A radar detection and identification system for low-altitude targets of unmanned aerial vehicles (UAVs)

Abstract

This invention discloses a low-altitude target radar detection and identification system for unmanned aerial vehicles (UAVs), belonging to the field of radar detection and target identification technology. It includes a quantum entangled laser emission module, a quantum multi-dimensional receiving module, a multi-source heterogeneous data fusion module, a quantum-classical hybrid computing module, an adaptive anti-interference module, and an environmental adaptive correction module. The quantum entangled laser emission module generates entangled photon pairs carrying orbital angular momentum (OAM) and emits them into the detection airspace. The quantum multi-dimensional receiving module demodulates the distance, angle, OAM attenuation spectrum, polarization state, and reflection spectrum of the echo signal through coincidence measurement. Employing quantum entanglement-OAM joint modulation technology, the system leverages the nonlocality of entangled photon pairs and the multi-dimensional information carrying capacity of nine orthogonal OAM modes, achieving a signal-to-noise ratio improvement of more than three times compared to traditional microwave radar. Single-photon-level coincidence measurement combined with quantum Fourier transform signal processing further enhances the signal-to-noise ratio by 3dB, enabling stable detection of weak targets.

Description

Technical Field

[0001] This invention relates to the field of radar detection and target recognition technology, and more specifically, to a radar detection and recognition system for low-altitude targets of unmanned aerial vehicles (UAVs). Background Technology

[0002] The disorderly flight of low-altitude unmanned aerial vehicles (flight altitude <1000 meters) has posed a serious security threat to key areas such as airport airspace, military restricted zones, and sites of major events. Existing detection and identification technologies suffer from the following core bottlenecks:

[0003] Detection performance limitations

[0004] Traditional microwave radar is limited by the diffraction limit, and its detection range for micro drones with an RCS < 0.1㎡ (such as the DJI Mavic series) is less than 1 kilometer. Moreover, its ground clutter suppression ratio is only 30dB, making it prone to target loss. Conventional lidar uses Gaussian beams and can only obtain target distance and intensity information. It cannot distinguish between drones and interference objects such as birds and balloons, resulting in a false alarm rate as high as 20%. Although quantum lidar has single-photon level sensitivity, current solutions have not solved the problem of quantum state decoherence caused by atmospheric turbulence, and the actual detection range is only 60% of the theoretical value.

[0005] Weak anti-interference ability

[0006] It is susceptible to active deception interference (such as injection of false target signals) and passive interference (such as chaff and smoke). Traditional encryption methods are difficult to eradicate due to the interceptability of classic communication. In urban environments, the multipath effect leads to signal delay extension, and the target positioning error can reach 10 meters.

[0007] Insufficient timeliness in identification

[0008] Existing deep learning algorithms rely on cloud computing centers, with end-to-end processing latency >500ms, which cannot meet the dynamic interception requirements of high-speed drones (speed >100km / h); feature extraction relies only on point clouds or optical images and does not utilize the target's physical properties (such as material, engine spectrum), resulting in a misclassification rate of >15% for similar targets.

[0009] Poor environmental adaptability

[0010] Under weather conditions such as rain and fog, the laser attenuation coefficient is >0.5dB / km, resulting in a sharp decrease in detection range; active radar cannot be deployed in electromagnetically silent areas, and the sensitivity is insufficient when relying on passive detection.

[0011] In view of this, the present invention is proposed to solve the above-mentioned technical problems. Summary of the Invention

[0012] The purpose of this invention is to provide a radar detection and identification system for low-altitude targets of unmanned aerial vehicles (UAVs) to solve the technical problems mentioned above.

[0013] To achieve the above objectives, the present invention provides the following technical solution:

[0014] A radar detection and identification system for low-altitude targets of unmanned aerial vehicles (UAVs) includes:

[0015] A quantum entangled laser emission module is used to generate entangled photon pairs carrying orbital angular momentum (OAM) and emit them into the detection space.

[0016] The quantum multidimensional receiving module measures the distance, angle, OAM attenuation spectrum, polarization state, and reflection spectrum of the demodulated echo signal according to the conformal measurement.

[0017] The multi-source heterogeneous data fusion module performs spatiotemporal alignment and feature fusion of quantum detection data with the outputs of phased array radar, infrared thermal imager, and radio frequency detector;

[0018] The quantum-classical hybrid computing module uses a quantum processor to accelerate the dimensionality reduction of high-dimensional features and combines it with an edge computing terminal to achieve real-time target classification and threat assessment.

[0019] The adaptive anti-interference module integrates a quantum key distribution link at the physical layer, uses random phase modulation at the signal layer, and employs cognitive frequency hopping at the protocol layer, thus constructing anti-interference protection across the entire link.

[0020] The environmental adaptive correction module dynamically adjusts the transmission parameters and optical system based on atmospheric parameter monitoring;

[0021] The quantum entangled laser emission module includes a spontaneous parametric downconversion source based on PPKTP crystal, a metasurface OAM encoder, and an adaptive emission control unit. The metasurface OAM encoder can load vortex phases with topological charge numbers l=0,±1,...,±8 to form 9 orthogonal OAM modes.

[0022] The quantum multidimensional receiver module uses a Sagnac interferometer to separate the signal light and idle light, and achieves coincidence measurement through a 9-channel superconducting nanowire single-photon detector to simultaneously acquire Stokes parameters and 400-1100nm reflection spectrum.

[0023] The quantum-classical hybrid computing module includes a 20-qubit superconducting processor and an FPGA+GPU heterogeneous edge terminal. The quantum processor runs a quantum support vector machine to achieve feature dimensionality reduction, while the classical terminal runs an improved PointNet++ algorithm, with an end-to-end processing latency of ≤100ms.

[0024] Furthermore, the environmental adaptive correction module acquires atmospheric parameters through a microwave radiometer and an aerosol lidar, driving a 1024-element MEMS deformable mirror to correct wavefront distortion, with a residual error < λ / 20.

[0025] Furthermore, the multi-source heterogeneous data fusion module employs federated Kalman filtering to weight and fuse quantum dot clouds, radar RCS, infrared radiation intensity, and radio frequency characteristics to construct an 18-dimensional target feature vector.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] (1) By adopting quantum entanglement-OAM joint modulation technology, the detection range of micro UAVs with RCS<0.1㎡ is increased to 3.2 kilometers through the nonlocality of entangled photon pairs and the multidimensional information carrying capacity of 9 orthogonal OAM modes, which is more than 3 times higher than that of traditional microwave radar; single photon level coincidence measurement combined with quantum Fourier transform signal processing improves the signal-to-noise ratio by 3dB, realizing stable detection of weak targets.

[0028] (2) Construct a full-link protection system of physical layer (quantum encryption) - signal layer (dynamic modulation) - protocol layer (cognitive frequency hopping), which still maintains a detection probability of more than 95% in the strong electromagnetic interference environment, which is 20dB higher than the traditional anti-interference scheme; the success rate of resisting deceptive interference reaches 99.99%, and the false alarm rate is reduced to less than 0.1%.

[0029] (3) The quantum-classical hybrid computing architecture improves the feature dimensionality reduction speed by 100 times. Combined with the improved PointNet++ algorithm, the end-to-end processing latency is controlled within 100ms, supporting parallel processing of 500 targets per second. Through 18-dimensional fusion features, the classification accuracy of drones / birds / balloons reaches 99%, which is 14% higher than the single sensor solution.

[0030] (4) The environmental adaptive correction module improves the detection range retention rate by 40% under level 8 wind force and only 28% under heavy rain weather by real-time inversion of atmospheric parameters and dynamic correction of 1024-element MEMS deformable mirror. This solves the problem of sudden performance drop of quantum radar under turbulent and severe weather conditions. Attached Figure Description

[0031] The accompanying drawings, which form part of this application, are used to provide a further understanding of the present invention. The illustrative embodiments and descriptions of the present invention are used to explain the present invention, but do not constitute an undue limitation of the present invention. Obviously, the drawings described below are merely some embodiments; those skilled in the art can obtain other drawings based on these drawings without creative effort. In the drawings:

[0032] Figure 1 This is a flowchart of the UAV low-altitude target radar detection and identification system provided in this embodiment of the application. Detailed Implementation

[0033] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0034] A radar detection and identification system for low-altitude targets of unmanned aerial vehicles (UAVs) includes:

[0035] The quantum entangled laser emission module generates entangled photon pairs carrying orbital angular momentum (OAM) and emits them into the probe space. The entangled light source unit generates an 808nm pump light using spontaneous parametric down-conversion (SPDC) based on a PPKTP crystal, producing entangled photon pairs of a 780nm signal light and an 840nm idler light (coincidence count rate > 10). 6 / s, entanglement degree > 0.95); Adaptive transmission control: dynamically adjusts pulse energy (10-100μJ) and repetition frequency (10-50kHz) according to atmospheric visibility (real-time monitoring by lidar), and automatically switches to Gaussian mode with l=0 in foggy weather to reduce scattering loss; the test object is a drone (DJI Mavic 3) with RCS=0.1㎡, and the baseline value is a detection distance of 3.2 kilometers in no wind; in an open area with wind force of level 8 (wind speed 17.2-20.7m / s) and visibility of 5 kilometers, the GB / T 19205-2017 wind speed measurement standard was used, and the test was repeated 30 times, with an average detection distance of 2.1 kilometers (retention rate 65.6%), which is 40% higher than the 1.5 kilometers (retention rate 46.9%) without the environmental correction module.

[0036] The quantum multidimensional receiving module measures the distance, angle, OAM attenuation spectrum, polarization state, and reflection spectrum of the demodulated echo signal through coincidence measurement. The entangled receiving subunit uses a Sagnac interferometer to separate the signal light from the idle light, achieving coincidence measurement through a superconducting nanowire single-photon detector (SNSPD, dark count <10Hz, time jitter <50ps), improving the signal-to-noise ratio by 3dB compared to traditional direct detection. The OAM demodulation unit, based on a spiral phase plate array and spatial filter, achieves parallel demodulation of 9 OAM modes (demodulation error <1%), extracting the target's radial distance, angular position, and OAM mode attenuation spectrum. The polarization-spectral joint detection integrates a polarization beam splitter and a grating spectrometer to simultaneously acquire the target's Stokes parameters (4 dimensions) and reflection spectrum (400-1100nm, 1nm resolution) for material identification (e.g., carbon fiber, plastic, metal).

[0037] The multi-source heterogeneous data fusion module performs spatiotemporal alignment and feature fusion of quantum detection data with the outputs of phased array radar, infrared thermal imager, and radio frequency detector; the collaborative sensor array includes an X-band phased array radar (360° omnidirectional scanning, azimuth resolution 0.5°), an infrared thermal imager (640×512 resolution, frame rate 30Hz), and a radio frequency detector (2.4 / 5.8GHz band, sensitivity -110dBm);

[0038] Spatiotemporal alignment algorithm: Based on GPS timestamps and lidar point cloud coordinates, Kalman filtering is used to correct the spatiotemporal deviations of each sensor (time synchronization error <10ns, spatial registration error <0.1m); Data fusion strategy: Federated Kalman filtering is used to achieve weighted fusion of quantum lidar point cloud (100 points / ㎡), radar echo (RCS value), infrared radiation intensity and radio frequency characteristics to construct an 18-dimensional feature vector of the target.

[0039] The quantum-classical hybrid computing module employs a quantum processor to accelerate dimensionality reduction of high-dimensional features, combined with edge computing terminals to achieve real-time target classification and threat assessment. The quantum acceleration unit, based on a 20-qubit superconducting processor (T1=50μs, T2=30μs), implements a quantum support vector machine for OAM pattern recognition (training speed is 100 times faster than classical SVM). The 20-qubit superconducting processor reduces the dimensionality of 512-dimensional quantum features using a quantum support vector machine, significantly improving time complexity. Down to The FPGA+GPU terminal processes the 18-dimensional features after dimensionality reduction in parallel, and the two interact with each other through the PCIe 4.0 interface (latency <10ms), which together improves the feature dimensionality reduction speed by 100 times compared with pure classical calculation.

[0040] Edge computing terminal: Equipped with an FPGA+GPU heterogeneous architecture (Xilinx Versal + NVIDIA A100), running an improved PointNet++ algorithm (introducing an attention mechanism) for real-time classification of fused features. The improved PointNet++ algorithm embeds a self-attention mechanism between the sampling layer and the feature aggregation layer, where the attention weights are calculated using the cosine similarity of the target feature vectors (such as OAM attenuation spectra and reflectance spectra), as shown in the formula: Where Q / K / V are the query, key, and value matrices of the quantum dot cloud features, respectively. The feature dimension was set to 18 to enhance the focus on the material and structural characteristics of the drones; the test sample included 1000 drones (including micro models such as DJI Mavic 3), 500 birds (sparrows, pigeons, etc.), and 300 balloons, tested in sunny weather with wind speed ≤3, showing a 14% improvement over the single infrared sensor solution (85% accuracy);

[0041] Dynamic load balancing: Automatically allocates quantum computing and classical computing resources according to the target density (0-500 units / square kilometer) to ensure that the single-target processing latency is <100ms;

[0042] The adaptive anti-interference module integrates a quantum key distribution link at the physical layer, random phase modulation at the signal layer, and cognitive frequency hopping at the protocol layer, constructing anti-interference protection across the entire link: The quantum key distribution (QKD) link (key generation rate > 1Mbps) encrypts control commands to resist man-in-the-middle attacks; random phase modulation (modulation rate 100MHz) ensures a spoofing interference success rate of <0.01%; the signal processing layer uses an adaptive clutter suppression algorithm based on deep belief networks (DBN), improving the suppression ratio of ground / sea clutter to 60dB; and the protocol layer uses dynamic frequency hopping (50-1000MHz band, hopping rate 1000 hops / second) combined with cognitive radio technology to avoid man-made interference frequency bands.

[0043] The environmental adaptive correction module dynamically adjusts the transmission parameters and optical system based on atmospheric parameter monitoring. Atmospheric parameter monitoring integrates a microwave radiometer (detecting temperature and humidity profiles) and an aerosol lidar (detecting extinction coefficient) to acquire the atmospheric transmission model in real time. Optical distortion compensation uses a 1024-element MEMS deformable mirror combined with a Shack-Hartmann wavefront sensor (1kHz sampling rate) to dynamically correct wavefront distortion caused by turbulence (residual error < λ / 20). Adaptation to severe weather automatically activates dual-band detection (780nm) in rainy or foggy conditions. The m+1550nm band is used to fuse penetration difference features through deep learning to maintain detection performance. The deep learning model adopts a dual-channel convolutional neural network (CNN), with the input being the penetration matrix (32×32 pixels) of the 780nm and 1550nm bands. It contains 3 convolutional layers (kernel size 3×3, stride 1), 2 pooling layers (max pooling, 2×2) and 1 fully connected layer (128 nodes). The training dataset contains 100,000 sets of dual-band penetration samples under rain and fog conditions (labeled with actual detection distance deviation). The Adam optimizer is used and converges after 100 iterations.

[0044] The data exchange frequency between the microwave radiometer and the aerosol lidar is 10Hz, and they are aligned by timestamps (synchronization error <50ms). A weighted average algorithm is used to fuse atmospheric parameters (temperature and humidity weighted at 0.3, extinction coefficient weighted at 0.4) to generate a real-time atmospheric transmission model, which serves as the basis for the calibration of the MEMS deformable mirror.

[0045] The quantum entangled laser emission module includes a spontaneous parametric downconversion light source based on PPKTP crystal, a metasurface OAM encoder, and an adaptive emission control unit. The metasurface OAM encoder can load vortex phases with topological charge numbers l=0,±1,...,±8 to form 9 orthogonal OAM modes.

[0046] The quantum multidimensional receiving module uses a Sagnac interferometer to separate the signal light from the idle light, and achieves coincidence measurement through a 9-channel superconducting nanowire single-photon detector to simultaneously acquire Stokes parameters and 400-1100nm reflection spectra.

[0047] The quantum-classical hybrid computing module includes a 20-qubit superconducting processor and an FPGA+GPU heterogeneous edge terminal. The quantum processor runs a quantum support vector machine to achieve feature dimensionality reduction, while the classical terminal runs an improved PointNet++ algorithm, with an end-to-end processing latency of ≤100ms.

[0048] The adaptive anti-interference module integrates a quantum key distribution link (key generation rate > 1Mbps), random phase modulation (rate 100MHz), and cognitive frequency hopping (1000 hops / second), and still maintains a detection probability of ≥95% when the interference power is > 20dB of the target signal.

[0049] The environmental adaptive correction module acquires atmospheric parameters through a microwave radiometer and an aerosol lidar, and drives a 1024-element MEMS deformable mirror to correct wavefront distortion, with a residual error of <λ / 20.

[0050] The multi-source heterogeneous data fusion module uses federated Kalman filtering to perform weighted fusion of quantum dot cloud, radar RCS, infrared radiation intensity and radio frequency characteristics to construct an 18-dimensional target feature vector.

[0051] Workflow

[0052] 1. Target Detection Phase

[0053] After the system is started, the quantum entangled laser emission module outputs 9 OAM-coded beams at a repetition frequency of 50kHz, covering a 3km × 3km airspace.

[0054] The receiving module acquires entangled photon pairs reflected from the target through coincidence measurement, and demodulates to obtain the distance (accuracy ±0.1 meters), angle (±0.01°), and OAM attenuation spectrum;

[0055] Data is collected synchronously by the collaborative sensors and transmitted to the fusion module after spatiotemporal alignment.

[0056] 2. Feature extraction and recognition stage

[0057] The quantum processor performs quantum Fourier transform on OAM spectrum, polarization state and spectral data to extract 512-dimensional quantum features;

[0058] The classic processor integrates features such as radar RCS and infrared radiation intensity, takes an improved PointNet++ model as input, and outputs the target category (fixed wing / rotor / multi-rotor), size (<0.5m / 0.5-2m / >2m), and material.

[0059] A threat level model (low / medium / high) is constructed based on 12 parameters (speed, trajectory, load, etc.).

[0060] 3. Interference resistance and environmental adaptability

[0061] Real-time monitoring of the electromagnetic environment; when interference power > -80dBm, QKD encryption and dynamic frequency hopping are automatically enabled.

[0062] The atmospheric lidar updates the extinction coefficient every 100ms, driving the MEMS deformable mirror to adjust 1024 micromirror units and correct wavefront distortion.

[0063] In rainy or foggy weather (visibility < 1 km), switch to the 1550 nm band and increase the pulse energy to 100 μJ.

[0064] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A radar detection and identification system for low-altitude targets of unmanned aerial vehicles (UAVs), characterized in that, include: A quantum entangled laser emission module is used to generate entangled photon pairs carrying orbital angular momentum (OAM) and emit them into the detection space. The quantum multidimensional receiving module measures the distance, angle, OAM attenuation spectrum, polarization state, and reflection spectrum of the demodulated echo signal according to the conformal measurement. The multi-source heterogeneous data fusion module performs spatiotemporal alignment and feature fusion of quantum detection data with the outputs of phased array radar, infrared thermal imager, and radio frequency detector; The quantum-classical hybrid computing module uses a quantum processor to accelerate the dimensionality reduction of high-dimensional features and combines it with an edge computing terminal to achieve real-time target classification and threat assessment. The adaptive anti-interference module integrates a quantum key distribution link at the physical layer, uses random phase modulation at the signal layer, and employs cognitive frequency hopping at the protocol layer, thus constructing anti-interference protection across the entire link. The environmental adaptive correction module dynamically adjusts the transmission parameters and optical system based on atmospheric parameter monitoring; The quantum entangled laser emission module includes a spontaneous parametric downconversion light source based on PPKTP crystal, a metasurface OAM encoder, and an adaptive emission control unit. The metasurface OAM encoder can load vortex phases with topological charge numbers l=0,±1,...,±8 to form 9 orthogonal OAM modes. The quantum multidimensional receiving module uses a Sagnac interferometer to separate the signal light and idle light, and achieves coincidence measurement through a 9-channel superconducting nanowire single-photon detector to simultaneously acquire Stokes parameters and 400-1100nm reflection spectrum. The quantum-classical hybrid computing module includes a 20-qubit superconducting processor and an FPGA+GPU heterogeneous edge terminal. The quantum processor runs a quantum support vector machine to achieve feature dimensionality reduction, while the classical terminal runs an improved PointNet++ algorithm. The end-to-end processing latency is ≤100ms.

2. The UAV low-altitude target radar detection and identification system according to claim 1, characterized in that, The environmental adaptive correction module acquires atmospheric parameters through a microwave radiometer and an aerosol lidar, and drives a 1024-element MEMS deformable mirror to correct wavefront distortion, with a residual error < λ / 20.

3. The UAV low-altitude target radar detection and identification system according to claim 1, characterized in that, The multi-source heterogeneous data fusion module uses federated Kalman filtering to perform weighted fusion of quantum dot cloud, radar RCS, infrared radiation intensity and radio frequency characteristics to construct an 18-dimensional target feature vector.

Images