Clustering-based unmanned aerial vehicle swarm threat degree evaluation method
By automatically clustering drone swarms and calculating threat levels based on cluster centers, the problem of low analysis efficiency in traditional methods is solved, enabling efficient threat assessment and countermeasure decisions for drone swarms.
Patent Information
- Application Number
- CN202510880490.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-27
- Publication Date
- 2025-11-07
AI Technical Summary
Traditional air defense threat analysis methods are inefficient at analyzing drone swarms, struggle to process multi-target information efficiently, and are prone to system crashes.
Clustering algorithms are used to automatically cluster drone swarms into different clusters, and threat levels are calculated based on the clustering results. Threat assessment is performed directly on the center of each cluster, abandoning the traditional calculation strategy that targets a single target.
It improves the information processing capabilities against drone swarm saturation attacks, enhancing the efficiency and accuracy of countermeasure decisions.
Smart Images
Figure CN120910592A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of unmanned aerial vehicles, and particularly relates to a clustering-based threat degree evaluation method for unmanned aerial vehicle swarms. BACKGROUND
[0002] Unmanned aerial vehicles and unmanned aerial vehicle swarms are playing an increasingly important role in the world today. When performing distributed tasks, they are often embodied in various forms of clustering, with a large number of flexible flight styles, including dive clusters, reconnaissance clusters, interference clusters, and replenishment clusters. One of the countermeasures is to use multi-platform cooperative countermeasures against unmanned aerial vehicle swarm targets. However, when facing a saturated attack by unmanned aerial vehicle swarms, the ground system has difficulty in efficiently processing multi-target information, and may even appear to be down. Since the unmanned aerial vehicles in a swarm are similar in attributes, combat modes, and combat tasks, the entire swarm can be analyzed as a unit.
[0003] Traditional air defense threat degree analysis methods are often directed at a single target, and the analysis efficiency for the entire unmanned aerial vehicle swarm is low. Unmanned aerial vehicle target threat degree evaluation is usually divided into three steps: threat factor extraction, threat degree calculation, and threat degree evaluation. Common threat degree evaluation methods mainly include multi-attribute decision-making, fuzzy set methods, rough set methods, and Bayesian methods. The fuzzy set method proposed by Nazih NY in 2004 has been widely used. Wang Qian completed a model for evaluating the threat degree of unmanned aerial vehicles against ground targets in air-to-ground time-sensitive mode by establishing an unmanned aerial vehicle threat degree index system, but did not consider the group characteristics of unmanned aerial vehicle swarms. SUMMARY
[0004] In order to overcome the deficiencies of the prior art, the application provides a clustering-based threat degree evaluation method for unmanned aerial vehicle swarms. Based on captured radar information of unmanned aerial vehicle swarms, a clustering algorithm is used to automatically cluster the tracks, and the unmanned aerial vehicle swarms are divided into different clusters. If it is a point track information, the K-means algorithm is used for clustering, and if it is a track information, the DBSCAN algorithm is used for clustering. Based on the clustering results, a conventional threat degree evaluation model is used to calculate the threat degree of each unmanned aerial vehicle cluster, and finally a suitable area shooting strategy is selected for anti-cluster shooting. This method discards the traditional strategy of sequentially calculating the threat degree for a single aerial target, allowing the ground countermeasure platform to directly target each unmanned aerial vehicle cluster with similar attributes in the entire swarm, thereby quickly evaluating the threat degree of a large number of incoming unmanned aerial vehicle swarm targets and improving the efficiency of anti-swarm auxiliary decision-making.
[0005] The technical solution adopted by the application to solve its technical problems is as follows:
[0006] Step 1: Establish a four-dimensional track sequence or point track matrix for the unmanned aerial vehicle swarm.
[0007] The UAV swarm radar track data acquired by the ground platform is processed into a four-dimensional track sequence or a point track sequence in a unified format, including time and three-dimensional coordinates, and is stored in a matrix manner;
[0008] Step 2: UAV swarm automatic clustering;
[0009] According to specific combat tasks and combat data information sources, a density-based clustering method and a Euclidean distance-based clustering method are respectively used to cluster the UAV swarm; specifically, the density-based clustering method is used to cluster the track, and the Euclidean distance-based clustering method is used to cluster the UAV point track;
[0010] Step 3: Based on the clustering results, a threat degree calculation method is used to calculate the threat degree of each cluster category;
[0011] The cluster center UAV after clustering is used as the input of the threat degree calculation method, that is, when calculating the threat degree, it is for each cluster center in the UAV swarm rather than a single UAV target.
[0012] Preferably, the step 1 is specifically:
[0013] Step 1-1: Coordinate conversion: normalize the latitude, longitude, pitch angle, height angle and distance of the UAV to a plane rectangular coordinate system;
[0014] Step 1-2: Data cleaning: delete data rows with null values or points with too large differences from the track, and add a timestamp to each point track, and add a UAV serial number to each point track and track;
[0015] Step 1-3: Matrix storage: store the processed UAV data into a matrix; the data is stored in the following manner:
[0016] data = [Num, Time, X, Y, Z]
[0017] Where data represents the data matrix, Num represents the UAV serial number vector, Time represents the timestamp vector; [X, Y, Z] represents the three-dimensional coordinate matrix of the UAV, and when it is a track matrix, the three-dimensional coordinate matrix of the UAV with the same serial number in [Num, Time, X, Y, Z] is linearly distributed, that is, it is a time sequence.
[0018] Preferably, the step 2 is specifically:
[0019] Step 2-1: Use the density-based clustering method DBSCAN to cluster the UAV track data;
[0020] 1) Select a point from the sample, given a radius epsilon and a minimum number of neighbors min_points, if the point satisfies that it has at least min_points neighbors within its neighborhood circle of radius epsilon, then the point is a core point;
[0021] 2) Recursively process all points in the neighborhood of the current point that have not been labeled as visited; if these unvisited points satisfy to be new core points, then mark them as visited and join the current cluster; at the same time, the points in the new cluster will also be taken as new core points and expanded;
[0022] 3) Repeat the above process until all samples are visited or there is no new core point, the final cluster is the clustering result generated by DBSCAN algorithm;
[0023] Step 2-2: adopt the mean clustering K-means to cluster the UAV track matrix;
[0024] 1) Select k data objects from the data as initial cluster centers;
[0025] 2) Calculate the distance of each cluster object to the cluster center;
[0026] 3) Recalculate each cluster center;
[0027] 4) Calculate the standard measure function, stop when the maximum iteration number is reached, otherwise continue to operate;
[0028] 5) In each iteration, reassign samples to the nearest cluster center, and update the cluster center to the average value of all samples in the category;
[0029] 6) Repeat the above steps until the condition is met.
[0030] Preferably, the step 3 is specifically:
[0031] Adopting the fuzzy set method, the threat degree of a target contains the threat degree factors of enemy distance, target speed and target type;
[0032] (1) Enemy distance factor u1: the distance d of the center UAV in a cluster of the UAV swarm to the protected target of our side;
[0033]
[0034] (2) Speed factor u2: the speed v of the center UAV in a cluster of the UAV swarm;
[0035]
[0036] (3) Target type factor u3: the type of a cluster center UAV in the UAV swarm, assuming that the target1 target type threat degree factor is the largest, the target2 is the second, and the others are the smallest;
[0037]
[0038] The membership degrees of each threat factor are calculated and weightedly summed, and finally the threat degree size evaluation is obtained.
[0039] A computer program, which enables a computer to execute the above-mentioned UAV swarm threat degree evaluation method.
[0040] An electronic device, comprising a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the above-mentioned UAV swarm threat degree evaluation method.
[0041] A computer readable storage medium, which stores a computer program, the computer program is executed by a processor to realize the above-mentioned UAV swarm threat degree evaluation method.
[0042] A chip, comprising a processor, which is used to call and run a computer program from a memory, so that the device installed with the chip executes the above-mentioned UAV swarm threat degree evaluation method.
[0043] A computer program product, which comprises a computer storage medium, the computer storage medium stores a computer program, the computer program comprises instructions executable by at least one processor, when the instructions are executed by the at least one processor, the above-mentioned UAV swarm threat degree evaluation method is realized.
[0044] The beneficial effects of the present application are as follows:
[0045] (1) High degree of automation, easy to implement;
[0046] The algorithm can be realized by adding under the existing ground air defense command platform. The threat degree evaluation algorithm based on clustering is automatically completed by software, and can be deployed in the command and control system of the anti-UAV platform.
[0047] (2) Improve the information processing capability for swarm saturation attack;
[0048] The method establishes a UAV swarm track database, and automatically clusters it into multiple clusters with the same attributes. When processing a large amount of UAV swarm data, it can be unified and summarized. For a cluster containing multiple UAVs with similar attributes, the same information processing means can be used, thereby improving the information processing capability for swarm saturation attack. BRIEF DESCRIPTION OF DRAWINGS
[0049] Figure 1 is a structure diagram of a UAV swarm;
[0050] Figure 2 is a track clustering diagram of a UAV swarm;
[0051] Figure 3 is a track clustering diagram of a UAV swarm. DETAILED DESCRIPTION
[0052] The present application is further illustrated below in conjunction with the accompanying drawings and examples.
[0053] The technical problem to be solved by the present application is that the traditional air defense threat degree analysis method has low analysis efficiency for a large number of UAV swarms.
[0054] To solve the above technical problem, the present application proposes a clustering-based UAV swarm threat degree evaluation method, comprising the following steps:
[0055] Step 1: Establishing a four-dimensional track sequence or track matrix of a UAV swarm;
[0056] The UAV swarm radar track data obtained by the ground platform is processed into a four-dimensional track sequence or track sequence in a unified format, including time and three-dimensional coordinates, and is stored in the form of a matrix.
[0057] Step 2: Automatic clustering of a UAV swarm;
[0058] According to specific combat tasks and combat data information sources, a density-based clustering method and a Euclidean distance-based clustering method are respectively used to cluster the UAV swarm. Specifically, the track is clustered using the density-based clustering method, and the UAV track is clustered using the Euclidean distance-based clustering method.
[0059] Step 3: Based on the cluster clustering result, the threat degree of each cluster category is calculated using a threat degree calculation method;
[0060] The UAV cluster center UAV after clustering is used as the input of the threat degree calculation method, that is, when the conventional threat degree calculation is performed, it is for each cluster center contained in the UAV swarm rather than a single UAV target.
[0061] Embodiment:
[0062] Step 1: Establishing a four-dimensional track sequence or track matrix of a UAV swarm;
[0063] (1) Coordinate conversion: The position data of the unmanned aerial vehicle detected by the ground platform is generally labeled by latitude and longitude, pitch angle-high-low angle-distance, etc. In order to facilitate subsequent clustering, processing and display calculations, it needs to be normalized to a plane rectangular coordinate system for subsequent processing.
[0064] (2) Data cleaning: delete data rows with null values or points with large differences from the track. At the same time, give each point a timestamp, and give each point and track a UAV serial number.
[0065] (3) Matrix storage: store the processed UAV data into a matrix. The data is stored in the following way.
[0066] data = [Num, Time, X, Y, Z]
[0067] Where data represents the data matrix, Num represents the UAV serial number vector, Time represents the timestamp vector. [X, Y, Z] represents the three-dimensional coordinate matrix of the UAV. When it is a track matrix, the three-dimensional coordinate matrix of the UAV with the same serial number is linearly distributed, that is, it is a time series.
[0068] Step 2: Use an adaptive clustering method to cluster the UAV track or point matrix, specifically:
[0069] (1) Use DBSCAN (density-based clustering method) to cluster the UAV track data. This clustering method does not need to know the approximate category of the swarm in advance, and is suitable for clustering tasks of track sequences. The main steps are: 1) Select a point from the sample, give a radius epsilon and a minimum number of neighbors min_points, if the point satisfies at least min_points neighbors in its radius epsilon neighborhood circle, then the point is taken as the core point; 2) Take the current point as the center, recursively process all unmarked points in its neighborhood. If these unvisited points meet the new core point, mark them as visited and join the current cluster. At the same time, the points in these newly added clusters will also be taken as new core points and expanded; 3) Repeat the above process until all samples are visited or there are no new core points, and the final cluster is the clustering result generated by DBSCAN algorithm.
[0070] (2) K-means (mean clustering) is used to cluster the UAV trajectory matrix, which is suitable for clustering operation with prior knowledge (K value). The main steps are: 1) select k data objects from the data as initial cluster centers; 2) calculate the distance of each cluster object to the cluster center; 3) calculate each cluster center again; 4) calculate the standard measure function, stop when the maximum iteration number is reached, otherwise continue operation; 5) in each iteration, reassign samples to the nearest cluster center, and update the cluster center to the average value of all samples in this category; 6) repeat the above steps until the condition is met.
[0071] The UAV swarm trajectory or track matrix is input into the clustering algorithm, and the output is the UAV serial number Num and the corresponding cluster number cluster_number, that is, the clustering result is obtained. As shown in Figure 1 , serial number 2 represents a UAV swarm, and serial number 1 represents a ground defense unit. After K-means trajectory clustering, each UAV cluster in the UAV swarm is separated, and each cluster can contain UAVs with similar attributes, as shown in Figure 2 . The schematic diagram of track clustering by DBSCAN is shown in Figure 3 , serial numbers 3-16 are the cluster categories of the UAV swarm after clustering.
[0072] Step 3: Use the threat degree calculation method to calculate the threat degree of each cluster of the UAV swarm. Taking the commonly used fuzzy set method as an example, the threat degree of a target should include the threat degree factors of enemy distance, target speed, and target type.
[0073] (1) Enemy distance factor: the distance of the target from the protected target. Here, it is set to have a threat degree within 6 km.
[0074]
[0075] (2) Speed factor: target speed. Here, it is set that the target speed greater than 300 m / s has the maximum threat degree factor.
[0076]
[0077] (3) Target type factor: target type. Assuming that target1 has the maximum threat degree factor, target2 has the second largest, and others have the smallest.
[0078]
[0079] This threat degree evaluation method is optimized by using the data preprocessing method based on the clustering result, which only needs to take the central UAV of the cluster as the target for threat degree calculation, that is:
[0080] (1) Friend-foe distance factor: the distance between the center UAV of a cluster in the UAV swarm and the protected target.
[0081]
[0082] (2) Speed factor: the speed of the center UAV of a cluster in the UAV swarm.
[0083]
[0084] (3) Target type factor: the type of the center UAV of a cluster in the UAV swarm.
[0085]
[0086] The threat degree is obtained by calculating the membership degrees of the threat factors by using the above membership functions and performing weighted summation.
[0087] For other threat degree evaluation methods, the single target to be evaluated is replaced by the center UAV of a cluster in the UAV swarm obtained in the second step.
Claims
1. A clustering-based method for evaluating threat level of a swarm of unmanned aerial vehicles, characterized in that, Comprise the following steps: Step 1: Establishing a four-dimensional track sequence or track matrix of UAV swarm; The UAV swarm radar track data obtained by the ground platform is processed into a four-dimensional track sequence or track sequence in a unified format, including time and three-dimensional coordinates, and is stored in the form of a matrix; Step 2: Automatic clustering of UAV swarm; According to specific combat tasks and combat data information sources, a density-based clustering method and a Euclidean distance-based clustering method are used to cluster the UAV swarm respectively; Specifically, the track is clustered by using the density-based clustering method, and the UAV track is clustered by using the Euclidean distance-based clustering method; Step 3: Based on the clustering results, the threat degree of each cluster is calculated by using the threat degree calculation method; The cluster center UAV after clustering is used as the input of the threat degree calculation method, that is, when calculating the threat degree, it is for each cluster center in the UAV swarm rather than a single UAV target. 2.The cluster-based threat assessment method for UAV swarm according to claim 1, wherein, The step 1 is specifically: Step 1-1: Coordinate conversion: normalize the longitude, latitude, pitch angle, height angle and distance of the UAV to the plane rectangular coordinate system; Step 1-2: Data cleaning: delete the data rows with null values or too large differences from the track, and add a timestamp to each track, and add a UAV number to each track and track; Step 1-3: Matrix storage: store the processed UAV data into a matrix; the data is stored in the following way: data=[Num,Time,X,Y,Z] Where data represents the data matrix, Num represents the UAV number vector, Time represents the timestamp vector; [X,Y,Z] represents the three-dimensional coordinate matrix of the UAV, when it is a track matrix, the three-dimensional coordinate matrix of the UAV with the same sequence number is linearly distributed, that is, it is a time sequence. 3.The cluster-based threat assessment method for UAV swarm according to claim 2, wherein, The step 2 is specifically: Step 2-1: Clustering the UAV track data by using the density-based clustering method DBSCAN; 1) Select a point from the sample, give a radius epsilon and a minimum number of neighbors min_points, if the point satisfies at least min_points neighbors in its radius epsilon neighborhood, then the point is a core point; 2) Take the current point as the center, recursively process all unmarked points in its neighborhood; if these unvisited points meet the new core point, mark them as visited and join the current cluster; at the same time, the points in these newly added clusters are also taken as new core points and expanded; 3) Repeat the above process until all samples are visited or there is no new core point, and the final cluster is the clustering result generated by DBSCAN algorithm; Step 2-2: Clustering the UAV track matrix by using the mean clustering K-means; 1) Select k data objects from the data as initial cluster centers; 2) Calculate the distance of each cluster object to the cluster center; 3) Recalculate each cluster center; 4) Calculate the standard measure function, stop when the maximum number of iterations is reached, otherwise continue operation; 5) In each iteration, reassign samples to the nearest cluster center and update the cluster center to the average of all samples in the class; 6) Repeat the above steps until the condition is met.
4. The cluster-based threat assessment method for UAV swarm according to claim 3, wherein, The step 3 is specifically: Adopting the fuzzy set method, the threat degree of a target contains the threat degree factors of enemy distance, target speed and target type; (1) Enemy distance factor u1: the distance d of the center UAV in a cluster of the UAV swarm to the protected target of our side; (2) Speed factor u2: the speed v of the center UAV in a cluster of the UAV swarm; (3) Target type factor u3: the type of the center UAV in a cluster of the UAV swarm, assuming that the target type threat degree factor of target1 is the largest, target2 is the second, and the others are the smallest; Calculate the membership degree of each threat factor and perform weighted summation to finally obtain the threat degree size evaluation.
5. A computer program, characterized in that, The computer program enables the computer to execute the method of any one of claims 1 to 4.
6. An electronic device, comprising: Comprise: A processor and a memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to enable the electronic device to execute the method of any one of claims 1 to 4.
7. A computer readable storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to realize the method of any one of claims 1 to 4.
8. A chip, characterized by Comprise: A processor, configured to call and run a computer program from a memory, so that the device installed with the chip executes the method of any one of claims 1 to 4.
9. A computer program product, characterised in that, The computer program product comprises a computer storage medium, the computer storage medium stores a computer program, the computer program comprises instructions executable by at least one processor, and when the instructions are executed by the at least one processor, the method of any one of claims 1 to 4 is realized.