Target classification identification and cluster target estimation method based on narrow-band radar detection

Abstract

The invention provides a target classification recognition and cluster target estimation method based on narrow-band radar detection, which detects and recognizes whether a target is a single target or a cluster target according to radar detection information in combination with statistical characteristics of a sum and difference signal monopulse ratio during target tracking, and further utilizes radar resource scheduling for the cluster target to estimate the target classification recognition and cluster target estimation. Scanning and tracking are carried out in a cluster target range by adopting dense beams, the number and boundaries of cluster targets are judged through the signal-to-noise ratio of echo signals in each beam, and finally the geometric center of the cluster targets is estimated through the boundaries of the cluster targets. According to the invention, the method does not need to employ an ultra-wideband radar with higher cost to carry out the classification and recognition of a target two-dimensional image, does not need to carry out the cooperative detection through photoelectric or other detection means, and achieves the classification and recognition and parameter estimation of ground moving personnel, vehicles, ultra-low-altitude unmanned aerial vehicles and unmanned aerial vehicle cluster targets. And estimating parameters of cluster targets in the same distance unit and the same angle unit.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of radar monitoring and detection, and in particular to a target estimation method. BACKGROUND

[0002] The detection and identification of typical "low, small and slow" targets such as ground moving personnel, vehicles, ultra-low altitude unmanned aerial vehicles and unmanned aerial vehicle clusters have always been a research hotspot and technical problem in the field of radar technology. In particular, with the lowering of the threshold of key technologies such as automatic control, positioning and navigation, inertial measurement and telemetry communication, unmanned aerial vehicle systems have gradually become low-cost and civilianized, and unmanned aerial vehicle technology has rapidly spread around the world, greatly reducing the technical threshold of large-area and multi-field applications of swarms. Unmanned aerial vehicle swarms have become a Damocles sword hovering in low altitude / ultra-low altitude. In a complex clutter environment on land, these targets, especially unmanned aerial vehicle clusters, pose a great challenge to radar detection. Cluster targets are generally considered to be in the same distance unit and the same beam of radar detection, so it is impossible to distinguish them through conventional range and angle resolution. Moreover, the targets in the cluster are flying in formation, and the speeds of the targets in the formation are almost identical, so it is also difficult to distinguish them through the speed difference of different targets in the cluster. Therefore, the detection, identification, number estimation and boundary estimation of unmanned aerial vehicle cluster targets are technical difficulties in the field of radar detection and classification.

[0003] The features of target classification and identification for low-resolution, narrowband radar (non-wideband imaging radar) usually include size, attitude, maneuverability, polarization, speed (Doppler) and micro-Doppler, etc. However, the "low, small and slow" targets such as ground moving personnel, vehicles, ultra-low altitude unmanned aerial vehicles and unmanned aerial vehicle clusters pose great difficulties to low-resolution, narrowband radar detection and feature extraction through detection due to their small size, slow speed and low flight altitude. At present, the classification and identification of general low-resolution, narrowband radars for ground and ultra-low altitude "low, small and slow" targets are mostly based on target height to first realize air-ground target classification, and then classify ground and air targets based on speed difference. There are also related technologies or algorithms for the classification and identification of helicopters and rotor unmanned aerial vehicles based on the micro-Doppler caused by the rotation of the aircraft propeller. However, there is no mature technology and algorithm for the identification, number estimation and distribution boundary estimation of unmanned aerial vehicles and unmanned aerial vehicle clusters that are in the same distance unit and the same beam of narrowband radar and have almost the same speed.

[0004] The shortcomings of the current classification and identification method of low resolution and narrow band radar for ground and air targets are that the height difference of the air target and the ground target is required to be large or the beam of the radar in the pitch is required to be narrow, so that the air target and the ground target cannot be separated or the cost of the radar is increased to realize the extremely narrow beam in the pitch and the extremely high resolution in the pitch angle. For the identification, number estimation and distribution boundary estimation of the unmanned aerial vehicle cluster targets in the same distance unit, the same beam and the same speed of the low resolution and narrow band radar, the sum-difference monopulse ratio of the echo data is used to confirm whether the target is the unmanned aerial vehicle cluster target, and then the number of the cluster targets is estimated by the Guass disc method to estimate the boundary and the geometric center. However, the algorithm is only suitable for the scene with high signal-to-noise ratio of the target echo, and the performance of the algorithm is sharply decreased when the number of the targets is large. SUMMARY

[0005] In order to overcome the shortcomings of the prior art, the application provides a target classification and identification and cluster target estimation method based on narrow band radar detection.

[0006] The technical scheme adopted by the application to solve the technical problem comprises the following specific steps:

[0007] First step: the radar detects the ground target and the ultra-low altitude target in the visual range, the ultra-low altitude target and the ground target are in the same radar pitch beam range, all the targets to be detected are in the radar power range, and all the targets in the cluster target are in the same distance resolution unit and the same beam of the radar; after the radar signal processing of radar signal transmission, echo reception, AD sampling, DDC (digital down conversion) and pulse compression, and MTD (monopulse time difference), the distance, speed, azimuth and pitch information of the target can be detected by using the constant false alarm detection (CFAR) detection technology;

[0008] Second step: the first step identification of the target is performed according to the speed of the target, when the detected target speed v is less than or equal to 5km / h, the target is directly determined as a ground moving person;

[0009] Third step: when the detected target speed v is greater than 5km / h, the distance of the target is further judged, the RCS of the ground vehicle is 5m 2 , the RCS of the small unmanned aerial vehicle is 0.03m 2 , and the RCS of the cluster target of the small unmanned aerial vehicle with a number of more than 10 is 0.2-0.3m 2 Therefore, for the same radar, the detection power for the ground vehicle is much greater than that for the small unmanned aerial vehicle, when the detected target distance R exceeds the detection distance range of the radar for the unmanned aerial vehicle and the unmanned aerial vehicle cluster target, the target is directly determined as a ground moving vehicle;

[0010] Fourth step: when the detected target distance R is within the target detection distance range of the UAV and the UAV cluster, further judge the signal-to-noise ratio of the detected target echo data;

[0011] If the signal-to-noise ratio of the detected target echo is greater than the set signal-to-noise ratio threshold, the target is directly determined as a ground moving vehicle, otherwise it is determined as a UAV or UAV cluster target;

[0012] The signal-to-noise ratio threshold is set according to the detection distance range of the UAV and the UAV cluster target, combined with the RCS of the UAV and the UAV cluster target, the ground vehicle target, and is corrected according to the actual detection and test of the target during radar debugging and test, and the signal-to-noise ratio threshold value is finally determined;

[0013] If the echo signal-to-noise ratio of the target at this time exceeds the threshold (S / N) Threshold , it is determined as a ground vehicle, otherwise, the fifth step is performed;

[0014] Fifth step: when the UAV and the UAV cluster target are determined, the radar performs TAS tracking processing on the UAV and the UAV cluster target, and then calculates the complex monopulse ratio of the difference signal and the sum signal of the tracking echo;

[0015] Sixth step: after obtaining the complex monopulse ratio of the target echo signal, the average of the complex monopulse ratio is calculated, if the average is a complex number and has a obvious imaginary part, it is determined that the target is a UAV cluster target, otherwise, it is determined that the target is a single UAV target;

[0016] Seventh step: when the target is determined as a UAV cluster target, record the signal-to-noise ratio SNR of the cluster target and the road echo data at this time, and further estimate the number, distribution boundary and geometric center of the cluster target distribution of the UAV cluster;

[0017] Eighth step: through radar resource scheduling, the tracking beam of the cluster target is directed from the boundary of the last tracking beam as the center of this beam, and dense scanning tracking or simultaneous multi-beam tracking is performed in the radar beam width from azimuth to elevation;

[0018] Ninth step: tracking processing is performed on each beam in the dense beam in turn, and the signal-to-noise ratio of the tracking echo data is calculated. The signal-to-noise ratio SNR multiplied by the threshold factor k is used as the detection threshold, which is corrected according to the actual detection and test of the target during radar debugging and test. If the signal-to-noise ratio of the detected target echo data is greater than the detection threshold, it is considered that there is a target in this beam, otherwise it is determined that there is no target in this beam;

[0019] Tenth step: the number of beams determined to have targets in the ninth step is the number of targets in the cluster target;

[0020] The tenth step is to count the target beam pointing in the tenth step, and extract the azimuth and pitch boundary beam pointing value, combine the target detection distance, and give the three-dimensional coordinates of the four most boundary targets as follows:

[0021]

[0022] Where, R i is the distance of the ith target; θ i is the azimuth angle of the ith target; is the azimuth angle of the ith target; that is, the boundary of the cluster target distribution can be estimated;

[0023] The twelfth step is to estimate the cluster target distribution boundary estimated in the tenth step The distance, azimuth and pitch information, that is, the geometric center of the cluster target can be estimated

[0024] In the fourth step, the signal-to-noise ratio of the target echo detection is calculated according to the radar range equation as follows:

[0025] The radar range equation is as follows:

[0026]

[0027] Where, P t is the peak transmit power; τ is the transmit pulse width; G t is the transmit gain; G r is the receive gain; n is the pulse accumulation number in MTD processing; σ is the target scattering cross section (RCS); λ is the transmit pulse wavelength; k is the Boltzmann constant 1.38×10 -23 J / deg; T0 is the ambient noise temperature; F n is the noise factor; (S / N) 0min is the minimum detectable signal-to-noise ratio of single pulse; L s is the system loss;

[0028] Through the radar range equation, the expression of the minimum detectable signal-to-noise ratio is as follows:

[0029]

[0030] Under the same conditions of P t , τ, G t , G r , n, λ, k, T0, F n and L s , the relationship between the theoretical signal-to-noise ratio of target detection and target distance R and target σ (RCS) is as follows:

[0031]

[0032] According to the above formula, when the detected target distance is within the target detection range of the UAV and UAV swarm, since the RCS of the ground vehicle is 5m... 2 The RCS of the small drone is 0.03m. 2 The RCS of a swarm of 10 or more small drones is 0.2–0.3 m. 2 The signal-to-noise ratio threshold is set as follows:

[0033]

[0034] Where ω is the signal-to-noise ratio threshold coefficient.

[0035] In actual detection and recognition, ω is rounded to the nearest integer between 15 and 20.

[0036] In the fifth step, the steps for calculating the complex single-pulse ratio of the difference signal and the sum signal of the tracking echo are as follows:

[0037]

[0038] Where Δ is the difference signal echo, Σ is the sum signal echo, m=1,2,…,M, is the number of targets detected by the radar; u m =sinθ m θ m s represents the azimuth angle value of the m-th target deviating from the radar normal; m The received signal echo of the m-th target; K is the angle discrimination slope, K = (Nπ) / 4, and N is the number of array elements in the radar azimuth direction; The complex indicator angle is given, and the ratio of complex to single pulses follows a complex Gaussian distribution. The mean and variance of the ratio of complex to single pulses are as follows:

[0039]

[0040] In the eighth step, the azimuth and elevation beam spacing is 1 / 15 to 1 / 20 of the beamwidth, and the number of azimuth and elevation beams is enough to fill one beamwidth, i.e., 15×15 to 20×20 beams. Figure 3 As shown, scheduling involves sequential scanning or simultaneous multi-beam tracking.

[0041] In the ninth step, the threshold factor k is taken as 0.5 to 0.8.

[0042] In the twelfth step, the geometric center of the cluster target is estimated:

[0043]

[0044] in, Let R be the geometric center of the cluster target, where R is the geometric center of the cluster target. imax Cluster target distribution boundary The maximum distance in R imin Cluster target distribution boundary The minimum distance in θ imax Cluster target distribution boundary The maximum azimuth value in θ imin Cluster target distribution boundary The minimum value of the orientation in the middle. Cluster target distribution boundary The maximum pitch value in the middle, Cluster target distribution boundary The minimum pitch value in the range.

[0045] An electronic device includes one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to perform the methods described above.

[0046] A computer-readable storage medium storing program code that can be invoked by a processor to perform the method described above.

[0047] The advantages of this invention lie in its simple algorithm, strong operability, and good real-time performance. It eliminates the need for more expensive ultra-wideband radar for target classification and identification based on two-dimensional images, and also eliminates the need for coordinated detection using photoelectric or other detection methods. Instead, it utilizes conventional information such as target detection range, velocity, signal-to-noise ratio, and single-pulse ratio to achieve classification, identification, and parameter estimation of ground-based moving personnel, vehicles, ultra-low-altitude UAVs, and UAV swarm targets. In particular, it estimates parameters such as the number, boundaries, and geometric center of swarm targets within the same distance and angle unit. Attached Figure Description

[0048] Figure 1 This is a flowchart of the classification and identification of ground targets and ultra-low-altitude cluster targets, as well as the parameter estimation process for cluster targets, according to the present invention.

[0049] Figure 2 This is a flowchart of the target parameter estimation process for a drone swarm.

[0050] Figure 3 This is a schematic diagram of beam scheduling when estimating the number of cluster targets. Detailed Implementation

[0051] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0052] This invention provides a method for classifying and identifying ground-moving targets such as people, vehicles, ultra-low-altitude drones, and drone swarms using low-resolution narrowband radar. It is applicable to the classification, identification, and parameter estimation of targets with speeds ≤10 km / h for people and >10 km / h for vehicles, drones, and drone swarms, provided that drone swarms are distributed within the same range and angle cells detected by the radar, and that the distribution of swarm targets in terms of range cannot be obstructed. Based on basic information such as target range, speed, and signal-to-noise ratio detected by the radar, and combined with the statistical characteristics of the sum and difference signal single-pulse ratios during target tracking, the method detects whether the target is a single target or a swarm. Then, for swarm targets, radar resource scheduling is used to scan and track the swarm target area using dense beams. The number and boundaries of the swarm targets are determined by the signal-to-noise ratio of the echo signals within each beam. Finally, the geometric center of the swarm targets is estimated based on the boundaries of the swarm targets.

[0053] The principle of this invention is that conventional low-resolution, narrowband radar can perform omnidirectional or directional detection of ground targets and ultra-low-altitude targets within the line-of-sight range. All targets to be detected are within the radar's effective range, and all targets within a cluster are within the same range resolution cell and the same beam. After basic processing such as radar signal transmission, echo reception, AD sampling, DDC (digital down-conversion), pulse compression, coherent accumulation (MTD), and constant false alarm rate (CFAR), information such as the target's range, velocity, azimuth, and elevation can be detected. Based on this basic detection information, firstly, ground personnel are classified and identified according to the speed of the detected targets; secondly, ground vehicles, drones, or drone swarms are classified and identified based on the speed and detection distance of the detected targets; thirdly, ground vehicles, drones, or drone swarms with overlapping speed ranges within the same distance range are classified and identified based on the speed, distance, and signal-to-noise ratio of the target echo data; then, the sum and difference single-pulse ratio of the echo data is calculated using the tracking information of drones or drone swarms to achieve the classification and identification of drones and drone swarms; finally, dense beam scanning tracking is performed on drone swarms, and the number of drone swarms is estimated based on the signal-to-noise ratio of each beam detection signal, thereby estimating the distribution boundary and geometric center of the swarm targets. The specific process is as follows: Figure 1 As shown.

[0054] The specific implementation steps of this invention are as follows:

[0055] Step 1: The radar detects ground targets and ultra-low-altitude targets within its line-of-sight range. Ultra-low-altitude targets and ground targets are within the same radar elevation beam. All targets to be detected are within the radar's effective range. All targets within a cluster are within the same radar range resolution cell and the same beam. After radar signal processing, including radar signal transmission, echo reception, AD sampling, DDC (digital down-conversion), pulse compression, and coherent accumulation (MTD), the range, velocity, azimuth, and elevation information of the target can be detected using constant false alarm rate (CFAR) detection technology.

[0056] The second step is to identify the target based on its speed. If the detected target speed v ≤ 5 km / h, the target is directly identified as a person moving on the ground.

[0057] Step 3: When the detected target speed v > 5 km / h, further determine the target distance. The RCS of the ground vehicle is 5m. 2 The RCS of the small drone is 0.03m. 2 The RCS of a swarm of 10 or more small drones is 0.2–0.3 m. 2 Therefore, for the same radar, the detection power for ground vehicles is much greater than the detection power for small drones. When the detected target distance R exceeds the radar's detection range for drones and drone swarms, the target is directly identified as a ground moving vehicle.

[0058] Step 4: When the detected target distance R is within the target detection range of the UAV and UAV swarm, the signal-to-noise ratio of the detected target echo data is further determined;

[0059] If the signal-to-noise ratio of the detected target echo is greater than the set signal-to-noise ratio threshold, the target is directly identified as a ground moving vehicle; otherwise, it is identified as a drone or drone swarm target.

[0060] The signal-to-noise ratio (SNR) threshold is set theoretically based on the detection range of UAVs and UAV swarm targets, combined with the RCS of UAVs, UAV swarm targets, and ground vehicle targets. During the radar debugging and testing phase, the SNR threshold is corrected based on actual target detection and testing, and the final SNR threshold value is determined.

[0061] The signal-to-noise ratio for target echo detection is calculated based on the radar range equation as follows:

[0062] The radar range equation is as follows:

[0063]

[0064] Among them, P tG is the peak transmit power; τ is the transmit pulse width; t For transmit gain; G r σ is the receiver gain; n is the pulse accumulation number during MTD processing; σ is the target scattering cross section (RCS); λ is the transmitted pulse wavelength; k is the Boltzmann constant 1.38 × 10⁻⁶. -23 J / deg; T0 is ambient noise temperature; F n Noise figure (S / N) 0min L represents the minimum detectable signal-to-noise ratio for a single pulse. s For system losses;

[0065] The expression for the minimum detectable signal-to-noise ratio is obtained from the radar range equation as follows:

[0066]

[0067] In P t , τ, G t G r , n, λ, k, T0, F n and L s Under the same conditions, the theoretical signal-to-noise ratio for target detection is related to the target distance R and the target σ(RCS) as follows:

[0068]

[0069] According to the above formula, when the detected target distance is within the target detection range of the UAV and UAV swarm, since the RCS of the ground vehicle is 5m... 2 The RCS of the small drone is 0.03m. 2 The RCS of a swarm of 10 or more small drones is 0.2–0.3 m. 2 The signal-to-noise ratio threshold is set as follows:

[0070]

[0071] Where ω is the signal-to-noise ratio threshold coefficient;

[0072] In actual detection and identification, ω is rounded to the nearest integer between 15 and 20. If the signal-to-noise ratio of the target's echo exceeds the threshold (S / N) at this time... Threshold If yes, it is determined to be a ground vehicle; otherwise, proceed to step five.

[0073] Step 5: When the target is identified as a drone or drone swarm, the radar performs TAS tracking on the drone or drone swarm target, and then calculates the ratio of the difference signal to the sum signal of the tracking echo, as shown below:

[0074]

[0075] Where Δ is the difference signal echo, Σ is the sum signal echo, m=1,2,…,M, is the number of targets detected by the radar; u m =sinθ m θ m s represents the azimuth angle value of the m-th target deviating from the radar normal; m The received signal echo of the m-th target; K is the angle discrimination slope, K = (Nπ) / 4, and N is the number of array elements in the radar azimuth direction; The complex indicator angle is given, and the ratio of complex to single pulses follows a complex Gaussian distribution. The mean and variance of the ratio of complex to single pulses are as follows:

[0076]

[0077] Step 6: After calculating the ratio of complex to single pulse in the target echo signal, calculate the mean of the complex to single pulse ratio. If the mean is a complex number and has a significant imaginary part, the target can be determined to be a drone swarm target; otherwise, the target is determined to be a single drone target.

[0078] Step 7: When the target is determined to be a drone swarm, record the signal-to-noise ratio (SNR) of the swarm target and the echo data at this time. Further estimate the number of drones in the swarm, their distribution boundaries, and the geometric center of the swarm target distribution. The estimation process is as follows: Figure 2 As shown;

[0079] Step 8: Through radar resource scheduling, the tracking beam for the cluster target is positioned so that the boundary of the previous frame's tracking beam serves as the center direction of the current beam. Within one radar beamwidth, dense scanning and tracking are performed sequentially from azimuth to elevation, or simultaneous multi-beam tracking is performed. Simultaneous multi-beam tracking yields better results. The beam spacing for azimuth and elevation is 1 / 15 to 1 / 20 of the beamwidth, and the number of azimuth and elevation beams is sufficient to fill one beamwidth, i.e., 15×15 to 20×20 beams. Figure 3 As shown, scheduling involves sequential scanning or simultaneous multi-beam tracking.

[0080] Step 9: Track each beam within the dense beam sequentially, calculate the signal-to-noise ratio (SNR) of the tracking echo data, and multiply the SNR by a threshold factor k as the detection threshold. k ranges from 0.5 to 0.8. The detection threshold is adjusted during radar debugging and testing based on actual target detection and experiments. If the SNR of the detected target echo data is greater than the detection threshold, it is considered that there is a target within this beam; otherwise, it is determined that there is no target within this beam.

[0081] Step 10: Count the number of beams that were identified as having targets in Step 9, which is the number of targets within the cluster target;

[0082] Step 11: Statistically analyze the beam pointing of targets identified in Step 10, and extract the beam pointing values ​​at the outermost boundaries of azimuth and elevation. Combined with the target detection range, the three-dimensional coordinates of the four outermost targets are given below:

[0083]

[0084] Among them, R i θ is the distance to the i-th target; i Let be the azimuth angle of the i-th target; Let be the azimuth angle of the i-th target; then the boundary of the cluster target distribution can be estimated.

[0085] Step 12: Based on the cluster target distribution boundary estimated in Step 11 By analyzing the distance, azimuth, and elevation information, the geometric center of the cluster target can be estimated.

[0086]

[0087] in, The geometric center of the cluster target.

[0088] The advantages of this invention are:

[0089] (1) Target classification and identification is achieved by combining information such as speed, distance, and signal-to-noise ratio of ground targets and ultra-low-altitude cluster targets based on low-resolution, narrow-band radar.

[0090] (2) The detection and identification of single targets and cluster targets of UAVs are realized based on the sum-difference single pulse ratio of target tracking information.

[0091] (3) Parameter estimation of cluster targets is achieved by combining radar dense beam tracking information, signal-to-noise ratio, beam pointing and distance information.

Claims

1. A target classification and identification method and cluster target estimation method based on narrowband radar detection, characterized in that... Includes the following steps: Step 1: The radar detects ground targets and ultra-low-altitude targets within its line-of-sight range. Ultra-low-altitude targets and ground targets are within the same radar elevation beam. All targets to be detected are within the radar's effective range. All targets within a cluster are within the same radar range resolution cell and the same beam. After radar signal processing, including radar signal transmission, echo reception, AD sampling, DDC, pulse compression, and coherent accumulation, the radar can detect the target's range, velocity, azimuth, and elevation information using constant false alarm rate (CFAR) detection technology. The second step is to identify the target based on its speed. If the detected target speed v ≤ 5 km / h, the target is directly identified as a person moving on the ground. Step 3: When the detected target speed v > 5 km / h, further determine the target distance. The RCS of the ground vehicle is 5m. 2 The RCS of the small drone is 0.03m. 2 The RCS of a swarm of 10 or more small drones is 0.2–0.3 m. 2 Therefore, for the same radar, the detection power for ground vehicles is much greater than the detection power for small drones. When the detected target distance R exceeds the radar's detection range for drones and drone swarms, the target is directly identified as a ground moving vehicle. Step 4: When the detected target distance R is within the target detection range of the UAV and UAV swarm, the signal-to-noise ratio of the detected target echo data is further determined; If the signal-to-noise ratio of the detected target echo is greater than the set signal-to-noise ratio threshold, the target is directly identified as a ground moving vehicle; otherwise, it is identified as a drone or drone swarm target. The signal-to-noise ratio (SNR) threshold is set theoretically based on the detection range of UAVs and UAV swarm targets, combined with the RCS of UAVs, UAV swarm targets, and ground vehicle targets. During the radar debugging and testing phase, the SNR threshold is corrected based on actual target detection and testing, and the final SNR threshold value is determined. If the target's echo signal-to-noise ratio exceeds the threshold (S / N) at this time. Threshold If yes, it is determined to be a ground vehicle; otherwise, proceed to step five. Step 5: When the target is identified as a drone or drone swarm, the radar performs TAS tracking on the drone or drone swarm target, and then calculates the ratio of the difference signal to the sum signal of the tracking echo. Step 6: After calculating the ratio of complex to single pulse in the target echo signal, calculate the mean of the complex to single pulse ratio. If the mean is a complex number and has a significant imaginary part, the target can be determined to be a drone swarm target; otherwise, the target is determined to be a single drone target. Step 7: When the target is determined to be a drone swarm target, record the signal-to-noise ratio (SNR) of the swarm target and the echo data at this time, and further estimate the number of drones, the distribution boundary, and the geometric center of the swarm target distribution; Step 8: By scheduling radar resources, the tracking beam for cluster targets is made to point from the boundary of the tracking beam in the previous frame as the center of the current beam. Within one radar beamwidth, dense scanning and tracking are performed sequentially from azimuth to elevation, or multiple beams are tracked simultaneously. Simultaneous multi-beam tracking is more effective. Step 9: Track each beam within the dense beam sequentially, calculate the signal-to-noise ratio (SNR) of the tracking echo data, and multiply the SNR by a threshold factor k to obtain the detection threshold. The detection threshold is adjusted during radar debugging and testing based on actual target detection and testing. If the SNR of the detected target echo data is greater than the detection threshold, it is considered that there is a target within this beam; otherwise, it is determined that there is no target within this beam. Step 10: Count the number of beams that were identified as having targets in Step 9, which is the number of targets within the cluster target; Step 11: Statistically analyze the beam pointing of targets identified in Step 10, and extract the beam pointing values ​​at the outermost boundaries of azimuth and elevation. Combined with the target detection range, the three-dimensional coordinates of the four outermost targets are given below: Among them, R i θ is the distance to the i-th target; i Let be the azimuth angle of the i-th target; Let be the azimuth angle of the i-th target; then the boundary of the cluster target distribution can be estimated. Step 12: Based on the cluster target distribution boundary estimated in Step 11 By analyzing the distance, azimuth, and elevation information, the geometric center of the cluster target can be estimated.

2. The target classification and identification and cluster target estimation method based on narrowband radar detection according to claim 1, characterized in that: In the fourth step, the signal-to-noise ratio for target echo detection is calculated based on the radar range equation as follows: The radar range equation is as follows: Among them, P t G is the peak transmit power; τ is the transmit pulse width; t For transmit gain; G r σ is the receiver gain; n is the pulse accumulation number during MTD processing; σ is the target scattering cross section (RCS); λ is the transmitted pulse wavelength; k is the Boltzmann constant 1.38 × 10⁻⁶. -23 J / deg; T0 is ambient noise temperature; F n Noise figure (S / N) 0min L represents the minimum detectable signal-to-noise ratio for a single pulse. s For system losses; The expression for the minimum detectable signal-to-noise ratio is obtained from the radar range equation as follows: In P t , τ, G t G r , n, λ, k, T0, F n and L s Under the same conditions, the theoretical signal-to-noise ratio for target detection is related to the target distance R and the target σ(RCS) as follows: According to the above formula, when the detected target distance is within the target detection range of the UAV and UAV swarm, since the RCS of the ground vehicle is 5m... 2 The RCS of the small drone is 0.03m. 2 The RCS of a swarm of 10 or more small drones is 0.2–0.3 m. 2 The signal-to-noise ratio threshold is set as follows: Where ω is the signal-to-noise ratio threshold coefficient.

3. The target classification and identification and cluster target estimation method based on narrowband radar detection according to claim 2, characterized in that: The signal-to-noise ratio threshold coefficient ω is an integer value between 15 and 20.

4. The target classification and identification and cluster target estimation method based on narrowband radar detection according to claim 1, characterized in that: In the fifth step, the steps for calculating the complex single-pulse ratio of the difference signal and the sum signal of the tracking echo are as follows: Where Δ is the difference signal echo, Σ is the sum signal echo, m=1,2,…,M, is the number of targets detected by the radar; u m =sinθ m θ m s represents the azimuth angle value of the m-th target deviating from the radar normal; m The received signal echo of the m-th target; K is the angle discrimination slope, K = (Nπ) / 4, and N is the number of array elements in the radar azimuth direction; The complex indicator angle is given, and the ratio of complex to single pulses follows a complex Gaussian distribution. The mean and variance of the ratio of complex to single pulses are as follows:

5. The target classification and identification and cluster target estimation method based on narrowband radar detection according to claim 1, characterized in that: In the eighth step, the beam spacing for azimuth and elevation is 1 / 15 to 1 / 20 of the beamwidth, and the number of azimuth and elevation beams is to fill one beamwidth, i.e., 15×15 to 20×20. The scheduling is to use sequential scanning tracking or simultaneous multi-beam tracking.

6. The target classification and identification and cluster target estimation method based on narrowband radar detection according to claim 1, characterized in that: In the ninth step, the threshold factor k is taken as 0.5 to 0.

8.

7. The target classification and identification and cluster target estimation method based on narrowband radar detection according to claim 1, characterized in that: In the twelfth step, the geometric center of the cluster target is estimated: in, Let R be the geometric center of the cluster target, where R is the geometric center of the cluster target. imax Cluster target distribution boundary The maximum distance in R imin Cluster target distribution boundary The minimum distance in θ imax Cluster target distribution boundary The maximum azimuth value in θ imin Cluster target distribution boundary The minimum value of the orientation in the middle. Cluster target distribution boundary The maximum pitch value in the middle, Cluster target distribution boundary The minimum pitch value in the range.

8. An electronic device, characterized in that, include: One or more processors; Memory; One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs being configured to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program code that can be invoked by a processor to execute the method as described in any one of claims 1-7.

Images