A UAV cluster collaborative control method and system

Abstract

The invention discloses a method and system for cooperative control of a drone cluster, and relates to the technical field of drone control. The method and system comprise the following steps: constructing a drone cluster topology structure; allocating the drones to corresponding clusters according to the positions of each drone, and determining a main cluster head and a secondary cluster head; in the drone cluster, disconnecting direct communication links between member drones in the cluster and direct communication links with all drones in other clusters, setting a cluster boundary range centered on the main cluster head, transmitting task data to the main cluster head and the secondary cluster head at the same time when executing a task, and when the main cluster head fails, the secondary cluster head takes over and reselects the secondary cluster head; using Kalman filtering to predict the motion state of the drone, and using an isolation forest algorithm to calculate an abnormality score, and if the abnormality score exceeds a preset score threshold, determining that the cluster is abnormal and readjusting the cluster structure, so as to avoid instability of the entire drone cluster due to the failure of an individual drone.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle control technology, and in particular to a method and system for cooperative control of a cluster of unmanned aerial vehicles. Background Art

[0002] With the rapid development of drone technology, drone swarm systems have shown great application potential in military reconnaissance, disaster monitoring, logistics distribution, agricultural plant protection and other fields due to their high efficiency, flexibility and collaborative operation. The efficient collaborative operation of drone swarms depends on advanced control technology and communication strategies. Most drone swarm control methods adopt centralized control, distributed control or hybrid control.

[0003] The centralized control method uses a central node to uniformly schedule and coordinate the behavior of the drone cluster. Although it can achieve global optimization, the failure of the central node will cause the entire system to paralyze. The distributed control method relies on local communication and collaboration between drones. Although it improves the robustness of the system, it lacks a global perspective and may lead to local optimization rather than global optimization. The hybrid control method performs global planning and scheduling through a central node or core controller, while allowing local communication and collaboration between drones. This fusion enables the system to maintain a global perspective while also having the ability to respond quickly to the local environment.

[0004] However, whether it is centralized control, distributed control or hybrid control, there is still a significant limitation: the failure or failure of any key node may trigger a chain reaction, resulting in the obstruction of the operation of the entire cluster, or at least causing some drones to malfunction, thus seriously affecting the stability and reliability of the drone cluster. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the shortcomings of the prior art, the present invention provides a method and system for collaborative control of drone clusters, which realizes efficient, stable and flexible collaborative control of drone clusters by dynamically constructing the topological structure of the drone cluster, adaptively selecting cluster heads and sub-cluster heads, and real-time monitoring and adjusting the cluster structure. This method not only improves the stability and reliability of the drone cluster, but also enhances the robustness and scalability of the system, providing a new solution for the collaborative operation of drone clusters in complex dynamic environments.

[0007] (II) Technical solution

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for cooperative control of a drone cluster, comprising the following steps:

[0009] Collect the location information and communication capability information of the drones in the drone cluster, regard each drone as a node in the communication network, and build the drone cluster topology structure;

[0010] Create an empty cluster information list, randomly select the initial main cluster head, assign the drones to the corresponding clusters according to the location of each drone, determine the main cluster head and the secondary cluster head, and update the cluster information list until all drones are assigned to the cluster;

[0011] In the UAV cluster, the direct communication links between the member UAVs in the cluster and the direct communication links with all the UAVs in other clusters are disconnected, and the cluster boundary range centered on the main cluster head is set. When executing a task, the task data is transmitted to the main cluster head and the secondary cluster head at the same time. When the main cluster head fails, the secondary cluster head takes over and reselects the secondary cluster head.

[0012] For each cluster, the dynamic real-time data of all drones in the cluster are collected, the Kalman filter is used to predict the drone's motion state, and the isolation forest algorithm is used to calculate the anomaly score. If the anomaly score exceeds the preset score threshold, the cluster is judged to be abnormal and the cluster structure is readjusted. If it does not exceed the threshold, it is judged to be normal and no processing is performed.

[0013] Furthermore, the topological structure of the drone cluster is constructed, including: assigning a unique identifier to each node, creating a node list containing all nodes and their basic attributes, the basic attributes including location, ID, and communication capability information, and calculating the distance between each pair of drones based on the location information of the drones;

[0014] According to the distance and communication range between each pair of drones, it is determined whether there is a direct communication link between the drones: if the distance between the two drones is less than or equal to the communication range between the two, it is considered that there is a direct communication link between the two, otherwise, it is considered that there is no direct communication link between the two;

[0015] Traverse all drones, build a communication link matrix based on the judgment results of the communication link, use graph theory methods to add the nodes in the node list to the graph, add edges to the graph based on the communication links between drones in the communication link matrix, and connect the corresponding nodes to build the drone cluster topology.

[0016] Furthermore, the distance between each pair of drones is calculated: ,in, represents the distance between UAV i and UAV j, and Represent the 3D coordinates of drone i and j respectively.

[0017] Furthermore, an empty cluster information list is created, including: creating an empty cluster information list for storing information of all clusters, the information of each cluster including cluster ID, primary cluster head ID, secondary cluster head ID and cluster member ID list, initializing the cluster state of each drone to: unassigned, indicating that the drone has not been assigned to any cluster.

[0018] Furthermore, according to the position of each drone, the drone is assigned to the corresponding cluster, including: randomly selecting a drone as the initial main cluster head, traversing the remaining drones, calculating the distance between each drone and the main cluster head, if the distance is less than a preset cluster radius, the drone is assigned to the cluster where the main cluster head is located, and the cluster state of the drone is updated to: allocated, checking the size of each cluster, if the number of members of any cluster is less than the preset number of cluster members, the cluster state of the cluster members of the cluster is updated to: unallocated, and re-selecting the main cluster head and assigning members for the cluster.

[0019] Furthermore, for each cluster, from the allocated cluster members, the cluster members whose distance from the main cluster head is within the preset distance range and whose communication link quality is the highest are selected as the secondary cluster heads. The preset distance satisfies the formula: ,in, Indicates the preset distance. The communication range of drone i, Rc represents the cluster radius, and the communication link quality formula is: ,in, represents the communication link quality of UAV i, M represents the number of communication parameters, represents the value of the fth communication parameter between UAV i and UAV j, represents the weight of the fth communication parameter.

[0020] Furthermore, each cluster is traversed, and for each pair of member drones in the cluster, the direct communication link between the two is disconnected by modifying the communication link matrix. The member drones are drones that do not include the main cluster head and the secondary cluster head. Each cluster is continued to be traversed, and for each member drone in the cluster, the communication link matrix is ​​modified to disconnect the communication link between it and all drones in other clusters. All drones in other clusters include the main cluster head, the secondary cluster head and the member drones.

[0021] Furthermore, for each cluster, the Kalman filter algorithm is used to combine the dynamic real-time data of the drone, including position information, speed information, and acceleration information, to predict the motion state of the drone in real time. The calculation formula is as follows:

[0022] ;

[0023] ;

[0024] ;

[0025] in, represents the predicted vector state, k represents the time, represents the probability of the nth state transferring to the mth state in the state transfer matrix A, represents the optimal estimated state vector at the previous moment, represents the control input at the previous moment, Represents the control input in the control input matrix B The nth component of the state vector The influence coefficient of the mth component of , m represents the number of rows of the matrix, and n represents the number of columns of the matrix.

[0026] Furthermore, the Isolation Forest algorithm is used to calculate the anomaly score of each cluster by taking the predicted motion state of the drone as input: ,in, represents the abnormal score of sample g, k represents the number of samples, h(g) represents the path length that sample g needs to go through from the root node to the leaf node, represents the expected value of the path length of sample g under h(g), and c(k) represents the average path length of the binary search tree.

[0027] A drone cluster collaborative control system, comprising:

[0028] The cluster topology building module collects the location information and communication capability information of the drones in the drone cluster, regards each drone as a node in the communication network, and builds the drone cluster topology structure;

[0029] The cluster partitioning module creates an empty cluster information list, randomly selects the initial main cluster head, assigns the drones to the corresponding clusters according to the location of each drone, determines the main cluster head and the secondary cluster head, and updates the cluster information list until all drones are assigned to the cluster;

[0030] The communication link adjustment module disconnects the direct communication links between the member drones in the cluster and the direct communication links with all drones in other clusters in the drone cluster, sets the cluster boundary range centered on the main cluster head, and transmits the task data to the main cluster head and the secondary cluster head at the same time when executing the task. When the main cluster head fails, the secondary cluster head takes over and reselects the secondary cluster head.

[0031] For each cluster, the control module collects the dynamic real-time data of all drones in the cluster, uses Kalman filtering to predict the motion status of the drone, and uses the isolation forest algorithm to calculate the anomaly score. If the anomaly score exceeds the preset score threshold, the cluster is judged to be abnormal and the cluster structure is readjusted. If it does not exceed the threshold, it is judged to be normal and no processing is performed.

[0032] (III) Beneficial effects

[0033] The present invention provides a method and system for cooperative control of a drone cluster, which has the following beneficial effects:

[0034] (1) In the process of coordinated control of UAV clusters, by constructing a communication link matrix and topological structure, the position and communication status of UAVs can be monitored in real time, so that each UAV has a clear communication relationship with other UAVs, which helps to quickly transmit information during the coordinated control process, reduce communication delays and errors, and improve the accuracy and efficiency of coordinated control.

[0035] (2) By assigning UAVs to different clusters and determining the main cluster head and the secondary cluster head, a hierarchical control structure can be formed. This structure helps to quickly take over the responsibilities of the secondary cluster head or other UAVs when a single UAV or cluster head fails, thereby maintaining the overall stability and reliability of the cluster.

[0036] (3) By disconnecting the direct communication links between member drones in the cluster and with drones in other clusters, communication redundancy and conflicts can be reduced, thereby improving the communication efficiency of the entire drone cluster and effectively reducing unnecessary communication resource consumption. Setting the cluster boundary range centered on the main cluster head helps to clarify the scope of authority and responsibility of each cluster and avoid confusion and conflicts during task execution. When the main cluster head fails, since the secondary cluster head has established a communication link with other member drones in the cluster, the secondary cluster head can be seamlessly connected, quickly take over the main cluster head and reselect a new secondary cluster head, ensuring that the operation of the cluster will not be interrupted by a single point of failure, thereby enhancing the stability and reliability of the cluster.

[0037] (4) By collecting the dynamic real-time data of all drones in the cluster and using Kalman filtering to predict the motion status of the drones, abnormal behaviors or failures of drones can be discovered in a timely manner. When the abnormality score exceeds the preset score threshold, it can be determined that there is an abnormality in the cluster and the cluster structure can be readjusted, thereby avoiding instability of the entire cluster due to the failure of individual drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a schematic diagram of the steps of the UAV cluster collaborative control method of the present invention;

[0039] Figure 2 This is a schematic diagram of the structure of the drone cluster of the present invention;

[0040] Figure 3 It is a schematic diagram of the structure of the UAV cluster collaborative control system of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] See also Figure 1-Figure 2 The present invention provides a method for cooperative control of a drone cluster, comprising the following steps:

[0043] Step 1: Collect the location information and communication capability information of the drones in the drone cluster, regard each drone as a node in the communication network, and build the drone cluster topology structure;

[0044] The step 1 comprises the following steps:

[0045] Step 101: collect the location information of each drone in the drone cluster through the GPS module or other positioning equipment carried by the drone, and collect the communication capability information between the drones through the communication system, including communication range, communication speed, signal stability, etc.;

[0046] Step 102: Consider each drone as a node in the communication network, assign a unique identifier (such as an ID) to each node, create a node list containing all nodes and their basic attributes (such as location, ID, communication capability, etc.), and calculate the distance between each pair of drones using a distance calculation formula (such as Euclidean distance) based on the location information of the drones;

[0047] Step 103: judging whether there is a direct communication link between the drones according to the distance and communication range between each pair of drones: if the distance between the two drones is less than or equal to the communication range between the two drones, it is considered that there is a direct communication link between them; otherwise, it is considered that there is no direct communication link between them;

[0048] Step 104: Traverse all drones, build a communication link matrix based on the judgment results of the communication link, use graph theory to add the nodes in the node list to the graph, add edges to the graph based on the communication links between drones in the communication link matrix, connect the corresponding nodes, and build a drone cluster topology structure;

[0049] It should be noted that the communication link matrix is ​​a two-dimensional matrix, in which the elements of the matrix represent whether there is a direct communication link between two drones. After the communication link matrix is ​​constructed, the communication relationship in the drone cluster is represented as a topological structure using graph theory. The nodes in the topological structure represent drones, and the edges represent the communication links between drones.

[0050] When using, combine the contents of step 101 to step 104:

[0051] In the process of collaborative control of drone clusters, by constructing a communication link matrix and topological structure, the position and communication status of drones can be monitored in real time, so that each drone has a clear communication relationship with other drones, which helps to quickly transmit information during the collaborative control process, reduce communication delays and errors, and improve the accuracy and efficiency of collaborative control.

[0052] Step 2: Create an empty cluster information list, randomly select the initial main cluster head, assign the drones to the corresponding clusters according to the location of each drone, determine the main cluster head and the secondary cluster head, and update the cluster information list until all drones are assigned to the cluster;

[0053] The step 2 comprises the following steps:

[0054] Step 201: Create an empty cluster information list to store the information of all clusters. The information of each cluster includes the cluster ID, the main cluster head ID, the secondary cluster head ID and the cluster member ID list. Initialize the cluster state of each drone to: unassigned, indicating that the drone has not been assigned to any cluster.

[0055] Step 202: Calculate the distance between each pair of drones: ,in, represents the distance between UAV i and UAV j, and Represent the three-dimensional coordinates of drones i and j respectively;

[0056] Step 203: Select a drone as the initial main cluster head by random selection or by using a weight-based selection (e.g., selecting the node with the highest score as the cluster head according to the performance score) or a geographic location-based selection (e.g., selecting the node at the center of the cluster as the cluster head). The cluster head is responsible for managing and coordinating the communication and task allocation within the cluster.

[0057] Step 204: traverse the remaining drones, calculate the distance between each drone and the main cluster head, if the distance is less than the preset cluster radius, assign the drone to the cluster where the main cluster head is located, and update the cluster state of the drone to: assigned, check the size of each cluster, if the number of members of any cluster is less than the preset number of cluster members, update the cluster state of the cluster members of the cluster to: unassigned, and reselect the main cluster head and assign members for the cluster;

[0058] It should be noted that the setting of cluster radius needs to take into account multiple factors. First, the cluster radius should be large enough to ensure effective communication and collaboration between drones in the same cluster. If the cluster radius is too small, communication between drones may be restricted, affecting the overall performance of the cluster.

[0059] However, the cluster radius cannot be set too large and cannot be larger than the communication range of the cluster head. If the cluster radius is too large, it may cause too many drones to be included in a cluster, which increases the difficulty of management and coordination. In addition, too large a cluster radius may also cause communication interference and conflicts between drones, and the cluster head cannot establish a communication link with each drone in the cluster, further affecting the stability and reliability of the cluster.

[0060] The setting of cluster radius needs to be comprehensively considered based on multiple factors such as the performance, communication capability, mission requirements and cluster size of the UAV. The optimal cluster radius setting should be determined through experiments and simulations to ensure that the UAV cluster can operate efficiently and stably.

[0061] Step 205: For each cluster, from the allocated cluster members, select the cluster member whose distance from the main cluster head is within the preset distance range and the drone with the highest communication link quality as the secondary cluster head. The preset distance satisfies the formula: ,in, Indicates the preset distance. The communication range of drone i, Rc represents the cluster radius, and the communication link quality formula is: ,in, represents the communication link quality of UAV i, M represents the number of communication parameters, represents the value of the fth communication parameter between UAV i and UAV j, represents the weight of the fth communication parameter;

[0062] It should be noted that the setting of the preset distance needs to balance multiple factors. First, the preset distance should be less than the communication range minus the cluster radius to ensure that the secondary cluster head can establish a communication link with all drones in the cluster. Second, the preset distance should be close enough to ensure that the secondary cluster head can respond quickly and take over its duties when the main cluster head fails or needs assistance. If the preset distance is too far, the secondary cluster head may not be able to communicate and collaborate with the main cluster head in a timely and effective manner, thereby affecting the overall performance and reliability of the cluster. On the other hand, the preset distance cannot be set too close. If the distance is too close, there may be too few drones to choose from, and it may be impossible to screen out a suitable secondary cluster head, thereby reducing the independence and reliability of the secondary cluster head as a candidate leader to some extent.

[0063] The setting of the preset distance needs to be comprehensively considered based on multiple factors such as the performance, communication capability, mission requirements, and cluster size of the UAV. A reasonable preset distance range can be determined through experiments, simulations, or judgment based on experience. This range should ensure that the secondary cluster head can respond quickly when the primary cluster head needs it, while maintaining a certain degree of independence and reliability.

[0064] The setting of weights reflects the importance of communication parameters in practical applications. A certain balance needs to be maintained to avoid excessive weights of one or several parameters, which may lead to the neglect of other parameters. The value of weights can be determined by methods such as data analysis and hierarchical analysis.

[0065] Step 206: Update the information of the main cluster head, the secondary cluster head and the members in each cluster to the cluster information list. If there are still unassigned drones, repeat steps 203 to 205 until all drones are assigned to the cluster;

[0066] When used, combine the contents of step 201 to step 206:

[0067] By assigning UAVs to different clusters and determining the main cluster head and the deputy cluster head, a hierarchical control structure can be formed. This structure helps to quickly take over the responsibilities of the deputy cluster head or other UAVs when a single UAV or cluster head fails, thereby maintaining the overall stability and reliability of the cluster.

[0068] Step 3: In the UAV cluster, disconnect the direct communication links between the member UAVs in the cluster and the direct communication links with all UAVs in other clusters, set the cluster boundary range centered on the main cluster head, and transmit the task data to the main cluster head and the secondary cluster head at the same time when executing the task. When the main cluster head fails, the secondary cluster head takes over and reselects the secondary cluster head;

[0069] The step three includes the following contents:

[0070] Step 301: for the currently constructed drone cluster topology, initialize a list for recording the communication link adjustment status, traverse each cluster, and for each pair of member drones in the cluster, the member drones are drones that do not include the main cluster head and the secondary cluster head, disconnect the direct communication link between the two by modifying the communication link matrix or directly deleting the corresponding edge in the topology structure;

[0071] Step 302: Continue to traverse each cluster, and for each member drone in the cluster, disconnect the communication link between it and all drones in all other clusters (including the main cluster head, secondary cluster head and member drones of other clusters) by modifying the communication link matrix or directly deleting the corresponding edge in the topology structure;

[0072] Step 303: Based on the position of the main cluster head and the preset communication range (effective communication distance after considering factors such as communication attenuation and environmental interference), a boundary range is set for each cluster with the main cluster head as the center: ,in, Indicates the radius of the boundary range, represents the communication range, λ represents the attenuation coefficient;

[0073] It should be noted that λ is a value less than or equal to 1, which is used to consider the influence of factors such as communication attenuation and environmental interference on the communication range. This coefficient can be determined through simulation experiments, field tests or reference to empirical data. The boundary range is used to limit the activity range of drones in each cluster to ensure that drones can effectively communicate and execute tasks within the cluster.

[0074] Step 304: When a task needs to be executed, the task data is transmitted to the main cluster head and the secondary cluster head at the same time. The main cluster head is responsible for decomposing the task into subtasks and assigning them to the member drones in the cluster for execution. When the main cluster head fails, the secondary cluster head immediately moves to the position of the main cluster head and becomes the new main cluster head, taking over the responsibility of task dispatching and intra-cluster communication coordination. At the same time, the mechanism of reselecting the secondary cluster head is triggered, and a new secondary cluster head is selected from the remaining cluster members by the method in step 205.

[0075] When used, combine the contents of step 301 to step 304:

[0076] By disconnecting the direct communication links between the member drones in the cluster and with the drones in other clusters, communication redundancy and conflicts can be reduced, thereby improving the communication efficiency of the entire drone cluster and effectively reducing unnecessary communication resource consumption. Setting the cluster boundary range centered on the main cluster head helps to clarify the scope of rights and responsibilities of each cluster and avoid confusion and conflicts during task execution. When the main cluster head fails, since the secondary cluster head has established a communication link with the member drones in the other clusters, the secondary cluster head can be seamlessly connected, quickly take over the main cluster head and reselect a new secondary cluster head to ensure that the operation of the cluster will not be interrupted by a single point of failure, thereby enhancing the stability and reliability of the cluster.

[0077] Step 4: For each cluster, collect dynamic real-time data of all drones in the cluster, use Kalman filtering to predict the drone's motion state, and use the isolation forest algorithm to calculate the anomaly score. If the anomaly score exceeds the preset score threshold, the cluster is judged to be abnormal and the cluster structure is readjusted. If it does not exceed the threshold, it is judged to be normal and no processing is required.

[0078] The step 4 comprises the following steps:

[0079] Step 401: For each cluster, obtain the dynamic real-time data such as the position information, velocity information, and acceleration information of all the drones in the cluster, and use the Kalman filter algorithm to combine the dynamic real-time data such as the position information, velocity information, and acceleration information of the drones to estimate and predict the motion state of the drones in real time. The calculation formula is as follows:

[0080] ;

[0081] ;

[0082] ;

[0083] in, represents the predicted vector state, k represents the time, represents the probability of the nth state transferring to the mth state in the state transfer matrix A, represents the optimal estimated state vector at the previous moment, represents the control input at the previous moment, Represents the control input in the control input matrix B The nth component of the state vector The influence coefficient of the mth component of , where m represents the number of rows of the matrix and n represents the number of columns of the matrix;

[0084] It should be noted that Kalman filtering is a recursive algorithm that uses linear algebra and probability statistics to estimate the state of a dynamic system from observation data containing noise. It can use the dynamic real-time data such as the position, velocity and acceleration of the drone as observation data, and use the Kalman filtering algorithm to estimate and predict the motion state of the drone in real time.

[0085] Step 402: Using the isolation forest algorithm, the predicted motion state of the drone is used as input to calculate the anomaly score of each cluster: ,in, represents the abnormal score of sample g, k represents the number of samples, h(g) represents the path length that sample g needs to go through from the root node to the leaf node, represents the expected value of the path length of sample g under h(g), and c(k) represents the average path length of the binary search tree;

[0086] It should be noted that Isolation Forest is a tree-based anomaly detection algorithm that partitions data points by randomly selecting features and feature values ​​until all data points are isolated (i.e., each data point is on a leaf node). Anomalies are usually easier to isolate, so they have shorter path lengths.

[0087] Step 403: pre-set a score threshold. If the abnormality score exceeds the score threshold, it is determined that the cluster has an abnormality, and the process returns to step 2 to readjust the cluster structure. If the abnormality score does not exceed the score threshold, it is determined that the cluster has no abnormality and no processing is performed.

[0088] It should be noted that the score threshold is usually set according to actual needs and the characteristics of the data set. In practical applications, a suitable threshold can be determined by analyzing the normal and abnormal distribution of the data set. Usually, this threshold is obtained through experiments and adjustments to ensure that the algorithm can accurately detect anomalies.

[0089] When used, combine the contents of step 401 to step 403:

[0090] By collecting the dynamic real-time data of all drones in the cluster and using Kalman filtering to predict the motion status of the drones, abnormal behaviors or failures of drones can be discovered in a timely manner. When the abnormal score exceeds the preset score threshold, it can be determined that the cluster is abnormal and the cluster structure can be readjusted, thereby avoiding instability of the entire cluster due to the failure of individual drones.

[0091] See also Figure 3 The present invention also provides a UAV cluster collaborative control system, including: a cluster topology building module, a cluster division module, a communication link adjustment module and a control module; wherein,

[0092] The cluster topology building module collects the location information and communication capability information of the drones in the drone cluster, regards each drone as a node in the communication network, and builds the drone cluster topology structure;

[0093] The cluster partitioning module creates an empty cluster information list, randomly selects the initial main cluster head, assigns the drones to the corresponding clusters according to the location of each drone, determines the main cluster head and the secondary cluster head, and updates the cluster information list until all drones are assigned to the cluster;

[0094] The communication link adjustment module disconnects the direct communication links between the member drones in the cluster and the direct communication links with all drones in other clusters in the drone cluster, sets the cluster boundary range centered on the main cluster head, and transmits the task data to the main cluster head and the secondary cluster head at the same time when executing the task. When the main cluster head fails, the secondary cluster head takes over and reselects the secondary cluster head.

[0095] For each cluster, the control module collects the dynamic real-time data of all drones in the cluster, uses Kalman filtering to predict the motion status of the drone, and uses the isolation forest algorithm to calculate the anomaly score. If the anomaly score exceeds the preset score threshold, the cluster is judged to be abnormal and the cluster structure is readjusted. If it does not exceed the threshold, it is judged to be normal and no processing is performed.

[0096] In the application, the several formulas involved are all calculated by removing dimensions and taking their numerical values. The formula is a formula of the most recent real situation obtained by collecting a large amount of data and performing software simulation. The coefficients in the formula are set by technicians in this field according to actual conditions.

[0097] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. A person of ordinary skill in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented in combination with electronic hardware, computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0098] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0099] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for cooperative control of a drone cluster, characterized in that: The following steps are involved: Collect the location information and communication capability information of the drones in the drone cluster, regard each drone as a node in the communication network, and build the drone cluster topology structure; Create an empty cluster information list, randomly select the initial main cluster head, assign the drones to the corresponding clusters according to the location of each drone, determine the main cluster head and the secondary cluster head, and update the cluster information list until all drones are assigned to the cluster; For each cluster, from the assigned cluster members, select the cluster members whose distance to the main cluster head is within the preset distance range and the drone with the highest communication link quality as the secondary cluster head. The preset distance satisfies the formula: d pr <Ra i -Rc, where d pr Indicates the preset distance, Ra i The communication range of drone i, Rc represents the cluster radius, and the communication link quality formula is: Among them, Score i represents the communication link quality of UAV i, M represents the number of communication parameters, and q f (i,j) represents the value of the fth communication parameter between UAV i and UAV j, ω f represents the weight of the fth communication parameter; In a drone cluster, the direct communication links between member drones in the cluster and the direct communication links with all drones in other clusters are disconnected, and the cluster boundary range centered on the main cluster head is set: R cl =R co *λ, where R cl Indicates the radius of the boundary range, R co represents the communication range, λ represents the attenuation coefficient; When the main cluster head fails, the secondary cluster head immediately moves to the position of the main cluster head and becomes the new main cluster head, and selects a new secondary cluster head from the remaining cluster members; When executing a task, the task data is transmitted to the main cluster head and the secondary cluster head at the same time. When the main cluster head fails, the secondary cluster head takes over and reselects the secondary cluster head; For each cluster, the dynamic real-time data of all drones in the cluster are collected, the Kalman filter is used to predict the drone's motion state, and the isolation forest algorithm is used to calculate the anomaly score. If the anomaly score exceeds the preset score threshold, the cluster is judged to be abnormal and the cluster structure is readjusted. If it does not exceed the threshold, it is judged to be normal and no processing is performed.

2. A UAV cluster collaborative control method according to claim 1, characterized in that: Build the drone cluster topology, including: Assign a unique identifier to each node and create a node list containing all nodes and their basic attributes. The basic attributes include location, ID, and communication capability information. Based on the location information of the drones, calculate the distance between each pair of drones. According to the distance and communication range between each pair of drones, it is determined whether there is a direct communication link between the drones: if the distance between the two drones is less than or equal to the communication range between the two, it is considered that there is a direct communication link between the two, otherwise, it is considered that there is no direct communication link between the two; Traverse all drones, build a communication link matrix based on the judgment results of the communication link, use graph theory methods to add the nodes in the node list to the graph, add edges to the graph based on the communication links between drones in the communication link matrix, and connect the corresponding nodes to build the drone cluster topology.

3. A UAV cluster collaborative control method according to claim 2, characterized in that: Calculate the distance between each pair of drones: Among them, d i,j represents the distance between UAV i and UAV j, (x i ,y i ,z i ) and (x j ,y j ,z j ) represent the 3D coordinates of UAV i and UAV j respectively.

4. The method for cooperative control of a drone cluster according to claim 1, characterized in that: Create an empty cluster information list, including: Create an empty cluster information list to store the information of all clusters. The information of each cluster includes the cluster ID, the main cluster head ID, the secondary cluster head ID and the cluster member ID list. Initialize the cluster state of each drone to: unassigned, indicating that the drone has not been assigned to any cluster.

5. The method for cooperative control of a drone cluster according to claim 4, characterized in that: Based on the location of each drone, the drones are assigned to the corresponding clusters, including: Randomly select a UAV as the initial main cluster head, traverse the remaining UAVs, calculate the distance between each UAV and the main cluster head, if the distance is less than the preset cluster radius, assign the UAV to the cluster where the main cluster head is located, and update the cluster state of the UAV to: allocated, check the size of each cluster, if the number of members in any cluster is less than the preset number of cluster members, update the cluster state of the cluster members of the cluster to: unallocated, and reselect the main cluster head and assign members for the cluster.

6. The method for cooperative control of a drone cluster according to claim 1, characterized in that: Traverse each cluster, and for each pair of member drones in the cluster, disconnect the direct communication link between them by modifying the communication link matrix. The member drones are the drones that do not include the main cluster head and the secondary cluster head; Continue to traverse each cluster, and for each member drone in the cluster, disconnect the communication link between it and all drones in other clusters by modifying the communication link matrix. All drones in other clusters include the main cluster head, deputy cluster heads and member drones.

7. The method for cooperative control of a drone cluster according to claim 1, characterized in that: For each cluster, the Kalman filter algorithm is used to combine the dynamic real-time data of the drone, including position information, speed information, and acceleration information, to predict the motion state of the drone in real time. The calculation formula is as follows: in, represents the predicted vector state, k represents the time, a mn represents the probability of the nth state transferring to the mth state in the state transfer matrix A, represents the optimal estimated state vector at the previous moment, u k-1 represents the control input at the previous moment, b mn represents the control input u in the control input matrix B k-1 The nth component of the state vector The influence coefficient of the mth component of , m represents the number of rows of the matrix, and n represents the number of columns of the matrix.

8. The method for cooperative control of a drone cluster according to claim 7, characterized in that: The Isolation Forest algorithm is used to predict the motion state of the drone and calculate the anomaly score of each cluster: Among them, S(g,k) represents the anomaly score of sample g, k represents the number of samples, h(g) represents the path length that sample g needs to go through from the root node to the leaf node, E(h(g)) represents the expected value of the path length of sample g under h(g), and c(k) represents the average path length of the binary search tree.

9. A drone swarm collaborative control system, used to implement the method described in any one of claims 1 to 8, characterized in that: include: The cluster topology building module collects the location information and communication capability information of the drones in the drone cluster, regards each drone as a node in the communication network, and builds the drone cluster topology structure; The cluster partitioning module creates an empty cluster information list, randomly selects the initial main cluster head, assigns the drones to the corresponding clusters according to the location of each drone, determines the main cluster head and the secondary cluster head, and updates the cluster information list until all drones are assigned to the cluster; The communication link adjustment module disconnects the direct communication links between the member drones in the cluster and the direct communication links with all drones in other clusters in the drone cluster, sets the cluster boundary range centered on the main cluster head, and transmits the task data to the main cluster head and the secondary cluster head at the same time when executing the task. When the main cluster head fails, the secondary cluster head takes over and reselects the secondary cluster head. For each cluster, the control module collects the dynamic real-time data of all drones in the cluster, uses Kalman filtering to predict the motion status of the drone, and uses the isolation forest algorithm to calculate the anomaly score. If the anomaly score exceeds the preset score threshold, the cluster is judged to be abnormal and the cluster structure is readjusted. If it does not exceed the threshold, it is judged to be normal and no processing is performed.

Images