Phased array radar for unmanned aerial vehicle detection based on multi-target tracking

Abstract

The invention discloses a phased array radar for unmanned aerial vehicle detection based on multi-target tracking, which comprises a radar antenna array, a signal processing unit, a control unit and a display unit, and is characterized in that the control unit comprises a motion state judgment module; a filtering switching module; a Kalman filtering module; a particle filtering module; and a data association module. The method can achieve the high-precision detection and positioning of an unmanned plane target, has a multi-target tracking capability, and improves the detection efficiency and accuracy of the unmanned plane.

Description

Technical Field

[0001] The present invention relates to the field of radar technology, and more specifically, to a phased array radar for detecting unmanned aerial vehicles based on multi-target tracking. Background Art

[0002] With the rapid development of drone technology, its application in military and civilian fields is becoming more and more extensive, but it also brings security risks. Traditional radar systems have limitations in multi-target tracking and detection, and it is difficult to quickly and accurately identify and track drone targets. Phased array radar technology has become an ideal choice for drone detection due to its advantages such as fast beam changes and strong multi-target tracking capabilities. However, when multiple drone clusters appear or appear at the same time as other targets, phased array radars must quickly distinguish, track and identify, which places extremely high demands on multi-target tracking algorithms and is prone to target loss or confusion. Summary of the invention

[0003] The purpose of the present invention is to provide a phased array radar for UAV detection based on multi-target tracking, which can achieve high-precision detection and positioning of UAV targets and has multi-target tracking capabilities, thereby improving the efficiency and accuracy of UAV detection.

[0004] In order to achieve these purposes and other advantages according to the present invention, a phased array radar for UAV detection based on multi-target tracking is provided, comprising: a radar antenna array, a signal processing unit, a control unit and a display unit, wherein the radar antenna array electronically controls the phase and amplitude of the channel feed signal of each antenna to achieve rapid changes in the antenna beam pointing, the signal processing unit is responsible for receiving and processing the echo signal returned by the radar antenna array, extracting the UAV target information, the control unit adjusts the beam pointing of the radar antenna array according to the UAV target information to achieve multi-target tracking, the display unit is used to display the detection results and tracking trajectories of the UAV targets, and the control unit comprises: A motion state judgment module, which is used to judge whether the motion state of each drone is a linear motion state or a nonlinear motion state by analyzing the target information of each drone extracted by the signal processing unit; A filter switching module sets a flag for each drone target to indicate the currently used filter module. When the motion state judgment module determines that the drone is in a linear motion state, the flag points to the Kalman filter module. When the motion state judgment module determines that the drone is in a nonlinear motion state, the flag points to the particle filter module. Kalman filter module, which is used to predict the position of the drone target at the next moment using the Kalman filter algorithm according to the current drone target information of the drone when the drone's flag position points to this module; The particle filter module is used to predict the position of the drone target at the next moment using the particle filter algorithm according to the current drone target information of the drone when the drone's flag bit points to this module; A data association module, which is used to match the latest information of each drone target extracted by the signal processing unit with the predicted position of the drone target using a nearest neighbor data association algorithm; The Kalman filter module is further used to calculate the Kalman gain using the matched UAV target information and the predicted position of the UAV target after the data association module completes the matching, and update the target state estimation and target state covariance of the UAV target; The particle filter module is also used to calculate the likelihood function of the drone target using the matched drone target information and the predicted position of the drone target after the data association module completes the matching. The likelihood function adopts Gaussian likelihood, updates the weight of the drone target according to the likelihood function value, and then resamples to remove drone targets with low weights and retain drone targets with high weights.

[0005] Preferably, the signal processing unit further includes a target recognition module for analyzing the extracted target information and accurately identifying the type of the drone in combination with the feature database of the drone.

[0006] Preferably, the drone target information includes the speed and position of the drone target, and the method in which the motion state judgment module judges whether the motion state of each drone is a linear motion state or a nonlinear motion state by analyzing each drone target information extracted by the signal processing unit includes: Based on the speed of each UAV target at two adjacent moments, the speed change and acceleration of the UAV target are calculated; Based on the flight trajectory of each UAV target, the trajectory curvature of the UAV target is calculated. The flight trajectory of the UAV target is obtained by integrating the position of the UAV target in the time series; The speed change of the drone target is compared with the preset speed change threshold, the acceleration is compared with the preset acceleration threshold, and the trajectory curvature is compared with the preset trajectory curvature threshold. If any of the indicators exceeds the threshold range, it is determined that the motion state of the drone is a nonlinear motion state.

[0007] Preferably, the signal processing unit receives and processes the echo signal returned by the radar antenna array, and the method for extracting the drone target information includes: The time difference between the transmitted signal and the echo signal is measured, and the distance between the drone target and the radar is calculated based on the propagation speed of the electromagnetic wave. The fast Fourier transform algorithm is used to convert the echo signal in the time domain into the frequency domain, extract the Doppler frequency shift information, and then calculate the radial velocity of the UAV target; By measuring the phase difference or amplitude difference of the echo signals received by different antennas, the azimuth and pitch angle of the UAV target are determined using the direction finding algorithm.

[0008] Preferably, the target recognition module acquires the echo signal and synthesizes the image of the UAV target using the inverse synthetic aperture radar imaging method, extracts the shape and size information of the UAV target based on the image of the UAV target, and then matches the shape and size information of the UAV target with the template features of various UAV types pre-stored in the feature database, thereby identifying the type of UAV.

[0009] Preferably, the signal processing unit first performs preprocessing when acquiring the echo signal, including: using a data filtering method to remove noise, then using an amplifier to amplify the echo signal, using a gain control circuit to dynamically adjust the gain of the echo signal, and then converting the echo signal into a discrete digital signal based on a preset sampling frequency and quantization accuracy.

[0010] The present invention has at least the following beneficial effects: The present invention combines Kalman filtering and particle filtering to comprehensively utilize their advantages. In the stage where the UAV's motion is relatively linear, the Kalman filter takes advantage of its efficient linear estimation to provide accurate state estimation; when the UAV exhibits nonlinear motion, the particle filter takes over to ensure accurate tracking of the target state. This complementary mechanism enables the algorithm to adapt to the diverse motion modes of the UAV and improve the overall state estimation accuracy; The nearest neighbor data association algorithm is simple and intuitive, and has a relatively low computational cost. It calculates the distance between the radar measurement data and the target predicted position, quickly assigns the measurement data to the nearest target, and can effectively match the radar echo data with the tracked target to avoid data association confusion, thereby accurately distinguishing different drone targets. And when combined with Kalman filtering and particle filtering, the accuracy of the nearest neighbor data association algorithm is further improved because these two filtering algorithms can provide more accurate target prediction positions; In addition, in actual environments, radar measurement data will be interfered by measurement noise and clutter. Kalman filtering can adjust state estimation according to the statistical characteristics of measurement noise (such as covariance) through its update mechanism, effectively suppressing the influence of measurement noise. Particle filtering also has unique advantages in dealing with noise. It can filter out noise interference to a certain extent by calculating the likelihood function and updating the weights of a large number of particles. This anti-interference ability enables the algorithm to stably track drone targets in complex electromagnetic environments or in the presence of meteorological clutter. The flight state of a drone swarm may change dynamically due to various factors, such as being affected by wind, changing flight missions, etc. The tracking algorithm that combines Kalman filtering, particle filtering and nearest neighbor data association algorithm can adapt to such dynamic changes and uncertainties. Whether it is an increase or decrease in the number of targets, a change in the target motion pattern, or the emergence of new interference factors, the algorithm can respond through its own adjustment mechanism (such as resampling of particle filtering, state update of Kalman filtering and rematching of data association algorithm) to maintain effective tracking of drone swarms.

[0011] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a schematic diagram of the structure of a phased array radar for detecting drones according to an embodiment of the present invention; Figure 2 It is a schematic diagram of the structure of each module in the control unit described in an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0014] It should be noted that the experimental methods described in the following embodiments are conventional methods unless otherwise specified, and the reagents and materials are commercially available unless otherwise specified; in the description of the present invention, the terms "lateral", "longitudinal", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0015] like Figure 1~2As shown, the present invention provides a phased array radar for UAV detection based on multi-target tracking, including: a radar antenna array, a signal processing unit, a control unit and a display unit. The radar antenna array electronically controls the phase and amplitude of the channel feed signal of each antenna to achieve rapid changes in the antenna beam pointing. The signal processing unit is responsible for receiving and processing the echo signal returned by the radar antenna array and extracting the UAV target information. The control unit adjusts the beam pointing of the radar antenna array according to the UAV target information to achieve multi-target tracking. The display unit is used to display the detection results and tracking trajectories of the UAV targets.

[0016] The control unit comprises: A motion state judgment module, which is used to judge whether the motion state of each drone is a linear motion state or a nonlinear motion state by analyzing the target information of each drone extracted by the signal processing unit; A filter switching module sets a flag for each drone target to indicate the currently used filter module. When the motion state judgment module determines that the drone is in a linear motion state, the flag points to the Kalman filter module. When the motion state judgment module determines that the drone is in a nonlinear motion state, the flag points to the particle filter module. Kalman filter module, which is used to predict the position of the drone target at the next moment using the Kalman filter algorithm according to the current drone target information of the drone when the drone's flag position points to this module; The particle filter module is used to predict the position of the drone target at the next moment using the particle filter algorithm according to the current drone target information of the drone when the drone's flag bit points to this module; A data association module, which is used to match the latest information of each drone target extracted by the signal processing unit with the predicted position of the drone target using a nearest neighbor data association algorithm; The Kalman filter module is further used to calculate the Kalman gain using the matched UAV target information and the predicted position of the UAV target after the data association module completes the matching, and update the target state estimation and target state covariance of the UAV target; The particle filter module is also used to calculate the likelihood function of the drone target using the matched drone target information and the predicted position of the drone target after the data association module completes the matching. The likelihood function adopts Gaussian likelihood, updates the weight of the drone target according to the likelihood function value, and then resamples to remove drone targets with low weights and retain drone targets with high weights.

[0017] Specifically, the radar antenna array consists of multiple miniaturized, high-performance antenna units, evenly distributed on a fixed frame. When the system starts, the control unit initializes the channel feed signal of each antenna to ensure that the antenna beam can cover the predetermined monitoring airspace.

[0018] The hardware equipment of the signal processing unit includes a high-performance digital signal processor (DSP) and a large-capacity memory to cope with the processing requirements of a large number of echo signals. In the initialization stage, various signal processing algorithms and data processing processes are loaded to prepare for receiving and processing radar echo signals.

[0019] The display unit is a high-resolution liquid crystal display screen with a display interface that includes a map background of the monitoring area, a display mark of the UAV target, and a display mode of the tracking trajectory.

[0020] When the phased array radar is working, the radar antenna array periodically transmits radar signals to the monitored airspace according to the scanning mode set by the control unit. When the radar signal encounters the drone target, an echo signal is generated. These echo signals are received by the radar antenna array and transmitted to the signal processing unit.

[0021] The signal processing unit performs a series of processing on the received echo signal. First, it removes noise through digital filtering technology to improve the signal-to-noise ratio. Then, it uses pulse compression, Doppler processing and other algorithms to extract the distance, speed, angle and other information of the drone target, and transmits this information to the control unit.

[0022] After receiving the drone target information extracted by the signal processing unit, the motion state judgment module starts to analyze the motion state of each drone. For example, for a drone, by calculating its speed change and acceleration at several consecutive time points, if the speed change is within the preset linear motion speed change threshold range and the acceleration is also close to zero, then the drone is judged to be in a linear motion state; conversely, if the speed change exceeds the threshold or the acceleration is obviously not zero, such as the drone suddenly turns or accelerates, it is judged to be in a nonlinear motion state. The filter switching module switches the flag bit of each drone target accordingly according to the result of the motion state judgment module. If a drone is judged to be in a linear motion state, the flag bit points to the Kalman filter module; if it is in a nonlinear motion state, the flag bit points to the particle filter module.

[0023] When the radar detects multiple drone targets for the first time, a set of particles (for particle filtering) and a Kalman filter state are initialized for each target. For example: for each target, 1,000 particles are randomly generated, and the initial positions of the particles are randomly distributed within a certain range according to the target positions detected by the radar for the first time, and the speeds are randomly set within a reasonable range of drone flight speeds. Each particle is assigned the same initial weight. At the same time, the initial state of the Kalman filter is set for each target, including the estimated values ​​of position and speed, which are also based on the radar's first detection data.

[0024] When the flag of a certain drone points to the Kalman filter module, the Kalman filter module uses the Kalman filter algorithm to predict the position of the drone target at the next moment based on the current drone target information, such as the current position and speed. For example, assuming that the drone moves in a straight line at a uniform speed, the Kalman filter module calculates the position prediction value at the next moment based on the state transfer matrix and the current state estimate. For example, the current position of the target and speed , assuming the target moves in a uniform straight line, the state transfer matrix , the state vector , through the formula (in, For the predicted state, is the state transition matrix, is the estimated state at the current moment) to predict its position at the next moment.

[0025] When the flag of a certain drone points to the particle filter module, the particle filter module uses the particle filter algorithm to predict the position according to the current drone target information. First, according to the current state and motion model of the particle, the state of each particle is transferred to predict the position of each particle at the next moment. Then, the possible position range of the drone target is estimated by the distribution of a large number of particles. For example, for the i-th particle, its position update formula is , ,in is the current location, is the current speed, is the time interval.

[0026] The data association module uses the nearest neighbor data association algorithm to match the latest drone target information extracted by the signal processing unit with the predicted position of the drone target. Specifically, the latest drone target information includes the latest drone position information. j , calculate the Euclidean distance between its predicted position and the latest position information of the surrounding drones. For example, for the target j Predicted location of and the latest drone location information ,distance ,when d When the minimum value is taken, the drone target j Associate with the drone closest to its predicted position to achieve the drone goal j Tracking.

[0027] For the drone whose flag points to the Kalman filter module, after the data association module completes the matching, the Kalman filter module uses the matched drone target information and the predicted position of the drone target to calculate the Kalman gain ,in, is the predicted state covariance, is the measurement matrix, is the measurement noise covariance. Update the target state estimate (where is the estimated state value at the next moment, It is the estimated state value at the next moment predicted based on the estimated state value at the current moment. is the latest drone target status, is the Kalman gain at the next moment), and updates the state covariance (in, is the identity matrix with the same dimension as the state covariance matrix), thereby obtaining a more accurate estimate of the target state.

[0028] For the drone whose flag points to the particle filter module, after the data association module completes the matching, the particle filter module uses the matched drone target information and the predicted position of the drone target to calculate the likelihood function of the drone target. Here, the likelihood function uses the Gaussian likelihood , the input parameters are and , It is a particle i The predicted state, in this embodiment, refers to the predicted position of the drone target, It is a particle i The observed state, in this embodiment, refers to the matched drone target information. The weight of each particle is updated according to the likelihood function value. The weight calculation formula is: (in, is the weight of particle i before updating, is the weight of particle i after update, is the likelihood function value) Particles with high weights indicate that their corresponding states are closer to the real state of the drone. Then resampling is performed to remove particles with low weights and retain particles with high weights to obtain a new set of particles to ensure that the particle distribution can better reflect the real state of the drone target.

[0029] The display unit displays the detection results and tracking tracks of drone targets in real time. Each detected drone is represented by a specific icon on the display interface, and its position on the map is updated in real time based on the location information provided by the control unit. At the same time, the tracking track of the drone is drawn on the map in the form of a line, which makes it easier for monitoring personnel to intuitively understand the flight path and movement trend of the drone.

[0030] It is not difficult to see from the above embodiments that the present invention combines Kalman filtering and particle filtering to comprehensively utilize their advantages. In the stage where the UAV's motion is relatively linear, the Kalman filter takes advantage of its efficient linear estimation and provides accurate state estimation; when the UAV exhibits nonlinear motion, the particle filter takes over to ensure accurate tracking of the target state. This complementary mechanism enables the algorithm to adapt to the diverse motion modes of the UAV and improve the overall state estimation accuracy; The nearest neighbor data association algorithm is simple and intuitive, and has a relatively low computational cost. It calculates the distance between the radar measurement data and the target predicted position, quickly assigns the measurement data to the nearest target, and can effectively match the radar echo data with the tracked target to avoid data association confusion, thereby accurately distinguishing different drone targets. And when combined with Kalman filtering and particle filtering, the accuracy of the nearest neighbor data association algorithm is further improved because these two filtering algorithms can provide more accurate target prediction positions; In addition, in actual environments, radar measurement data will be interfered by measurement noise and clutter. Kalman filtering can adjust state estimation according to the statistical characteristics of measurement noise (such as covariance) through its update mechanism, effectively suppressing the influence of measurement noise. Particle filtering also has unique advantages in dealing with noise. It can filter out noise interference to a certain extent by calculating the likelihood function and updating the weights of a large number of particles. This anti-interference ability enables the algorithm to stably track drone targets in complex electromagnetic environments or in the presence of meteorological clutter. The flight state of a drone swarm may change dynamically due to various factors, such as being affected by wind, changing flight missions, etc. The tracking algorithm that combines Kalman filtering, particle filtering and nearest neighbor data association algorithm can adapt to such dynamic changes and uncertainties. Whether it is an increase or decrease in the number of targets, a change in the target motion pattern, or the emergence of new interference factors, the algorithm can respond through its own adjustment mechanism (such as resampling of particle filtering, state update of Kalman filtering and rematching of data association algorithm) to maintain effective tracking of drone swarms.

[0031] In another embodiment, the signal processing unit further includes a target recognition module for analyzing the extracted target information and accurately identifying the type of the drone in combination with a feature database of the drone.

[0032] Specifically, the target recognition module acquires the echo signal and synthesizes the image of the UAV target using the inverse synthetic aperture radar imaging method, extracts the shape and size information of the UAV target based on the image of the UAV target, and then matches the shape and size information of the UAV target with the template features of various UAV types pre-stored in the feature database, thereby identifying the type of UAV.

[0033] The target recognition module uses the inverse synthetic aperture radar imaging method to synthesize the image of the drone using its own motion characteristics. As the distance and angle of different parts of the drone relative to the radar change during flight, Doppler shift is generated. A series of echo signals are processed by accurately measuring these Doppler shifts and the delay of the echo signal. For example, multiple sets of echo signal data are collected within a short time window. Range-Doppler processing is performed on these data, and the echo signals at different times are arranged and calculated according to the dimensions of distance and Doppler shift. Through complex algorithms, the two-dimensional image outline of the drone target is gradually constructed. After a series of signal processing and image reconstruction steps, a clearer drone target image is finally obtained, which can distinguish the general structure of the drone's fuselage, wings, tail, etc.

[0034] Based on the generated drone target image, the target recognition module uses image analysis technology to extract its shape and size information. The edge contour of the drone image is determined using edge detection algorithms, such as the Canny algorithm. Through contour analysis, the general shape of the drone can be identified. For example, it is found that the drone has a slender body and symmetrically distributed wings, and it is initially judged to be a common fixed-wing drone shape.

[0035] In terms of size measurement, the key dimensions of the drone are measured through the mapping relationship between pre-calibrated radar parameters and image pixels and actual physical dimensions. For example, the length of the drone fuselage is measured to occupy a certain number of pixels in the image. Combined with the known conversion ratio of pixels to actual length, the actual length of the fuselage is calculated to be about 1.2 meters; the width of the wings can also be measured in the image by a similar method, and is calculated to be about 2 meters.

[0036] The target recognition module matches the shape and size information of the extracted drone target with the template features of various drone types pre-stored in the feature database. The feature database contains template information of common civilian, commercial and potential security threat drones. Each drone type has its own specific shape and size range and other feature descriptions.

[0037] For example, the measured fuselage length of 1.2 meters and wing width of 2 meters, as well as the shape features of the fixed wing, are compared with the templates in the database. After one-by-one matching and similarity calculation, it is found that the features of the drone are closest to the template features of a common civilian quad-rotor fixed-wing hybrid drone. The fuselage length of this drone recorded in the feature database ranges from 1 to 1.5 meters, the wing width ranges from 1.8 to 2.2 meters, and the shape features also match the currently identified drone. Through this matching method, the type of the drone is finally identified as a civilian quad-rotor fixed-wing hybrid drone of this model.

[0038] It is not difficult to find from the above embodiments that the drone target image generated by the inverse synthetic aperture radar imaging method of the present invention can present the outline and structure of the target in detail. The shape and size information extracted based on this image is very accurate. For example, the bending angle of the drone wing, the curvature of the fuselage and other subtle shape features, as well as the dimensional accuracy of the millimeter level can be accurately measured. This makes it possible to accurately find the most suitable among many similar drone types when matching with the template features, greatly improving the accuracy of recognition. Compared with other methods that only rely on simple feature recognition, it can effectively distinguish drones with similar appearance but belonging to different models. Inverse synthetic aperture radar uses electromagnetic wave echo imaging, and its working principle is not affected by lighting conditions. Whether it is strong light during the day or dark environment at night, it can stably obtain echo signals and perform imaging. At the same time, under severe weather conditions, such as heavy rain, dense fog, etc., radar waves have stronger penetration ability than optical means. This means that the target recognition module can continue to operate under various complex meteorological conditions, ensuring uninterrupted monitoring and identification of drones, which is a scene that requires all-day security monitoring.

[0039] In another embodiment, the drone target information includes the speed and position of the drone target, and the method in which the motion state judgment module judges whether the motion state of each drone is a linear motion state or a nonlinear motion state by analyzing the drone target information extracted by the signal processing unit includes: Based on the speed of each UAV target at two adjacent moments, the speed change and acceleration of the UAV target are calculated; Based on the flight trajectory of each UAV target, the trajectory curvature of the UAV target is calculated. The flight trajectory of the UAV target is obtained by integrating the position of the UAV target in the time series; The speed change of the drone target is compared with the preset speed change threshold, the acceleration is compared with the preset acceleration threshold, and the trajectory curvature is compared with the preset trajectory curvature threshold. If any of the indicators exceeds the threshold range, it is determined that the motion state of the drone is a nonlinear motion state.

[0040] For example, for the motion of a drone in a two-dimensional plane, the velocity can be decomposed into velocity components in the x-axis and y-axis directions: v x and v y , by calculating and The change of speed on the x-axis and y-axis can be obtained, where k and k- 1 represents two adjacent moments. The acceleration can be obtained by and Calculated, is the time interval.

[0041] For a plane curve, the curvature s The formula can be To calculate, where and is the velocity component, and is the acceleration component.

[0042] Set the thresholds for velocity change, acceleration, and curvature. If or , or , ( , , , , is a pre-set threshold) if any of the above situations occurs, the UAV’s motion is considered to be nonlinear.

[0043] These thresholds can be adjusted according to the actual performance and application scenarios of the drone. For example, for small consumer drones, the speed change threshold may be relatively small because their flight speed and maneuverability are limited; while for high-performance military drones, the speed change threshold may be set larger. The determination of the curvature threshold also needs to consider the type of drone and the mission scenario. For example, a drone performing a reconnaissance mission in an urban environment may need to detect trajectory curvature more sensitively, so the curvature threshold can be set relatively low.

[0044] It is not difficult to find from the above embodiments that the present invention can directly reflect the speed change of the drone at adjacent moments by calculating the speed change and acceleration. For example, when the drone suddenly accelerates or decelerates, the speed change will increase significantly, and the acceleration will also deviate from the normal range. This makes the system very sensitive to the dynamic adjustment of the speed of the drone, and even small but critical speed changes can be captured. Calculating the trajectory curvature provides a new dimension for judging the motion state of the drone. During the flight of the drone, if its flight trajectory is curved, such as turning, circling, etc., the trajectory curvature will change accordingly. Compared with relying only on speed information, the trajectory curvature can more intuitively reflect the curvature of the drone's path, thereby more comprehensively judging whether its motion is linear. For a multi-target tracking system, it is crucial to accurately judge the motion state of the drone. When the system recognizes that the drone is in a nonlinear motion state, it can automatically switch to a tracking algorithm that is more suitable for nonlinear motion, such as a particle filter algorithm, thereby improving the accuracy and stability of tracking. The system can automatically judge the motion state of the drone according to a preset threshold without manual intervention. This autonomous decision-making ability enables the system to quickly respond to the motion changes of the drone and take timely measures in emergency situations.

[0045] In another embodiment, the signal processing unit receives and processes the echo signal returned by the radar antenna array, and the method for extracting the drone target information includes: The time difference between the transmitted signal and the echo signal is measured, and the distance between the drone target and the radar is calculated by combining the propagation speed of electromagnetic waves. Assume that the time of transmitting the signal is t 1 The signal is received at time t 2 , then the time difference Δt = t 2 -t 1 , it is known that the propagation speed of electromagnetic waves in air is approximately c=3×10 8 m / s, then the distance between the drone target and the radar is d=(c×Δt) / 2; Using the fast Fourier transform algorithm, the echo signal in the time domain is converted to the frequency domain, the Doppler frequency shift information is extracted, and then the radial velocity of the drone target is calculated. Assume that the frequency of the transmitted signal is f 0 , the received echo signal frequency is f 1 , Doppler frequency shift Δf = f 1 -f 2 , it is known that the wavelength of the radar transmission signal is λ, according to the relationship between Doppler frequency shift and radial velocity, v=(λ×Δf) / 2; By measuring the phase difference or amplitude difference of the echo signals received by different antennas, the azimuth and pitch angle of the UAV target are determined using the direction finding algorithm. The lateral algorithm can use a multiple signal classification algorithm.

[0046] It is not difficult to find from the above embodiments that the present invention calculates the distance by measuring the time difference between the transmission and the echo signal, with extremely high accuracy. The azimuth and pitch angles are determined by using the phase difference or amplitude difference of the signals received by different antennas, which can accurately determine the direction of the drone. The fast Fourier transform algorithm converts the time domain echo signal into the frequency domain, and quickly and accurately extracts the Doppler frequency shift information to calculate the radial velocity. This process is completed in a very short time, and the speed data of the drone can be updated in real time.

[0047] In another embodiment, the signal processing unit first performs preprocessing when acquiring the echo signal, including: using a data filtering method to remove noise, then using an amplifier to amplify the echo signal, using a gain control circuit to dynamically adjust the gain of the echo signal, and then converting the echo signal into a discrete digital signal based on a preset sampling frequency and quantization accuracy.

[0048] In actual environments, the echo signals received by radar are often mixed with various noises, such as thermal noise from the electronic equipment itself, electromagnetic interference from the external environment, etc. The data filtering method is to use specific algorithms to suppress or eliminate these noise components and improve the quality of the signal. For example, the commonly used Kalman filter is based on the linear system state equation. Through the two steps of prediction and update, it continuously optimizes the estimation based on the prior knowledge of the signal and the current measurement value, and effectively separates the noise in the target echo signal; there is also the Wiener filter, which uses the minimum mean square error as the criterion and determines the filter coefficient by calculating the correlation function between the input signal and the expected output signal, thereby filtering out the noise part that is different from the target signal characteristics.

[0049] After removing the noise, the proportion of useful information in the echo signal increases relatively, making the subsequent extraction of target information, such as the analysis of distance, speed, angle and other features, more accurate and reliable. For example, in the processing of long-distance and weak echo signals, if denoising is not performed, the existence of the target and related parameters may not be accurately judged due to noise masking. After denoising, the target echo can be clearly distinguished, laying the foundation for subsequent processing. It provides purer input data for a series of algorithms that rely on signal features, such as target recognition and tracking, and improves the execution effect and accuracy of these algorithms.

[0050] The echo signals received by the radar are usually weak, and the role of the amplifier is to amplify the strength of these weak signals. It uses active devices such as transistors in electronic circuits to increase the voltage or power of the input echo signal according to a certain magnification factor. The amplified signal is easier to be operated and analyzed by subsequent circuits and processing units. Many signal processing algorithms and devices have certain requirements for the strength of the input signal. Weak signals may not be able to drive the subsequent circuits to work normally or cannot reach the effective range of algorithm processing. After amplification, these requirements can be met to ensure the smooth progress of the entire signal processing process. For drone targets at long distances or with low reflection cross-sectional areas, the echo signal itself has a very low strength. Amplification can make these originally imperceptible signals obvious, which helps to improve the radar's detection capabilities for such targets and expand the radar's effective detection range, so that potential drone threats can be discovered earlier.

[0051] The gain control circuit monitors the strength of the echo signal in real time and automatically adjusts the gain of the amplifier according to preset rules or algorithms. For example, when a strong echo signal is detected, the gain is reduced; when the echo signal is weak, the gain is increased. It can compare the actual received signal strength with the reference level based on the feedback mechanism through circuit elements such as comparators, and then output a control signal to adjust the gain parameters of the amplifier. Without dynamic gain adjustment, when a strong echo signal is received, the amplifier may cause the output signal to exceed its linear operating range, resulting in signal saturation, distortion, and loss of some information. Dynamic gain control can ensure that the signal is always in the appropriate intensity range, and the target echo signal, regardless of distance, strength or weakness, can enter the subsequent processing link in the best state to ensure the integrity and accuracy of the signal. In complex actual scenarios, the distance, size, material, etc. of the drone target are different, and the reflected echo signal strength varies greatly. At the same time, environmental factors such as weather changes will also affect the signal strength. Dynamic gain control can automatically adapt to these changing conditions, so that the radar system can stably and efficiently process the echo signal under various conditions, enhancing the versatility and robustness of the system.

[0052] In the process of converting the echo signal into a discrete digital signal based on the preset sampling frequency and quantization accuracy, sampling is to take the value of the continuous analog echo signal at a certain time interval (determined by the preset sampling frequency) to discretize it in time. For example, setting the sampling frequency to 1000Hz means collecting 1000 sample points of the analog signal per second. Quantization is to divide the amplitude of the sampled analog signal into a finite number of discrete levels according to a certain quantization accuracy (usually expressed in bits, such as 8 bits, 16 bits, etc.), and represent the amplitude of each sampling point with a corresponding digital code, so as to realize the discretization of the signal in amplitude and finally complete the conversion from analog signal to digital signal. Most of the modern signal processing is based on digital technology. After converting the echo signal into a digital signal, various digital signal processing algorithms, such as digital filtering, fast Fourier transform, etc., can be easily used for subsequent analysis and calculation. Moreover, digital signals are easy to store in the computer's storage device, which is convenient for subsequent data call, viewing and long-term historical data analysis. Digital signal processing can use high-speed digital processors, FPGA and other hardware devices to implement complex and efficient algorithms through programming, which can achieve higher processing accuracy and speed than analog signal processing. At the same time, signal processing parameters can be adjusted more flexibly in the digital domain, and the signal processing process can be precisely controlled, further improving the accuracy and timeliness of UAV target information extraction.

[0053] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A phased array radar for drone detection based on multi-target tracking, comprising: A radar antenna array, a signal processing unit, a control unit and a display unit. The radar antenna array electronically controls the phase and amplitude of the channel feed signal of each antenna to achieve rapid changes in the antenna beam pointing. The signal processing unit is responsible for receiving and processing the echo signal returned by the radar antenna array and extracting the UAV target information. The control unit adjusts the beam pointing of the radar antenna array according to the UAV target information to achieve multi-target tracking. The display unit is used to display the detection results and tracking trajectories of the UAV targets. It is characterized in that the control unit includes: A motion state judgment module, which is used to judge whether the motion state of each drone is a linear motion state or a nonlinear motion state by analyzing the target information of each drone extracted by the signal processing unit; A filter switching module sets a flag for each drone target to indicate the currently used filter module. When the motion state judgment module determines that the drone is in a linear motion state, the flag points to the Kalman filter module. When the motion state judgment module determines that the drone is in a nonlinear motion state, the flag points to the particle filter module. Kalman filter module, which is used to predict the position of the drone target at the next moment using the Kalman filter algorithm according to the current drone target information of the drone when the drone's flag position points to this module; The particle filter module is used to predict the position of the drone target at the next moment using the particle filter algorithm according to the current drone target information of the drone when the drone's flag bit points to this module; A data association module, which is used to match the latest information of each drone target extracted by the signal processing unit with the predicted position of the drone target using a nearest neighbor data association algorithm; The Kalman filter module is further used to calculate the Kalman gain using the matched UAV target information and the predicted position of the UAV target after the data association module completes the matching, and update the target state estimation and target state covariance of the UAV target; The particle filter module is also used to calculate the likelihood function of the drone target using the matched drone target information and the predicted position of the drone target after the data association module completes the matching. The likelihood function adopts Gaussian likelihood, updates the weight of the drone target according to the likelihood function value, and then resamples to remove drone targets with low weights and retain drone targets with high weights.

2. The phased array radar for drone detection based on multi-target tracking as claimed in claim 1, characterized in that: The signal processing unit also includes a target recognition module, which is used to analyze the extracted target information and accurately identify the type of the drone in combination with the feature database of the drone.

3. The phased array radar for drone detection based on multi-target tracking as claimed in claim 1, characterized in that: The drone target information includes the speed and position of the drone target. The method in which the motion state judgment module judges whether the motion state of each drone is a linear motion state or a nonlinear motion state by analyzing the drone target information extracted by the signal processing unit includes: Based on the speed of each UAV target at two adjacent moments, the speed change and acceleration of the UAV target are calculated; Based on the flight trajectory of each UAV target, the trajectory curvature of the UAV target is calculated. The flight trajectory of the UAV target is obtained by integrating the position of the UAV target in the time series; The speed change of the drone target is compared with the preset speed change threshold, the acceleration is compared with the preset acceleration threshold, and the trajectory curvature is compared with the preset trajectory curvature threshold. If any of the indicators exceeds the threshold range, it is determined that the motion state of the drone is a nonlinear motion state.

4. The phased array radar for drone detection based on multi-target tracking as claimed in claim 1, characterized in that: The signal processing unit receives and processes the echo signal returned by the radar antenna array, and the method for extracting the target information of the UAV includes: The time difference between the transmitted signal and the echo signal is measured, and the distance between the drone target and the radar is calculated based on the propagation speed of the electromagnetic wave. The fast Fourier transform algorithm is used to convert the echo signal in the time domain into the frequency domain, extract the Doppler frequency shift information, and then calculate the radial velocity of the UAV target; By measuring the phase difference or amplitude difference of the echo signals received by different antennas, the azimuth and pitch angle of the UAV target are determined using the direction finding algorithm.

5. The phased array radar for drone detection based on multi-target tracking as claimed in claim 2, characterized in that: The target recognition module acquires the echo signal and synthesizes the image of the UAV target using the inverse synthetic aperture radar imaging method, extracts the shape and size information of the UAV target based on the image of the UAV target, and then matches the shape and size information of the UAV target with the template features of various UAV types pre-stored in the feature database, thereby identifying the type of UAV.

6. The phased array radar for drone detection based on multi-target tracking as claimed in claim 1, characterized in that: When the signal processing unit obtains the echo signal, it first performs preprocessing, including: using a data filtering method to remove noise, then using an amplifier to amplify the echo signal, using a gain control circuit to dynamically adjust the gain of the echo signal, and then converting the continuous analog echo signal into a discrete digital echo signal based on a preset sampling frequency and quantization accuracy.

Images