A method for identifying the scale of a UAV group based on multi-dimensional super-resolution analysis

Abstract

The application discloses a method for identifying the scale of a UAV group based on multi-dimensional super-resolution analysis, constructs a distance-azimuth-pitch-Doppler four-order tensor model of a target echo, constructs a four-order tensor decomposition model of the echo signal based on PARAFAC decomposition, performs parameter estimation on the echo signal to obtain parameter information of the azimuth angle, the pitch angle, the speed and the distance of each point target, merges the motion trend of the cluster targets to obtain the aggregation degree of each point target, establishes a cluster target connected graph model to obtain the motion direction corresponding to each group, and geometrically reconstructs the formation shape of the cluster targets to complete the estimation of the number and shape of the UAV cluster formation. The application fully utilizes multi-dimensional feature information of the target to construct a distance-azimuth-pitch-Doppler four-dimensional joint domain, geometrically reconstructs the cluster targets, and realizes the scale estimation and identification of the UAV cluster targets.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing, specifically relating to a method for identifying the size of a drone swarm based on multidimensional super-resolution analysis. Background Technology

[0002] In modern warfare, air strikes account for an increasingly large proportion of attacks, with incoming drone swarms often operating in formation. The ability to accurately identify the number and shape of drone swarms within a short time is crucial for further threat intent assessment and the development of appropriate operational plans. Traditional target discrimination methods rely solely on angular information from target echo signals. However, their performance suffers when the angular differences between targets are insufficient for differentiation. Considering that radar echo signals contain multiple dimensions of information, such as range and Doppler information, combining these dimensions for multi-target discrimination will improve the performance for densely packed targets. Furthermore, research on super-resolution targeting of swarms primarily focuses on identifying the number of targets, with limited analysis of dynamic changes and shape characteristics. However, shape recognition of drone swarms is crucial, especially when drone formations are prone to separation and merging, significantly impacting situational awareness and intent assessment.

[0003] Therefore, how to use multi-dimensional feature information to reconstruct the formation number and shape of UAV swarms is of great research significance. Summary of the Invention

[0004] Purpose of the invention: This invention aims to solve the problems of drone swarm size estimation and formation type identification, and proposes a method for drone swarm size identification based on multidimensional super-resolution analysis, which performs geometric reconstruction of swarm targets to achieve drone swarm target size estimation and identification.

[0005] Technical Solution: This invention provides a method for identifying the size of a drone swarm based on multidimensional super-resolution analysis, comprising the following steps:

[0006] (1) Construct a fourth-order tensor model of the target echo's range-azimuth-elevation-Doppler;

[0007] (2) Construct a fourth-order tensor decomposition model of the echo signal based on PARAFAC decomposition;

[0008] (3) Based on the decomposition model, the parameters of the echo signal are estimated to obtain the parameter information of the four dimensions of azimuth, pitch, velocity and distance of each target point;

[0009] (4) The movement trends of clustered targets are merged to obtain the clustering degree of each target;

[0010] (5) Establish a cluster target connectivity graph model and use clustering to divide point targets with the same direction of movement into a group, that is, each group has a corresponding direction of movement;

[0011] (6) Geometric reconstruction of the shape of the cluster target formation to complete the estimation of the number and shape of the UAV cluster formation.

[0012] Furthermore, the implementation process of step (1) is as follows:

[0013] There are P targets within a dense target group in the far field region of the array, and these P targets are all located within the same beamwidth, Doppler cell, and range cell. The targets are similar in azimuth, elevation, Doppler frequency, and range. If the sampled data of the target echoes are arranged in the form of a fourth-order tensor, then the (k1,k2,k3,k4)th element of the fourth-order tensor is represented as:

[0014]

[0015] In the formula, A j Let R be the amplitude of the echo from the j-th target. j R is the distance to the j-th target. ref θ is the distance between the reference target and the receiving array element. j Let β be the azimuth angle of the j-th target. j Let be the elevation angle of the j-th target, d be the element spacing, and f be the pitch angle. dj Let K be the Doppler frequency of the j-th target, and K be the frequency modulation slope.

[0016] Furthermore, the implementation process of step (2) is as follows:

[0017] Using PARAFAC decomposition, the fourth-order tensor decomposition model of the echo signal is represented as:

[0018]

[0019] in, Represents a four-dimensional echo signal. and Let represent the azimuth vector, elevation vector, slow-time sampling vector, and fast-time sampling vector of the j-th target, respectively.

[0020] When noise is considered, the fourth-order tensor decomposition model of the echo signal is expressed as:

[0021]

[0022] in, This represents the noise tensor.

[0023] Furthermore, the implementation process of step (3) is as follows:

[0024] Considering the uniqueness of the PARAFAC decomposition, the only differences between the decomposed matrix factors and the true matrix factors are the magnitude and the column vector order; for the j-th target, the estimated values ​​of its azimuth, elevation, velocity, and range are as follows:

[0025]

[0026]

[0027]

[0028]

[0029] Where angle represents the phase angle. and These represent the p-th column of the factor estimate of each dimension vector of the target.

[0030] Furthermore, the implementation process of step (4) is as follows:

[0031] Let target i be a point target within the neighborhood of target j, v i and v j Calculate the motion similarity based on their velocity vectors:

[0032]

[0033] The trajectory similarity between each target is defined as:

[0034]

[0035]

[0036] Where R(i,j) represents the trajectory from point target i to point target j, and P(i,j) represents the set of all trajectories with path length l between point target i and point target j;

[0037] For a single point target i, its clustering degree on a path of length l is defined as:

[0038]

[0039] Where C represents the set of cluster targets.

[0040] Furthermore, the implementation process of step (5) is as follows:

[0041] The clustering degree of each target point corresponds to its clustering degree matrix Z, which is defined as:

[0042] Z = (IW)-1 -I

[0043] Where I is the identity matrix, W is the adjacency matrix corresponding to cluster target C; the adjacency matrix G of the connected graph model is defined as:

[0044]

[0045] Where ε is the connectivity threshold, the nodes are connected if the correlation between nodes between trajectories is greater than the threshold, and the value is 1 otherwise. According to the connected graph model, the clustering method is used to divide point targets with the same direction of movement into a group, that is, each group has a corresponding direction of movement.

[0046] Furthermore, the implementation process of step (6) is as follows:

[0047] Within each group obtained in step (5), point targets are connected in pairs based on node number and relative distance to reconstruct the geometry of the UAV formation; combined with the typical flight pattern of the UAV formation, the formation type of each cluster is matched to realize the estimation of UAV swarm size and formation type identification.

[0048] Beneficial effects: Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Most existing methods only use the distance, orientation, and Doppler dimension information of the target for identification and classification, while the present invention makes full use of the multi-dimensional spatial feature information of the target to accurately analyze the characteristics of each target in the UAV swarm; 2. Most existing methods only identify the number of target swarms, while the present invention further reconstructs the shape of the UAV formation, thereby realizing the size estimation and formation type identification of the UAV swarm, which lays the foundation for further situational awareness and threat intent assessment. Attached Figure Description

[0049] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0050] The present invention will now be described in further detail with reference to the accompanying drawings.

[0051] This invention discloses a method for UAV swarm size identification based on multidimensional super-resolution analysis, belonging to the field of radar signal processing, such as... Figure 1 As shown, the specific steps are as follows:

[0052] Step 1: Construct a fourth-order tensor model of the target echo's range-azimuth-elevation-Doppler.

[0053] Suppose there are P targets within a dense cluster of targets in the far field of the array, and these P targets are all located within the same beamwidth, Doppler cell, and range cell, and the targets are similar in azimuth, elevation, Doppler frequency, and range. If the sampled data of the target echoes are arranged in the form of a fourth-order tensor, then the (k1,k2,k3,k4)th element of the fourth-order tensor can be represented as:

[0054]

[0055] Among them, A j Let R be the amplitude of the echo from the j-th target. j R is the distance to the j-th target. ref θ is the distance between the reference target and the receiving array element. j Let β be the azimuth angle of the j-th target. j Let be the elevation angle of the j-th target, d be the element spacing, and f be the pitch angle. dj Let K be the Doppler frequency of the j-th target, and K be the frequency modulation slope.

[0056] Step 2: Construct a fourth-order tensor decomposition model of the echo signal based on PARAFAC decomposition.

[0057] Using PARAFAC decomposition, the above equation can be expressed as:

[0058]

[0059] in, Represents a four-dimensional echo signal. and Let represent the azimuth vector, elevation vector, slow-time sampling vector, and fast-time sampling vector of the j-th target, respectively.

[0060] When noise is taken into account, the fourth-order tensor decomposition model of the echo can be expressed as:

[0061]

[0062] in, This represents the noise tensor.

[0063] Step 3: Based on the decomposition model, perform parameter estimation on the echo signal to obtain parameter information of four dimensions of each target: azimuth, elevation, velocity, and distance.

[0064] The decomposition model obtained in step 2 contains parameter information of the target in the azimuth, pitch, velocity, and range dimensions. Considering the uniqueness of the PARAFAC decomposition, the only differences between the decomposed matrix factors and the true matrix factors are the amplitude and the column vector order. For the j-th target, the estimated values ​​of its azimuth, pitch, velocity, and range are derived as follows:

[0065]

[0066]

[0067]

[0068]

[0069] Where angle represents the phase angle. and These represent the p-th column of the factor estimate of each dimension vector of the target.

[0070] Step 4: Merge the movement trends of clustered targets to obtain the clustering degree of each target.

[0071] Considering the motion correlation between individual targets in a cluster, their motion trends mainly consist of two states: merging and separation. Based on the four-dimensional parameter information of each point target obtained in step (3), namely azimuth, elevation, Doppler, and range, let target i be a point target within the neighborhood of target j, v i and v j Let each be its velocity vector. Then, the motion similarity can be calculated:

[0072]

[0073] The trajectory similarity between each target is defined as:

[0074]

[0075]

[0076] Where R(i,j) represents the trajectory from point target i to point target j, and P(i,j) represents the set of all trajectories with path length l between point target i and point target j.

[0077] For a single point target i, its clustering degree on a path of length l is defined as:

[0078]

[0079] Where C represents the set of cluster targets.

[0080] Step 5: Establish a cluster target connectivity graph model. Use clustering to divide point targets with the same direction of movement into a group, that is, each group has a corresponding direction of movement.

[0081] Based on the clustering degree of each target obtained in step 4, its clustering degree matrix Z can be derived, defined as:

[0082] Z = (IW) -1 -I

[0083] Where I is the identity matrix, and W is the adjacency matrix corresponding to cluster target C. Then the adjacency matrix G of the connected graph is defined as:

[0084]

[0085] Here, ε is the connectivity threshold. Nodes are considered connected if the correlation between trajectories is greater than this threshold (value 1), otherwise it is 0. Based on the connected graph model, clustering is used to group point targets with the same direction of motion into groups, meaning each group has a corresponding direction of motion.

[0086] Step 6: Geometric reconstruction of the shape of the swarm target formation to complete the estimation of the number and shape of the UAV swarm formation.

[0087] Within each group obtained in step 5, point targets are connected pairwise based on node number and relative distance, ultimately reconstructing the geometry of the UAV formation. Combining this with typical UAV formation flight patterns, the formation type of each cluster can be determined, thus achieving UAV swarm size estimation and formation type identification.

Claims

1. A method for identifying the size of a drone swarm based on multidimensional super-resolution analysis, characterized in that, Includes the following steps: (1) Construct a fourth-order tensor model of the target echo's range-azimuth-elevation-Doppler; (2) Construct a fourth-order tensor decomposition model of the echo signal based on PARAFAC decomposition; (3) Based on the decomposition model, the parameters of the echo signal are estimated to obtain the parameter information of the four dimensions of azimuth, pitch, velocity and distance of each target point; (4) The movement trends of clustered targets are merged to obtain the clustering degree of each target; (5) Establish a cluster target connectivity graph model and use clustering to divide point targets with the same direction of movement into a group, that is, each group has a corresponding direction of movement; (6) Geometric reconstruction of the formation shape of the cluster target to complete the estimation of the number and shape of the UAV cluster formation; The implementation process of step (1) is as follows: There are P targets within a dense target group in the far field region of the array, and these P targets are all located within the same beamwidth, Doppler cell, and range cell. The targets are similar in azimuth, elevation, Doppler frequency, and range. If the sampled data of the target echoes are arranged in the form of a fourth-order tensor, then the (k1,k2,k3,k4)th element of the fourth-order tensor is represented as: In the formula, A j Let R be the amplitude of the echo from the j-th target. j R is the distance to the j-th target. ref θ is the distance between the reference target and the receiving array element. j Let β be the azimuth angle of the j-th target. j Let d be the elevation angle of the j-th target, d be the element spacing, K be the frequency modulation slope, and v be the frequency angle of the j-th target. j Let be the velocity vector of target j.

2. The method for UAV swarm size identification based on multidimensional super-resolution analysis according to claim 1, characterized in that, The implementation process of step (2) is as follows: Using PARAFAC decomposition, the fourth-order tensor decomposition model of the echo signal is represented as: in, Represents a four-dimensional echo signal. and Let represent the azimuth vector, elevation vector, slow-time sampling vector, and fast-time sampling vector of the j-th target, respectively. When noise is considered, the fourth-order tensor decomposition model of the echo signal is expressed as: in, This represents the noise tensor.

3. The method for UAV swarm size identification based on multidimensional super-resolution analysis according to claim 2, characterized in that, The implementation process of step (3) is as follows: Considering the uniqueness of the PARAFAC decomposition, the only differences between the decomposed matrix factors and the true matrix factors are the magnitude and the column vector order; for the j-th target, the estimated values ​​of its azimuth, elevation, velocity, and range are as follows: Where angle represents the phase angle. and These represent the p-th column of the factor estimate of each dimension vector of the target.

4. The method for UAV swarm size identification based on multidimensional super-resolution analysis according to claim 1, characterized in that, The implementation process of step (4) is as follows: Let target i be a point target within the neighborhood of target j, v i and v j Calculate the motion similarity based on their velocity vectors: The trajectory similarity between each target is defined as: Where R(i,j) represents the trajectory from point target i to point target j, and P(i,j) represents the set of all trajectories with path length l between point target i and point target j; For a single point target i, its clustering degree on a path of length l is defined as: Where C represents the set of cluster targets.

5. The method for UAV swarm size identification based on multidimensional super-resolution analysis according to claim 1, characterized in that, The implementation process of step (5) is as follows: The clustering degree of each target point corresponds to its clustering degree matrix Z, which is defined as: Z=(IW) -1 -AND Where I is the identity matrix, W is the adjacency matrix corresponding to cluster target C; the adjacency matrix G of the connected graph model is defined as: Where ε is the connectivity threshold, the nodes are connected if the correlation between nodes between trajectories is greater than the threshold, and the value is 1 otherwise. According to the connected graph model, the clustering method is used to divide point targets with the same direction of movement into a group, that is, each group has a corresponding direction of movement.

6. The method for UAV swarm size identification based on multidimensional super-resolution analysis according to claim 1, characterized in that, The implementation process of step (6) is as follows: Within each group obtained in step (5), point targets are connected in pairs based on node number and relative distance to reconstruct the geometry of the UAV formation; combined with the typical flight pattern of the UAV formation, the formation type of each cluster is matched to realize the estimation of UAV swarm size and formation type identification.

Images