Separation event discrimination method based on HRRP extended target tracking

Abstract

This invention discloses a separation event discrimination method based on HRRP extended target tracking, comprising the following steps: S1: acquiring HRRP data and preprocessing the HRRP data to eliminate translation sensitivity; S2: extracting feature points from the HRRP data; S3: estimating the target state using an extended target tracking method based on the feature points; S4: establishing a separation event discrimination criterion, and determining the separation event time based on the separation event discrimination criterion and the target state. This invention can determine the time of target separation events in real time using radar observation information, solving the problems of numerous echo signal processing steps and difficulties in real-time application; this invention does not require the original radar echo, only using HRRP data, reducing the real-time transmission bandwidth requirements of the radar, and can process and discriminate target separation events in real time. It can be applied to existing broadband radars, and the discrimination result can accurately reflect the occurrence time of the separation event.

Description

Technical Field

[0001] This invention relates to the field of radar broadband imaging and processing technology, and in particular to a separation event discrimination method based on HRRP extended target tracking. Background Technology

[0002] During space launches, rockets experience separation events such as stage separation and fairing jettison. Traditional separation event detection primarily relies on calculating telemetry data transmitted from the onboard computer to obtain the relevant separation timing. However, when noise interference or telemetry signals fail to transmit, the separation event cannot be detected. Ground-based measurement equipment, particularly optical devices, has limited observation distance and is greatly affected by weather conditions, making it difficult to play a role in separation event detection.

[0003] The problem of using radar measurement information to determine target separation has been addressed in existing research, which focuses on extracting information about multiple targets from echo signals. This involves separating the measurement information of each target based on the micro-Doppler frequency caused by different micro-motion forms such as vibration, self-selection, or precession. This method has the following drawbacks: (1) The amount of echo signal data used is large, making it difficult to transmit in real time. It can only be processed after pulse compression and envelope alignment. (2) The narrowband echo signal used has poor range resolution (range resolution is inversely proportional to bandwidth), resulting in less target information contained in the echo. (3) The purpose of this method is to separate and extract information about multiple targets from the echo signal after a separation event occurs, and it cannot determine the time of the separation event. Summary of the Invention

[0004] To address the aforementioned problems, this invention aims to provide a separation event discrimination method based on HRRP extended target tracking, which can use HRRP (High Resolution One-Dimensional Range Profile) to determine the timing of target separation events.

[0005] The technical solution of the present invention is as follows:

[0006] A method for determining separation events based on HRRP extended target tracking includes the following steps:

[0007] S1: Acquire HRRP data and preprocess the HRRP data to eliminate translation sensitivity;

[0008] S2: Extract feature points from the HRRP data;

[0009] S3: Estimate the target state using the extended target tracking method based on the feature points;

[0010] S4: Establish separation event discrimination criteria, and determine the separation event time based on the separation event discrimination criteria and the target state.

[0011] Preferably, in step S1, a correlation alignment method is used for preprocessing to eliminate translation sensitivity.

[0012] Preferably, when using the correlation alignment method for preprocessing, a baseline HRRP sample is first selected, then the maximum sliding correlation coefficient between the current HRRP sample and the baseline sample is calculated, and finally the sliding HRRP sample corresponding to the current HRRP sample is determined based on the maximum sliding correlation coefficient, thereby achieving alignment between the current HRRP sample and the baseline HRRP sample.

[0013] Preferably, the baseline HRRP sample is the average of all aligned samples preceding the current time step, and the maximum moving correlation coefficient is calculated using the following formula:

[0014]

[0015] x mean =mean(x1(τ1),...,x i-1 (τ i-1 i = 2, 3, ...

[0016] In the formula: d i x represents the maximum moving correlation coefficient between the i-th HRRP sample and the baseline sample; mean x represents the average of the 1 to i-1 aligned samples, where the translation distance of the first sample is 0; i (τ) represents x that has been translated by τ distance units. i sample.

[0017] Preferably, when extracting feature points from the HRRP data, the extreme points of scattering intensity that are not less than 60% of the highest intensity are extracted.

[0018] Preferably, step S3, which uses the extended target tracking method to estimate the target state, specifically includes the following sub-steps:

[0019] S31: Establish an elliptic parameter model to describe the state of the extended target; the state r of the center position parameter of the elliptic parameter model is represented as:

[0020]

[0021] Where: x, y, Let represent the horizontal and vertical positions and velocities of the ellipse center, respectively; T represents the transpose of the column vector; the ellipse shape parameter state p of the ellipse parameter model is represented as:

[0022] p = [α, l1, l2] T (3)

[0023] In the formula: α represents the rotation angle of the ellipse; l1 represents the major semi-axis; l2 represents the minor semi-axis;

[0024] In the elliptical parameter model, the offset of a certain scattering point z from the center is represented by h1 and h2, where h1 represents the offset in the direction of the major semi-axis and h2 represents the offset in the direction of the minor semi-axis, and the magnitudes of both offsets are between [-1, 1].

[0025] S32: Establish a measurement model; the measured value of the scattering point z is expressed as:

[0026]

[0027] Considering the influence of measurement noise, the measurement equation is expressed as:

[0028] y = Hr + Sh + v (5)

[0029]

[0030] In the formula: y represents the measured value; H represents the measurement observation matrix; v represents the measurement noise;

[0031] S33: Establish a tracking filter model, and estimate the target state based on the tracking filter model; specifically:

[0032] The state parameters of all measurement points at time k are sequentially updated; the position, shape parameters, and covariance matrix of point i at time k are denoted as follows: and The state of point i before the measurement value at time k is denoted as: and The predicted value at time k is denoted as

[0033] The position parameters of point i are updated using the following formula:

[0034]

[0035] In the formula: This represents the estimated value of the position parameter measurement at point i at time k; This represents the cross-covariance matrix between the location parameters and the measurements; Represents the measurement error matrix; These are measured values;

[0036] The shape parameters of point i are updated using the following formula:

[0037]

[0038]

[0039] In the formula: The cross-covariance matrix representing the virtual measurement and shape parameters; Represents the covariance matrix of the virtual measurement; Indicates virtual measurement; Indicates the estimated value of a virtual measurement; Represents the Kronecher multiplier;

[0040] The position parameters and covariance matrix at the next moment are predicted using the following formula:

[0041]

[0042] In the formula: This represents the position parameter state transition matrix of a linear system. The location parameter covariance matrix represents zero-mean Gaussian noise;

[0043] The shape parameters and state covariance matrix are predicted using the following formula:

[0044]

[0045] In the formula: The shape parameter state transition matrix of a linear system; n k The number of measurement points at time k; The shape parameter covariance matrix represents zero-mean Gaussian noise.

[0046] Preferably, in step S4, the separation event discrimination criterion is:

[0047]

[0048] In the formula: Δ represents the ellipse increase rate; D represents the separation event discrimination threshold; L1 represents the major semi-axis of the ellipse at the current moment; This represents the semi-major axis of the ellipse at the previous moment.

[0049] The beneficial effects of this invention are:

[0050] This invention eliminates the need for raw radar echoes and utilizes only HRRP data to process and identify target separation events in real time, reducing the real-time transmission bandwidth requirements of the radar. It can be applied to existing broadband radars, and the identification results can accurately reflect the moment of the separation event. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A schematic diagram of the center position parameters and ellipse parameters of the elliptic parametric model;

[0053] Figure 2 This is a schematic diagram for separating events;

[0054] Figure 3 This is a schematic diagram of the event discrimination result for Task 1 in a specific implementation example;

[0055] Figure 4 This is a schematic diagram of the event discrimination result for Task 2 in a specific implementation example. Detailed Implementation

[0056] The present invention will be further described below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and technical features described in this application can be combined with each other. It should also be pointed out that, unless otherwise indicated, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terms "comprising" or "including" and similar words used in this invention refer to elements or objects preceding the word that encompass the elements or objects listed following the word and their equivalents, without excluding other elements or objects.

[0057] This invention provides a method for determining separation events based on HRRP extended target tracking, comprising the following steps:

[0058] S1: Acquire HRRP data and preprocess the HRRP data to eliminate translation sensitivity.

[0059] High-resolution one-dimensional range profile (HRRP) is the vector projection of the sub-echoes of target scattering points onto the radar line of sight (LOS) under conditions of large transmit bandwidth and target size much larger than the radar range resolution cell. It has advantages such as ease of acquisition, immunity to weather conditions, and rich content of target structural information (target size, scattering point distribution, etc.), providing the possibility for real-time identification of target separation events.

[0060] In actual radar measurements, a range profile is typically constructed by extracting a data vector containing the target within a single gate window from the target echo. For maneuvering targets, the relative positions of the target range profile within the gate window cannot remain consistent, leading to translational sensitivity in HRRP. Therefore, preprocessing is necessary to eliminate this translational sensitivity.

[0061] In one specific embodiment, a correlation alignment method is used for preprocessing to eliminate translation sensitivity. First, a baseline HRRP sample is selected, which is the average of all aligned samples at the current time. Then, the maximum sliding correlation coefficient between the current HRRP sample and the baseline sample is calculated. Finally, the slidable HRRP sample corresponding to the current HRRP sample is determined based on the maximum sliding correlation coefficient, thus achieving alignment between the current HRRP sample and the baseline HRRP sample. The maximum sliding correlation coefficient is calculated using the following formula:

[0062]

[0063] x mean =mean(x1(τ1),...,x i-1 (τ i-1 i = 2, 3, ...

[0064] In the formula: d i x represents the maximum moving correlation coefficient between the i-th HRRP sample and the baseline sample; mean x represents the average of the 1 to i-1 aligned samples, where the translation distance of the first sample is 0; i (τ) represents x that has been translated by τ distance units. i sample.

[0065] S2: Extract feature points from the HRRP data.

[0066] HRRP reflects the distribution and intensity variation of the scattering coefficient along the radar line-of-sight, containing information about the target's scattering characteristics. To more accurately represent the characteristics of a single frame of HRRP and suppress background noise and clutter interference, this invention extracts feature points from each frame of HRRP data. In a specific embodiment, when extracting feature points from the HRRP data, extreme scattering points with scattering intensity not less than 60% of the highest intensity are extracted.

[0067] S3: Based on the feature points, estimate the target state using the extended target tracking method.

[0068] In a specific embodiment, estimating the target state using the extended target tracking method specifically includes the following sub-steps:

[0069] S31: Establish an elliptic parameter model to describe the state of the extended target;

[0070] like Figure 1 As shown, the center position parameter r of the elliptic parametric model is represented as:

[0071]

[0072] Where: x, y, These represent the horizontal and vertical positions and velocities of the ellipse center, respectively; T represents the transpose of the column vector.

[0073] The elliptical shape parameter state p of the elliptical parameter model is represented as follows:

[0074] p = [α, l1, l2] T (3)

[0075] In the formula: α represents the rotation angle of the ellipse; l1 represents the major semi-axis; l2 represents the minor semi-axis;

[0076] In the elliptic parameter model, the offset of a certain scattering point z from the center is represented by h1 and h2, where h1 represents the offset in the direction of the major semi-axis and h2 represents the offset in the direction of the minor semi-axis, and the magnitudes of both offsets are between [-1, 1].

[0077] S32: Establish a measurement model; the measured value of the scattering point z is expressed as:

[0078]

[0079] Considering the influence of measurement noise, the measurement equation is expressed as:

[0080] y = Hr + Sh + v (5)

[0081]

[0082] In the formula: y represents the measured value; H represents the measurement observation matrix; v represents the measurement noise;

[0083] S33: Establish a tracking filter model, and estimate the target state based on the tracking filter model; specifically:

[0084] The state parameters of all measurement points at time k are sequentially updated; the position, shape parameters, and covariance matrix of point i at time k are denoted as follows: and The state of point i before the measurement value at time k is denoted as: and The predicted value at time k is denoted as

[0085] The position parameters of point i are updated using the following formula:

[0086]

[0087] In the formula: This represents the estimated value of the position parameter measurement at point i at time k; This represents the cross-covariance matrix between the location parameters and the measurements; Represents the measurement error matrix; These are measured values;

[0088] The shape parameters of point i are updated using the following formula:

[0089]

[0090]

[0091] In the formula: The cross-covariance matrix representing the virtual measurement and shape parameters; Represents the covariance matrix of the virtual measurement; Indicates virtual measurement; Indicates the estimated value of a virtual measurement; Represents the Kronecher multiplier;

[0092] The position parameters and covariance matrix at the next moment are predicted using the following formula:

[0093]

[0094] In the formula: This represents the position parameter state transition matrix of a linear system. The location parameter covariance matrix represents zero-mean Gaussian noise;

[0095] The shape parameters and state covariance matrix are predicted using the following formula:

[0096]

[0097] In the formula: The shape parameter state transition matrix of a linear system; n k The number of measurement points at time k; The shape parameter covariance matrix represents zero-mean Gaussian noise.

[0098] Traditional point target tracking assumes that the tracked target generates at most one measurement at any given time, and then uses a target association algorithm to generate a stable trajectory for the target to estimate its motion state. With the increased resolution of modern measurement equipment, this assumption no longer holds. At any given time, the same target may return multiple measurement information from different equivalent scattering centers, thus exhibiting the target's scalability. In the above embodiment, the extended target tracking method can more accurately estimate the target state.

[0099] S4: Establish separation event discrimination criteria, and determine the separation event time based on the separation event discrimination criteria and the target state.

[0100] In one specific embodiment, the separation event discrimination criterion is:

[0101]

[0102] In the formula: Δ represents the ellipse increase rate; D represents the separation event discrimination threshold; L1 represents the major semi-axis of the ellipse at the current moment; This represents the semi-major axis of the ellipse at the previous moment.

[0103] like Figure 2 As shown, after the separation occurs, the number of targets is updated to 2, and then updated according to the extended target tracking method.

[0104] In a specific embodiment, the separation event discrimination method based on HRRP extended target tracking described in this invention is used for separation event discrimination. Specifically:

[0105] HRRP data from two missions were selected, relevance alignment was performed, and feature points were extracted. Based on this, a virtual measurement was constructed, and the elliptical shape parameters of the expanded target were updated iteratively using filtering. The separation event time was output according to the designed separation event discrimination criterion. The separation event discrimination effect is as follows: Figure 3 and Figure 4 As shown in the figure, each point represents an extracted feature point, the light gray ellipse represents one target, and the dark gray ellipse represents another separated target.

[0106] from Figure 3 and Figure 4 It can be seen that the separation event occurred in frame 58 for Task 1 and in frame 11 for Task 2. When the feature points are relatively dispersed, the ellipse area is larger, while when the feature points are relatively clustered, the ellipse area is smaller. This indicates that the ellipse parameter model designed by the method of this invention can fully reflect the distribution characteristics and changes of HRRP feature points, and the parameter update process is good. From the perspective of the separation event discrimination time, it is consistent with the feature point expansion time, and it also has a certain anti-interference ability against noise. The method of this invention demonstrates good robustness.

[0107] In summary, this invention can accurately determine the timing of separation events using only HRRP data. Compared with existing technologies, this invention represents a significant advancement.

[0108] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A separation event discrimination method based on HRRP extended target tracking, characterized in that, The method comprises the following steps: S1: obtaining HRRP data and pre-processing the HRRP data to eliminate translation sensitivity; S2: extracting feature points of the HRRP data; S3: estimating a target state by using an extended target tracking method according to the feature points; The step of estimating the target state by using the extended target tracking method comprises the following sub-steps: S31: establishing an elliptic parameter model to describe the state of the extended target by using the elliptic parameter model; a center position parameter state r of the elliptic parameter model is expressed as: （2） where: x, y, z, vx, vy, vz represent the position and velocity of the center of the ellipse in the x, y, z directions, respectively; T represents the transpose of a column vector; and , represent the position and velocity of the center of the ellipse in the x, y, z directions, respectively; T represents the transpose of a column vector; and An elliptic shape parameter state p of the elliptic parameter model is expressed as: （3） In the formula, α represents a rotation angle of the ellipse; l1 represents a long semi-axis; and l2 represents a short semi-axis; A displacement of a scattering point z in the elliptic parameter model from the center is expressed by h1 and h2, wherein h1 represents a displacement in the long semi-axis direction, h2 represents a displacement in the short semi-axis direction, and the sizes of the two displacements are both in the range of [-1, 1]; S32: establishing a measurement model; a measurement value of the scattering point z is expressed as: （4） Considering the influence of measurement noise, a measurement equation is expressed as: （5） （6） In the formula, y represents the measurement value; H represents a measurement observation matrix; and v represents measurement noise; S33: establishing a tracking filter model to estimate the target state according to the tracking filter model; specifically: The state parameters of all measurement points at time k are sequentially updated; the position, shape parameters and covariance matrix of point i at time k are respectively denoted as , and , The state before the measurement value of point i at time k is updated is denoted as , and , The predicted value at time k is denoted as ; A position parameter of the i-th point is updated by using the following formula: （7） wherein: represents an estimate of the position parameter measurement at time k and point i; represents the cross-covariance matrix of the position parameters with the measurements; represents the error matrix of the measurements; is the measurement value; A shape parameter of the i-th point is updated by using the following formula: （8） （9） wherein: represents the cross-covariance matrix of virtual measurements and shape parameters; represents the covariance matrix of virtual measurements; represents a virtual measurement; represents an estimate of a virtual measurement; represents a Kronecher multiplier; A position parameter and a covariance matrix of the next moment are predicted by using the following formula: （10） where: represents a position parameter state transition matrix of a linear system; represents a position parameter covariance matrix of zero-mean Gaussian noise; A shape parameter and a state covariance matrix are predicted by using the following formula: （11） where: represents a shape parameter state transition matrix of a linear system; n k is the number of measurement points at time k; represents a shape parameter covariance matrix of zero-mean Gaussian noise; S4: establishing a separation event discrimination criterion to discriminate a separation event time in combination with the target state according to the separation event discrimination criterion; The separation event discrimination criterion is: （12） where Δ represents the ellipse growth rate; D represents the separation event discrimination threshold; represents the semi-major axis of the ellipse at the current time instant; represents the semi-major axis of the ellipse at the previous time instant.

2. The separation event discrimination method based on HRRP extended target tracking according to claim 1, characterized in that, In step S1, a correlation alignment method is used for pre-processing to eliminate translation sensitivity.

3. The separation event discrimination method based on HRRP extended target tracking according to claim 2, characterized in that, When the correlation alignment method is used for pre-processing, a reference HRRP sample is first selected, then a maximum sliding correlation coefficient between a current HRRP sample and the reference HRRP sample is calculated, and finally a sliding HRRP sample corresponding to the current HRRP sample is determined according to the maximum sliding correlation coefficient to realize alignment of the current HRRP sample and the reference HRRP sample.

4. The separation event discrimination method based on HRRP extended target tracking according to claim 3, characterized in that, The reference HRRP sample is an average value of all alignment samples before the current moment, and the maximum sliding correlation coefficient is calculated by using the following formula: （1） where: d i represents the maximum sliding correlation coefficient between the ith HRRP sample and the reference HRRP sample; x mean represents the average value of the 1 to i-1 aligned samples, where the first sample is translated by 0 distance units; x i represents the x i sample translated by τ distance units.

5. The separation event discrimination method based on HRRP extended target tracking according to claim 1, characterized in that, In step S2, when the feature points of the HRRP data are extracted, extreme value points with scattering intensity not lower than 60% of the highest intensity are extracted.

Images