Radar signal anti-jamming processing method and system based on space-time attention mechanism

Abstract

The application provides a radar signal anti-interference processing method and system based on a space-time attention mechanism. The method comprises: mapping original horizontal polarization channel signals and original vertical polarization channel signals to an interference suppression space to generate a three-dimensional feature tensor comprising an amplitude component, a polarization phase difference component and an interference suppression residual component; dividing a radar observation area into multiple layers of target blocks in a spatial dimension according to the amplitude component; calculating space-time correlation weights between adjacent layers of target blocks through a space-time attention mechanism; screening multiple target blocks that satisfy a preset continuous motion condition according to the space-time correlation weights; and decoding the motion mode of the multiple target blocks using a long short-term memory network to generate an anti-interference target trajectory coordinate and a velocity vector. The application combines polarization features and a space-time attention mechanism, improves the anti-interference effect of radar signal processing in a strong interference environment, and thus improves the detection accuracy of multiple target motion trajectories.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar signal processing, and in particular to a radar signal anti-interference processing method and system based on a space-time attention mechanism. BACKGROUND

[0002] In a radar multi-target tracking scene under a strong electromagnetic interference environment, the following three core requirements need to be met: first, the ability to resist strong suppression jamming to ensure target signal stability; second, to avoid confusion of the cross tracks of multiple moving targets; and third, to maintain tracking accuracy when the target performs complex maneuvering movements such as sharp turns and high acceleration changes.

[0003] The current mainstream existing scheme adopts a polarization filtering and trajectory prediction joint framework, and the main steps are as follows: extracting the amplitude ratio, phase difference features, etc. of the horizontal and vertical polarization channels of the target echo; and suppressing the interference outside the working frequency band through an adaptive filter; finally, a sliding window is used to intercept the motion segment of the target, and the motion segment is input into a long short-term memory network to predict the future trajectory coordinates of the target.

[0004] However, the existing scheme has the following three key defects: first, the interference signals and target polarization features are severely mixed in the two-dimensional plane, and some interference components remain after adaptive filtering, which have a negative impact on subsequent trajectory reconstruction; second, the spatial correlation is rigid, and a fixed sliding window is used to divide the target trajectory segment, resulting in weak correlation between adjacent motion segments, which easily causes the tracking trajectory to break when the target performs complex maneuvering movements; third, when multiple target trajectories are close or cross, the existing scheme is limited in its ability to continuously track a single target, and it is difficult to effectively distinguish the cross tracks. SUMMARY

[0005] The present application provides a radar signal anti-interference processing method and system based on a space-time attention mechanism to solve the problem that existing technologies cannot cleanly separate interference at the feature level, resulting in residual interference and ultimately causing the tracking trajectory to break and the tracks to be confused.

[0006] In a first aspect, the present application provides a radar signal anti-interference processing method based on a space-time attention mechanism, comprising:

[0007] obtaining a raw horizontal polarization channel signal and a raw vertical polarization channel signal of a radar;

[0008] mapping the raw horizontal polarization channel signal and the raw vertical polarization channel signal into an interference suppression space to generate a three-dimensional feature tensor containing an amplitude component, a polarization phase difference component, and an interference suppression residual component;

[0009] According to the amplitude component, a radar observation area is divided into multiple layers of target blocks in a spatial dimension, and a spatio-temporal correlation weight between adjacent layers of target blocks is calculated through a spatio-temporal attention mechanism;

[0010] According to the spatio-temporal correlation weight, multiple target blocks satisfying a preset continuous motion condition are screened out, and a long short-term memory network is used to decode a motion mode of the multiple target blocks to generate a target trajectory coordinate and a velocity vector after anti-interference.

[0011] Optionally, the mapping of the original horizontal polarization channel signal and the original vertical polarization channel signal into an interference suppression space to generate a three-dimensional feature tensor including an amplitude component, a polarization phase difference component and an interference suppression residual component comprises:

[0012] The quadrature phase difference of the original horizontal polarization channel signal and the original vertical polarization channel signal is calculated;

[0013] According to the cosine component and the sine component of the quadrature phase difference, an orthogonal basis vector of the interference suppression space is constructed, and the orthogonal basis vector includes an energy response principal axis, a polarization difference component axis and an interference residual separation axis;

[0014] The original horizontal polarization channel signal is projected onto the energy response principal axis to generate an amplitude component, and the original vertical polarization channel signal is projected onto the polarization difference component axis to generate a polarization phase difference component;

[0015] According to the amplitude component and the polarization phase difference component, a projection difference amount of the original horizontal polarization channel signal and the original vertical polarization channel signal in a plane formed by the energy response principal axis and the polarization difference component axis is calculated;

[0016] The projection difference amount is projected onto the interference residual separation axis to generate an interference suppression residual component;

[0017] The amplitude component, the polarization phase difference component and the interference suppression residual component are integrated along three coordinate axes of the orthogonal basis vector to generate a three-dimensional feature tensor.

[0018] Optionally, the projecting of the projection difference amount onto the interference residual separation axis to generate an interference suppression residual component comprises:

[0019] An amplitude compression operation is performed on the projection difference amount to generate a normalized residual amplitude;

[0020] According to the normalized residual amplitude, a residual density distribution is generated using a distribution function template;

[0021] The residual density distribution is discretized in the interference residual separation axis to generate a residual vector sequence;

[0022] tensor expanding the residual vector sequence along the interference residual separation axis to generate an interference suppression residual component.

[0023] Optionally, the generating the residual density distribution according to the normalized residual amplitude and using the distribution function template comprises:

[0024] scanning the time domain waveform of the normalized residual amplitude to identify pulse rising edge and falling edge positions to generate an amplitude fluctuation quantization index;

[0025] According to the amplitude fluctuation quantization index, a distribution function template is matched from a pre-set residual distribution function template library;

[0026] According to the pulse rising edge and falling edge positions, statistical moment features are extracted from the normalized residual amplitude, and shape parameters of the matched distribution function template are calculated based on the statistical moment features of the normalized residual amplitude;

[0027] Based on the normalized residual amplitude, a residual density distribution is generated using the matched distribution function template with the shape parameters.

[0028] Optionally, the generating the residual density distribution based on the normalized residual amplitude and using the matched distribution function template with the shape parameters comprises:

[0029] According to the radar pulse repetition period, the normalized residual amplitude is divided into a plurality of pulse window residual sequences;

[0030] Parallel computing operations are performed on each of the pulse window residual sequences using the matched distribution function template with the shape parameters to generate window residual density distributions;

[0031] The window residual density distributions of all windows are superimposed along the pulse sequence dimension to generate a pulse cumulative residual distribution;

[0032] The pulse cumulative residual distribution is mapped to a range-Doppler plane to generate a residual density distribution.

[0033] Optionally, the filtering out, according to the spatio-temporal correlation weight, a plurality of target blocks satisfying a pre-set continuous motion condition and decoding the motion mode of the plurality of target blocks using a long short-term memory network to generate an anti-interference target trajectory coordinate and velocity vector comprises:

[0034] The spatio-temporal correlation weight is analyzed for motion continuity to generate a motion continuity confidence index representing the motion correlation strength of a target block between consecutive observation frames;

[0035] The target blocks whose motion continuity confidence index is in a threshold interval of a pre-set continuous motion condition are filtered out to form a target block set;

[0036] inputting the historical phase sequence of the target block set in the Doppler spectrum into a long short-term memory network for motion mode decoding to generate a target motion vector field;

[0037] decomposing the target motion vector field into a radial velocity component and a tangential velocity component, and performing joint trajectory and velocity calculation based on the radial velocity component and the tangential velocity component to generate an anti-interference target trajectory coordinate and a velocity vector.

[0038] Optionally, the amplitude component, the polarization phase difference component and the interference suppression residual component are integrated along three coordinate axes of the orthogonal basis vector to generate a three-dimensional feature tensor, including:

[0039] performing an alignment operation on the amplitude component in an energy response principal axis direction to generate a first axial feature component;

[0040] performing an alignment operation on the polarization phase difference component in a polarization difference component axis direction to generate a second axial feature component;

[0041] performing a mapping operation on the interference suppression residual component in an interference residual separation axis direction to generate a third axial feature component;

[0042] performing a three-dimensional tensor synthesis operation on the first axial feature component, the second axial feature component and the third axial feature component to generate a three-dimensional feature tensor.

[0043] In a second aspect, the present application provides a radar signal anti-interference processing system based on a space-time attention mechanism, including:

[0044] an acquisition module configured to acquire original horizontal polarization channel signals and original vertical polarization channel signals of a radar;

[0045] a mapping module configured to map the original horizontal polarization channel signals and the original vertical polarization channel signals into an interference suppression space to generate a three-dimensional feature tensor including an amplitude component, a polarization phase difference component and an interference suppression residual component;

[0046] a division module configured to divide a radar observation area into multiple layers of target blocks in a spatial dimension according to the amplitude component, and calculate space-time correlation weights between adjacent layers of target blocks through a space-time attention mechanism;

[0047] a screening module configured to screen multiple target blocks satisfying a preset continuous motion condition according to the space-time correlation weights, and decode motion modes of the multiple target blocks through a long short-term memory network to generate an anti-interference target trajectory coordinate and a velocity vector.

[0048] In a third aspect, the present application provides a computing device, comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component, to implement the radar signal anti-interference processing method based on the space-time attention mechanism as any one of the first aspect.

[0049] In a fourth aspect, the present application provides a computer storage medium, which stores a computer program, and when the computer program is executed by a computer, a radar signal anti-interference processing method based on the space-time attention mechanism as any one of the first aspect is implemented.

[0050] In the present application, a radar signal anti-interference processing method based on the space-time attention mechanism is provided, which comprises: acquiring original horizontal polarization channel signals and original vertical polarization channel signals of a radar; mapping the original horizontal polarization channel signals and the original vertical polarization channel signals into an interference suppression space to generate a three-dimensional feature tensor containing an amplitude component, a polarization phase difference component and an interference suppression residual component; dividing a radar observation area into multiple layers of target blocks in a spatial dimension according to the amplitude component, and calculating space-time correlation weights between adjacent layers of target blocks through a space-time attention mechanism; screening multiple target blocks that meet a preset continuous motion condition according to the space-time correlation weights, and decoding a motion mode of the multiple target blocks by using a long short-term memory network to generate an anti-interference target trajectory coordinate and a velocity vector.

[0051] In the present application, by synchronously collecting horizontal and vertical polarization channel signals, the full polarization scattering characteristics of a target are retained to provide an original information basis for interference suppression; the dual-polarization signals are mapped into an interference suppression space to generate a three-dimensional tensor that fuses the amplitude, the polarization phase difference and the interference suppression residual, so as to realize physical decoupling of interference signals and target features; the target blocks are divided in layers based on the amplitude component, and the correlation weights of adjacent layers of blocks are adaptively calculated through a space-time attention mechanism to solve the problem of false correlation caused by the crossing of dense target trajectories; the continuous motion target blocks are screened according to the correlation weights, and a long short-term memory network is used to decode a complex motion mode, so as to finally output an anti-interference high-precision trajectory coordinate and a velocity vector. Finally, the robust reconstruction of multiple target trajectories and the accurate analysis of motion states in a strong interference scene are realized.

[0052] Further, the application constructs an interference suppression space with energy response principal axis, polarization difference axis and interference residual separation axis as base vectors by calculating the quadrature phase difference of the original horizontal and vertical polarization channel signals; projects the horizontal channel signal to the energy response axis to generate an amplitude component, and projects the vertical channel signal to the polarization difference axis to generate a polarization phase difference component; calculates the projection difference based on the two components, and after amplitude compression, residual density distribution generation and discretization processing, expands the interference suppression residual component along the interference residual separation axis; finally, integrates the three components along the orthogonal base vector coordinate axis into a three-dimensional feature tensor. The three-dimensional feature tensor constructed by the application realizes interference suppression capability improvement through triple decoupling: the energy response principal axis carries target intensity information, the polarization difference axis separates target polarization scattering characteristics, and the interference residual separation axis uses residual density distribution transformation to compress the interference signal to an independent dimension, fundamentally solving the coupling problem of interference and target signal in traditional two-dimensional polarization features, and providing a physically interpretable feature base for subsequent processing.

[0053] These and other aspects of the present application will become more fully understood from the following detailed description. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0055] Figure 1 A flow chart of a radar signal anti-interference processing method based on a space-time attention mechanism provided by an embodiment of the present application;

[0056] Figure 2 A structural schematic diagram of a radar signal anti-interference processing system based on a space-time attention mechanism provided by an embodiment of the present application;

[0057] Figure 3 A structural schematic diagram of a computing device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0058] In order to make those skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0059] In some of the processes described in this specification and in the claims and in the accompanying drawings, multiple processes are described in a particular order. However, it should be understood that, in accordance with the application, the processes can be performed in an order other than that which is described. For example, processes can be performed in an order other than that specifically described, or processes can be performed concurrently or with partial concurrence. Moreover, some processes can be performed automatically, in response to a determination by one or more suitable components. In addition, the processes described with reference to the example implementations can include more processes or fewer processes than those described. It should also be understood that the processes described with reference to the example implementations can be performed by different components of the example implementations than those specifically described.

[0060] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0061] To solve the problem that the prior art cannot cleanly peel off interference at the feature level, resulting in residual interference, and ultimately causing the breaking of the tracking trajectory and the confusion of the trajectory, the embodiments of the present application provide a radar signal anti-interference processing method based on a space-time attention mechanism. The method adopts the following concept: in view of the pain points of easy breaking and misassociation of multi-target trajectories in a strong interference scene, the embodiments of the present application adopt a phased framework of feature decoupling, dynamic association and motion decoding, construct an interference suppression space at the polarization signal layer, decouple the dual-polarization signal into amplitude, phase difference and interference residual components with clear physical meaning, and form an anti-interference feature base; at the space-time modeling layer, the target block is divided based on the amplitude layer, the adjacent layer block association is dynamically learned through the space-time attention mechanism, and the fragmentation bottleneck of the trajectory caused by the fixed threshold is broken through; at the trajectory generation layer, the continuous motion target block is screened, the long short-term memory network is used to decode the complex motion mode, and the end-to-end reconstruction of the anti-interference trajectory is realized.

[0062] Figure 1 A flowchart of a radar signal anti-interference processing method based on a space-time attention mechanism provided by the embodiments of the present application is shown in FIG. 1, and the method comprises the following steps. Figure 1

[0063] S11, obtaining a raw horizontal polarization channel signal and a raw vertical polarization channel signal of a radar.

[0064] ​The original horizontal polarization channel signal refers to a radar horizontal polarization antenna received echo signal, containing target scattering energy and environmental interference, and is used to extract electromagnetic wave polarization information in the horizontal direction. The original vertical polarization channel signal refers to a radar vertical polarization antenna received echo signal, which forms an orthogonal polarization pair with the horizontal channel, and is used to analyze the scattering characteristics of the target to the vertical polarization electromagnetic wave.

[0065] In the embodiment of the present application, first, the original horizontal polarization channel signal and the original vertical polarization channel signal are obtained by the radar receiver. The two signals are composed of electromagnetic wave echo data collected by the radar antenna horizontal polarization port and the vertical polarization port.

[0066] S12, map the original horizontal polarization channel signal and the original vertical polarization channel signal to the interference suppression space to generate a three-dimensional feature tensor containing an amplitude component, a polarization phase difference component and an interference suppression residual component.

[0067] The interference suppression space refers to a three-dimensional vector space constructed based on the orthogonal phase difference, and the three axes are respectively used to separate the target intensity, the polarization feature and the residual interference component. The amplitude component is the projection value of the original horizontal polarization channel signal on the energy response main axis, reflecting the energy intensity characteristics of the target echo. The polarization phase difference component is the projection value of the original vertical polarization channel signal on the polarization difference axis, representing the phase difference characteristics of the target horizontal and vertical polarization echoes. The interference suppression residual component is the mapping result of the projection difference on the interference residual separation axis, used to capture the residual interference noise after polarization decomposition. The three-dimensional feature tensor is a data structure integrating the amplitude component, the polarization phase difference component and the interference suppression residual component along the three coordinate axes of the orthogonal basis vector, constituting the input feature of the subsequent processing.

[0068] In the embodiment of the present application, first, the original horizontal polarization channel signal and the original vertical polarization channel signal are mapped to the interference suppression space: the orthogonal phase difference of the two signals is calculated, the cosine component and the sine component are extracted to construct the orthogonal basis vector containing the energy response main axis, the polarization difference axis and the interference residual separation axis; then the horizontal signal is projected onto the energy response main axis to generate the amplitude component, and the vertical signal is projected onto the polarization difference axis to generate the polarization phase difference component; then the projection difference of the two signals on the energy-polarization plane is calculated based on the amplitude component and the polarization phase difference component; finally, the projection difference is projected onto the interference residual separation axis to generate the interference suppression residual component, and the three components are integrated to form a three-dimensional feature tensor.

[0069] S13, according to the amplitude component, divide the radar observation area into multiple layers of target blocks in the spatial dimension, and calculate the spatio-temporal correlation weight between adjacent layers of target blocks through the spatio-temporal attention mechanism.

[0070] The spatial dimension refers to a distance-azimuth two-dimensional plane of a radar observation region, and is used to divide a spatial coordinate system of a target block. The radar observation region refers to a range of space to be monitored covered by a radar beam, and contains potential spatial positions of all targets to be detected. The multi-layer target block refers to a grid unit divided in the spatial dimension, and each layer corresponds to an azimuth slice in a specific range gate. The target block refers to a minimum processing unit with similar motion characteristics in a single layer, and contains a target echo set in a local region. The adjacent layer target block refers to two target block levels continuously in the range dimension, and is used to analyze the motion continuity of a target across range gates. The spatio-temporal correlation weight refers to a correlation strength value between target blocks calculated through an attention mechanism, and reflects a probability of position migration of a target between continuous frames.

[0071] In the embodiment of the present application, first, a multi-layer target block is divided in the spatial dimension of the amplitude component in the radar observation region, and each layer of the target block corresponds to a specific range-azimuth unit. Then, the correlation between adjacent layer target blocks is calculated through a spatio-temporal attention mechanism. The motion features of the target block between continuous frames are extracted, the coupling relationship between the spatial position and the time evolution is quantified by using an attention weight matrix, and a spatio-temporal correlation weight representing the correlation strength of the target is generated.

[0072] S14, according to the spatio-temporal correlation weight, a plurality of target blocks satisfying a preset continuous motion condition are screened out, and a long short-term memory network is used to decode the motion mode of the plurality of target blocks to generate a target trajectory coordinate and a velocity vector after anti-interference.

[0073] The preset continuous motion condition is used to determine whether a target is a real moving target or random noise or residual interference. The plurality of target blocks refer to a target block set formed by screening, representing the correlation echoes of the same target in multiple spatial layers. The long short-term memory network refers to a recurrent neural network with a forget gate, an input gate, and an output gate structure, which is used to model the long-term dependence relationship of the target motion sequence. It should be noted that the structure of the network is not limited in the embodiment of the present application. The motion mode refers to the speed change law of the target block set in the spatio-temporal evolution, including acceleration, turning, and other dynamic characteristics. The target trajectory coordinate refers to a target spatial position sequence after anti-interference processing, which is represented by a three-dimensional space coordinate. The velocity vector refers to a description of the instantaneous speed of the target motion, which includes the vector composition result of the radial velocity component and the tangential velocity component.

[0074] In the embodiment of the present application, first, the target blocks whose motion continuity confidence reaches the preset continuous motion condition are screened out according to the spatio-temporal correlation weight to form a plurality of target block sets. Then, the historical phase sequence of the target block in the Doppler spectrum is input into the long short-term memory network, and the acceleration and turning features of the target are decoded through the gating mechanism. Finally, the target trajectory coordinate and the velocity vector after anti-interference are output, wherein the velocity vector contains radial and tangential components.

[0075] The following is a specific example: a certain air defense surveillance radar tracks three high-speed aircraft in a strong electromagnetic interference environment: aircraft No. 1 maintains a constant speed straight flight at an altitude of 10,000 meters; aircraft No. 2 performs continuous figure-eight maneuvers at low altitude; aircraft No. 3 crosses the trajectory of a flock of migratory birds. The radar first synchronously collects the original horizontal polarization channel signal, which contains strong scattering energy from the metal parts of the aircraft, and the original vertical polarization channel signal, which is sensitive to the phase modulation characteristics of the rotor or rudder surface. Then the dual-polarized signal is mapped to the interference suppression space, and the quadrature phase difference between the two signals is calculated, for example, aircraft No. 2 phase difference Δφ = 32°, to construct a three-dimensional orthogonal coordinate system composed of energy response principal axis, polarization difference axis and interference residual separation axis. Project the horizontal signal onto the energy axis to generate the amplitude component, such as aircraft No. 1 amplitude 0.92; project the vertical signal onto the polarization axis to generate the polarization phase difference component, such as aircraft No. 2 phase modulation amount 0.78. Based on the two components, calculate the projection difference and map it to the residual axis to generate the interference suppression residual component, for example, aircraft No. 3 residual 0.21, effectively suppressing bird flock interference; finally integrated into a 128x128x3 three-dimensional feature tensor. Based on the amplitude component, the observation space is divided into 20 layers of distance gates, each layer spaced 5 kilometers apart, and each layer is divided into 8x8 azimuth grids to form multiple layers of target blocks. For aircraft No. 2 in the target blocks of adjacent distance layers, the spatio-temporal attention mechanism calculates the spatio-temporal correlation of its displacement vector 0.3 km, -1.2 km and speed change 120 m / s, outputs the spatio-temporal correlation weight 0.88, which represents a high confidence correlation; while the correlation weight of aircraft No. 3 with the bird flock is only 0.12, effectively distinguishing the cross trajectory. According to the weight, select the target blocks that meet the continuous motion condition, for example, aircraft No. 1 motion continuity confidence 0.94 is greater than the threshold 0.8, forming a multi-target block set. The Doppler history phase sequence [1.2, 0.8,..., -0.3] of aircraft No. 2 in the last 5 frames is input into the long short-term memory network, and the motion vector field 650 m / s, -320 m / s is decoded and generated; finally decomposed into radial velocity component 580 m / s and tangential velocity component 420 m / s, output the anti-interference trajectory coordinate sequence such as (12.3, 8.7), (11.8, 8.1) and the synthesized velocity vector, reducing the speed error. This scheme improves the interference suppression, and the number of aircraft No. 2 maneuver trajectory breaks is zero, and the misassociation rate of the cross trajectory is reduced.

[0076] By performing S11-S14, the embodiment of the present application generates a three-dimensional feature tensor by decoupling and mapping the dual-polarized signal in the interference suppression space, improving the separation capability of the target and the interference; the spatio-temporal attention mechanism based on spatially layered target blocks accurately captures the target motion correlation; combined with the long short-term memory network for dynamic modeling of continuous motion targets, effectively overcoming the trajectory breaking problem in strong interference environment, realizing stable and reliable target trajectory reconstruction and motion parameter estimation.

[0077] In a possible embodiment, S12, the original horizontal polarization channel signal and the original vertical polarization channel signal are mapped into an interference suppression space to generate a three-dimensional feature tensor containing an amplitude component, a polarization phase difference component and an interference suppression residual component, including:

[0078] Step 121, calculate the quadrature phase difference of the original horizontal polarization channel signal and the original vertical polarization channel signal.

[0079] Wherein, the quadrature phase difference refers to the arctangent value of the real part divided by the imaginary part of the phase difference of the horizontal and vertical polarization channel signals, reflecting the phase response difference of the target scatterer to the quadrature polarization electromagnetic wave.

[0080] In the embodiments of the present application, first, the ratio of the real part and the imaginary part of the phase difference of the original horizontal polarization channel signal and the original vertical polarization channel signal is calculated to obtain the quadrature phase difference which characterizes the quadrature characteristic of the dual-polarized signal.

[0081] Step 122, construct the quadrature basis vector of the interference suppression space according to the cosine component and the sine component of the quadrature phase difference, and the quadrature basis vector includes an energy response principal axis, a polarization difference component axis and an interference residual separation axis.

[0082] Wherein, the cosine component is the projection coefficient of the quadrature phase difference in the real axis direction, which is used to construct the unit vector basis of the energy response principal axis. The sine component is the projection coefficient of the quadrature phase difference in the imaginary axis direction, which is used as the mathematical basis for constructing the polarization difference component axis. The quadrature basis vector refers to a set of mutually perpendicular unit vectors composed of the energy response principal axis, the polarization difference component axis and the interference residual separation axis, which is used to define the coordinate system of the interference suppression space. The energy response principal axis refers to the coordinate axis with the cosine component as the direction reference, which is used to extract the energy intensity feature of the target echo. The polarization difference component axis refers to the coordinate axis with the sine component as the direction reference, which is used to separate the phase difference feature of the horizontal and vertical polarization signals. The interference residual separation axis refers to the coordinate axis generated by the vector cross product of the energy response principal axis and the polarization difference component axis, which is used to capture the residual interference component after orthogonal projection.

[0083] In the embodiments of the present application, first, the cosine component and the sine component of the quadrature phase difference are extracted as the basic elements; second, the cosine component is used to define the energy response principal axis, and the sine component is used to define the polarization difference component axis; finally, the interference residual separation axis perpendicular to the first two axes is generated by vector cross product to construct the complete three-dimensional quadrature basis vector.

[0084] Step 123, project the original horizontal polarization channel signal to the energy response principal axis to generate the amplitude component, and project the original vertical polarization channel signal to the polarization difference component axis to generate the polarization phase difference component.

[0085] In the embodiment of the present application, first, the original horizontal polarization channel signal is dot multiplied with the unit vector of the energy response principal axis to generate an amplitude component reflecting the echo intensity of the target; second, the original vertical polarization channel signal is projected to the unit vector of the polarization difference axis to generate a polarization phase difference component representing the polarization difference.

[0086] Step 124, according to the amplitude component and the polarization phase difference component, the projection difference quantity of the original horizontal polarization channel signal and the original vertical polarization channel signal in the plane formed by the energy response principal axis and the polarization difference axis is calculated.

[0087] The projection difference quantity refers to the vector modulus of the projection of the dual-polarized signal on the energy-polarization plane, reflecting the deviation degree of the target echo from the ideal polarization model.

[0088] In the embodiment of the present application, first, the vector coordinates of the amplitude component and the polarization phase difference component are calculated; second, the projection vector Euclidean distance of the horizontal and vertical polarization signals in the plane formed by the energy response principal axis and the polarization difference axis is calculated to generate the projection difference quantity representing the signal difference.

[0089] Step 125, the projection difference quantity is projected to the interference residual separation axis to generate an interference suppression residual component.

[0090] In the embodiment of the present application, first, the projection difference quantity is dot multiplied with the unit vector of the interference residual separation axis; second, the amplitude fluctuation is suppressed by absolute value limiting; finally, the interference suppression residual component is generated by logarithmic transformation.

[0091] Step 126, the amplitude component, the polarization phase difference component and the interference suppression residual component are integrated along the three coordinate axes of the orthogonal basis vector to generate a three-dimensional feature tensor.

[0092] The coordinate axis refers to the three direction dimensions of the orthogonal basis vector, including the energy response principal axis, the polarization difference axis and the interference residual separation axis.

[0093] In the embodiment of the present application, first, the amplitude component is arranged along the energy response principal axis to form the first dimension; second, the polarization phase difference component is arranged along the polarization difference axis to form the second dimension; finally, the interference suppression residual component is expanded along the interference residual separation axis to form the third dimension, and the three-dimensional feature tensor is synthesized.

[0094] The following is a specific example: when a certain air defense surveillance radar processes low-altitude maneuvering aircraft, first calculate the orthogonal phase difference between the horizontal polarization signal real part 0.65, imaginary part 0.12 and the vertical polarization signal real part -0.21, imaginary part 0.73. Through the real and imaginary part product operation and the arctangent function calculation, a phase difference value of 122.8 degrees is obtained, which reflects the modulation effect of the aircraft rudder deflection on the electromagnetic wave scattering characteristics. Then, based on the phase difference, a three-dimensional orthogonal coordinate system is constructed: take the cosine component -0.54 to define the energy response principal axis direction [-0.54, 0, 0], take the sine component 0.84 to define the polarization difference axis direction [0, 0.84, 0], and generate the interference residual separation axis direction [0, 0, -1] through vector cross product. For the target flying at high altitude at a constant speed, the horizontal polarization signal [0.92, 0.15] is projected onto the energy response principal axis direction through dot product, obtaining the amplitude component 0.50 representing the target intensity; at the same time, the vertical polarization signal [-0.08, 0.42] is projected onto the polarization difference axis direction, generating the phase difference component 0.35 reflecting the polarization characteristics. For the target intersecting the bird group trajectory, calculate its projection point coordinates (-0.21, 0) and (0, 0.38) in the energy-polarization plane, and obtain the projection difference 0.43 by the Euclidean distance formula. Project the difference onto the residual separation axis, and after amplitude limiting and logarithmic transformation processing, finally generate the interference suppression residual component 0.15. All feature components along the three coordinate axes are integrated into a 128x128x3 three-dimensional feature tensor: at the target space position (60, 75), the energy layer is assigned a value of 0.50, the polarization layer is assigned a value of 0.35, and the residual layer is assigned a value of 0.00; while the residual layer of the intersecting target is raised to 0.15, effectively separating the bird group interference characteristics. The processing process takes only 2 milliseconds under the acceleration of a graphics processing unit, realizing the physical interpretability of feature decoupling, that is, the energy layer highlights the metal fuselage reflection, the polarization layer captures the dynamic modulation characteristics of the rudder, and the residual layer isolates the natural interference, establishing a highly discriminative feature base for subsequent target tracking.

[0095] By performing steps 121-126, the embodiment of the present application constructs a three-dimensional orthogonal space to realize signal decoupling: the energy axis direction retains the essential characteristics of the target, the polarization axis direction extracts discriminative differences, and the residual axis direction separates noise interference; the finally generated three-dimensional feature tensor fuses multi-dimensional information, providing a highly discriminative input base for subsequent processing.

[0096] In one possible embodiment, step 125, projecting the projection difference onto the interference residual separation axis to generate an interference suppression residual component, comprises:

[0097] Step a1, performing amplitude compression operation on the projection difference to generate a normalized residual amplitude.

[0098] The amplitude compression operation refers to a processing procedure of applying amplitude limitation and non-linear transformation to the projection difference quantity, including threshold truncation to eliminate extreme value interference and logarithmic scaling to enhance weak signal visibility. The normalized residual amplitude refers to a standardized residual signal output after the amplitude compression operation, and the numerical range thereof is constrained in a unit interval, which is used to unify the measurement benchmark of different intensity interference.

[0099] In the embodiments of the present application, the amplitude compression operation is first performed on the projection difference quantity, that is, the abnormal high amplitude points are truncated by a preset upper threshold, then the truncated amplitude is non-linearly scaled by a logarithmic function, and finally the normalized residual amplitude is generated by dividing the reference energy value.

[0100] Step a2, generating a residual density distribution by using a distribution function template according to the normalized residual amplitude.

[0101] The distribution function template refers to a mathematical model in a preset probability density function library, which is constructed based on a Weibull distribution or a Rayleigh distribution, and is used to fit the statistical characteristics of the residual amplitude. The residual density distribution refers to a result of probability modeling of the normalized residual amplitude by using the distribution function template, and reflects the distribution law of the residual signal in a statistical sense.

[0102] In the embodiments of the present application, the time domain waveform of the normalized residual amplitude is first scanned to identify the pulse feature, then the amplitude fluctuation quantitative index is calculated according to the positions of the rising edge and the falling edge of the pulse, subsequently the distribution function template is matched from a preset template library, then the statistical moment feature of the residual amplitude is extracted to calculate the template shape parameter, and finally the residual density distribution is generated by using the distribution function template with the parameter.

[0103] Step a3, discretizing the residual density distribution in the interference residual separation axial direction to generate a residual vector sequence.

[0104] The discretization refers to a process of converting the continuous residual density distribution into a finite number of discrete points in the interference residual separation axial direction, and the vectorization of the continuous signal is realized by equidistant sampling. The residual vector sequence refers to a set of probability density values arranged in the order of the axial coordinates after discretization, which constitutes a one-dimensional structured residual representation.

[0105] In the embodiments of the present application, first, equidistant coordinate points are set in the interference residual separation axial direction, then the probability density values of the residual density distribution at each coordinate point are calculated, and subsequently the probability density values are arranged in the order of the coordinates to generate the residual vector sequence.

[0106] Step a4, tensor expanding the residual vector sequence along the interference residual separation axial direction to generate an interference suppression residual component.

[0107] The tensor expansion refers to a dimension replication operation of the residual vector sequence along an interference residual separation axis, and the three-dimensional integration with other components is realized through a Kronecker product.

[0108] In the embodiment of the present application, the residual vector sequence is first copied and expanded along the interference residual separation axis, secondly, the dimension tensor product operation is performed with the amplitude component and the polarization phase difference component, and finally, the interference suppression residual component with a three-dimensional structure is generated.

[0109] The following is a specific example: first, the onboard radar system performs amplitude compression on the projection difference quantity of a certain low-altitude unmanned aerial vehicle target: set the amplitude upper limit threshold to truncate the burst interference, and generate a normalized residual amplitude after logarithmic scaling and division by the background noise mean. Secondly, analyze the pulse waveform characteristics of the residual amplitude, match the Weibull distribution template and calculate the shape parameter to generate a residual density probability model. Then, set discrete coordinate points to sample the probability density along the interference residual separation axis to form a residual vector sequence. Finally, the sequence is expanded along the residual axis to a three-dimensional space, and the energy and polarization dimensions are synthesized to generate the final interference suppression residual component.

[0110] By performing steps a1-a4, the embodiment of the present application suppresses abnormal interference pulses by amplitude compression, fits the residual characteristics by using a statistical distribution model to enhance the anti-rising ability; the discrete processing realizes the structured conversion of the continuous signal, and finally the tensor expansion ensures the integrity of the three-dimensional characteristics, which provides a robust guarantee for interference suppression.

[0111] In a possible embodiment, step a2, according to the normalized residual amplitude, generates a residual density distribution using a distribution function template, including:

[0112] Step a21, scan the time domain waveform of the normalized residual amplitude, identify the positions of the pulse rising edge and falling edge, and generate an amplitude fluctuation quantization index.

[0113] The time-domain waveform refers to a curve form of normalized residual amplitude changing with time, and reflects amplitude fluctuation characteristics of the interference signal. The identified pulse rising edge position and falling edge position directly determine the calculation reference of the amplitude fluctuation quantitative index. Firstly, the rising edge position is taken as a starting point and the falling edge position is taken as a terminal point to define a pulse effective action interval. Secondly, a ratio of a maximum change of the normalized residual amplitude in the interval to a time span is calculated to generate an amplitude change rate reflecting a steepness of the interference pulse. Finally, the amplitude change rate is multiplied by a pulse width normalization coefficient to output the amplitude fluctuation quantitative index representing the interference impact strength. The pulse rising edge refers to a transition section of the residual amplitude in the time-domain waveform from a low level to a high level, and is used to locate a starting time of the interference pulse. The falling edge refers to a transition section of the residual amplitude in the time-domain waveform from a high level to a low level, and is used to locate an ending time of the interference pulse. The position refers to a coordinate point of the pulse rising edge and the falling edge on a time axis, and is marked by a sampling point serial number. The amplitude fluctuation quantitative index refers to a scalar value calculated based on the amplitude change rate between the rising edge and the falling edge, and is used to measure the steepness of the interference pulse.

[0114] In the embodiment of the present application, firstly, the time-domain waveform of the normalized residual amplitude is scanned, and a signal slope mutation point is detected through a difference operation. Secondly, the pulse rising edge position is identified as a turning point from a low amplitude to a high amplitude, and the falling edge position is identified as a turning point from a high amplitude to a low amplitude. Finally, an amplitude change rate between the rising edge and the falling edge is calculated to generate the amplitude fluctuation quantitative index.

[0115] Step a22, according to the amplitude fluctuation quantitative index, matching a distribution function template from a preset residual distribution function template library.

[0116] The preset residual distribution function template library refers to a pre-stored statistical model set, and includes probability density function prototypes such as Rayleigh distribution and Weibull distribution. The physical meaning of the distribution function template is a mathematical abstract model of the statistical characteristics of the interference residual, and the content of the distribution function template includes a probability density function form such as Weibull distribution or Rayleigh distribution defining the probability distribution law of the residual amplitude, an adjustable shape parameter and a scale parameter controlling the distribution form, and a mapping rule relating the amplitude fluctuation quantitative index and the distribution type. The core difference between different templates lies in the distribution type and the essence of the physical meaning: the distribution function is suitable for modeling uniform background noise generated by a large number of weak scatterers, the distribution function is characterized by a single scale parameter and no shape parameter, and the Weibull distribution is suitable for a few strong scatterers dominated by burst fluctuation interference. The shape parameter is sensitive to the steepness of the pulse, and the two correspond to different interference physical generation mechanisms.

[0117] In the embodiment of the present application, firstly, the value range of the amplitude fluctuation quantization index is determined; secondly, a matching distribution type is selected from a preset residual distribution function template library, a Rayleigh distribution template is matched when the index is lower than a threshold, and a Weibull distribution template is matched when the index is higher than the threshold; and finally, the selected distribution function template is output.

[0118] Step a23, according to the pulse rising edge and falling edge positions, statistical moment features are extracted from the normalized residual amplitude, and shape parameters of the matching distribution function template are calculated based on the statistical moment features of the normalized residual amplitude.

[0119] The statistical moment features refer to numerical features for describing the statistical characteristics of the residual amplitude, including a mean value reflecting a central position and a variance reflecting a dispersion degree. The shape parameters refer to key variables for controlling the curve shape of the distribution function, such as a shape coefficient in the Weibull distribution for determining the skewness of the distribution.

[0120] In the embodiment of the present application, firstly, the effective section of the normalized residual amplitude is intercepted based on the pulse rising edge and falling edge positions; secondly, the statistical moment features of the section are extracted, including a first-order origin moment mean value and a second-order central moment variance; and finally, the statistical moment features are input into a parameter calculator of the distribution function template, and the shape parameters are output.

[0121] Step a24, based on the normalized residual amplitude, the residual density distribution is generated by using the matching distribution function template with the shape parameters.

[0122] In the embodiment of the present application, firstly, the normalized residual amplitude is input as an independent variable; secondly, the matching distribution function template with the shape parameters is called; and finally, the probability density value corresponding to each amplitude point is calculated, and the residual density distribution function is generated.

[0123] The following is a specific example: firstly, the normalized residual amplitude waveform of a ship-borne radar system scanning a ship target in a sea clutter environment is scanned, the pulse rising edge position and falling edge position are identified, and the amplitude fluctuation quantization index is calculated. Secondly, the Weibull distribution function template is matched from the template library according to the index. Then, the mean value and variance statistical moment features of the residual amplitude are extracted in the interval defined by the pulse edge, and the shape parameters of the template are calculated. Finally, the normalized residual amplitude is input into the Weibull distribution template with the shape parameters, and the residual density probability distribution function is generated.

[0124] By performing steps a21 to a24, the embodiment of the present application dynamically matches the optimal statistical model through waveform features, accurately extracts residual characteristics by using pulse edge information, adjusts the distribution shape by combining statistical moment parameterization, realizes adaptive modeling of a complex interference environment, and improves the physical rationality of the residual density distribution.

[0125] In a possible embodiment, step a24, generating the residual density distribution with the matched distribution function template with shape parameters based on the normalized residual amplitudes, comprises:

[0126] Step b1, dividing the normalized residual amplitudes into a plurality of pulse window residual sequences according to a radar pulse repetition period.

[0127] Wherein, the radar pulse repetition period refers to a fixed time interval between adjacent two pulses emitted by the radar, determines the maximum detection distance and serves as a signal segmentation reference. The pulse window residual sequence refers to a data segment of the normalized residual amplitudes within a single pulse period, contains residual information of a specific distance gate.

[0128] In the embodiments of the present application, first, the pulse repetition period parameter of the radar system is acquired, second, the continuous normalized residual amplitudes are cut into a plurality of equal-length segments with the period length as the time window, and finally each segment forms a pulse window residual sequence.

[0129] Step b2, performing parallel computing operations on each pulse window residual sequence with the matched distribution function template with shape parameters to generate a window residual density distribution.

[0130] Wherein, the parallel computing operation refers to a technology of simultaneously performing the same operation on a plurality of data windows with a multi-core processor, and the distribution function calculation is accelerated by graphic processor thread parallelism. The window residual density distribution refers to a probability density function of the residual amplitudes within a single pulse window, reflects local statistical characteristics.

[0131] In the embodiments of the present application, first, the matched distribution function template with shape parameters is loaded, second, the probability density calculation is independently performed on each pulse window residual sequence on the graphic processor, and finally the corresponding window residual density distribution function is generated for each window.

[0132] Step b3, superimposing the window residual density distributions of all windows along the pulse sequence dimension to generate a pulse cumulative residual distribution.

[0133] Wherein, the pulse sequence dimension refers to an index axis of the pulse window arranged in time sequence, used for organizing the stacking direction of the multi-window data. The pulse cumulative residual distribution refers to a comprehensive probability model formed by superimposing all window density distributions along the pulse sequence dimension.

[0134] In the embodiments of the present application, first, all window residual density distributions are aligned along the pulse time sequence dimension, second, the density values of the corresponding distance units are added point by point, and finally the pulse cumulative residual distribution representing the overall statistical characteristics is generated.

[0135] Step b4, mapping the pulse cumulative residual distribution to the range-doppler plane to generate a residual density distribution.

[0136] The distance Doppler plane is a two-dimensional coordinate system with target distance as the horizontal axis and Doppler frequency shift as the vertical axis, and is used for spatial frequency domain signal representation. The association of the distance Doppler plane with the orthogonal basis vector is that the distance Doppler plane integrates the three-dimensional information of the orthogonal basis vector through coordinate transformation, the distance dimension corresponds to the target echo intensity characteristic in the energy response main axis direction, the Doppler dimension integrates the phase dynamic change characteristic in the polarization difference axis direction, and the noise statistical characteristic captured by the interference residual separation axis in the pulse accumulation residual distribution is encoded as the amplitude intensity of each coordinate point on the distance Doppler plane, so as to uniformly map the energy, polarization and residual characteristics of orthogonal space decomposition to the classical radar signal representation domain.

[0137] In the embodiment of the application, first, a distance Doppler plane coordinate system is established, second, the distance dimension of the pulse accumulation residual distribution is mapped as the horizontal coordinate of the plane, and the Doppler frequency is mapped as the vertical coordinate, and finally a two-dimensional residual density distribution is generated through bilinear interpolation.

[0138] The following is a specific example: first, the on-board radar system divides the normalized residual amplitude of a certain cruise missile target into a plurality of pulse window residual sequences according to the pulse repetition period. Second, the Weibull distribution calculation is performed in parallel on the graphics processor to generate the residual density distribution of each window. Then, the density functions of all windows are superimposed along the pulse sequence dimension to form a pulse accumulation residual distribution. Finally, the distribution is mapped to the distance Doppler plane to generate a spatialized residual density distribution containing distance-velocity interference characteristics.

[0139] By performing steps b1-b4, the embodiment of the application realizes time-domain localized analysis of interference through pulse segmentation, and parallel calculation improves processing efficiency; pulse dimension superposition enhances statistical robustness, and finally mapping to the distance Doppler plane forms a spatially analyzable interference distribution representation.

[0140] In a possible embodiment, S14, according to the spatio-temporal correlation weight, a plurality of target blocks satisfying a preset continuous motion condition are screened out, and a long short-term memory network is used to decode the motion mode of the plurality of target blocks to generate an anti-interference target trajectory coordinate and a velocity vector, including:

[0141] Step 141, motion continuity analysis is performed on the spatio-temporal correlation weight to generate a motion continuity confidence index representing the motion correlation strength of the target block between consecutive observation frames.

[0142] The motion continuity analysis refers to analyzing state transition probability of the target block between continuous observation frames by a Markov chain model, reflecting physical rationality of a target motion trajectory. The continuous observation frame refers to a signal snapshot sequence collected by the radar at a fixed time interval, constituring a time reference for target motion analysis. The motion correlation strength refers to a parameter quantifying correlation of the target block between frames, including displacement vector similarity and acceleration continuity. The motion continuity confidence index refers to a normalized scalar value, ranging from 0 to 1, and the greater the value, the more the target motion trajectory conforms to the physical law.

[0143] In the embodiment of the application, firstly, the motion continuity analysis is performed on the space-time correlation weight: the displacement change of the target block between the continuous observation frames is extracted, secondly, the displacement direction consistency coefficient and the velocity change smoothness are calculated, and then the motion continuity confidence index representing the motion correlation strength is generated by fusing the two indexes.

[0144] Step 142, screening the target block whose motion continuity confidence index is in a threshold interval of a preset continuous motion condition to form a target block set.

[0145] The threshold interval refers to a preset confidence effective range, the lower limit of which excludes random noise interference, and the upper limit of which filters out the maneuvering mutation target. The target block set refers to a group of spatial correlation units formed by screening, representing the echo set of the same target in multiple distance gates.

[0146] In the embodiment of the application, firstly, the upper and lower limits of the threshold interval of the preset continuous motion condition are set, secondly, the motion continuity confidence index is compared with the interval, then the target block whose index value is in the interval is screened, and finally the target block set containing the associated targets of multiple spatial layers is aggregated.

[0147] Step 143, inputting the historical phase sequence of the target block set in the Doppler spectrum into the long short-term memory network for motion pattern decoding to generate a target motion vector field.

[0148] The Doppler spectrum refers to a frequency-amplitude distribution of the target block echo signal generated by Fourier transform, reflecting the radial motion characteristics. The historical phase sequence refers to a sequence of the main lobe phase of the target block in the Doppler spectrum changing with time, carrying the target acceleration information. The target motion vector field refers to the structured data output by the long short-term memory network, including the velocity vector distribution of the spatial position.

[0149] In the embodiment of the application, firstly, the historical phase sequence in the Doppler spectrum is extracted from the target block set, secondly, it is input into the long short-term memory network, the abnormal value is filtered through the forget gate, the state is updated through the input gate, and the instantaneous velocity vector is generated through the output gate, and finally the target motion vector field is synthesized.

[0150] Step 144, decompose the target motion vector field into radial velocity component and tangential velocity component, based on the radial velocity component and the tangential velocity component, perform trajectory and velocity joint solution, and generate the target trajectory coordinates and velocity vector after anti-interference.

[0151] Wherein, the radial velocity component refers to the projection component of the target motion vector in the direction of the radar beam, used to calculate the target range rate. The tangential velocity component refers to the plane component of the target motion vector perpendicular to the beam direction, which determines the target azimuth change. The trajectory and velocity joint solution refers to the Kalman filtering process of synchronously optimizing the position coordinates and velocity vector, ensuring the physical consistency of the motion parameters.

[0152] In the embodiments of the present application, first, the target motion vector field is decomposed into the radial velocity component in the direction of the radar line of sight and the tangential velocity component in the vertical direction, second, the range change is calculated based on the radial component integration, and the azimuth change is calculated based on the tangential component, and finally the anti-interference target trajectory coordinates and the synthesized velocity vector are fused and generated.

[0153] The following is a specific example: first, the spatio-temporal correlation weight of a certain low-altitude unmanned aerial vehicle target block is analyzed for motion continuity, generating a motion continuity confidence index. Second, the target blocks with index values in the threshold interval of 0.8 to 0.95 are selected to form a set. Then, the historical phase sequence of the set in the Doppler spectrum is input into the long short-term memory network to decode and generate the target motion vector field. Finally, the vector field is decomposed into radial velocity component and tangential velocity component, and the anti-interference three-dimensional trajectory coordinates and velocity vector are output through joint solution.

[0154] By performing steps 141-144, the embodiments of the present application eliminate non-actual existing targets through motion continuity analysis, accurately model complex maneuvers using long short-term memory network, and realize trajectory reconstruction by combining radial and tangential velocity decomposition, effectively overcoming the problem of trajectory breakage under strong interference.

[0155] In one possible embodiment, step 126, integrate the amplitude component, the polarization phase difference component and the interference suppression residual component along the three coordinate axes of the orthogonal basis vector to generate a three-dimensional feature tensor, including:

[0156] Step c1, perform alignment operation on the amplitude component in the energy response principal axis direction to generate the first axial feature component.

[0157] Wherein, the alignment operation refers to the mathematical processing process of transforming the feature data to the reference axial direction, including coordinate rotation to eliminate directional deviation and scale normalization to eliminate dimensional difference. The first axial feature component refers to the standardized data layer generated after the amplitude component is aligned in the energy response principal axis direction, reflecting the distribution characteristics of the target echo intensity in the spatial dimension.

[0158] In the embodiment of the present application, first, the original coordinate value of the amplitude component in the energy response principal axis direction is obtained, second, the coordinate transformation is performed with the unit vector of the axis direction as the reference, then the scale difference between the distance units is eliminated through least square fitting, and finally the first axis direction feature component with uniform scale is generated.

[0159] Step c2, performing alignment operation on the polarization phase difference component in the polarization difference differential axis direction to generate the second axis direction feature component.

[0160] The second axis direction feature component refers to the phase difference distribution layer formed after the polarization phase difference component is aligned in the polarization difference differential axis direction, and represents the continuous change of the polarization characteristics in the observation area.

[0161] In the embodiment of the present application, first, the distribution data of the polarization phase difference component in the polarization difference differential axis direction is read, second, the cosine value of the included angle between the component and the reference direction of the axis direction is calculated, then the data is aligned to the positive direction of the axis direction through rotation transformation, and finally the second axis direction feature component with consistent direction is generated.

[0162] Step c3, performing mapping operation on the interference suppression residual component in the interference residual separation axis direction to generate the third axis direction feature component.

[0163] The mapping operation refers to the processing of converting the continuous probability distribution into discrete axis direction coordinates, and the data structure is realized through domain compression and value range quantization. The third axis direction feature component refers to the discretized residual layer generated by the mapping operation of the interference suppression residual component, which carries the spatial distribution information of the noise statistical characteristics.

[0164] In the embodiment of the present application, first, the probability density function of the interference suppression residual component in the interference residual separation axis direction is determined, second, the function value is normalized to the preset dynamic range, then the linear mapping is converted into discrete coordinate value, and finally the third axis direction feature component with matched dimension is generated.

[0165] Step c4, performing three-dimensional tensor synthesis operation on the first axis direction feature component, the second axis direction feature component and the third axis direction feature component to generate a three-dimensional feature tensor.

[0166] The three-dimensional tensor synthesis operation refers to the process of constructing a high-order tensor according to the orthogonal relationship of the three axis direction feature components, and the dimension fusion is realized through the Kronecker product.

[0167] In the embodiment of the present application, first, the first axis direction feature component is taken as the X-axis data layer in the three-dimensional space, second, the second axis direction feature component is taken as the Y-axis data layer, then the third axis direction feature component is taken as the Z-axis data layer, and finally the three-dimensional feature tensor is synthesized through the Kronecker product operation.

[0168] The following is a specific example: first, the energy response principal axis alignment operation is performed on the amplitude component of a certain aircraft target to generate a first axial feature component. Second, the polarization phase difference component is rotated in the polarization difference axis to generate a second axial feature component. Then, the interference suppression residual component is mapped into discrete coordinate values in the residual separation axis to generate a third axial feature component. Finally, the three feature layers are fused by a three-dimensional tensor synthesis operation to generate a three-dimensional feature tensor for target recognition.

[0169] By performing steps c1-c4, the embodiment of the present application realizes the spatial consistency expression of multi-source features through axial alignment, converts continuous statistics into computable data structures through mapping and conversion, and finally constructs a three-dimensional information carrier compatible with energy, polarization and noise features through tensor synthesis, thereby providing a structured input for a deep learning model.

[0170] Figure 2 The structure diagram of a radar signal anti-interference processing system based on a spatio-temporal attention mechanism provided by the embodiment of the present application is shown in FIG. Figure 2 The system comprises:

[0171] The acquisition module 21 is configured to acquire original horizontal polarization channel signals and original vertical polarization channel signals of a radar.

[0172] The mapping module 22 is configured to map the original horizontal polarization channel signals and the original vertical polarization channel signals into an interference suppression space to generate a three-dimensional feature tensor comprising an amplitude component, a polarization phase difference component and an interference suppression residual component.

[0173] The division module 23 is configured to divide a radar observation area into multiple layers of target blocks in a spatial dimension according to the amplitude component, and calculate spatio-temporal correlation weights between adjacent layers of target blocks through a spatio-temporal attention mechanism.

[0174] The screening module 24 is configured to screen multiple target blocks satisfying a preset continuous motion condition according to the spatio-temporal correlation weights, and decode motion patterns of the multiple target blocks using a long short-term memory network to generate target trajectory coordinates and velocity vectors after anti-interference.

[0175] Figure 2 The radar signal anti-interference processing system based on the spatio-temporal attention mechanism can perform Figure 1 The radar signal anti-interference processing method based on the spatio-temporal attention mechanism has the same implementation principles and technical effects as those of the embodiment shown in FIG. For the specific manner in which each module, unit of the radar signal anti-interference processing system based on the spatio-temporal attention mechanism performs operations, a detailed description has been made in the embodiment related to the method, which will not be described in detail here.

[0176] In one possible design,Figure 2 The radar signal anti-interference processing system based on the spatio-temporal attention mechanism of the illustrated embodiment can be implemented as a computing device, such as Figure 3 As shown, the computing device can include a storage component 31 and a processing component 32.

[0177] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called for execution by the processing component 32.

[0178] The processing component 32 is configured to perform the following processes: obtaining original horizontal polarization channel signals and original vertical polarization channel signals of a radar; mapping the original horizontal polarization channel signals and the original vertical polarization channel signals into an interference suppression space to generate a three-dimensional feature tensor containing an amplitude component, a polarization phase difference component, and an interference suppression residual component; dividing a radar observation area into multiple layers of target blocks in a spatial dimension according to the amplitude component, calculating spatio-temporal correlation weights between adjacent layers of target blocks through a spatio-temporal attention mechanism; screening multiple target blocks that satisfy a preset continuous motion condition according to the spatio-temporal correlation weights, and decoding a motion pattern of the multiple target blocks using a long short-term memory network to generate an anti-interference target trajectory coordinate and a velocity vector.

[0179] The processing component 32 can include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic elements for executing the above method.

[0180] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or nonvolatile storage devices or a combination thereof, such as a random access memory (RAM), a static random access memory (SRAM), an electrically erasable programmable read only memory (EEPROM), an erasable programmable read only memory (EPROM), a programmable read only memory (PROM), a read only memory (ROM), a magnetic storage, a flash memory, a magnetic disk, or a compact disk.

[0181] Of course, the computing device can also necessarily include other components, such as an input / output interface, a display component, a communication component, etc.

[0182] The input / output interface provides an interface between the processing component and peripheral interface modules, which can be output devices, input devices, etc.

[0183] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0184] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform, and the computing device can be a cloud server, and the processing component, the storage component, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0185] The embodiments of the present application also provide a computer storage medium storing a computer program, and the computer program can implement the above-mentioned Figure 1 The embodiment shown provides a radar signal anti-interference processing method based on a space-time attention mechanism.

[0186] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0187] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0188] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and necessary general hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of software products, and the computer software products can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and include a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0189] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A radar signal anti-jamming processing method based on a spatiotemporal attention mechanism, characterized in that, include: Acquire the original horizontal polarization channel signal and the original vertical polarization channel signal of the radar; The original horizontal polarization channel signal and the original vertical polarization channel signal are mapped into the interference suppression space to generate a three-dimensional feature tensor containing amplitude components, polarization phase difference components and interference suppression residual components. Based on the amplitude components, the radar observation area is divided into multiple target blocks in the spatial dimension, and the spatiotemporal correlation weights between adjacent target blocks are calculated through a spatiotemporal attention mechanism. Based on the spatiotemporal correlation weights, multiple target blocks that meet the preset continuous motion conditions are selected, and the motion patterns of the multiple target blocks are decoded using a long short-term memory network to generate anti-interference target trajectory coordinates and velocity vectors. The process of mapping the original horizontally polarized channel signal and the original vertically polarized channel signal into the interference suppression space to generate a three-dimensional feature tensor containing amplitude components, polarization phase difference components, and interference suppression residual components includes: Calculate the orthogonal phase difference between the original horizontal polarization channel signal and the original vertical polarization channel signal; Based on the cosine and sine components of the orthogonal phase difference, an orthogonal basis vector for the interference suppression space is constructed. The orthogonal basis vector includes the principal axis of energy response, the polarization difference axis, and the interference residual separation axis. The original horizontal polarization channel signal is projected onto the principal axis of the energy response to generate an amplitude component, and the original vertical polarization channel signal is projected onto the polarization differential axis to generate a polarization phase difference component. Based on the amplitude component and the polarization phase difference component, calculate the projection difference between the original horizontal polarization channel signal and the original vertical polarization channel signal on the plane formed by the principal axis of the energy response and the polarization difference axis; The projection difference is projected onto the interference residual separation axis to generate an interference suppression residual component. The amplitude component, the polarization phase difference component, and the interference suppression residual component are integrated along the three coordinate axes of the orthogonal basis vector to generate a three-dimensional feature tensor.

2. The method according to claim 1, characterized in that, The step of projecting the projection difference onto the interference residual separation axis to generate interference suppression residual components includes: The projection difference is subjected to amplitude compression to generate a normalized residual amplitude. Based on the normalized residual magnitude, a residual density distribution is generated using a distribution function template; Discretize the residual density distribution along the interference residual separation axis to generate a residual vector sequence; The residual vector sequence is tensor-expanded along the interference residual separation axis to generate interference suppression residual components.

3. The method according to claim 2, characterized in that, The step of generating a residual density distribution using a distribution function template based on the normalized residual magnitude includes: The time-domain waveform of the normalized residual amplitude is scanned to identify the positions of the rising and falling edges of the pulse, and an amplitude fluctuation quantification index is generated. Based on the amplitude fluctuation quantification index, a distribution function template is matched from a preset residual distribution function template library; Based on the positions of the rising and falling edges of the pulse, statistical moment features are extracted from the normalized residual amplitude. Based on the statistical moment features of the normalized residual amplitude, the shape parameters of the matching distribution function template are calculated. Based on the normalized residual magnitude, a residual density distribution is generated using a matching distribution function template with the shape parameters.

4. The method according to claim 3, characterized in that, The step of generating a residual density distribution based on the normalized residual amplitude using a matching distribution function template with the shape parameters includes: According to the radar pulse repetition period, the normalized residual amplitude is divided into multiple pulse window residual sequences; Parallel computation is performed on each of the pulse window residual sequences using a matching distribution function template with the shape parameters to generate a window residual density distribution; The window residual density distributions of all windows are superimposed along the pulse sequence dimension to generate the pulse cumulative residual distribution; The pulse accumulation residual distribution is mapped onto the distance Doppler plane to generate a residual density distribution.

5. The method according to claim 1, characterized in that, The step of selecting multiple target blocks that meet preset continuous motion conditions based on the spatiotemporal correlation weights, and decoding the motion patterns of the multiple target blocks using a long short-term memory network to generate anti-interference target trajectory coordinates and velocity vectors includes: Motion continuity analysis is performed on the spatiotemporal correlation weights to generate a motion continuity confidence index that characterizes the motion correlation strength of the target block between consecutive observation frames; Target blocks that fall within the threshold range of a preset continuous motion condition, based on the motion continuity confidence index, are selected to form a target block set; The historical phase sequence of the target block set in the Doppler spectrum is input into a long short-term memory network for motion pattern decoding to generate a target motion vector field. The target motion vector field is decomposed into radial velocity components and tangential velocity components. Based on the radial velocity components and tangential velocity components, the trajectory and velocity are jointly calculated to generate the target trajectory coordinates and velocity vector after interference resistance.

6. The method according to claim 1, characterized in that, The step of integrating the amplitude component, the polarization phase difference component, and the interference suppression residual component along the three coordinate axes of the orthogonal basis vectors to generate a three-dimensional feature tensor includes: Alignment operations are performed on the amplitude components along the principal axis of the energy response to generate a first axial characteristic component; Alignment operation is performed on the polarization phase difference components along the polarization difference axis to generate a second axial characteristic component; A mapping operation is performed on the interference suppression residual component along the interference residual separation axis to generate a third axial characteristic component; Perform a three-dimensional tensor synthesis operation on the first axial feature component, the second axial feature component, and the third axial feature component to generate a three-dimensional feature tensor.

7. A radar signal anti-jamming processing system based on a spatiotemporal attention mechanism, characterized in that, A radar signal anti-jamming processing method based on a spatiotemporal attention mechanism as described in any one of claims 1 to 6, comprising: The acquisition module is used to acquire the original horizontal polarization channel signal and the original vertical polarization channel signal of the radar; The mapping module is used to map the original horizontal polarization channel signal and the original vertical polarization channel signal to the interference suppression space to generate a three-dimensional feature tensor containing amplitude components, polarization phase difference components and interference suppression residual components. The partitioning module is used to divide the radar observation area into multiple target blocks in the spatial dimension according to the amplitude component, and to calculate the spatiotemporal correlation weight between adjacent target blocks through a spatiotemporal attention mechanism. The filtering module is used to filter out multi-target blocks that meet the preset continuous motion conditions according to the spatiotemporal correlation weights, and use a long short-term memory network to decode the motion mode of the multi-target blocks to generate anti-interference target trajectory coordinates and velocity vectors.

8. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement a radar signal anti-jamming processing method based on a spatiotemporal attention mechanism as described in any one of claims 1 to 6.

9. A computer storage medium, characterized in that, The system contains a computer program that, when executed by a computer, implements a radar signal anti-jamming processing method based on a spatiotemporal attention mechanism as described in any one of claims 1 to 6.

Images