An unmanned aerial vehicle target recognition method based on track features

Abstract

The application provides a UAV target recognition method based on track characteristics, uses radar track data, calculates track straightness error characteristic values, and judges whether the track type is a UAV target through a threshold judgment method. Step one, a radar is erected, started and operated in a full airspace search mode, and historical track data of the radar at T time is extracted; step two, an ideal straight line is fitted according to linear least square fitting of each dimension information in the track data, deviation values of each dimension information of the target relative to the ideal straight line are calculated, and track straightness error characteristic values are obtained; step three, a large number of historical track data of UAV targets and other targets are used for training, and a threshold judgment matrix is constructed; step four, the track straightness error characteristic values of N tracks are compared with the threshold judgment matrix respectively, when the characteristic values are all less than the judgment matrix, the UAV target is judged, and a UAV target recognition result is output; when the characteristic values are greater than or equal to the judgment matrix, other targets are judged.

Description

Technical Field

[0001] This invention belongs to the field of radar target tracking and identification, specifically relating to a method for UAV target identification based on flight path features. Background Technology

[0002] With the rapid development of drone technology and the expansion of the market, drones are being used more and more widely in various fields. Radar can provide high-precision position and velocity information and can conduct continuous detection for extended periods under different time periods and environmental conditions. Therefore, radar identification technology has a significant advantage in countering drone targets.

[0003] Research on radar identification technology based on UAV targets typically employs complex techniques such as signal processing, pattern recognition, and machine learning. These methods have high computational demands in multi-target environments, poor fault tolerance for abnormal data, and high requirements for both radar system hardware and software.

[0004] The existing patent "202010040144[P][2023-06-16] A method for classifying and identifying external radiation source radar targets based on track feature extraction" extracts key data features from radar track data and constructs a recurrent neural network model to achieve the classification and identification of aerial targets. However, this method requires a large amount of data for training, and the model has poor interpretability.

[0005] Low-speed, slow-moving, and small-sized detection radars typically output target trajectory information such as range, azimuth, elevation, and speed. Analysis of the trajectories of aerial drones and other targets reveals significant differences in their trajectory characteristics. When designing drones and other aircraft targets, crucial parameters such as lateral deviation, directional deviation, and flight speed must be considered to ensure successful mission completion; therefore, their flight trajectories possess a certain degree of determinism and predictability. Summary of the Invention

[0006] This invention proposes a UAV target recognition method based on track features. It uses radar track data to calculate the straightness error feature value of the track and uses a threshold decision method to determine whether the track type is a UAV target.

[0007] The present invention is achieved through the following technical solution.

[0008] A method for UAV target recognition based on flight path features includes the following steps:

[0009] Step 1: Set up and power on the radar and run it in full-airspace search mode to extract the radar's historical flight track data at time T;

[0010] Step 2: Based on the linear least squares fitting line of each dimension of the track data as the ideal line, calculate the deviation value of each dimension of the target information relative to the ideal line, thereby obtaining the track straightness error characteristic value;

[0011] Step 3: Use a large amount of historical flight path data of UAV targets and other targets to train and construct a threshold decision matrix;

[0012] Step 4: Compare the linearity error feature values ​​of the N tracks with the threshold decision matrix. When the feature values ​​are all less than the decision matrix, the target is determined to be a UAV target, and the UAV target recognition result is output. When the feature values ​​are greater than or equal to the decision matrix, the target is determined to be another target.

[0013] Beneficial effects of this invention:

[0014] 1. This invention only requires processing radar track data, extracting track features, and using a threshold decision method to achieve preliminary identification of multiple batches of targets in the full airspace search mode.

[0015] 2. The model of this invention is simple, has low computational cost, does not require data training, and does not require modification of radar hardware and software systems, thus having strong universality.

[0016] 3. This invention is mainly used to distinguish the flight paths of UAV targets from those of other targets. Attached Figure Description

[0017] Figure 1 This is a flowchart of the UAV target recognition method based on flight path features according to the present invention;

[0018] Figure 2 This is a diagram showing the real-time target recognition results of the UAV on the radar display and control interface. Detailed Implementation

[0019] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the drawings are merely exemplary and are intended to illustrate the principles and spirit of the present invention, and are not intended to limit the scope of the present invention.

[0020] like Figure 1 As shown, the present invention provides a UAV target recognition method based on flight path features, which specifically includes the following steps:

[0021] Step 1: Set up and power on the radar, and run it in full-airspace search mode to extract the radar's historical flight track data at time T; details are as follows:

[0022] (1) Set the radar scan data rate to Rdata, unit: Hz;

[0023] (2) At time T, extract N batches of UAV trajectory data and other target trajectory data. The length of each batch of trajectory data is L, and the corresponding time range is (T-L+1 T-L+2 T) / Rdata, in seconds;

[0024] (3) Extract the distance R, azimuth α, pitch β and velocity V information sequences from N batches of track data;

[0025] In this embodiment, the information sequence of each dimension of the i-th track is defined as follows:

[0026] The distance sequence of the i-th track is defined as:

[0027] R i =[R i1 R i2 ... R iL (1)

[0028] The azimuth sequence of the i-th track is defined as:

[0029] α i =[α i1 α i2 ... α iL (2)

[0030] The pitch angle sequence of the i-th track is defined as:

[0031] β i =[β i1 β i2 ... β iL (3)

[0032] The velocity sequence of the i-th track is defined as:

[0033] V i =[V i1 V i2 ... V iL (4)

[0034] Step 2: Based on the linear least squares fitted line of each dimension of the track data as the ideal line, calculate the deviation of each dimension of the target information from the ideal line, thereby obtaining the track straightness error characteristic value; specifically as follows:

[0035] (1) Define the length L of each batch of track data as the independent variable x;

[0036] x = [1 2 ... L] (5)

[0037] (2) The distance sequence R of the i-th track i Define it as the dependent variable, and set the independent variable x and the dependent variable R.i The functional relationship between them is:

[0038] R i =k Ri x+b Ri (6)

[0039] In the formula, k Ri Let b represent the slope of the linear least squares fit of the distance sequence for the i-th track. Ri The intercept of the linear least squares fit of the distance sequence of the i-th track is represented;

[0040] (3) Define the optimization function Q;

[0041]

[0042] (4) Using the minimum of the optimization function Q as a constraint, calculate k Ri and b Ri The best estimate;

[0043] (5) Based on the calculated parameter k Ri b Ri Substituting the best estimate into formula (5), we obtain the linear least squares fitting data R′ of the distance sequence of the i-th track. i Defined as

[0044] R′ i =k Ri x+b Ri (8)

[0045] (6) Calculate the deviation matrix h of the distance sequence of the i-th track. Ri Defined as

[0046] h Ri =R′ i -R i (9)

[0047] (7) Calculate the standard deviation σ of the deviation matrix of the distance sequence of the i-th track. Ri Defined as

[0048]

[0049] in,

[0050] (8) The azimuth sequence α of the i-th track i As the dependent variable input, repeat steps (2) to (7) to obtain the standard deviation σ of the deviation matrix of the azimuth sequence of the i-th track. ai ;

[0051] (9) The pitch angle sequence β of the i-th tracki As the dependent variable input, repeat steps (2) to (7) to obtain the standard deviation σ of the deviation matrix of the pitch angle sequence of the i-th track. βi ;

[0052] (10) The velocity sequence V of the i-th track i As the dependent variable input, repeat steps (2) to (7) to obtain the standard deviation σ of the deviation matrix of the velocity sequence of the i-th track. Vi ;

[0053] (11) Take the distance, azimuth, pitch and velocity sequences of the other N-1 tracks as dependent variables and repeat steps (2) to (10) to obtain the eigenvalue matrix E of all tracks;

[0054]

[0055] Step 3: Train the system using historical flight path data from a large number of UAV targets and other targets to construct a threshold decision matrix Thr; the specific formula is as follows:

[0056]

[0057] Step 4: Compare the linearity error feature values ​​of the N tracks with the threshold decision matrix. When the feature values ​​are all less than the decision matrix, the target is determined to be a UAV target, and the UAV target recognition result is output. When the feature values ​​are greater than or equal to the decision matrix, the target is determined to be another target.

[0058] To demonstrate the effectiveness of this invention, the following experiments were conducted.

[0059] This embodiment utilizes radar-measured UAV trajectory data and trajectory data of other targets to extract target trajectory feature values ​​in real time, and uses a pre-trained threshold decision matrix to determine the target type. The real-time UAV target recognition results on the radar display and control interface are as follows: Figure 2 As shown, the drone icon indicates the drone track, while unmarked tracks are other tracks.

[0060] The results of drone target identification are shown in Table 1, based on statistical analysis of drone data and other target data over a period of one month.

[0061] Table 1. Target Identification Results of Unmanned Aerial Vehicles

[0062] Probability of correct identification 83％ Identify the probability of missed alarms 17％ Identifying false alarm probability 10％

[0063] This experiment verified the effectiveness of UAV target track recognition using the method proposed in this invention. The experimental results show that the UAV target recognition method based on track features provided by this invention can effectively identify UAV targets and can serve as an initial screening process for further precise target identification.

[0064] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for UAV target recognition based on flight path features, characterized in that, Includes the following steps: Step 1: Install and power on the radar, and run it in full-airspace search mode to extract... Real-time radar historical track data; Step 2: Based on the linear least squares fitting line of each dimension of the track data as the ideal line, calculate the deviation value of each dimension of the target information relative to the ideal line, thereby obtaining the track straightness error characteristic value; Step 3: Use a large amount of historical flight path data of UAV targets and other targets to train and construct a threshold decision matrix; Step 4: Separately The straightness error feature value of the flight path is compared with the threshold decision matrix. When the feature value is less than the decision matrix, it is determined to be a UAV target and the UAV target recognition result is output; when it is greater than or equal to the decision matrix, it is determined to be another target.

2. The UAV target recognition method based on flight path features as described in claim 1, characterized in that, Step one is as follows: (1) Set the radar scan data rate Unit: Hz; (2) In At any given time, extract drone trajectory data and other target trajectory data. There are 1 batch, and the length of each batch of track data is 1. Corresponding time range: Unit: seconds; (3) Extraction Distance in each batch of track data Azimuth Pitch angle ,speed Information sequence.

3. A method for UAV target recognition based on flight path features as described in claim 1 or 2, characterized in that, Step two is as follows: (1) The length of each batch of track data Defined as independent variable ; (5) (2) The first Distance sequence of the track Define it as the dependent variable and set the independent variable. With dependent variable The functional relationship between them is: (6) In the formula, Indicates the first The slope of the linear least squares fit of the distance sequence of the track. Indicates the first The intercept of the linear least squares fit of the distance sequence of the track; (3) Define the optimization function ; (7) (4) The optimization function Minimum as a constraint, calculate and The best estimate; (5) Based on the calculated parameters , Substituting the best estimate into formula (5), we obtain the first... Linear least squares fitting data of the distance sequence of the track Defined as (8) (6) Calculate the first Deviation matrix of the distance sequence of the track Defined as (9) (7) Calculate the first Standard deviation of the deviation matrix of the distance sequence of the track Defined as (10) in, ; (8) The first Azimuth sequence of the flight path As the input of the dependent variable, repeat steps (2) to (7) to obtain the first... Standard deviation of the deviation matrix of the azimuth sequence of the track ; (9) The first Pitch angle sequence of the flight path As the input of the dependent variable, repeat steps (2) to (7) to obtain the first... Standard deviation of the deviation matrix of the pitch angle sequence of the flight path ; (10) The first velocity sequence of the track As the input of the dependent variable, repeat steps (2) to (7) to obtain the first... Standard deviation of the deviation matrix of the velocity sequence of the track ; (11) Each of the other The distance, azimuth, pitch, and velocity sequences of each track are used as dependent variables. Steps (2) to (10) are repeated to obtain the eigenvalue matrix of all tracks. ; (11)。 4. The UAV target recognition method based on flight path features as described in claim 3, characterized in that, The threshold decision matrix The specific formula is as follows: (12)。 5. A method for UAV target recognition based on flight path features as described in claim 3 or 4, characterized in that, No. The information sequence of each dimension of the track is defined as follows: No. The distance sequence of a track is defined as: (1) No. The azimuth sequence of a flight path is defined as: (2) No. The pitch angle sequence of a flight path is defined as: (3) No. The velocity sequence of a track is defined as: (4)。

Images