A low-altitude unmanned aerial vehicle cooperative identification method based on multi-source information fusion

Abstract

The application discloses a low-altitude unmanned aerial vehicle cooperative identification method based on multi-source information fusion, solves the problems of high false alarm rate of single detection of an existing low-altitude unmanned aerial vehicle detection system, incapability of cooperation of equipment, and lack of complete defense closed loop, preliminarily screens suspected unmanned aerial vehicle targets through a millimeter wave radar, identifies unmanned aerial vehicle communication signals and measures directions by using spectrum equipment, cooperatively schedules photoelectric equipment to target and track targets, realizes intelligent discrimination of photoelectric targets by relying on an AI model, fuses multi-source data through a dynamic weighted fusion algorithm, determines unmanned aerial vehicle targets after space-time correlation analysis, predicts target trajectories in combination with a Kalman filter model, divides multiple levels of threats according to flight characteristics, matches corresponding warning and countermeasures strategies, feeds back optimization of countermeasure effects, forms a complete low-altitude defense closed loop, and the application is characterized in that multiple sensors are complementary and cooperative, false alarm rate is effectively reduced, target identification precision and tracking stability in a complex environment are improved, intelligent hierarchical prevention and control is realized, and the application is suitable for multiple types of key area low-altitude safety protection scenes.

Description

Technical Field

[0001] This invention relates to the field of low-altitude security defense and intelligent detection technology, and in particular to a collaborative identification method for low-altitude unmanned aerial vehicles based on multi-source information fusion. Background Technology

[0002] With the rapid development of the low-altitude economy and drone technology, drones have been widely used in logistics transportation, aerial photography and mapping, power line inspection, agricultural plant protection and other fields. However, at the same time, security issues such as illegal flights, intrusions, delivery of dangerous goods, and reconnaissance of sensitive areas by low-altitude drones have become increasingly prominent, posing significant hidden dangers to airports, energy facilities, border and coastal defense, key industrial parks and urban public safety.

[0003] Currently, low-altitude UAV detection systems on the market mainly employ a single detection method, including radar detection, radio spectrum detection, or photoelectric identification. Single detection methods have the following problems: 1) When relying solely on radar detection, it is easily interfered with by low-speed, small targets such as birds, kites, and balloons, resulting in a high false alarm rate; 2) When relying solely on spectrum detection, the ability to identify drones that are flying silently, autonomously, or using non-standard communication protocols is limited; 3) Relying solely on photoelectric recognition presents challenges such as short detection range, insufficient nighttime recognition capability, and difficulty in target locking; 4) Existing systems mostly operate independently, lacking a collaborative mechanism between detection devices, making it impossible to share target information and jointly determine targets; 5) Existing systems typically only have target detection capabilities and lack the ability to predict UAV flight trajectories, assess threats, and coordinate automatic countermeasures, making it difficult to form a complete low-altitude defense closed loop.

[0004] Therefore, there is an urgent need for a method for low-altitude UAV collaborative detection and intelligent identification that can integrate multi-source information such as radar, spectrum and photoelectric data to achieve multi-device collaborative detection, intelligent identification, trajectory prediction and automatic countermeasure linkage. Summary of the Invention

[0005] The purpose of this invention is to provide a low-altitude unmanned aerial vehicle (UAV) collaborative identification method based on multi-source information fusion, in order to solve the problems of high false alarm rate, insufficient target identification capability, inability of multiple devices to work collaboratively, weak target tracking capability, and inability to form a complete defense closed loop in existing low-altitude defense systems. This invention constructs a low-altitude target collaborative detection mechanism by fusing multi-source detection information such as millimeter-wave radar, radio spectrum detection, and photoelectric identification, so as to realize joint judgment, intelligent identification, trajectory prediction, threat assessment, and automatic countermeasure linkage of UAV targets, thereby improving the detection accuracy, identification efficiency, and automation level of the low-altitude defense system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a collaborative identification method for low-altitude unmanned aerial vehicles based on multi-source information fusion, comprising the following steps: S1. Initial Target Detection: Millimeter-wave three-dimensional active radar is used to scan the target airspace in real time to obtain the spatial coordinates, distance, azimuth, pitch angle, flight speed, motion trajectory and target size information of the aerial target; based on the target flight characteristics, suspected UAV targets are preliminarily identified, and the spatial position and motion parameters of the suspected UAV targets obtained by screening are sent to the multi-source information fusion module; S2. Spectrum Signal Detection: The system scans for radio signals in the 2.4GHz and 5.8GHz bands within the target area using radio spectrum detection equipment. It identifies UAV communication signals by combining protocol feature recognition and frequency band feature recognition. When a suspected UAV communication signal is detected, the system generates a spectrum alarm message and uses a direction finding algorithm to estimate the target signal's azimuth, outputting the corresponding target azimuth angle information. The system also sends the spectrum direction finding results, protocol recognition results, and timestamps to the multi-source information fusion module. The frequency, bandwidth, protocol frame structure and modulation characteristics of the communication signal are analyzed, the UAV communication protocol is identified and the signal location is estimated, and the protocol identification result, direction finding result and timestamp information are sent to the multi-source information fusion module. S3. Coordinated Scheduling and Target Guidance: When radar or spectrum equipment detects a suspected UAV target, it coordinates the optoelectronic equipment according to the target's spatial coordinates and timestamp; it converts the three-dimensional coordinates (x, y, z) output by the radar into gimbal control angles, and achieves automatic pointing, automatic tracking, and automatic zooming through a combination of PID control and visual feedback tracking; when target tracking fails, it executes a target recovery strategy: when there are multiple suspected targets, it selects the target to track based on the priority of the comprehensive score. S4. Photoelectric AI Intelligent Recognition: After the photoelectric device acquires the target video image, it extracts and analyzes the target's appearance, contour, motion, and infrared thermal imaging features through a convolutional neural network target detection model to identify quadcopter drones, FPV drones, and other low-altitude drone targets, while also distinguishing birds, kites, balloons, and other non-drone low-altitude targets; it outputs the target category, confidence level, bounding box position, motion direction, and tracking status; and sends the image recognition results to the multi-source information fusion module. S5. Multi-source information fusion judgment: Multi-source fusion analysis is performed on radar detection information, spectrum detection information and photoelectric recognition results, including radar target motion feature analysis, radio spectrum feature analysis, photoelectric image target feature analysis, multi-source target spatiotemporal correlation analysis and target joint confidence analysis; a dynamic weighted fusion algorithm is used to calculate the target comprehensive confidence, and the weights of each data source are dynamically adjusted according to the environment type. When the comprehensive confidence is greater than 0.8, the target is confirmed to be an unmanned aerial vehicle (UAV). S6, Track prediction and threat assessment: predicting the target flight position by combining the target's historical track, velocity and acceleration based on the Kalman filter prediction model; dividing the target into four threat levels: low, medium, high and extremely high according to the target's flight area, flight direction, flight speed and flight behavior; S7, Automatic alarm and countermeasure linkage: triggering corresponding alarm and countermeasure operations according to the target threat level, and monitoring the countermeasure effect in real time; when the countermeasure effect does not meet the preset conditions, automatically adjusting the countermeasure strategy and regenerating control instructions, forming a low-altitude defense closed loop of detection-identification-fusion-prediction-evaluation-countermeasure-feedback.

[0007] Further, the judgment conditions for preliminary screening of suspected UAV targets in step S1 are: the target flight speed satisfies 0<V<35m / s, where V is the target flight speed; the flight altitude satisfies 0<H<300m, where H is the target flight altitude; the size satisfies 0.2<S<0.8m, where S is the equivalent size of the target; the radar cross section of the target conforms to the characteristics of low-slow-small aircraft, the periodic modulation characteristics of rotor wings and the UAV flight track characteristics of hovering / fixed-point circling / small-range maneuvering are detected, and bird targets with periodic wing-flapping micro-Doppler characteristics are eliminated; when the target simultaneously satisfies the three basic conditions of flight speed, flight altitude and target size, and satisfies at least two of rotor micro-Doppler characteristics, flight behavior characteristics and radar cross section characteristics, it is marked as a suspected UAV target.

[0008] Further, the UAV communication protocols identified in step S2 include DJI OcuSync communication protocol, WiFi FPV image transmission protocol, and RemoteID protocol; the spectrum direction finding error range is ±3°.

[0009] Further, the calculation formula for the horizontal angle of the gimbal in step S3 is , the calculation formula for the pitch angle of the gimbal is , where x, y, z are the three-dimensional coordinates of the target output by the radar, θ is the horizontal angle of the gimbal, ϕ is the pitch angle of the gimbal; the photoelectric equipment includes a visible light camera and an infrared thermal imaging device, and H.265 video streaming is used to collect real-time images.

[0010] Further, in step S3, the recovery strategies when target tracking fails are in order: returning to the last appearance position of the target, performing expanded range search in the target area, calling radar and spectrum equipment for re-guidance, and re-predicting the target position according to the historical track.

[0011] Further, in step S5, the calculation formula of the dynamic weighted fusion algorithm is: ; where: P is the comprehensive confidence, is the radar identification confidence; Confidence level for spectrum identification; For photoelectric recognition confidence level; , , These are the dynamic weights of the corresponding data sources; the weight adjustment rules are as follows: increase the infrared photoelectric weight in nighttime environments, increase the radar weight in strong electromagnetic interference environments, and decrease the spectrum weight in spectrum-quiet target scenarios. When P>0.8, the system confirms that the target is a UAV target.

[0012] Furthermore, in step S6, the Kalman filter prediction model establishes the target state vector: , in, , , For the target spatial location, , , For the target velocity component; The system predicts the target location based on the state transition model: ; Where is the target state at the current moment; A is the state transition matrix; B is the control matrix; For control input; This is system noise; The flight behaviors include: the target approaching a no-fly zone; the target hovering for an extended period of time; and the target approaching a key area at high speed.

[0013] Furthermore, in step S7, the tiered handling rules are as follows: low-threat targets only trigger alarm prompts, medium-threat targets activate radio jamming equipment, high-threat targets activate navigation decoy equipment, and extremely high-threat targets activate directional suppression equipment; the countermeasure effect evaluation indicators include whether the target returns to base, whether it makes an emergency landing, whether the communication link is lost, whether the flight trajectory deviates from the predetermined route, and whether it leaves the warning area.

[0014] The beneficial effects of the present invention are: (1) High detection reliability: The present invention uses millimeter-wave three-dimensional radar, radio spectrum detection equipment and optoelectronic equipment to work together. By utilizing the information complementarity between different sensors, it can effectively solve the problems of detection blind spots, weak environmental adaptability and target misjudgment that exist in a single detection method, and improve the probability of UAV target detection in complex low-altitude environments. (2) Strong anti-interference capability: This invention integrates the target's spatial location, motion trajectory, radio communication characteristics, image texture characteristics and infrared thermal characteristics to comprehensively identify non-UAV low-altitude targets such as birds, kites and balloons, thereby reducing false alarms caused by environmental factors and improving identification accuracy; (3) Strong target recognition capability: This invention uses AI large model target detection to analyze visible light and infrared images, which can realize the classification and recognition of different types of low-altitude UAV targets. At the same time, it combines radar micro-Doppler features and communication protocol features to improve the recognition capability of UAVs in different flight states and different communication modes. (4) High degree of equipment collaboration and automation: This invention realizes data sharing between radar, spectrum equipment and optoelectronic equipment through target information fusion module, automatically calculates the pointing angle of optoelectronic equipment according to the target position, and realizes automatic target acquisition and continuous tracking by combining closed-loop tracking algorithm, reducing manual operation and improving system response speed; (5) Strong threat assessment capability: This invention combines the target's flight altitude, speed, flight path direction, regional attributes and historical movement trajectory, and uses a trajectory prediction model to predict the target's future behavior, and executes a graded response strategy based on the target's threat level, thereby improving the initiative of low-altitude security protection; (6) Strong system scalability: The present invention adopts a modular design, and the various detection devices and processing modules are connected through data interfaces. New sensing devices or algorithm models can be added according to actual application needs. It is suitable for various low-altitude safety protection scenarios such as airports, energy facilities, border areas, and key urban areas. This invention can improve the accuracy, real-time performance, and intelligence of low-altitude UAV detection and identification systems, reduce false alarm rates, and enhance the safety management capabilities of UAVs in complex environments. Attached Figure Description

[0015] The present invention will be further described below with reference to the accompanying drawings: Figure 1 This is a schematic diagram of the overall structure of a low-altitude unmanned aerial vehicle (UAV) collaborative detection and intelligent identification system that integrates multi-source information. Detailed Implementation

[0016] The present invention will be further described in detail below with reference to embodiments and specific implementation methods: Example 1

[0017] like Figure 1 As shown, the overall architecture of the low-altitude UAV collaborative identification method system of the present invention consists of millimeter-wave radar, spectrum detection equipment, optoelectronic equipment, gimbal control system, multi-source information fusion module, AI intelligent identification module, trajectory prediction module, threat assessment module, decision and linkage control module, and execution and feedback module. The whole system operates according to the closed-loop logic of multi-source detection-collaborative scheduling-AI identification-data fusion-trajectory prediction-threat assessment-tiered countermeasures.

[0018] When the system is in operation, the millimeter-wave 3D active radar continuously scans the designated airspace around the clock, collecting the distance, azimuth, pitch, 3D spatial coordinates, flight speed, flight altitude, equivalent size, radar cross section, and micro-Doppler features of all flying targets within the airspace. The system filters targets based on preset thresholds: low, slow, and small targets with flight speeds of 0-35 m / s, flight altitudes of 0-300 m, and equivalent sizes of 0.2-0.8 m are selected. The system combines radar cross section, rotor micro-Doppler features, hovering / circling flight trajectory features, etc., to determine the aircraft's attributes. At the same time, it identifies and eliminates bird-specific periodic wing-beating micro-Doppler features, and uploads the parameters of suspected UAV targets that meet the judgment criteria to the multi-source information fusion module.

[0019] The radio spectrum detection equipment simultaneously scans the two major operating frequency bands of drones, 2.4GHz and 5.8GHz, analyzes the signal frequency, bandwidth, frame structure, and modulation method, and identifies mainstream drone communication protocols such as DJI OcuSync, WiFi FPV image transmission, and Remote ID. When a suspected drone communication signal is detected, a direction-finding algorithm is used to calculate the signal's azimuth, with the direction-finding error controlled within ±3°. Data such as protocol type, frequency, bandwidth, target azimuth, and timestamp are uploaded to the multi-source information fusion module, and a spectrum alarm is generated simultaneously.

[0020] After receiving a suspected target signal, the multi-source information fusion module extracts the target's three-dimensional coordinates output by the radar. The system calculates the horizontal and vertical angles of the gimbal using geometric formulas, driving the gimbal control system to adjust the angles. The optoelectronic equipment, equipped with a visible light camera and an infrared thermal imaging device, acquires real-time images of the target area using H.265 video streams. The system employs PID control combined with visual feedback to form a closed-loop tracking system, ensuring continuous target lock. If tracking is lost, the system sequentially executes four recovery strategies: reverting to the last position, expanding the search range, radar / spectrum secondary guidance, and historical trajectory prediction and positioning. When multiple suspected targets exist in the airspace, the system scores targets based on distance, threat level, flight direction, identification confidence, and motion status, prioritizing the tracking of high-scoring targets.

[0021] The real-time images acquired by the optoelectronic equipment are input into an AI target recognition model built on a convolutional neural network. The model extracts the target's outline, dynamic features, and infrared thermal imaging features, and classifies and identifies quadcopter drones, FPV drones, and other types of drones. At the same time, it accurately distinguishes interfering targets such as birds, kites, and balloons. The model outputs information such as target category, recognition confidence, image bounding box, direction of motion, and tracking status, and uploads it to the fusion module.

[0022] The multi-source information fusion module aggregates radar motion characteristic data, spectrum communication characteristic data, and photoelectric image characteristic data, and combines timestamps and spatial coordinates to complete the spatiotemporal correlation of multi-source data. A formula is used...

[0023] The system calculates the overall confidence level and dynamically adjusts the weights based on the on-site environment: increases the weight of infrared photoelectric sensors in nighttime scenarios, increases the weight of radar sensors in scenarios with strong electromagnetic interference, and reduces the weight of the spectrum for communication-silent drones; when the overall confidence level P>0.8, the system officially determines that the target is a real drone.

[0024] The trajectory prediction module constructs a Kalman filter model and establishes the target state vector. , in: , , The target spatial location; , , For the target velocity component; Based on the state transition equation , in: A represents the target state at the current moment; B represents the state transition matrix; and C represents the control matrix. For control input; This is system noise; By combining the target's historical location, speed, and acceleration to predict its future flight trajectory, the threat assessment module combines the target's behavior, such as whether it is approaching a no-fly zone, hovering for a long time, or approaching a key area at high speed, with parameters such as flight speed and flight area, to classify the target into four threat levels: low, medium, high, and extremely high.

[0025] The system performs tiered responses based on threat level: low-threat targets only issue audible and visual / background alarms; medium-threat targets activate radio jamming equipment to cut off the drone's remote control and image transmission links; high-threat targets activate navigation decoy equipment to mislead the drone's navigation system; and extremely high-threat targets activate directional suppression equipment to suppress signals across the entire area. The system monitors the target's return to base, forced landing, communication interruption, trajectory deviation, and departure from the warning zone in real time to assess the effectiveness of countermeasures. If countermeasures fail, the system automatically adjusts the countermeasure mode and parameters and continues to execute response actions. After this step is completed, the data is fed back to the front-end detection module, forming a complete low-altitude defense closed loop.

[0026] This invention can be deployed independently or used in a network, and is suitable for various low-altitude security scenarios such as airports, border and coastal defense, petrochemical energy facilities, government parks, and urban core areas. It features high detection accuracy, strong anti-interference ability, high degree of automation, and flexible deployment. Example 2

[0027] In this embodiment, the system is deployed in the vicinity of the airport's perimeter warning area to monitor the airspace with a radius of 10km and an altitude of less than 300m.

[0028] When the system is working, the millimeter-wave radar first scans the target airspace in real time. The millimeter-wave radar is a three-dimensional active millimeter-wave radar that can acquire target distance, azimuth angle, pitch angle, flight speed and motion trajectory information in real time. It can also perform preliminary detection of low, slow and small targets. After the radar detects a suspected target, it sends the target's spatial coordinates and motion parameters to the multi-source information fusion module.

[0029] The radar detected a flying object: flight speed 12m / s, altitude 50m, size approximately 0.4m. The target's radar cross-section matches that of a UAV, and rotor micro-Doppler modulation characteristics were detected. In addition, the target exhibited hovering and turning behavior. The system determined that it met the basic motion parameters and two of the rotor and behavioral characteristics, and marked it as a suspected UAV target.

[0030] The spectrum detection equipment simultaneously scans the radio signals in the target area and detects suspected drone image transmission and remote control signals in the 2.4GHz and 5.8GHz frequency bands. After protocol feature analysis, it matches the DJI OcuSync protocol and the direction finding azimuth angle is 135°±2°. The system then sends this information to the multi-source information fusion module.

[0031] After receiving the radar coordinates and spectral azimuth information, the system determines the target's location as x=500m, y=500m, z=50m. Through coordinated scheduling and control, the system controls the optoelectronic equipment for linked identification, and the pan-tilt-zoom (PTZ) angle is adjusted. =45°, pitch angle With a range of approximately 4.0°, the gimbal automatically turns to the target area, and PID control combined with visual feedback enables automatic tracking and automatic zooming.

[0032] The optoelectronic equipment uses H.265 video stream to acquire images of the target area in real time and sends the image data to the AI ​​intelligent recognition module. The AI ​​intelligent recognition module extracts the target shape features based on a deep learning image recognition model, and the recognition result is a quadcopter drone with a confidence level of 0.94.

[0033] The multi-source information fusion module performs spatiotemporal correlation analysis on the millimeter-wave radar detection results (confidence level 0.85), spectrum detection results (confidence level 0.90), and photoelectric recognition results (confidence level 0.94). The radar target and the spectrum target have an azimuth angle difference of 2.5° and a time difference of 0.2s, thus they are determined to be the same target. Dynamic weighted fusion is used: at this time, it is evening, and the default weights are... =0.4, =0.3, =0.3; Calculate the overall confidence level: P = 0.4 * 0.85 + 0.3 * 0.90 + 0.3 * 0.94 = 0.34 + 0.27 + 0.282 = 0.892 > 0.8, the system confirms the target is a drone.

[0034] The trajectory prediction module uses a Kalman filter to predict the target's position at the next moment based on the target's 10 most recent historical trajectory points. The current state vector is X=[500,500,50,12,5,0], dt=0.1s. This means the target will continue to move northeast.

[0035] The target is currently about 1.2km from the airport boundary, flying towards the runway, and at an altitude of 50m, which is below the minimum safe altitude for approach routes. According to behavioral rules, it is approaching a key area at high speed, and the threat level is assessed as high.

[0036] Due to the high threat level, the system activated the drone jamming equipment to cut off the signal, forcing the drone to return to base. At the same time, it pushed alarm information to the management personnel and highlighted the target trajectory and threat level on the situation display interface.

[0037] During the countermeasure process, the system monitors the target's flight status and trajectory changes in real time. After the countermeasure command is executed, the target changes its course and enters a return-to-home flight state, gradually moving away from the key protection area. Subsequently, the target's flight trajectory deviates from the original intrusion direction and flies away from the preset warning area. The system comprehensively judges the countermeasure effect based on indicators such as the target's course change, flight distance, communication status, and whether it has left the protection area. When the target meets the preset safety conditions, the countermeasure is deemed successful.

[0038] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A collaborative identification method for low-altitude unmanned aerial vehicles (UAVs) based on multi-source information fusion, characterized in that, Includes the following steps: S1. Initial Target Detection: Millimeter-wave three-dimensional active radar is used to scan the target airspace in real time to obtain the spatial coordinates, distance, azimuth, pitch angle, flight speed, trajectory and size information of the aerial target; Based on the target flight characteristics, the suspected drone targets are initially identified, and the spatial position and motion parameters of the suspected drone targets obtained from the screening are sent to the multi-source information fusion module. S2. Spectrum Signal Detection: The system scans for radio signals in the 2.4GHz and 5.8GHz bands within the target area using radio spectrum detection equipment. It identifies UAV communication signals by combining protocol feature recognition and frequency band feature recognition. When a suspected UAV communication signal is detected, the system generates a spectrum alarm message and uses a direction finding algorithm to estimate the target signal's azimuth, outputting the corresponding target azimuth angle information. The system also sends the spectrum direction finding results, protocol recognition results, and timestamps to the multi-source information fusion module. The frequency, bandwidth, protocol frame structure and modulation characteristics of the communication signal are analyzed, the UAV communication protocol is identified and the signal location is estimated, and the protocol identification result, direction finding result and timestamp information are sent to the multi-source information fusion module. S3. Coordinated Scheduling and Target Guidance: When radar or spectrum equipment detects a suspected UAV target, it coordinates the optoelectronic equipment according to the target's spatial coordinates and timestamp; it converts the three-dimensional coordinates (x, y, z) output by the radar into gimbal control angles, and achieves automatic pointing, automatic tracking, and automatic zooming through a combination of PID control and visual feedback tracking; when target tracking fails, it executes a target recovery strategy: when there are multiple suspected targets, it selects the target to track based on the priority of the comprehensive score. S4. Photoelectric AI Intelligent Recognition: After the photoelectric device acquires the target video image, it extracts and analyzes the target's appearance, contour, motion, and infrared thermal imaging features through a convolutional neural network target detection model to identify quadcopter drones, FPV drones, and other low-altitude drone targets, while also distinguishing birds, kites, balloons, and other non-drone low-altitude targets; it outputs the target category, confidence level, bounding box position, motion direction, and tracking status; and sends the image recognition results to the multi-source information fusion module. S5. Multi-source information fusion judgment: Multi-source fusion analysis is performed on radar detection information, spectrum detection information and photoelectric recognition results, including radar target motion feature analysis, radio spectrum feature analysis, photoelectric image target feature analysis, multi-source target spatiotemporal correlation analysis and target joint confidence analysis; a dynamic weighted fusion algorithm is used to calculate the target comprehensive confidence, and the weights of each data source are dynamically adjusted according to the environment type. When the comprehensive confidence is greater than 0.8, the target is confirmed to be an unmanned aerial vehicle (UAV). S6. Trajectory Prediction and Threat Assessment: Based on the Kalman filter prediction model combined with the target's historical trajectory, speed, and acceleration, the target's flight position is predicted; based on the target's flight area, flight direction, flight speed, and flight behavior, the target is divided into four threat levels: low, medium, high, and extremely high. S7. Automatic alarm and countermeasure linkage: triggering corresponding alarm and countermeasure operations according to the target threat level, and monitoring the countermeasure effect in real time; when the countermeasure effect does not meet the preset conditions, automatically adjusting the countermeasure strategy and regenerating control instructions, forming a low-altitude defense closed-loop of detection-identification-fusion-prediction-evaluation-countermeasure-feedback-.

2. The low-altitude UAV cooperative identification method based on multi-source information fusion according to claim 1, characterized in that, The judgment conditions for preliminary screening of suspected unmanned aerial vehicle (UAV) targets in step S1 are: The target flight speed satisfies 0<V<35m / s, where V is the target flight speed; the flight altitude satisfies 0<H<300m, where H is the target flight altitude; the size satisfies 0.2<S<0.8m, where S is the equivalent size of the target; the radar cross section of the target conforms to the characteristics of low, slow and small aircraft, the periodic modulation characteristics of rotors and the flight path characteristics of UAVs such as hovering / fixed-point circling / small-range maneuvering are detected, and bird targets with micro-Doppler characteristics of periodic wing flapping are eliminated; when the target simultaneously satisfies the three basic conditions of flight speed, flight altitude and target size, and satisfies at least two of the rotor micro-Doppler characteristics, flight behavior characteristics and radar cross section characteristics, it is marked as a suspected UAV target.

3. The low-altitude UAV cooperative identification method based on multi-source information fusion according to claim 1, characterized in that, The UAV communication protocols identified in step S2 include DJI OcuSync communication protocol, WiFi FPV video transmission protocol and Remote ID protocol; the spectrum direction finding error range is ±3°.

4. The low-altitude UAV cooperative identification method based on multi-source information fusion according to claim 1, characterized in that, The formula for calculating the horizontal angle of the gimbal in step S3 is as follows: The formula for calculating the gimbal pitch angle is: Where x, y, z are the target's three-dimensional coordinates output by the radar, θ is the horizontal angle of the gimbal, and ϕ is the gimbal's elevation angle; the optoelectronic equipment includes a visible light camera and an infrared thermal imaging device, which uses H.265 video stream to acquire real-time images.

5. The low-altitude UAV cooperative identification method based on multi-source information fusion according to claim 1, characterized in that, In step S3, the recovery strategies when target tracking fails are in order: returning to the last position where the target appeared, performing expanded range search in the target area, invoking radar and spectrum equipment for re-guidance, and re-predicting the target position based on historical tracks.

6. The low-altitude UAV cooperative identification method based on multi-source information fusion according to claim 1, characterized in that, In step S5, the calculation formula for the dynamic weighted fusion algorithm is as follows: ; Where: P is the overall confidence level. To determine the confidence level for radar identification; Confidence level for spectrum identification; For photoelectric recognition confidence level; , , These are the dynamic weights of the corresponding data sources; the weight adjustment rules are as follows: increase the infrared photoelectric weight in nighttime environments, increase the radar weight in strong electromagnetic interference environments, and decrease the spectrum weight in spectrum-quiet target scenarios. When P>0.8, the system confirms that the target is a UAV target.

7. The low-altitude UAV cooperative identification method based on multi-source information fusion according to claim 1, characterized in that, In step S6, the Kalman filter prediction model establishes a target state vector: ， in, , , For the target spatial location, , , For the target velocity component; The system predicts the target position according to the state transition model: ； in, A represents the target state at the current moment; B represents the state transition matrix; and C represents the control matrix. For control input; This is system noise; The flight behaviors include: the target approaches the no-fly area; the target hovers for a long time; the target approaches the key area at high speed.

8. The low-altitude UAV cooperative identification method based on multi-source information fusion according to claim 1, characterized in that, In step S7, the hierarchical disposal rules are: only an alarm prompt is implemented for low-threat targets, radio interference equipment is activated for medium-threat targets, navigation deception equipment is activated for high-threat targets, and directional suppression equipment is activated for extremely high-threat targets; Countermeasure effect evaluation indicators include whether the target returns, whether it forces landing, whether the communication link disappears, whether the flight track deviates from the planned route, and whether it leaves the warning area.

Images