Dynamic unmanned aerial vehicle cluster multi-radar cooperative detection tracking method and system based on target resolution

Abstract

The invention discloses a dynamic unmanned aerial vehicle cluster multi-radar cooperative detection tracking method and system based on target resolution. The method specifically comprises the following steps: constructing an observation scene of multiple radars for an unmanned aerial vehicle cluster; carrying out observation group division and geometric relationship calculation on a multi-radar observation scene, and carrying out cluster target electromagnetic scattering characteristic calculation and multi-radar time-varying scattering center modeling; constructing a multi-radar signal level echo model, performing beam forming on echo data, constructing a distance-azimuth spectrum and a distance-pitch spectrum, and realizing cluster target resolution and parameter output; and generating single-station dynamic measurement data of each radar, completing spatial registration and fusion preprocessing of multi-station measurement data, and outputting a collaborative accurate sensing result. The system comprises a multi-radar observation scene modeling module, a cluster target characteristic calculation module, a dense target distinguishing detection module and a multi-radar cooperative sensing tracking module. According to the invention, high-precision and multi-dimensional radar sensing of the unmanned aerial vehicle cluster is realized, the detection performance is optimized, and the tracking quality is improved.

Description

Technical Field

[0001] This invention relates to the fields of electromagnetic computing technology and radar sensing technology, and in particular to a dynamic UAV swarm multi-radar cooperative detection and tracking method and system based on target resolution. Background Technology

[0002] With the rapid popularization of civilian drone technology and the continuous advancement of swarm control methods, drone swarms have shown broad application prospects in civilian fields such as aerial photography, logistics distribution, agricultural plant protection, and urban air traffic management. Drone swarms typically consist of multiple drones flying collaboratively within a limited airspace. They not only possess highly dynamic motion characteristics, complex and varied spatial configurations, and collaborative behavior among individuals, but also exhibit electromagnetic scattering characteristics significantly different from those of a single target. The spatial distribution, relative motion relationships, and electromagnetic coupling effects of the individuals within the swarm mean that the overall radar echo is no longer a simple superposition of individual scattering, but rather a complex collective response that evolves continuously over time, resulting in significant time-varying and unstable actual observed signals.

[0003] At the radar modeling level, traditional echo models are mostly based on the assumption of isolated targets or simplified point scatterers, making it difficult to accurately describe the electromagnetic scattering behavior of UAV swarms under conditions of coordinated maneuvering and configuration changes. Especially in high-density formation scenarios, factors such as obstruction effects, mutual influence between targets, and changes in viewing angle significantly affect the distribution and amplitude characteristics of the scattering center, resulting in a large deviation between the echoes constructed based on idealized models and actual observation results. Therefore, constructing a radar echo model that reflects the dynamic structure and group electromagnetic effects of UAV swarms based on target characteristics has become an important prerequisite for improving the accuracy of swarm target perception.

[0004] At the target detection level, single radar systems also face significant challenges when dealing with UAV swarms. Limited by angular resolution and observation angle, single radar systems are prone to problems such as indistinguishable targets, increased false alarms and missed detections under high-density, small-angle-spaced conditions, making it difficult to reliably acquire high-quality measurement information reflecting the swarm structure. Even with high-resolution processing methods, under complex electromagnetic backgrounds and target energy coupling conditions, detection results are still easily affected by factors such as main lobe broadening and side lobe leakage, thus limiting subsequent tracking performance. Therefore, conducting detection optimization research at the single radar level for swarm target resolution is a crucial foundational step in achieving refined perception of UAV swarms.

[0005] In multi-radar cooperative perception, multi-station radar systems theoretically possess multi-view advantages, helping to alleviate the limitations of single-radar perspectives and insufficient resolution. However, existing multi-radar processing methods mostly focus on point-level or track-level fusion, typically assuming that a single radar can provide reliable and independent target detection results. This makes it difficult to address the inherent uncertainties and error accumulation issues in the detection phase of UAV swarm scenarios. Without close coordination between detection and tracking, or without fully utilizing multi-radar observation information at the signal level, it is difficult to realize the potential advantages of multi-radar systems in perceiving dense swarm targets. Therefore, researching multi-radar cooperative detection and tracking methods for UAV swarms and achieving cooperative optimization from detection to tracking is a key direction for improving the overall perception capability of swarm targets.

[0006] In summary, given the high dynamism, dense distribution, and complex electromagnetic characteristics of UAV swarms under radar observation, it is urgent to conduct systematic research from multiple levels, including target characteristic modeling, single radar detection optimization, and multi-radar collaborative detection and tracking, to provide theoretical support and technical foundation for achieving high-precision, multi-dimensional radar perception of UAV swarms. Summary of the Invention

[0007] The purpose of this invention is to provide a multi-radar cooperative detection and tracking method and system for dynamic UAV swarms based on target resolution, which has high model description accuracy and strong adaptability to complex dynamic scenarios.

[0008] The technical solution to achieve the purpose of this invention is: a dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution, comprising the following steps:

[0009] Step 1: Determine the configuration and flight status parameters of the UAV swarm, set up multi-radar deployment information, generate virtual host flight paths and execute follower formation cooperative control strategies to obtain a swarm motion model that satisfies dynamic constraints, and construct a multi-radar observation scenario for the UAV swarm.

[0010] Step 2: Divide the observation scene into observation groups and solve the geometric relationship to obtain the line-of-sight information and corresponding pose parameters of each radar cluster. Using the UAV CAD model and radar system parameters as input, carry out the coupling analysis between occlusion and target under the line-of-sight constraint, calculate the electromagnetic scattering characteristics of cluster targets and construct a multi-radar time-varying scattering center model.

[0011] Step 3: Construct a multi-radar signal-level echo model and generate observation echo data for each radar. Perform beamforming on the observation echo data, identify the main beam and form a local angular domain. Construct range-azimuth spectrum and range-elevation spectrum in the local angular domain respectively. Achieve cluster target resolution and parameter output through joint parameter estimation.

[0012] Step 4: Based on the results of resolving cluster targets using joint parameter estimation, generate single-station dynamic measurement data for each radar. Complete the spatial registration and fusion preprocessing of multi-station measurement data in a common coordinate system, and then perform signal-level multi-radar cooperative tracking and state estimation to continuously output cooperative sensing results.

[0013] A target-resolution-based dynamic UAV swarm multi-radar cooperative detection and tracking system, comprising:

[0014] The multi-radar observation scenario modeling module determines the UAV swarm formation configuration and flight state parameters, sets the multi-radar deployment information, generates virtual host flight paths and executes follower formation cooperative control strategies to obtain a swarm motion model that satisfies dynamic constraints, and constructs a multi-radar observation scenario for the UAV swarm.

[0015] The cluster target characteristic calculation module divides the observation scene into observation groups and solves the geometric relationship, obtains the line-of-sight information of each radar to the cluster and the corresponding pose parameters, and takes the UAV CAD model and radar system parameters as input. Under the line-of-sight constraint, it performs the coupling analysis between occlusion and target, calculates the electromagnetic scattering characteristics of the cluster target and constructs a multi-radar time-varying scattering center model.

[0016] The dense target resolution and detection module constructs a multi-radar signal-level echo model and generates observation echo data from each radar. It performs beamforming on the observation echo data, confirms the main beam and forms a local angular domain. Within the local angular domain, it constructs range-azimuth and range-elevation spectra respectively. Through joint parameter estimation, it achieves cluster target resolution and parameter output.

[0017] The multi-radar cooperative sensing and tracking module generates single-station dynamic measurement data for each radar based on the results of joint parameter estimation to distinguish cluster targets. It then completes spatial registration and fusion preprocessing of multi-station measurement data in a common coordinate system, and performs signal-level multi-radar cooperative tracking and state estimation, continuously outputting cooperative sensing results.

[0018] Compared with the prior art, the significant advantages of this invention are: (1) Radar echo modeling based on target characteristics, signal-level echo modeling is closer to the real scene; (2) Single radar joint parameter estimation based on local angular domain realizes target resolution of dynamic UAV clusters, optimizes detection performance and improves tracking quality; (3) Signal-level multi-radar collaborative detection and tracking makes tracking more accurate in complex scenes; (4) Realizes high-precision, multi-dimensional radar perception of UAV clusters, providing theoretical support and technical foundation for multi-radar collaborative detection and tracking of dynamic UAV clusters. Attached Figure Description

[0019] Figure 1This is a flowchart of the dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution, which is based on the present invention.

[0020] Figure 2 This is a schematic diagram of the performance parameters of the UAV target and the simulation parameters of the radar system in an embodiment of the present invention.

[0021] Figure 3 This is a schematic diagram of a scenario in which multiple radars observe a cluster of drones in an embodiment.

[0022] Figure 4 This is a schematic diagram of the configuration distribution model of the drone swarm in the embodiment.

[0023] Figure 5 This is a graph showing the dynamic RCS data of targets within the UAV cluster under radar 1 in the embodiment, as well as the radar echo pulse compression result curve.

[0024] Figure 6 This is a graph showing the dynamic RCS data of targets within the UAV cluster under radar 2 in the embodiment, as well as the radar echo pulse compression result curve.

[0025] Figure 7 This is a graph showing the dynamic RCS data of targets within the UAV cluster under radar 3 in the embodiment, as well as the radar echo pulse compression result curve.

[0026] Figure 8 This is a schematic diagram of the target resolution results of the UAV swarm based on joint parameter estimation in the embodiment.

[0027] Figure 9 This is a schematic diagram of the spatiotemporal alignment and spatial registration results of the multiple radars on both sides in the embodiment.

[0028] Figure 10 This is a schematic diagram of the multi-station data fusion and preprocessing results in the embodiment.

[0029] Figure 11 This is a schematic diagram of the signal-level multi-radar collaborative detection and tracking results in the embodiment.

[0030] Figure 12 This is a curve comparing the results of multi-radar collaborative precision sensing with the results of traditional single-radar tracking in this embodiment. Detailed Implementation

[0031] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0032] like Figure 1 As shown, the present invention provides a dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution, comprising the following steps:

[0033] Step 1: Determine the configuration and flight status parameters of the UAV swarm, set up multi-radar deployment information, generate virtual host flight paths and execute follower formation cooperative control strategies to obtain a swarm motion model that satisfies dynamic constraints, and construct a multi-radar observation scenario for the UAV swarm.

[0034] Step 2: Divide the observation scene into observation groups and solve the geometric relationship to obtain the line-of-sight information and corresponding pose parameters of each radar cluster. Using the UAV CAD model and radar system parameters as input, carry out the coupling analysis between occlusion and target under the line-of-sight constraint, calculate the electromagnetic scattering characteristics of cluster targets and construct a multi-radar time-varying scattering center model.

[0035] Step 3: Construct a multi-radar signal-level echo model and generate observation echo data for each radar. Perform beamforming on the observation echo data, identify the main beam and form a local angular domain. Construct range-azimuth spectrum and range-elevation spectrum in the local angular domain respectively. Achieve cluster target resolution and parameter output through joint parameter estimation.

[0036] Step 4: Based on the results of resolving cluster targets using joint parameter estimation, generate single-station dynamic measurement data for each radar. Complete the spatial registration and fusion preprocessing of multi-station measurement data in a common coordinate system, and then perform signal-level multi-radar cooperative tracking and state estimation to continuously output cooperative sensing results.

[0037] As a specific example, step 1 is as follows:

[0038] Step 1.1: Based on the performance parameter constraints of the UAV swarm target, determine the UAV swarm formation configuration and flight state parameter range to provide constraints for motion modeling;

[0039] Step 1.2: Set up multi-radar deployment information under a unified coordinate system, determine the location of each radar, and form a multi-radar spatial deployment scheme for UAV swarm observation;

[0040] Step 1.3: Generate virtual host tracks and design and execute follower formation cooperative control strategies to obtain a cluster motion model that satisfies dynamic constraints;

[0041] Step 1.4: Based on the multi-radar spatial deployment scheme and swarm motion model, establish the observation geometric relationship between the multi-radar and UAV swarm in the common coordinate system to form a time-series observation scenario of the UAV swarm by the multi-radar, providing scenario input for target characteristic modeling and collaborative detection and tracking.

[0042] As a specific example, step 1.3 is as follows:

[0043] Step 1.3.1: Input drone parameters and initial virtual host location. and the initial state of each follower The drone parameters include the number of drones in formation, relative offset parameters, and speed parameters.

[0044] Step 1.3.2: Recursively update the virtual host status according to the preset track command, generate the virtual host track, and determine the expected formation position of each follower based on the current position of the virtual host and the formation relative offset parameter, as shown in the following formula:

[0045]

[0046]

[0047]

[0048] ,

[0049] in, For virtual hosts at any time The heading angle, For virtual hosts at any time The heading angle, For virtual hosts at any time angular velocity of heading, For discrete time steps, For virtual hosts at any time pitch angle, For a moment Given the pitch angle command, For virtual hosts at any time Location, For virtual hosts at any time Location, For virtual hosts at any time speed, Indicates the first The relative offset vectors of each follower in the formation, where , and The first One follower Relative offset in the three axes Indicates the first A follower at a moment The expected formation position This indicates the orbital angle determined by the virtual host's heading angle. Axis rotation matrix;

[0050] Step 1.3.3: Based on the positional error between the current position and the desired formation position of each follower, and the velocity error between the current velocity of each follower and the virtual host, construct the formation cooperative control law, generate the control input for each follower, and dynamically update the position and velocity states of each follower to obtain a cluster motion model that satisfies the dynamic constraints, as shown in the following formula:

[0051] ,

[0052]

[0053]

[0054] in, Indicates the first The actual position vector of each follower Indicates the first The actual velocity vector of each follower Represents the virtual host velocity vector. Indicates the first The positional error of each follower Indicates speed error; Indicates time Follower-controlled input, , These represent the position error gain matrix and the velocity error gain matrix, respectively. Indicates the virtual host reference control variable. Represents gravitational acceleration; , and These represent the heading angle, pitch angle, and roll angle of the nth follower, respectively. This represents the transpose of the coordinate transformation rotation matrix. , and These represent the longitudinal, lateral, and vertical accelerations of the nth follower, respectively.

[0055] The inherent correlation of the cluster motion model is reflected in the fact that the virtual host track determines the overall motion trend of the cluster, the relative offset relationship of the formation determines the expected formation position of each follower, and the error feedback control input drives the dynamic update of the follower state, thus realizing the correlation modeling between the overall motion trend of the cluster, the spatial configuration of the formation and the temporal evolution of the followers.

[0056] Step 1.3.4: To ensure that control commands remain within the physical execution capabilities of the UAV and to avoid collisions within the formation, it is necessary to saturate and limit the control inputs, and to monitor and correct dangerous maneuvers and formation spacing in real time. Specifically:

[0057] (1) When the control input exceeds the preset range, saturation processing is performed, specifically for the roll angle. The amplitude limiting formula is as follows:

[0058]

[0059] in, This indicates that the command controls the input. This indicates that the input limit is controlled. Represents a saturation function;

[0060] (2) Monitor flight safety from the perspective of maneuver overload and define the maximum overload coefficient. Real-time calculation of the vertical acceleration of the follower When satisfied In such cases, the control inputs may be further limited or the maneuver commands may be reprogrammed to prevent the drone from being damaged or out of control due to excessive overload.

[0061] (3) Real-time monitoring of the relative distance between targets; when the relative distance between targets... Less than the safe distance Right now At that time, the desired formation position is corrected;

[0062] Final output virtual host location sequence and the state sequence of each follower .

[0063] As a specific example, step 2 is as follows:

[0064] Step 2.1: Based on the multi-radar observation scenario, the UAV swarm targets are grouped for observation to obtain the set of observed targets corresponding to each radar at the current observation time;

[0065] Step 2.2: Based on the observation grouping results, calculate the observation geometric relationship between multiple radars and each set of observed targets to obtain the radar line-of-sight information and the pose distribution of the targets corresponding to each radar.

[0066] Step 2.3: Using the UAV CAD model and corresponding radar system parameters as input, conduct occlusion relationship analysis and electromagnetic coupling analysis between targets under line-of-sight and pose constraints.

[0067] Step 2.4: Based on the results of the occlusion relationship analysis and the electromagnetic coupling analysis between targets, calculate the electromagnetic scattering characteristics of the UAV swarm and construct a time-varying scattering center model under multiple radar views. The expression is as follows:

[0068]

[0069] in, Indicates the first The first radar unit The observation group at the observation time The composite scattering field Indicates the radar number, Indicates the observation group number, Indicates the observation time. Indicates the first Radar of the first Each observation group at the observation time The set of scattering centers It is the first The amplitude of each scattering center For the scattering center index, The imaginary unit, It is the speed at which electromagnetic waves propagate in free space. For the first The carrier frequency of the radar It is the first The three-dimensional position vectors of each scattering center in the body coordinate system; Represents an exponential function; The unit vector of the radar line-of-sight direction and , , These are the azimuth and pitch angles in the body coordinate system, respectively.

[0070] As a specific example, step 3 is as follows:

[0071] Step 3.1: Based on the time-varying scattering center model and radar system parameters, construct a multi-radar signal-level echo model and generate the corresponding observation echo data for each radar.

[0072] Step 3.2: Perform beamforming processing on the observed echo data, confirm the main beam direction through spectral peak search, and extract the local angular domain used in subsequent processing;

[0073] Step 3.3: Construct the range-azimuth spectrum and range-elevation spectrum in the local angular domain, normalize the two spectra, and perform product consistency fusion in the local angular domain to achieve consistency constraints and pairing of the peak relationships of range, azimuth and elevation.

[0074] Step 3.4: Perform joint parameter estimation based on the range-azimuth spectrum and the range-elevation spectrum, and output the range, azimuth and elevation parameters of the cluster targets to achieve target resolution and parameter output results.

[0075] As a specific example, step 3.1, which involves constructing a multi-radar signal-level echo model based on the time-varying scattering center model and radar system parameters, and generating the corresponding observation echo data for each radar, is detailed as follows:

[0076] Step 3.1.1: Based on the three-dimensional scattering center model, extract key target characteristic parameters for radar echo construction, including the time-varying radar cross section and the equivalent scattering centroid position, as the core input parameters for echo modeling. The expression is:

[0077]

[0078]

[0079] in, Indicates the first Radar of the first The first observation group The target at the observation time radar cross section, Indicates the first Radar of the first The first observation group The complex number of echoes corresponding to each target For the first Radar of the first The first observation group The set of three-dimensional scattering centers corresponding to each target; This represents the direction vector of the target relative to the radar. They represent the first In the observation group, the target is relative to the first The azimuth and elevation angles of the radar; Indicates the first Radar of the first The first observation group Consider the equivalent position of each target after occlusion;

[0080] Step 3.1.2: Using the target characteristic parameters, generate the target echo signal corresponding to each radar. The echo modeling process comprehensively considers signal propagation delay, Doppler modulation caused by target motion, range propagation attenuation, antenna pattern gain, and receiver additive noise, thereby achieving an accurate characterization of the electromagnetic response behavior of UAV swarm targets under multi-radar observation conditions. The expression is as follows:

[0081]

[0082] in, For the first Radar at observation time The received echo signal The index of the equivalent scattering unit within the observation group, used to characterize the first... The equivalent scattering unit corresponding to each target after modeling at the scattering center; The number of observation groups, The number of equivalent scattering units within each observation group. In order to transmit signals, For the first Radar and the first Within the observation group, the first Round-trip propagation delay between equivalent scattering units For the first Within the observation group, the first Doppler frequency shift corresponding to each equivalent scattering unit For carrier wavelength, For the first Radar and the first Within the observation group, the first Radar cross section of an equivalent scattering element For the first Radar and the first Within the observation group, the first The instantaneous distance of an equivalent scattering unit. For beamforming weight vectors, for The conjugate transpose of . The azimuth and elevation angles are represented as The array guide vector, The first Within the observation group, the first The equivalent scattering unit relative to the first The azimuth and elevation angles of the radar. For the first Noise model on the radar array channel, For the first Radar at observation time The multi-channel array receives the echo signal.

[0083] As a specific example, step 3.3 involves constructing a range-azimuth spectrum and a range-elevation spectrum within the local angular domain, normalizing the two spectra, and performing product consistency fusion within the local angular domain to achieve consistency constraints and pairing of the peak relationships of range, azimuth, and elevation. The details are as follows:

[0084] Step 3.3.1: Form the range-azimuth spectrum within the local angular domain. With distance-elevation spectrum The expressions are as follows:

[0085]

[0086]

[0087] in, For distance variables, It is the azimuth angle. The pitch angle, To determine the azimuth angle corresponding to the main beam in beamforming, The elevation angle corresponding to the main beam is determined for beamforming. , Each of the corresponding angles , The array guide vector, , They are respectively , The conjugate transpose of . Indicates distance unit Observation time The noise subspace matrix is ​​constructed from the received data. for The conjugate transpose of;

[0088] Step 3.3.2: To ensure consistent constraints and pairing of the peak values ​​for distance, azimuth, and elevation, the two spectra are normalized, with the following expressions:

[0089]

[0090]

[0091] in, This is the normalized distance-azimuth spectrum. The normalized distance-elevation spectrum. and These represent the maximum value of the spectral amplitude in the direction angle dimension and the elevation angle dimension, respectively;

[0092] Step 3.3.3: Perform product-consistency fusion on the normalized spectrum within the local angular domain to construct a joint spectral domain representation, expressed as:

[0093] ,

[0094] in, The joint consistency spectrum function of range-azimuth-elevation. , These are the center azimuth and elevation angles determined by the main beam, respectively. , These represent the azimuth and elevation ranges of the local angular domain, respectively. The local angular domain range defined for the construction of the joint spectrum.

[0095] As a specific example, step 4 is as follows:

[0096] Step 4.1: Within each radar station, based on the results of resolving cluster targets using joint parameter estimation, generate dynamic measurement data for each station that is updated over time. The measurement data includes range, azimuth, and elevation information, thereby improving the cluster target resolution capability and optimizing detection performance.

[0097] Step 4.2: Map the dynamic measurements of each radar station to a common coordinate system to complete the time alignment and spatial registration of the multi-station measurement data, forming a multi-station measurement set that can be associated and fused under a unified geometric framework;

[0098] Step 4.3: Perform fusion preprocessing on the registered multi-station measurements, including outlier measurement suppression, point aggregation and merging, and cross-station consistency processing, to obtain high-quality fused measurement input for collaborative tracking;

[0099] Step 4.4: Based on the fusion measurement input, perform signal-level multi-radar cooperative tracking and state estimation, and continuously output cooperative perception results that can characterize the location, spatial distribution, shape and size information of the UAV swarm.

[0100] As a specific example, step 4.4 describes the signal-level multi-radar cooperative tracking and state estimation based on fused measurement input, continuously outputting cooperative perception results that characterize the location, spatial distribution, shape, and size information of the UAV swarm, as follows:

[0101] Step 4.4.1: Define the cooperative tracking state of the UAV swarm as a state vector containing position and extended attributes. At each observation time, each radar obtains the corresponding local measurement parameters, expressed as:

[0102]

[0103] in, For the first Radar at observation time Local measurement vectors; , and The first Radar at observation time Target distance measurement, azimuth measurement, and elevation angle measurement. , The first Radar at observation time Shape characteristics and scale characteristics;

[0104] Step 4.4.2: Based on the geometric relationship of multi-radar observations, establish an observation model between the target state and radar measurements to characterize the mapping relationship of the target state in each radar coordinate system. The expression is:

[0105]

[0106] in, No. Radar at observation time The measurement vector, For the first The nonlinear measurement function corresponding to the radar; To observe the noise term, satisfy the following conditions: , This indicates that the mean is 0 and the covariance matrix is... Gaussian distribution, for The corresponding noise covariance matrix;

[0107] Step 4.4.3: For each radar, construct the signal-level measurement likelihood function of a single radar based on the local measurements obtained at the current observation time. ;

[0108] Step 4.4.4: Based on the single-radar signal-level likelihood functions constructed for each radar, a multi-radar joint update function is formed to achieve the fusion of multi-radar observation information at the signal level, obtaining the continuous state estimation result of the cluster target. The expression is:

[0109]

[0110] in, This is the state estimation vector for the collaborative tracking of cluster targets. For the collaborative tracking state vector of the cluster target, For the number of radars, For the first Radar at observation time The weighting coefficients, For the global state up to the th A physical mapping model of the local state of the radar; For extrapolated prior information used in joint updates, These are the state mean and state covariance matrix from the extrapolated prior information, respectively.

[0111] Step 4.4.5: Based on the cooperative state estimation, the overall shape and scale parameters of the cluster are stably estimated through a time-series recursive method, so that the output cooperative sensing results can not only reflect the motion state of the cluster center, but also continuously characterize the spatial structural evolution characteristics of the cluster. The expression is:

[0112]

[0113] in, , These represent the predicted and updated values ​​of the scale matrix, respectively. As a smoothing factor, The location of the group centroid after merging. The global coordinates of the local measurement point;

[0114] Eigenvalue decomposition is performed on the updated shape matrix to obtain the axial directions and axial lengths of the ellipsoid, expressed as:

[0115]

[0116] in, The eigenvector matrix, It is a diagonal matrix composed of eigenvalues, used to characterize the spatial scale and shape features of the cluster along the principal axis, thereby describing the spatial distribution pattern of the group of targets;

[0117]

[0118]

[0119] in, For the observation time The scale measurement weighted value, For the observation time Time-smoothed size estimate The previous observation time Time-smoothed size estimate This indicates the size measurement value of the corresponding radar. Indicates the index of discrete observation counts. For the weights corresponding to the measurements, the index pairs Representing the number of discrete observations and the th observation, respectively. Radar index, index set This represents the sample set obtained after removing outlier size measurements. This is the time smoothing factor.

[0120] This invention also provides a target-resolution-based dynamic UAV swarm multi-radar cooperative detection and tracking system, which is used to implement the aforementioned target-resolution-based dynamic UAV swarm multi-radar cooperative detection and tracking method, including:

[0121] The multi-radar observation scenario modeling module determines the UAV swarm formation configuration and flight state parameters, sets the multi-radar deployment information, generates virtual host flight paths and executes follower formation cooperative control strategies to obtain a swarm motion model that satisfies dynamic constraints, and constructs a multi-radar observation scenario for the UAV swarm.

[0122] The cluster target characteristic calculation module divides the observation scene into observation groups and solves the geometric relationship, obtains the line-of-sight information of each radar to the cluster and the corresponding pose parameters, and takes the UAV CAD model and radar system parameters as input. Under the line-of-sight constraint, it performs the coupling analysis between occlusion and target, calculates the electromagnetic scattering characteristics of the cluster target and constructs a multi-radar time-varying scattering center model.

[0123] The dense target resolution and detection module constructs a multi-radar signal-level echo model and generates observation echo data from each radar. It performs beamforming on the observation echo data, confirms the main beam and forms a local angular domain. Within the local angular domain, it constructs range-azimuth and range-elevation spectra respectively. Through joint parameter estimation, it achieves cluster target resolution and parameter output.

[0124] The multi-radar cooperative sensing and tracking module generates single-station dynamic measurement data for each radar based on the results of joint parameter estimation to distinguish cluster targets. It then completes spatial registration and fusion preprocessing of multi-station measurement data in a common coordinate system, and performs signal-level multi-radar cooperative tracking and state estimation, continuously outputting cooperative sensing results.

[0125] The present invention also provides a mobile terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the aforementioned target-resolution-based dynamic UAV swarm multi-radar cooperative detection and tracking method.

[0126] Example

[0127] This embodiment uses a simulation experiment based on a target resolution-based dynamic UAV swarm multi-radar cooperative detection and tracking method provided by the present invention. Figure 2 This is a schematic diagram showing the performance parameters of the UAV target and the simulation parameters of the radar system. Figure 3 This is a scenario where multiple radars observe a swarm of drones. Figure 4 It is a configuration and distribution model of drone swarms. Figure 5 These are the dynamic RCS data of targets within the UAV swarm under radar 1 and the results of radar echo pulse compression. Figure 6 These are the dynamic RCS data of targets within the UAV swarm under Radar 2 and the results of radar echo pulse compression. Figure 7 It is the dynamic RCS data of targets within the UAV swarm under radar 3 and the radar echo pulse compression results. Figure 8 The results of UAV swarm target discrimination based on joint parameter estimation are presented for radar 1 to radar 3, showing that it can effectively distinguish each target in the swarm under dense target conditions, and improve the discriminability of single radar detection and the quality of parameter estimation. Figure 9 The spatiotemporal alignment and spatial registration results of multi-radar measurements are presented, which can provide a unified benchmark for subsequent fusion and collaborative processing; Figure 10The results of multi-station data fusion and preprocessing are presented, demonstrating that fusion preprocessing can suppress abnormal measurements and improve the measurement quality of collaborative tracking. Based on this, Figure 11 The results of signal-level multi-radar cooperative detection and tracking are presented, which can stably output cooperative sensing results such as cluster location and spatial distribution; Figure 12 The accuracy comparison between multi-radar collaborative perception results and traditional single-radar tracking results is presented, demonstrating the ability to accurately track UAV swarms and showcasing the advantages over traditional processes.

[0128] This invention achieves high-precision, multi-dimensional radar perception of UAV swarms, optimizes detection performance, improves tracking quality, and provides theoretical support and technical foundation for multi-radar collaborative detection and tracking of dynamic UAV swarms.

[0129] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution, characterized in that, Includes the following steps: Step 1: Determine the configuration and flight status parameters of the UAV swarm, set up multi-radar deployment information, generate virtual host flight paths and execute follower formation cooperative control strategies to obtain a swarm motion model that satisfies dynamic constraints, and construct a multi-radar observation scenario for the UAV swarm. Step 2: Divide the observation scene into observation groups and solve the geometric relationship to obtain the line-of-sight information and corresponding pose parameters of each radar cluster. Using the UAV CAD model and radar system parameters as input, carry out the coupling analysis between occlusion and target under the line-of-sight constraint, calculate the electromagnetic scattering characteristics of cluster targets and construct a multi-radar time-varying scattering center model. Step 3: Construct a multi-radar signal-level echo model and generate observation echo data for each radar. Perform beamforming on the observation echo data, identify the main beam and form a local angular domain. Construct range-azimuth spectrum and range-elevation spectrum in the local angular domain respectively. Achieve cluster target resolution and parameter output through joint parameter estimation. Step 4: Based on the results of resolving cluster targets using joint parameter estimation, generate single-station dynamic measurement data for each radar. Complete the spatial registration and fusion preprocessing of multi-station measurement data in a common coordinate system, and then perform signal-level multi-radar cooperative tracking and state estimation to continuously output cooperative sensing results.

2. The dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution according to claim 1, characterized in that, Step 1 is described in detail as follows: Step 1.1: Based on the performance parameter constraints of the UAV swarm target, determine the UAV swarm formation configuration and flight state parameter range to provide constraints for motion modeling; Step 1.2: Set up multi-radar deployment information under a unified coordinate system, determine the location of each radar, and form a multi-radar spatial deployment scheme for UAV swarm observation; Step 1.3: Generate virtual host tracks and design and execute follower formation cooperative control strategies to obtain a cluster motion model that satisfies dynamic constraints; Step 1.4: Based on the multi-radar spatial deployment scheme and swarm motion model, establish the observation geometric relationship between the multi-radar and UAV swarm in the common coordinate system to form a time-series observation scenario of the UAV swarm by the multi-radar, providing scenario input for target characteristic modeling and collaborative detection and tracking.

3. The dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution according to claim 2, characterized in that, Step 1.3 is as follows: Step 1.3.1: Input drone parameters and initial virtual host location. and the initial state of each follower The drone parameters mentioned above include the number of drones in formation, relative offset parameters, and speed parameters. Step 1.3.2: Recursively update the virtual host status according to the preset track command, generate the virtual host track, and determine the expected formation position of each follower based on the current position of the virtual host and the formation relative offset parameter, as shown in the following formula: ， in, For virtual hosts at any time The heading angle, For virtual hosts at any time The heading angle, For virtual hosts at any time angular velocity of heading, For discrete time steps, For virtual hosts at any time pitch angle, For a moment Given pitch angle command, For virtual hosts at any time Location, For virtual hosts at any time Location, For virtual hosts at any time speed, Indicates the first The relative offset vectors of each follower in the formation, where , and The first One follower Relative offset in the three axes Indicates the first A follower at a moment The expected formation position This indicates the orbital angle determined by the virtual host's heading angle. Axis rotation matrix; Step 1.3.3: Based on the positional error between the current position and the desired formation position of each follower, and the velocity error between the current velocity of each follower and the virtual host, construct the formation cooperative control law, generate the control input for each follower, and dynamically update the position and velocity states of each follower to obtain a cluster motion model that satisfies the dynamic constraints, as shown in the following formula: ， in, Indicates the first The actual position vector of each follower Indicates the first The actual velocity vector of each follower Represents the virtual host velocity vector. Indicates the first The positional error of each follower Indicates speed error; Indicates time Follower-controlled input, , These represent the position error gain matrix and the velocity error gain matrix, respectively. Indicates the virtual host reference control variable. Represents gravitational acceleration; , and These represent the heading angle, pitch angle, and roll angle of the nth follower, respectively. This represents the transpose of the coordinate transformation rotation matrix. , and These represent the longitudinal, lateral, and vertical accelerations of the nth follower, respectively. The inherent correlation of the cluster motion model is reflected in the fact that the virtual host track determines the overall motion trend of the cluster, the relative offset relationship of the formation determines the expected formation position of each follower, and the error feedback control input drives the dynamic update of the follower state, thus realizing the correlation modeling between the overall motion trend of the cluster, the spatial configuration of the formation and the temporal evolution of the followers. Step 1.3.4: To ensure that control commands remain within the physical execution capabilities of the UAV and to avoid collisions within the formation, it is necessary to saturate and limit the control inputs, and to monitor and correct dangerous maneuvers and formation spacing in real time. Specifically: (1) When the control input exceeds the preset range, saturation processing is performed, specifically for the roll angle. The amplitude limiting formula is as follows: in, This indicates that the command controls the input. This indicates that the input limit is controlled. Represents a saturation function; (2) Monitor flight safety from the perspective of maneuver overload and define the maximum overload coefficient. Real-time calculation of the vertical acceleration of the follower When satisfied In such cases, the control inputs may be further limited or the maneuver commands may be reprogrammed to prevent the drone from being damaged or out of control due to excessive overload. (3) Real-time monitoring of the relative distance between targets; when the relative distance between targets... Less than the safe distance Right now At that time, the desired formation position is corrected; Final output virtual host location sequence and the state sequence of each follower .

4. The dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution according to claim 3, characterized in that, Step 2 is described in detail below: Step 2.1: Based on the multi-radar observation scenario, the UAV swarm targets are grouped for observation to obtain the set of observed targets corresponding to each radar at the current observation time; Step 2.2: Based on the observation grouping results, calculate the observation geometric relationship between multiple radars and each set of observed targets to obtain the radar line-of-sight information and the pose distribution of the targets corresponding to each radar. Step 2.3: Using the UAV CAD model and corresponding radar system parameters as input, conduct occlusion relationship analysis and electromagnetic coupling analysis between targets under line-of-sight and pose constraints. Step 2.4: Based on the results of the occlusion relationship analysis and the electromagnetic coupling analysis between targets, calculate the electromagnetic scattering characteristics of the UAV swarm and construct a time-varying scattering center model under multiple radar views. The expression is as follows: in, Indicates the first Radar of the first Each observation group at the observation time The composite scattering field Indicates the radar number, Indicates the observation group number, Indicates the observation time. Indicates the first Radar of the first Each observation group at the observation time The set of scattering centers It is the first The amplitude of each scattering center For the scattering center index, The imaginary unit, It is the speed at which electromagnetic waves propagate in free space. For the first The carrier frequency of the radar It is the first The three-dimensional position vectors of each scattering center in the body coordinate system; Represents an exponential function; The unit vector of the radar line-of-sight direction and , , These are the azimuth and pitch angles in the body coordinate system, respectively.

5. The dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution according to claim 4, characterized in that, Step 3 is described in detail below: Step 3.1: Based on the time-varying scattering center model and radar system parameters, construct a multi-radar signal-level echo model and generate the corresponding observation echo data for each radar. Step 3.2: Perform beamforming processing on the observed echo data, confirm the main beam direction through spectral peak search, and extract the local angular domain used in subsequent processing; Step 3.3: Construct the range-azimuth spectrum and range-elevation spectrum in the local angular domain, normalize the two spectra, and perform product consistency fusion in the local angular domain to achieve consistency constraints and pairing of the peak relationships of range, azimuth and elevation. Step 3.4: Perform joint parameter estimation based on the range-azimuth spectrum and the range-elevation spectrum, and output the range, azimuth and elevation parameters of the cluster targets to achieve target resolution and parameter output results.

6. The dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution according to claim 5, characterized in that, Step 3.1, which involves constructing a multi-radar signal-level echo model based on the time-varying scattering center model and radar system parameters, and generating the corresponding observation echo data for each radar, is detailed below: Step 3.1.1: Based on the three-dimensional scattering center model, extract key target characteristic parameters for radar echo construction, including the time-varying radar cross section and the equivalent scattering centroid position, as the core input parameters for echo modeling. The expression is: in, Indicates the first Radar of the first The first observation group The target at the observation time radar cross section, Indicates the first Radar of the first The first observation group The complex number of echoes corresponding to each target For the first Radar of the first The first observation group The set of three-dimensional scattering centers corresponding to each target; This represents the direction vector of the target relative to the radar. They represent the first In the observation group, the target is relative to the first The azimuth and elevation angles of the radar; Indicates the first Radar of the first The first observation group Consider the equivalent position of each target after occlusion; Step 3.1.2: Using the target characteristic parameters, generate the target echo signal corresponding to each radar. In the echo modeling process, the signal propagation delay, Doppler modulation caused by target motion, range propagation attenuation, antenna pattern gain, and receiver additive noise factors are comprehensively considered to achieve an accurate characterization of the electromagnetic response behavior of UAV swarm targets under multi-radar observation conditions. The expression is as follows: in, For the first radar at observation time The received echo signal; The index of the equivalent scattering unit within the observation group, used to characterize the first... The equivalent scattering unit corresponding to each target after modeling at the scattering center; The number of observation groups, The number of equivalent scattering units within each observation group. In order to transmit signals, For the first Radar and the first Within the observation group, the first Round-trip propagation delay between equivalent scattering units For the first Within the observation group, the first Doppler frequency shift corresponding to each equivalent scattering unit For carrier wavelength, For the first Radar and the first Within the observation group, the first Radar cross section of an equivalent scattering element For the first Radar and the first Within the observation group, the first The instantaneous distance of an equivalent scattering unit. For beamforming weight vectors, for The conjugate transpose of . The azimuth and elevation angles are represented as The array guide vector, The first Within the observation group, the first The equivalent scattering unit relative to the first The azimuth and elevation angles of the radar. For the first Noise model on the radar array channel, For the first Radar at observation time The multi-channel array receives the echo signal.

7. The dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution according to claim 6, characterized in that, Step 3.3 involves constructing range-azimuth and range-elevation spectra within the local angular domain, normalizing the two spectra, and performing product consistency fusion within the local angular domain to achieve consistency constraints and pairing of the peak relationships of range, azimuth, and elevation. The details are as follows: Step 3.3.1: Form the range-azimuth spectrum within the local angular domain. With distance-elevation spectrum The expressions are as follows: in, For distance variables, It is the azimuth angle. The pitch angle, To determine the azimuth angle corresponding to the main beam in beamforming, The elevation angle corresponding to the main beam is determined for beamforming. , Each of the corresponding angles , The array guide vector, , They are respectively , The conjugate transpose of . Indicates distance unit Observation time The noise subspace matrix is ​​constructed from the received data. for The conjugate transpose of; Step 3.3.2: To ensure consistent constraints and pairing of the peak values ​​for distance, azimuth, and elevation, the two spectra are normalized, with the following expressions: in, This is the normalized distance-azimuth spectrum. The normalized distance-elevation spectrum. and These represent the maximum value of the spectral amplitude in the direction angle dimension and the elevation angle dimension, respectively; Step 3.3.3: Perform product-consistency fusion on the normalized spectrum within the local angular domain to construct a joint spectral domain representation, expressed as: , in, The joint consistency spectrum function of range-azimuth-elevation. , These are the center azimuth and elevation angles determined by the main beam, respectively. , These represent the azimuth and elevation ranges of the local angular domain, respectively. The local angular domain range defined for the construction of the joint spectrum.

8. The dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution according to claim 7, characterized in that, Step 4 is described in detail below: Step 4.1: Within each radar station, based on the results of resolving cluster targets using joint parameter estimation, generate dynamic measurement data for each station that is updated over time. The measurement data includes range, azimuth, and elevation information, thereby improving the cluster target resolution capability and optimizing detection performance. Step 4.2: Map the dynamic measurements of each radar station to a common coordinate system to complete the time alignment and spatial registration of the multi-station measurement data, forming a multi-station measurement set that can be associated and fused under a unified geometric framework; Step 4.3: Perform fusion preprocessing on the registered multi-station measurements, including outlier measurement suppression, point aggregation and merging, and cross-station consistency processing, to obtain high-quality fused measurement input for collaborative tracking; Step 4.4: Based on the fusion measurement input, perform signal-level multi-radar cooperative tracking and state estimation, and continuously output cooperative perception results that can characterize the location, spatial distribution, shape and size information of the UAV swarm.

9. The dynamic UAV swarm multi-radar cooperative detection and tracking method based on target resolution according to claim 8, characterized in that, Step 4.4 describes the execution of signal-level multi-radar cooperative tracking and state estimation based on fused measurement input, continuously outputting cooperative perception results that characterize the location, spatial distribution, shape, and size information of the UAV swarm, as detailed below: Step 4.4.1: Define the cooperative tracking state of the UAV swarm as a state vector containing position and extended attributes. At each observation time, each radar obtains the corresponding local measurement parameters, expressed as: in, For the first Radar at observation time Local measurement vectors; , and The first Radar at observation time Target distance measurement, azimuth measurement, and elevation angle measurement. , The first Radar at observation time Shape characteristics and scale characteristics; Step 4.4.2: Based on the geometric relationship of multi-radar observations, establish an observation model between the target state and radar measurements to characterize the mapping relationship of the target state in each radar coordinate system. The expression is: in, For the first Radar at observation time The measurement vector, For the first The nonlinear measurement function corresponding to the radar; To observe the noise term, satisfy the following conditions: , This indicates that the mean is 0 and the covariance matrix is... Gaussian distribution, for The corresponding noise covariance matrix; Step 4.4.3: For each radar, construct the signal-level measurement likelihood function of a single radar based on the local measurements obtained at the current observation time. ; Step 4.4.4: Based on the single-radar signal-level likelihood functions constructed for each radar, a multi-radar joint update function is formed to achieve the fusion of multi-radar observation information at the signal level, obtaining the continuous state estimation result of the cluster target. The expression is: in, For the cooperative tracking state estimation vector of the cluster target, For the collaborative tracking state vector of the cluster target, For the number of radars, For the first Radar at observation time The weighting coefficients, For the global state up to the th A physical mapping model of the local state of the radar; For extrapolation prior information used in joint updates, These are the state mean and state covariance matrix from the extrapolated prior information, respectively. Step 4.4.5: Based on the cooperative state estimation, the overall shape and scale parameters of the cluster are stably estimated through a time-series recursive method, so that the output cooperative sensing results can not only reflect the motion state of the cluster center, but also continuously characterize the spatial structural evolution characteristics of the cluster. The expression is: in, , These represent the predicted and updated values ​​of the scale matrix, respectively. As a smoothing factor, The location of the group centroid after merging. The global coordinates of the local measurement point; Eigenvalue decomposition is performed on the updated shape matrix to obtain the axial directions and axial lengths of the ellipsoid, expressed as: in, The eigenvector matrix, It is a diagonal matrix composed of eigenvalues, used to characterize the spatial scale and shape features of the cluster along the principal axis, thereby describing the spatial distribution pattern of the group of targets; in, For the observation time The scale measurement weighted value, For the observation time Time-smoothed size estimate, The previous observation time Time-smoothed size estimate, This indicates the size measurement value of the corresponding radar. Indicates the index of discrete observation counts. For the weights of the corresponding measurements, the index pairs Representing the number of discrete observations and the th observation, respectively. Radar index, index set This represents the sample set obtained after removing outlier size measurements. This is the time smoothing factor.

10. A dynamic UAV swarm multi-radar cooperative detection and tracking system based on target resolution, characterized in that, This system is used to implement the target resolution-based dynamic UAV swarm multi-radar cooperative detection and tracking method according to any one of claims 1 to 9, comprising: The multi-radar observation scenario modeling module determines the UAV swarm formation configuration and flight state parameters, sets the multi-radar deployment information, generates virtual host flight paths and executes follower formation cooperative control strategies to obtain a swarm motion model that satisfies dynamic constraints, and constructs a multi-radar observation scenario for the UAV swarm. The cluster target characteristic calculation module divides the observation scene into observation groups and solves the geometric relationship, obtains the line-of-sight information of each radar to the cluster and the corresponding pose parameters, and takes the UAV CAD model and radar system parameters as input. Under the line-of-sight constraint, it performs the coupling analysis between occlusion and target, calculates the electromagnetic scattering characteristics of the cluster target and constructs a multi-radar time-varying scattering center model. The dense target resolution and detection module constructs a multi-radar signal-level echo model and generates observation echo data from each radar. It performs beamforming on the observation echo data, confirms the main beam and forms a local angular domain. Within the local angular domain, it constructs range-azimuth and range-elevation spectra respectively. Through joint parameter estimation, it achieves cluster target resolution and parameter output. The multi-radar cooperative sensing and tracking module generates single-station dynamic measurement data for each radar based on the results of joint parameter estimation to distinguish cluster targets. It then completes spatial registration and fusion preprocessing of multi-station measurement data in a common coordinate system, and performs signal-level multi-radar cooperative tracking and state estimation, continuously outputting cooperative sensing results.

Images