Multi-sensor data acquisition and fusion method for unmanned aerial vehicle detection countering system

Abstract

The invention relates to the technical field of unmanned aerial vehicle detection countering, and particularly discloses a multi-sensor data acquisition and fusion method for an unmanned aerial vehicle detection countering system, and the method comprises the following steps: S1, reasonably arranging a plurality of sensors in a target region, and forming an omnibearing monitoring network; s2, performing corresponding data preprocessing according to the type of the collected data; s3, extracting feature vectors from the preprocessed data of each sensor based on feature association, and judging whether the feature vectors are from the same unmanned aerial vehicle or not by using cosine similarity; a weighted voting method is adopted for decision-making fusion, and the state and the threat degree of the unmanned aerial vehicle are judged by integrating voting values of the sensors; and S4, threats are evaluated according to the fusion data, and different countering means, such as long-distance radio frequency interference, middle and short-distance laser blinding and short-distance physical capture, are adopted according to the distance of the unmanned aerial vehicle. Through cooperative deployment and data acquisition of multiple sensors, the advantages of the sensors are integrated, the detection precision and reliability of the unmanned aerial vehicle are remarkably improved, and the leak detection and false alarm rate is reduced.

Description

Technical Field

[0001] The present invention relates to the field of UAV detection and countermeasure technology, and in particular to a multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system. Background Art

[0002] With the rapid development of drone technology, its application areas continue to expand, playing a vital role in many fields, including aerial photography, logistics and distribution, agricultural plant protection, and power inspections. However, the widespread use of drones has also brought with it a series of serious security issues. In sensitive areas such as airports, military bases, government agencies, and large-scale event venues, the illegal intrusion of drones can lead to information leaks, disrupt normal activities, and even threaten the safety of personnel and facilities. For example, drones carrying cameras and illegally filming sensitive areas can lead to the theft of confidential information. Drones flying near airports can interfere with the normal takeoff and landing of aircraft, seriously threatening aviation safety. According to relevant statistics, security incidents caused by illegal drone flights have been increasing year by year in recent years, posing a significant challenge to social security and stability.

[0003] Currently, drone detection and countermeasure systems have become a key means of addressing these security issues. However, existing detection and countermeasure technologies suffer from numerous drawbacks. From a sensor perspective, single-sensor detection methods have significant limitations. Take radar, for example. While it can detect drones at relatively long distances and obtain approximate location and velocity information, it is susceptible to electromagnetic interference. In urban environments, numerous electronic devices and communication base stations generate complex electromagnetic signals, which can interfere with radar operations, significantly increasing false alarm rates and making it difficult for the system to accurately identify target drones. Other reflective objects may be misidentified as drones, wasting countermeasure resources. Photoelectric sensors, such as cameras and infrared sensors, can provide relatively clear images at close range, helping to identify drone models and appearance characteristics, but they are severely limited by weather conditions. Inclement weather conditions such as haze, heavy rain, and dust storms significantly affect light propagation, causing camera images to plummet in quality or even become incapable of forming images. Infrared sensors also experience significant reductions in detection accuracy, making it impossible to effectively detect and identify drones. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system, which improves the detection and countermeasure capabilities of UAVs through multi-sensor collaborative work and advanced data processing and fusion technology.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] The multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system includes the following steps:

[0007] S1: Rationally deploy multiple sensors in the target area to form a comprehensive monitoring network;

[0008] S2: Perform corresponding data preprocessing according to the type of collected data;

[0009] S3: Based on feature association, feature vectors are extracted from the pre-processed data of each sensor, and cosine similarity is used to determine whether they come from the same drone. A weighted voting method is used for decision-level fusion, and the voting values ​​of each sensor are combined to determine the drone status and threat level.

[0010] S4: Assess threats based on fused data and take different countermeasures based on the distance to the drone, including long-range radio frequency interference, medium- and short-range laser blinding, and close-range physical capture.

[0011] As a further solution of the present invention: the sensor includes a microwave radar, a laser radar, a photoelectric composite sensor and a radio frequency sensor.

[0012] As a further solution of the present invention: in said S2, data collection and data preprocessing are included, wherein data collection specifically includes:

[0013] Microwave radar data acquisition: with scanning period T r Scan the airspace and obtain the distance R of the drone r , azimuth angle θ r , pitch angle φ r , radial velocity V r , echo signal strength I r and phase Φ r ;

[0014] LiDAR data acquisition: Based on the target position provided by the microwave radar, at the measurement time t l Collect distance data R l,i , angle data θ l,i and φ l,i , i=1,2,...,n, n is the number of measurement points, constructing 3D point cloud data;

[0015] Photoelectric composite sensor data acquisition: HD camera with frame rate f c Acquisition of optical image sequence {I c,j}, j = 1, 2, ..., m, m is the number of image frames, the infrared thermal imager synchronously collects thermal imaging data {T i,j};

[0016] RF sensor data acquisition: monitors a specific frequency band and collects signal frequency f when a drone communication signal is detected s Bandwidth B s , modulation mode Ms , signal strength P s And the rate of change of the signal over time dPs / dt.

[0017] As a further solution of the present invention: the data preprocessing includes:

[0018] Microwave radar data preprocessing: Adaptive filtering algorithm is used to remove noise and clutter;

[0019] Assume the original radar signal is S r (t), the filtered signal is

[0020] Doppler filtering is used to enhance the moving target signal, and the spherical coordinates (R r ,θ r ,φ r ) is converted to rectangular coordinates (X r ,Y r ,Z r ), the conversion formula is:

[0021]

[0022] At the same time, time synchronization is performed to ensure that the time is consistent with other sensor data;

[0023] LiDAR data preprocessing: denoising the 3D point cloud data and downsampling using a voxelization algorithm;

[0024] Assume that the number of original point cloud data points is N0, the number after downsampling is N1, and the downsampling rate is r s =N1 / N0;

[0025] Spatial alignment of lidar data with other sensor data through coordinate transformation;

[0026] Photoelectric composite sensor data preprocessing: grayscale, filtering and enhancement of camera images;

[0027] Let the original color image be I c (x,y), grayscaled to I gray (x,y), the grayscale formula using the weighted average method is:

[0028] I gray (x,y)=0.299R(x,y)+0.587G(x,y)+0.114B(x,y),

[0029] Where R(x,y), G(x,y), and B(x,y) are the RGB components of the color image respectively;

[0030] Use image segmentation algorithm to extract the contour features of drone;

[0031] Perform temperature correction and pseudo-color processing on infrared thermal imager data to enhance thermal feature display;

[0032] RF sensor data preprocessing: demodulate the received RF signal and restore the communication data;

[0033] Using spectrum analysis technology, let the spectrum of the signal be S(f), and get it through fast Fourier transform Extract signal characteristic parameters;

[0034] Establish a RF signal feature database, compare the collected signal features with the database, and determine the type of drone.

[0035] As a further solution of the present invention: in said S3, data fusion based on feature association and decision-level fusion are included, wherein the data fusion based on feature association specifically includes: Extract feature vectors from pre-processed data of each sensor; Microwave radar characteristic vector F r =[R r ,V r ]; LiDAR feature vector F l =[R l,avg ,θ l,avg ,φ l,avg , shape features], where R l,avg represents the distance average, θ l,avg Indicates the average azimuth angle, φ l,avg represents the average pitch angle; Photoelectric composite sensor feature vector F p =[visual features, thermal features]; RF sensor feature vector F s =[f s ,B s ,M s ]; The association between feature vectors is determined by calculating the cosine similarity. The cosine similarity formula is:

[0036] When cos(F i ,F j )≥λ, the feature vectors are considered to come from the same UAV target, λ is the set threshold, 0<λ<1.

[0037] As a further solution of the present invention: the decision-level fusion specifically includes:

[0038] Assume the microwave radar weight is w r , the lidar weight is wl , the weight of the photoelectric composite sensor is w p , the weight of the RF sensor is w s , and w r +w l +w p +w s =1;

[0039] Assume that the voting value of each sensor is v r 、v l 、v p 、v s , the final decision value:

[0040] V=Σ k＝r,l,p,s w k v k ;

[0041] Determine the drone's status and threat level based on V.

[0042] As a further solution of the present invention: S4 specifically includes:

[0043] Assess the threat level of drones based on the fused data and select countermeasures if a threat is posed;

[0044] For long-range drones, the distance R>R th1 , R th1 is the long distance threshold, using radio frequency interference, the transmission power P j Satisfy P j ≥P s +G j- L p , where G j is the interference antenna gain, L p is the propagation loss;

[0045] For medium and short range drones R th2 <R≤R th1 , Rth2 is the medium and short distance threshold, which starts the laser blinding system;

[0046] For UAVs approaching the target area, R≤R th2 , using a physical capture device.

[0047] The beneficial effects of the present invention are as follows: multi-sensor collaborative deployment and data collection, integrating the advantages of each sensor, significantly improving the accuracy and reliability of drone detection, and reducing missed detection and false alarm rates; innovative data preprocessing and fusion algorithms, deeply mining the complementarity of multi-sensor data, providing accurate basis for countermeasure decision-making, and enhancing the response speed and effect of the countermeasure system; improving the adaptability of drone detection and countermeasure systems in complex environments, and ensuring public safety and the safety of important facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The present invention will be further described below with reference to the accompanying drawings.

[0049] Figure 1 It is a flow chart of the multi-sensor data acquisition and fusion method for the UAV detection and countermeasure system of the present invention. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0051] See also Figure 1 As shown, the present invention is a multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system, comprising the following steps:

[0052] Multi-sensor coordinated deployment: Microwave radar, lidar, optoelectronic composite sensors (including high-definition cameras and infrared thermal imagers), and radio frequency sensors are strategically deployed within the target monitoring area. Each sensor has a clear role: microwave radar is used to quickly detect the drone's approximate position and speed from a long distance; lidar provides high-precision ranging; optoelectronic composite sensors acquire the drone's optical and thermal imaging signatures at medium and short ranges; and radio frequency sensors intercept the drone's radio frequency signals.

[0053] Data collection:

[0054] Microwave radar data acquisition: Microwave radar uses a scanning period T r Scan the airspace and obtain the distance R of the drone each time r , azimuth angle θ r , pitch angle φ r and radial velocity V r . At the same time, the radar echo signal strength I is collected r and phase Φ r , that is, the collected data can be expressed as {R r ,θ r ,φ r ,V r ,I r ,Φ r}.

[0055] LiDAR data acquisition: Based on the target position provided by the microwave radar, the LiDAR measures the distance to the target. l Within, a series of distance data R is collected l,i(i=1,2,...,n, n is the number of measurement points) and the corresponding angle data θ l,i 、φ l,i , construct 3D point cloud data {(R l,i ,θ l,i ,φ l,i )}.

[0056] Photoelectric composite sensor data acquisition: HD camera with frame rate f c The optical image is collected and the infrared thermal imager works synchronously. p The image sequence captured by the camera is {I c,j}(j=1,2,...,m, m is the number of image frames), record the appearance characteristics of the UAV; the thermal imaging data collected by the infrared thermal imager is {T i,j}, reflecting the temperature distribution of the drone.

[0057] RF sensor data collection: RF sensors monitor specific frequency bands and collect signal frequency f when drone communication signals are detected. s Bandwidth B s , modulation mode M s , signal strength P s And the rate of change of the signal over time dPs / dt.

[0058] Data preprocessing:

[0059] Microwave radar data preprocessing: Adaptive filtering algorithm is used to remove noise and clutter. Assume that the original radar signal is S r (t), the filtered signal is Doppler filtering is used to enhance the moving target signal, and the spherical coordinates (R r ,θ r ,φ r ) is converted to rectangular coordinates (X r ,Y r ,Z r ), the conversion formula is:

[0060]

[0061] At the same time, time synchronization is performed to ensure that the time is consistent with other sensor data.

[0062] LiDAR data preprocessing: denoise the 3D point cloud data and use voxelization algorithm to downsample. Assume that the number of original point cloud data points is N0, the number after downsampling is N1, and the downsampling rate is r s = N1 / N0. The LiDAR data is spatially aligned with other sensor data through coordinate transformation.

[0063] Photoelectric composite sensor data preprocessing: grayscale, filtering and enhancement of camera images. Let the original color image be I c (x,y), grayscaled to I gray (x, y), the grayscale formula using weighted average method is I gray (x,y)=0.299R(x,y)+0.587G(x,y)+0.114B(x,y), where R(x,y), G(x,y), and B(x,y) are the RGB components of the color image. Image segmentation algorithms are used to extract the drone's contour features. Thermal imager data is temperature-corrected and pseudo-colored to enhance thermal signatures.

[0064] RF sensor data preprocessing: demodulate the received RF signal and restore the communication data. Using spectrum analysis technology, let the spectrum of the signal be S(f), and obtain it through fast Fourier transform (FFT) Extract signal characteristic parameters. Build a RF signal characteristic database, compare the collected signal characteristics with the database, and determine the type of drone.

[0065] Data Fusion: Data fusion based on feature association: Extract feature vectors from the pre-processed data of each sensor. Microwave radar feature vector F r =[R r ,V r ], the laser radar feature vector F l =[R l,avg ,θ l,avg ,φ l,avg , shape features](R l,avg ,θ l,avg 、φ l,avg is the average value of distance, azimuth and pitch angle), the characteristic vector F of the photoelectric composite sensor p =[visual features, thermal features], RF sensor feature vector F s =[f s ,B s ,M s ]. The association between feature vectors is determined by calculating the cosine similarity. The cosine similarity formula is: ; When cos(F i ,F j )≥λ (λ is the set threshold, 0<λ<1), the feature vectors are considered to come from the same UAV target.

[0066] Decision-level fusion: Using weighted voting, let the microwave radar weight be w r , the lidar weight is w l , the weight of the photoelectric composite sensor is wp , the weight of the RF sensor is w s , and w r +w l +w p +w s = 1. Each sensor votes on the drone’s recognition results, location information, and threat assessment. Let the voting value of each sensor be v r 、v l 、v p 、v s , the final decision value V = Σ k＝r,l,p,s w k v k . Judge the status and threat level of the drone based on V.

[0067] Countermeasure decision: Evaluate the threat level of drones based on the fused data. If a threat is posed, select countermeasures based on its position (X, Y, Z), speed V, type, and other information. For long-range drones (distance R>R th1 , R th1 is the long distance threshold), using radio frequency interference, the transmission power P j Satisfy P j ≥P s +G j -L p , where G j is the interference antenna gain, L p is the propagation loss; for short-range and medium-range UAVs (R th2 <R≤R th1 , R th2 is the medium and short distance threshold), the laser blinding system is activated; for UAVs approaching the target area (R≤R th2 ), using a physical capture device.

[0068] In a specific embodiment:

[0069] Data collection implementation:

[0070] Microwave radar data acquisition: Microwave radar scanning period T r = 2s, the distance R of the drone is obtained in a certain scan r =800m, azimuth angle θ r =45°, pitch angle φ r =20°, radial velocity V r =15m / s, echo signal strength I r =30dBm, phase Φ r =0.5π.

[0071] LiDAR data collection: l=15s, the laser radar collects data of n=1000 measurement points, such as R l,1 =802m,θ l,1 =44.5°,φ l,1 =20.5°, etc., to construct three-dimensional point cloud data.

[0072] Photoelectric composite sensor data acquisition: HD camera frame rate f c =25fps, at t p = m = 250 frames of imagery captured within 10 seconds. The thermal imager simultaneously collects thermal imaging data, showing areas with higher engine temperatures.

[0073] RF sensor data acquisition: UAV communication signal is detected, frequency f s =2.42GHz, bandwidth B s =10MHz, modulation mode M s =ASK, signal strength P s =-50dBm, signal change rate dPs / dt = 2dBm / s.

[0074] Data preprocessing implementation:

[0075] Microwave radar data preprocessing: Adaptive filtering and Doppler filtering are performed on the radar signal, and the spherical coordinates (800, 45°, 20°) are converted to rectangular coordinates (X r ,Y r ,Z r ), calculated X r ≈514.23m, Y r ≈514.23m, Z r ≈751.75m.

[0076] LiDAR data preprocessing: denoising and voxel downsampling of 3D point cloud data, downsampling rate r s = 0.1. After coordinate transformation, the data space is aligned with other sensor data.

[0077] Photoelectric composite sensor data preprocessing: grayscale, filtering, and enhancement of camera images to extract drone contour features. Temperature correction and pseudo-color processing of infrared thermal imager data.

[0078] RF sensor data preprocessing: Demodulate the RF signal, extract characteristic parameters through spectrum analysis, and compare them with the database to determine whether the drone is a certain model of consumer-grade product.

[0079] Data fusion implementation:

[0080] Data fusion based on feature association: Extract feature vectors of each sensor, such as microwave radar F r=[800,15]. Calculate the cosine similarity, assuming λ = 0.8, if cos(F r ,F l )≥0.8 and other conditions are met, it is determined that the data comes from the same drone.

[0081] Decision-level fusion: Assume w r =0.3, w l =0.2, w p =0.3, w s =0.2. The voting value of each sensor v r =0.8, v l =0.7, v p =0.9, v s =0.8, the final decision value V = 0.3×0.8+0.2×0.7+0.3×0.9+0.2×0.8=0.81, judging that the drone is a high-risk target.

[0082] Countermeasure decision implementation: It is known that the distance between the drone and the airport is R = 900m > R th1 =850m, using radio frequency interference. Assume that the interference antenna gain G j =15dB, propagation loss L p =30dB, then the interference transmission power P J ≥-50+15-30=-65dBm, the radio frequency jammer is activated to transmit jamming signals, successfully blocking the drone's communication link and eliminating the threat.

[0083] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system, characterized in that: The following steps are involved: S1: Rationally deploy multiple sensors in the target area to form a comprehensive monitoring network; S2: Perform corresponding data preprocessing according to the type of collected data; S3: Based on feature association, feature vectors are extracted from the pre-processed data of each sensor, and cosine similarity is used to determine whether they come from the same drone. A weighted voting method is used for decision-level fusion, and the voting values ​​of each sensor are combined to determine the drone status and threat level. S4: Assess threats based on fused data and take different countermeasures based on the distance to the drone, including long-range radio frequency interference, medium- and short-range laser blinding, and close-range physical capture.

2. The multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system according to claim 1 is characterized in that: The sensors include microwave radar, laser radar, photoelectric composite sensor and radio frequency sensor.

3. The multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system according to claim 2 is characterized in that: In said S2, data collection and data preprocessing are included, wherein data collection specifically includes: Microwave radar data acquisition: with scanning period T r Scan the airspace and obtain the distance R of the drone r , azimuth angle θ r , pitch angle φ r , radial velocity V r , echo signal strength I r and phase Φ r ; LiDAR data acquisition: Based on the target position provided by the microwave radar, at the measurement time t l Collect distance data R l,i , angle data θ l,i and φ l,i , i=1,2,...,n, n is the number of measurement points, constructing 3D point cloud data; Photoelectric composite sensor data acquisition: HD camera with frame rate f c Acquisition of optical image sequence {I c,j }, j = 1, 2, ..., m, m is the number of image frames, the infrared thermal imager synchronously collects thermal imaging data {T i,j }; RF sensor data acquisition: monitors a specific frequency band and collects signal frequency f when a drone communication signal is detected s Bandwidth B s , modulation mode M s , signal strength P s And the rate of change of the signal over time dPs / dt.

4. The multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system according to claim 3 is characterized in that: The data preprocessing includes: Microwave radar data preprocessing: Adaptive filtering algorithm is used to remove noise and clutter; Assume the original radar signal is S r (t), the filtered signal is Doppler filtering is used to enhance the moving target signal, and the spherical coordinates (R r ,θ r ,φ r ) is converted to rectangular coordinates (X r ,Y r ,Z r ), the conversion formula is: At the same time, time synchronization is performed to ensure that the time is consistent with other sensor data; LiDAR data preprocessing: denoising the 3D point cloud data and downsampling using a voxelization algorithm; Assume that the number of original point cloud data points is N0, the number after downsampling is N1, and the downsampling rate is r s =N1 / N0; Spatial alignment of lidar data with other sensor data through coordinate transformation; Photoelectric composite sensor data preprocessing: grayscale, filtering and enhancement of camera images; Let the original color image be I c (x,y), grayscaled to I gray (x,y), the grayscale formula using the weighted average method is: I gray (x,y)=0.299R(x,y)+0.587G(x,y)+0.114B(x,y), Where R(x,y), G(x,y), and B(x,y) are the RGB components of the color image respectively; Use image segmentation algorithm to extract the contour features of drone; Perform temperature correction and pseudo-color processing on infrared thermal imager data to enhance thermal feature display; RF sensor data preprocessing: demodulate the received RF signal and restore the communication data; Using spectrum analysis technology, let the spectrum of the signal be S(f), and get it through fast Fourier transform Extract signal characteristic parameters; Establish a RF signal feature database, compare the collected signal features with the database, and determine the type of drone.

5. The multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system according to claim 1, characterized in that: In said S3, data fusion based on feature association and decision-level fusion are included, wherein the data fusion based on feature association specifically includes: Extract feature vectors from pre-processed data of each sensor; Microwave radar characteristic vector F r =[R r ,V r ]; LiDAR feature vector F l =[R l,avg ,θ l,avg ,φ l,avg , shape features], where R l,avg represents the distance average, θ l,avg Indicates the average azimuth angle, φ l,avg represents the average pitch angle; Photoelectric composite sensor feature vector F p =[visual characteristics, thermal characteristics]; RF sensor feature vector F s =[f s ,B s ,M s ]; The association between feature vectors is determined by calculating the cosine similarity. The cosine similarity formula is: When cos(F i ,F j )≥λ, the feature vectors are considered to come from the same UAV target, λ is the set threshold, 0<λ<1.

6. The multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system according to claim 5, characterized in that: The decision-level fusion specifically includes: Assume the microwave radar weight is w r , the lidar weight is w l , the weight of the photoelectric composite sensor is w p , the weight of the RF sensor is w s , and w r +w l +w p +w s =1; Assume that the voting value of each sensor is v r 、v l 、v p 、v s , the final decision value: V=∑ k＝r,l,p,s w k v k ; Determine the drone's status and threat level based on V.

7. The multi-sensor data acquisition and fusion method for a UAV detection and countermeasure system according to claim 1, characterized in that: The S4 specifically includes: Assess the threat level of drones based on the fused data and select countermeasures if a threat is posed; For long-range drones, the distance R>R th1 , R th1 is the long distance threshold, using radio frequency interference, the transmission power P j Satisfy P j ≥P s +G j- L p , where G j is the interference antenna gain, L p is the propagation loss; For medium and short range drones R th2 <R≤R th1 , Rth2 is the medium and short distance threshold, which starts the laser blinding system; For UAVs approaching the target area, R≤R th2 , using a physical capture device.

Images