reduce formation trajectory sway and repeated control corrections, reduce abnormal energy consumption, and improve the overall stability, coordination, and mission execution reliability in the UAV swarm linkage control process.

[0158] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention should still fall within the scope of the claims of the present invention.

Claims

1. A method for coordinated control of unmanned aerial vehicle (UAV) swarms based on multi-objective optimization, characterized in that, Includes the following steps: The system acquires the position, speed, formation adjacency, communication adjacency, and current mission direction of each UAV in the UAV swarm, and constructs a unidirectional hierarchical structure of the UAV swarm within the current control cycle. Based on the unidirectional hierarchical structure, a periodic freezing reference surface corresponding to the current control cycle is generated based on the full group state information at the end of the previous control cycle, so that the periodic freezing reference surface and the unidirectional hierarchical structure together serve as the basic constraints for the UAV swarm linkage control in the current control cycle. Based on the unidirectional hierarchical structure and the periodic frozen reference surface, the control correction relationship of each UAV layer is determined sequentially from the upstream layer to the downstream layer. This enables the downstream UAV to complete the control correction of the current layer according to the periodic frozen reference surface and the correction result transmitted from the upstream layer, and restricts the control correction result of the current layer from being transmitted back to the upstream layer within the current control cycle to obtain the correction result. Based on the control correction results, under the constraints of a unidirectional hierarchical structure and a periodically frozen reference surface, multi-objective optimization is performed on each layer of UAVs to obtain the linkage control results of each UAV in the current control cycle. The multi-objective optimization includes mission advancement objectives, formation maintenance objectives, obstacle avoidance and safety objectives, communication connectivity objectives, and energy consumption control objectives. Based on the linkage control results, each UAV is controlled to complete the coordinated flight within the current control cycle. After the current control cycle ends, the overall state information of the UAV swarm is updated. Then, based on the updated overall state information, the unidirectional hierarchical structure and the cycle-freezing reference surface corresponding to the next control cycle are reconstructed to enter the linkage control process of the next control cycle.

2. The UAV swarm coordinated control method based on multi-objective optimization according to claim 1, characterized in that, The steps to construct a unidirectional hierarchical structure for a drone swarm within the current control cycle are as follows: Obtain the position vector of each UAV in the UAV swarm, and calculate the current mission propulsion direction vector based on the mission target point position and the geometric center position of the UAV swarm, and perform normalization processing to obtain the unit mission propulsion direction vector; Project the position vector of each UAV onto the propulsion direction vector of the unit task to obtain the propulsion projection value of each UAV; A comprehensive adjacency matrix is ​​constructed based on the formation adjacency matrix and the communication adjacency matrix. Based on the propulsion projection value of each UAV and the comprehensive adjacency matrix, the first layer of nodes with predetermined adjacency coverage capability at the propulsion front is determined. The set of nodes is expanded layer by layer according to the adjacency relationship with the node set of the previous layer and the propulsion projection value constraint, to obtain a hierarchical node set consisting of the first layer to the Mth layer. The control effects between adjacent layers are limited to propagation along the increasing layer number direction, forming a unidirectional hierarchical structure within the current control cycle.

3. The UAV swarm coordinated control method based on multi-objective optimization according to claim 2, characterized in that, The steps for using the periodically frozen reference surface and the unidirectional hierarchical structure together as the basic constraints for the coordinated control of the UAV swarm within the current control cycle are as follows: Obtain the position vector and velocity vector of each UAV in the UAV swarm at the end of the previous control cycle, and use them as the frozen reference position and frozen reference velocity of the corresponding UAV in the current control cycle. The freezing center of the entire swarm is calculated based on the freezing reference position of each UAV, and then the freezing center position corresponding to the node set of each layer is calculated by combining the unidirectional hierarchical structure. Calculate the frozen main propulsion direction vector for the current control cycle based on the mission target point position vector and the frozen group center, and normalize the frozen main propulsion direction vector to obtain the frozen unit propulsion direction vector; The distance between adjacent frozen layers is calculated based on the freezing center location of adjacent layers, and the freezing reference velocity of each layer is calculated based on the freezing reference velocity of each UAV in each layer, and the basic constraints are determined.

4. The UAV swarm coordinated control method based on multi-objective optimization according to claim 3, characterized in that, The steps to determine the basic constraints are as follows: The frozen reference position of each UAV, the frozen reference speed of each UAV, the frozen center of the entire group, the frozen center position of each layer, the frozen unit's propulsion direction vector, the frozen reference speed of each layer, and the frozen layer distance between adjacent layers are collectively constituted as the periodic frozen reference surface of the current control cycle. The periodic freezing reference surface and the unidirectional hierarchical structure are jointly determined as the basic constraints for the coordinated control of the UAV swarm within the current control cycle.

5. The UAV swarm coordinated control method based on multi-objective optimization according to claim 4, characterized in that, To restrict the control correction results of the current layer from being propagated back to the upstream layer within the current control cycle, the steps to obtain the correction results are as follows: For each UAV in the first layer, calculate the position deviation vector between its frozen reference position and current position, and the velocity deviation vector between its frozen reference velocity and current velocity. Then, add the position deviation vector, velocity deviation vector and frozen unit thrust direction vector and perform modulus normalization to obtain the control correction amount for each UAV in the first layer. The layer correction center amount of the first layer is calculated based on the control correction amount of each UAV in the first layer; then, for each UAV in any downstream layer, the position deviation vector between the frozen reference position and the current position, the velocity deviation vector between the frozen reference speed and the current speed, and the actual distance between the current position of the UAV and the frozen center of the previous layer are calculated respectively, and the interlayer distance deviation is calculated based on the actual distance and the corresponding interlayer distance. The correction results are further determined based on the control correction values ​​of each UAV in the first layer and the inter-layer distance deviation.

6. The UAV swarm coordinated control method based on multi-objective optimization according to claim 5, characterized in that, The steps to further determine the correction results based on the control correction values ​​of each UAV in the first layer and the inter-layer distance deviation are as follows: The position deviation vector, velocity deviation vector, layer correction center value of the previous layer, and inter-layer distance deviation correction term along the direction of frozen unit advance are added together and then normalized to obtain the control correction value of each UAV in the current downstream layer. The control correction values ​​for all UAVs within the current control cycle are calculated layer by layer from the first layer to the last layer. Within the current control cycle, only the layer correction center values ​​of the previous layer are allowed to participate in the calculation of the control correction values ​​of the next layer, and the control correction values ​​of the next layer are not allowed to participate in the recalculation of the previous layer. The set of control correction values ​​is output as the correction result of the current control cycle.

7. The UAV swarm coordinated control method based on multi-objective optimization according to claim 6, characterized in that, The steps for performing multi-objective optimization on each layer of UAVs to obtain the coordinated control results of each UAV within the current control cycle are as follows: Based on the unidirectional hierarchical structure, the generated periodically frozen reference surface, and the obtained set of control correction results; Based on the current control cycle duration, establish the predicted position and predicted velocity of each UAV at the end of the current control cycle; The control inputs of each UAV are limited to control inputs along the direction of the corresponding control correction result; based on the predicted position and control inputs, mission advancement objectives, formation maintenance objectives, obstacle avoidance and safety objectives, communication connectivity objectives, and energy consumption control objectives are constructed respectively.

8. The UAV swarm coordinated control method based on multi-objective optimization according to claim 7, characterized in that, The mission advancement objectives, formation maintenance objectives, obstacle avoidance and safety objectives, communication connectivity objectives, and energy consumption control objectives are as follows: The mission advancement target is used to characterize the distance between the predicted position of each UAV and the mission target point; the formation maintenance target is used to characterize the deviation between the predicted distance and the expected formation distance between UAVs with formation adjacency; the obstacle avoidance safety target is used to characterize the distance between the predicted position of each UAV and each obstacle; the communication connectivity target is used to characterize the overrange amount when the predicted distance between UAVs with communication adjacency exceeds the effective communication distance threshold; and the energy consumption control target is used to characterize the modulus of the control input of each UAV. The multi-objective optimization is performed sequentially according to the constraints, following the order of obstacle avoidance and safety objectives, communication connectivity objectives, formation maintenance objectives, mission advancement objectives, and finally energy consumption control objectives. After each objective is solved, the current optimal feasible solution set for that objective is used as the basis for solving the next objective.

9. The UAV swarm coordinated control method based on multi-objective optimization according to claim 8, characterized in that, The constraints are as follows: During the solution process, the following constraints are satisfied: the control input of the (m+1)th layer is determined only based on the control correction result of the mth layer and is not reversed based on the result of the downstream layer. The control input of each UAV is solved with the corresponding freezing reference position, freezing reference speed, freezing main propulsion direction, freezing center of each layer and distance between freezing layers as reference benchmarks. The control input direction of each UAV is consistent with the direction of the corresponding control correction result and satisfies the preset maximum control input amplitude constraint. The optimized solution yields the control input combinations for each UAV, which are then output as the coordinated control results for each UAV within the current control cycle.

10. A multi-objective optimization-based unmanned aerial vehicle (UAV) swarm coordinated control system, used to implement the multi-objective optimization-based UAV swarm coordinated control method according to any one of claims 1-9, characterized in that, The system includes: Structure building module: Obtains the position status, speed status, formation adjacency, communication adjacency, and current mission direction of each UAV in the UAV swarm, and constructs the unidirectional hierarchical structure of the UAV swarm within the current control cycle; Basic constraint module: Based on the unidirectional hierarchical structure, the periodic freezing reference surface corresponding to the current control cycle is generated based on the full group state information at the end of the previous control cycle, so that the periodic freezing reference surface and the unidirectional hierarchical structure together serve as the basic constraints for the UAV swarm linkage control in the current control cycle. Correction module: Based on a unidirectional hierarchical structure and a periodically frozen reference surface, the control correction relationship of each UAV layer is determined sequentially from the upstream layer to the downstream layer. This enables the downstream UAV to complete the control correction of the current layer according to the periodically frozen reference surface and the correction results transmitted from the upstream layer, and restricts the control correction results of the current layer from being transmitted back to the upstream layer within the current control cycle to obtain the correction result. Target optimization module: Based on the control correction results, under the constraints of a unidirectional hierarchical structure and a periodically frozen reference surface, multi-objective optimization is performed on each layer of UAVs to obtain the linkage control results of each UAV in the current control cycle. The multi-objective optimization includes mission advancement objectives, formation maintenance objectives, obstacle avoidance and safety objectives, communication connectivity objectives, and energy consumption control objectives. Control module: Based on the linkage control results, it controls each UAV to complete the coordinated flight within the current control cycle, and updates the overall state information of the UAV swarm after the current control cycle ends. Then, based on the updated overall state information, it reconstructs the unidirectional hierarchical structure and the cycle-freezing reference surface corresponding to the next control cycle, so as to enter the linkage control process of the next control cycle.

Images