Intelligent identification method of radar active jammer types based on multi-domain information fusion

Abstract

The present invention provides a radar active interference type intelligent identification method based on multi-domain information fusion, comprising: extracting a time-frequency distribution matrix of a radar active interference signal through a pseudo-Wigner distribution, and acquiring signal feature data of the signal at the same time; wherein the signal feature at least includes a frequency spectrum, a power spectrum, a real part and an imaginary part of the signal; analyzing and cropping a non-main feature area of ​​the time-frequency distribution matrix according to a preset ratio, and determining an image to be classified and identified; performing information fusion on the signal feature data of the signal and the image to be classified and identified, and acquiring a multi-domain information fusion matrix; and performing interference type identification on the multi-domain information fusion matrix based on a preset residual network classifier.

Description

Technical Field

[0001] The present invention relates to the technical field of radar signal processing, and in particular to a method for intelligently identifying radar active interference types based on multi-domain information fusion. Background Art

[0002] Traditional radar interference type recognition methods mainly include likelihood-based and feature-based methods. The former first establishes a mathematical model of active interference and calculates its likelihood function, and then determines the interference type by comparing it with a certain threshold. In order to distinguish the types of interference, many researchers have conducted in-depth research on extracting interference signal features based on time, frequency, time-frequency and other transform domains, and established relevant theories for analyzing radar interference features. However, the likelihood-based method requires mathematical modeling based on prior knowledge, has high computational complexity and limited application scope; the manual feature extraction in the feature-based method requires a lot of manpower and material resources, is expensive, and has poor timeliness.

[0003] In recent years, deep learning can extract features in a hierarchical manner. In deep learning models, CNN can automatically learn discriminative features and invariant features from data, and has achieved great success in many fields. In particular, CNN has also achieved certain success in the field of radar interference type recognition. Depending on the form of input data, the CNN models currently used for radar interference type recognition are mainly 1D-CNN and 2D-CNN. Since the time domain and frequency domain features of radar echo signals are clear and intuitive, it is a more common strategy to combine the one-dimensional features of the original signal with the 1D-CNN model for interference type recognition. Compared with 2D-CNN, 1D-CNN has obvious advantages in computational efficiency, but its application scenarios are limited to the processing of sequence data, and it is difficult to cope with the analysis of high-dimensional data.

[0004] 2D-CNN is the most typical convolutional neural network structure, which has significant advantages in image data processing. The active interference type recognition method based on 2D-CNN needs to convert the one-dimensional signal data into a two-dimensional form, and then implement effective feature extraction and type recognition. In existing research, there are two typical ways to realize the conversion of data from one-dimensional to two-dimensional form, namely data matrix conversion and time-frequency representation conversion. The former rearranges the one-dimensional feature data of the signal into a two-dimensional data matrix as the model input. The latter converts the one-dimensional echo signal into a two-dimensional time-frequency distribution (TFD) matrix for analysis, which can effectively extract the comprehensive feature information of the signal in the time domain and frequency domain, and is conducive to a more comprehensive distinction of radar active interference types. At present, many scholars and institutions are committed to the research of radar interference type recognition based on time-frequency analysis and deep learning. Using the 2D-CNN model to analyze TFD data can better realize the automatic extraction of radar active interference features, but the cost of obtaining rich information is a heavier computational burden.

[0005] In response to the above problems, this paper proposes a radar active jammer intelligent recognition model based on multi-domain information fusion through the research and analysis of common feature fusion and data enhancement methods. The aim is to give new information to the signal time-frequency distribution matrix through the 1D-CNN and 2D-CNN data matrix fusion method, thereby improving the accuracy of jammer type recognition. Summary of the invention

[0006] The present invention provides a radar active interference type intelligent identification method based on multi-domain information fusion to solve the above problems.

[0007] The present invention provides a method for intelligently identifying radar active interference types based on multi-domain information fusion, comprising:

[0008] Through the pseudo-Wigner distribution, the signal characteristic data of the signal is obtained, and at the same time, the time-frequency distribution matrix of the radar active interference signal is extracted; among them,

[0009] The signal characteristic data at least includes the frequency spectrum, power spectrum, real part and imaginary part of the signal;

[0010] Analyze and crop the non-main feature area of ​​the time-frequency distribution matrix according to a preset ratio, and determine the image to be classified and identified;

[0011] Performing information fusion on the signal feature data of the signal and the image to be classified and identified to obtain a multi-domain information fusion matrix;

[0012] Based on a preset residual network classifier, interference type identification is performed on the multi-domain information fusion matrix.

[0013] As an embodiment of the present technical solution, the obtaining of signal characteristic data of a signal through pseudo-Wigner distribution includes:

[0014] Receive a radar active interference signal x(t) through a radar receiver, and obtain a frequency spectrum of the radar active interference signal x(t) varying with time;

[0015]

[0016] Among them, |X t (ω)| 2 It represents the distribution of signal energy of radar active jammer x(t) in the frequency domain, where ω represents frequency;

[0017] Calculating the instantaneous power spectrum of the signal power through the frequency spectrum;

[0018]

[0019] Where t represents time, P x(t,ω) represents the distribution of signal power in the frequency domain at time t;

[0020] The received radar active jammer signal is divided into an orthogonal phase channel and an in-phase channel to obtain the real part and the imaginary part of the signal.

[0021] As an embodiment of the present technical solution, the received radar active interference signal is divided into an orthogonal phase channel and an in-phase channel, and the real part and the imaginary part of the signal are obtained, including:

[0022] In the in-phase channel, the received signal is mixed with the local oscillator signal, and the sum frequency component of the mixed received signal and the local oscillator signal is filtered out through a low-pass filter to obtain the real part of the signal;

[0023] In the orthogonal phase channel, the local oscillator signal is phase-shifted to a preset value, and the phase-shifted local oscillator signal is mixed with the received signal. The sum frequency component of the mixed received signal and the local oscillator signal is filtered out through a low-pass filter to obtain the imaginary part of the signal.

[0024] As an embodiment of the present technical solution, the time-frequency distribution matrix of the extracted radar active interference signal includes:

[0025] Extract the real part of the echo signal, fill it with zero, normalize it and transform it into a matrix to obtain a non-negative observation sample matrix U 1 ;

[0026] Extract the imaginary part of the echo signal, fill it with zero, normalize and transform the matrix to obtain the non-negative observation sample matrix U 2 ;

[0027] Calculate the frequency spectrum of the echo signal, normalize and transform the frequency spectrum, and obtain a non-negative observation sample matrix U 3 ;

[0028] Calculate the power spectrum of the echo signal, normalize and transform the power spectrum, and obtain the non-negative observation sample matrix U 4 ;

[0029] Based on the preset pseudo-Wigner distribution, process and calculate the non-negative observation sample matrix U 1 , non-negative observation sample matrix U 2 , non-negative observation sample matrix U 3 and the non-negative observation sample matrix U 4 , extract the time-frequency distribution matrix of radar active interference signal.

[0030] As an embodiment of the present technical solution, the information fusion of the signal feature data of the signal and the image to be classified and identified to obtain a multi-domain information fusion matrix includes:

[0031] Acquire a time-frequency distribution matrix, preprocess and normalize the time-frequency distribution matrix, and determine a processing matrix;

[0032] Through the image to be classified, a non-negative observation sample matrix is ​​collected; among them,

[0033] The non-negative observation sample matrix at least includes U 1 , U 2 , U 3 and U 4 ;

[0034] The processing matrix and non-negative observation sample matrix U 1 , U 2 , U 3 and U 4 Perform fusion, perform information fusion, and obtain a multi-domain information fusion matrix.

[0035] As an embodiment of the present technical solution, the processing matrix and the non-negative observation sample matrix U 1 , U 2 , U 3 and U 4 Perform fusion, perform information fusion, and obtain a multi-domain information fusion matrix, including:

[0036] The processing matrix and non-negative observation sample matrix U 1 Perform fusion and obtain the reconstructed matrix X 01 ;

[0037] The reconstructed matrix X 01 and the non-negative observation sample matrix U 2 Perform fusion and obtain the reconstructed matrix X 02 ;

[0038] The reconstructed matrix X 02 and the non-negative observation sample matrix U 3 Perform fusion and obtain the reconstructed matrix X 03 ;

[0039] The reconstructed matrix X 03 and the non-negative observation sample matrix U 4 Perform fusion and obtain the reconstructed matrix X 04 ;

[0040] For the reconstruction matrix X 01 , reconstruct the matrix X 02 , reconstruct the matrix X 03 and reconstruct the matrix X 04 Perform information fusion and obtain the multi-domain information fusion matrix.

[0041] As an embodiment of the present technical solution, the processing matrix and the non-negative observation sample matrix U 1 Perform fusion and obtain the reconstructed matrix X 01 ,include:

[0042] Splitting the processing matrix to obtain a first sub-fusion matrix;

[0043] Sub-matrix x 12 Reassign the non-negative observation sample matrix U 1 , determine the first target matrix;

[0044] The processing matrix and the first target matrix are merged and concatenated to determine the reconstruction matrix X 01 .

[0045] As an embodiment of the present technical solution, the reconstruction matrix X 01 and the non-negative observation sample matrix U 2 Perform fusion and obtain the reconstruction matrix, including:

[0046] The reconstructed matrix X 01 Perform segmentation to obtain the second sub-fusion matrix;

[0047] Through the preset hyperparameters, the reconstruction matrix X is controlled when cropping and patching 01 The distance between the sub-matrix x and the second sub-fusion matrix, and based on the distance, the sub-matrix x 12 Reassign the non-negative observation sample matrix U 2 , determine the first target matrix;

[0048] The processing matrix and the first target matrix are merged and concatenated to determine the reconstruction matrix X 02 .

[0049] As an embodiment of the present technical solution, the reconstruction matrix X 03 and the non-negative observation sample matrix U 4 Perform fusion and obtain the reconstructed matrix X 04 ,include:

[0050] The reconstructed matrix X 03 Perform segmentation to obtain the third sub-fusion matrix;

[0051] For the third sub-fusion matrix x 22 Reassign the non-negative observation sample matrix U 4 , and perform matrix reconstruction to obtain the reconstructed matrix X 04 .

[0052] As an embodiment of the present technical solution, the residual network classifier adopts ResNet18, ResNet50, WideResNet28_2 and WideResNet50_2.

[0053] Other features and advantages will be described in the following description, and partly become apparent from the description, or be understood by practicing the invention. The purpose and other advantages of the invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0054] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0056] In the attached picture:

[0057] Figure 1 It is a flow chart of a method for intelligently identifying radar active interference types based on multi-domain information fusion in an embodiment of the present invention;

[0058] Figure 2 It is a flow chart of a method for intelligently identifying radar active interference types based on multi-domain information fusion in an embodiment of the present invention;

[0059] Figure 3 It is a radar active interference characteristic diagram after multi-domain information fusion in an intelligent identification method of radar active interference type based on multi-domain information fusion in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0061] It should be noted that when a component is referred to as being "fixed to" or "disposed on" another component, it can be directly on the other component or indirectly on the other component. When a component is referred to as being "connected to" another component, it can be directly or indirectly connected to the other component.

[0062] It should be understood that the orientation or position relationship indicated by terms such as "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside" and "outside" are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0063] In addition, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations, and "multiple" means two or more, unless otherwise clearly and specifically limited. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0064] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

[0065] Embodiment 1:

[0066] according to Figure 1 As shown, an embodiment of the present invention provides a method for intelligently identifying radar active interference types based on multi-domain information fusion, including:

[0067] Through the pseudo-Wigner distribution, the signal characteristic data of the signal is obtained, and at the same time, the time-frequency distribution matrix of the radar active interference signal is extracted; among them,

[0068] The signal characteristic data at least includes the frequency spectrum, power spectrum, real part and imaginary part of the signal;

[0069] Analyze and crop the non-main feature area of ​​the time-frequency distribution matrix according to a preset ratio, and determine the image to be classified and identified;

[0070] Performing information fusion on the signal feature data of the signal and the image to be classified and identified to obtain a multi-domain information fusion matrix;

[0071] Based on a preset residual network classifier, interference type identification is performed on the multi-domain information fusion matrix.

[0072] The working principle and beneficial effects of the above technical solution are:

[0073] This technical solution proposes an intelligent identification method for radar active jammer types based on multi-domain information fusion, which is used to improve the accuracy of active jammer type identification under the condition of limited computing power and resource allocation of missile-borne radar. First, complete the mathematical modeling of radar active jammers and use simulation to generate signal samples to extract signal feature data; second, extract the time-frequency distribution characteristics of the signal and its one-dimensional features such as its real part, imaginary part, frequency spectrum and power spectrum; third, use the multi-domain information fusion algorithm to realize the fusion of the one-dimensional and two-dimensional features of the signal; finally, use the residual network classifier to perform radar active jammer type identification simulation. After multi-domain information fusion, the recognition rate of radar active jammer types has been greatly improved.

[0074] Embodiment 2:

[0075] The technical solution provides an embodiment, wherein the signal characteristic data of a signal is obtained through pseudo-Wigner distribution, including:

[0076] Receive a radar active interference signal x(t) through a radar receiver, and obtain a frequency spectrum of the radar active interference signal x(t) varying with time;

[0077]

[0078] Among them, |X t (ω)| 2 It represents the distribution of the signal energy of the radar active interference signal x(t) in the frequency domain τ, where ω represents the frequency;

[0079] Calculating the instantaneous power spectrum of the signal power through the frequency spectrum;

[0080]

[0081] Where t represents time, P x (t,ω) represents the distribution of signal power in the frequency domain at time t;

[0082] The received radar active jammer signal is divided into an orthogonal phase channel and an in-phase channel to obtain the real part and the imaginary part of the signal.

[0083] The working principle and beneficial effects of the above technical solution are:

[0084] The radar receiver receives the radar active interference signal, which is a random signal. Since the Fourier transform of the random signal does not converge, it is described by the power spectrum. t (ω)| 2 represents the distribution of signal energy in the frequency domain (until time t). Therefore, |Xt (ω)| 2 By taking the derivative with respect to time, we can obtain the distribution of the signal power in the frequency domain at time t, which is the instantaneous power spectrum.

[0085] Embodiment 3:

[0086] The technical solution provides an embodiment, wherein the received radar active interference signal is divided into an orthogonal phase channel and an in-phase channel, and the real part and the imaginary part of the signal are obtained, including:

[0087] In the in-phase channel, the received signal is mixed with the local oscillator signal, and the sum frequency component of the mixed received signal and the local oscillator signal is filtered out through a low-pass filter to obtain the real part of the signal;

[0088] In the orthogonal phase channel, the local oscillator signal is phase-shifted to a preset value, and the phase-shifted local oscillator signal is mixed with the received signal. The sum frequency component of the mixed received signal and the local oscillator signal is filtered out through a low-pass filter to obtain the imaginary part of the signal.

[0089] The working principle and beneficial effects of the above technical solution are:

[0090] When the radar receiver is processing, it is necessary to transform the information-carrying part of the radar signal to the baseband. The received signal is divided into two channels, the lower channel is the in-phase channel, and the upper channel is the quadrature channel. In the in-phase channel of the receiver, the received signal is mixed with the local oscillator signal to obtain a mixed output. After the low-pass filter removes the sum frequency component, only the real part of the modulation term is left. In the quadrature phase (Q) channel of the receiver, the received signal is mixed with the local oscillator signal, but the local oscillator signal needs to be phase-shifted by 90 before mixing to obtain a mixed output. After the low-pass filter removes the sum frequency component, only the modulation term and the imaginary part are left. The signals of the I and Q channels are analyzed at the same time, that is, the real and imaginary parts of the analytical signal are examined at the same time.

[0091] Embodiment 4:

[0092] The technical solution provides an embodiment, wherein the time-frequency distribution matrix of the extracted radar active interference signal includes:

[0093] Extract the real part of the echo signal, fill it with zero, normalize it and transform it into a matrix to obtain a non-negative observation sample matrix U 1 ;

[0094] Extract the imaginary part of the echo signal, fill it with zero, normalize and transform the matrix to obtain the non-negative observation sample matrix U 2 ;

[0095] Calculate the frequency spectrum of the echo signal, normalize and transform the frequency spectrum, and obtain a non-negative observation sample matrix U 3 ;

[0096] Calculate the power spectrum of the echo signal, normalize and transform the power spectrum, and obtain the non-negative observation sample matrix U 4 ;

[0097] Based on the preset pseudo-Wigner distribution, process and calculate the non-negative observation sample matrix U 1 , non-negative observation sample matrix U 2 , non-negative observation sample matrix U 3 and the non-negative observation sample matrix U 4 , extract the time-frequency distribution matrix of radar active interference signal.

[0098] Embodiment 5:

[0099] The technical solution provides an embodiment, wherein information fusion is performed on the signal feature data of the signal and the image to be classified and identified to obtain a multi-domain information fusion matrix, including:

[0100] Acquire a time-frequency distribution matrix, preprocess and normalize the time-frequency distribution matrix, and determine a processing matrix;

[0101] Through the image to be classified, a non-negative observation sample matrix is ​​collected; among them,

[0102] The non-negative observation sample matrix at least includes U 1 , U 2 , U 3 and U 4 ;

[0103] The processing matrix and non-negative observation sample matrix U 1 , U 2 , U 3 and U 4 Perform fusion, perform information fusion, and obtain a multi-domain information fusion matrix.

[0104] Embodiment 6:

[0105] The technical solution provides an embodiment of the present invention, wherein the processing matrix and the non-negative observation sample matrix U 1 , U 2 , U 3 and U 4 Perform fusion, perform information fusion, and obtain a multi-domain information fusion matrix, including:

[0106] The processing matrix and non-negative observation sample matrix U 1 Perform fusion and obtain the reconstructed matrix X 01 ;

[0107] The reconstructed matrix X 01 and the non-negative observation sample matrix U 2Perform fusion and obtain the reconstructed matrix X 02 ;

[0108] The reconstructed matrix X 02 and the non-negative observation sample matrix U 3 Perform fusion and obtain the reconstructed matrix X 03 ;

[0109] The reconstructed matrix X 03 and the non-negative observation sample matrix U 4 Perform fusion and obtain the reconstructed matrix X 04 ;

[0110] For the reconstruction matrix X 01 , reconstruct the matrix X 02 , reconstruct the matrix X 03 and reconstruct the matrix X 04 Perform information fusion and obtain the multi-domain information fusion matrix.

[0111] Embodiment 7:

[0112] The technical solution provides an embodiment of the present invention, wherein the processing matrix and the non-negative observation sample matrix U 1 Perform fusion and obtain the reconstructed matrix X 01 ,include:

[0113] Splitting the processing matrix to obtain a first sub-fusion matrix;

[0114] Sub-matrix x 12 Reassign the non-negative observation sample matrix U 1 , determine the first target matrix;

[0115] The processing matrix and the first target matrix are merged and concatenated to determine the reconstruction matrix X 01 .

[0116] Embodiment 8:

[0117] This technical solution provides an embodiment, wherein the reconstructed matrix X 01 and the non-negative observation sample matrix U 2 Perform fusion and obtain the reconstruction matrix, including:

[0118] The reconstructed matrix X 01 Perform segmentation to obtain the second sub-fusion matrix;

[0119] Through the preset hyperparameters, the reconstruction matrix X is controlled when cropping and patching 01 The distance between the sub-matrix x and the second sub-fusion matrix, and based on the distance, the sub-matrix x 12 Reassign the non-negative observation sample matrix U 2, determine the first target matrix;

[0120] The processing matrix and the first target matrix are merged and concatenated to determine the reconstruction matrix X 02 .

[0121] Embodiment 9:

[0122] This technical solution provides an embodiment in which the reconstructed matrix X 03 and the non-negative observation sample matrix U 4 Perform fusion and obtain the reconstructed matrix X 04 ,include:

[0123] The reconstructed matrix X 03 Perform segmentation to obtain the third sub-fusion matrix;

[0124] For the third sub-fusion matrix x 22 Reassign the non-negative observation sample matrix U 4 , and perform matrix reconstruction to obtain the reconstructed matrix X 04 .

[0125] Embodiment 10:

[0126] The present technical solution provides an embodiment, wherein the residual network classifier adopts ResNet18, ResNet50, WideResNet28_2 and WideResNet50_2.

[0127] The working principle and beneficial effects of the above technical solution are:

[0128] The radar active interference intelligent recognition model based on multi-domain information fusion can effectively improve the accuracy of radar active interference type recognition, which is improved by 0.47% under the ResNet18 classifier, 1.12% under the ResNet50 classifier, 0.23% under the Wi deResNet28_2 classifier, and 0.64% under the Wi deResNet50_2 classifier. The present invention has higher interference type recognition accuracy and high engineering application value.

[0129] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program code.

[0130] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0131] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0133] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for intelligent identification of radar active interference types based on multi-domain information fusion, characterized in that: include: The signal characteristic data of the signal is obtained through pseudo-Wigner distribution, and at the same time, the time-frequency distribution matrix of the radar active interference signal is extracted; wherein the signal characteristic data at least includes the frequency spectrum, power spectrum, real part and imaginary part of the signal; The time-frequency distribution matrix of the extracted radar active interference signal includes: Extract the real part of the echo signal, fill the real part with zero, normalize and transform the matrix, and obtain a non-negative observation sample matrix U1; Extract the imaginary part of the echo signal, fill the imaginary part with zero, normalize and transform the matrix, and obtain a non-negative observation sample matrix U2; Calculate the frequency spectrum of the echo signal, normalize and transform the frequency spectrum into a matrix, and obtain a non-negative observation sample matrix U3; Calculate the power spectrum of the echo signal, normalize and transform the power spectrum into a matrix, and obtain a non-negative observation sample matrix U4; Based on the preset pseudo-Wigner distribution, the non-negative observation sample matrix U1, the non-negative observation sample matrix U2, the non-negative observation sample matrix U3 and the non-negative observation sample matrix U4 are processed and calculated to extract the time-frequency distribution matrix of the radar active interference signal; Analyze and crop the non-main feature area of ​​the time-frequency distribution matrix according to a preset ratio, and determine the image to be classified and identified; Information fusion is performed on the signal feature data of the signal and the image to be classified and identified to obtain a multi-domain information fusion matrix, specifically including: Acquire a time-frequency distribution matrix, preprocess and normalize the time-frequency distribution matrix, and determine a processing matrix; Through the image to be classified, a non-negative observation sample matrix is ​​collected; among them, The non-negative observation sample matrix at least includes U1, U2, U3 and U4; The processing matrix and the non-negative observation sample matrix U1, U2, U3 and U4 are fused to perform information fusion and obtain a multi-domain information fusion matrix; The processing matrix and the non-negative observation sample matrices U1, U2, U3 and U4 are fused to perform information fusion to obtain a multi-domain information fusion matrix, including: The processing matrix and the non-negative observation sample matrix U1 are fused to obtain the reconstruction matrix X 01 ; The reconstructed matrix X 01 Fuse with the non-negative observation sample matrix U2 to obtain the reconstruction matrix X 02 ; The reconstructed matrix X 02 Fuse with the non-negative observation sample matrix U3 to obtain the reconstruction matrix X 03 ; The reconstructed matrix X 03 Fuse with the non-negative observation sample matrix U4 to obtain the reconstruction matrix X 04 ; For the reconstruction matrix X 01 , reconstruct the matrix X 02 , reconstruct the matrix X 03 and reconstruct the matrix X 04 Perform information fusion to obtain a multi-domain information fusion matrix; Based on a preset residual network classifier, interference type identification is performed on the multi-domain information fusion matrix.

2. The method for intelligently identifying the type of radar active interference based on multi-domain information fusion according to claim 1, characterized in that: The method of obtaining signal characteristic data of the signal through pseudo-Wigner distribution includes: Receive a radar active interference signal x(t) through a radar receiver, and obtain a frequency spectrum of the radar active interference signal x(t) varying with time; Among them, |X t (ω)| 2 It represents the distribution of signal energy of radar active jammer x(t) in the frequency domain, where ω represents frequency; Calculating the instantaneous power spectrum of the signal power through the frequency spectrum; Where t represents time, P x (t,ω) represents the distribution of signal power in the frequency domain at time t; The received radar active jammer signal is divided into an orthogonal phase channel and an in-phase channel to obtain the real part and the imaginary part of the signal.

3. The method for intelligently identifying the type of radar active interference based on multi-domain information fusion as claimed in claim 2, characterized in that: The received radar active jammer signal is divided into an orthogonal phase channel and an in-phase channel, and the real part and the imaginary part of the signal are obtained, including: In the in-phase channel, the received signal is mixed with the local oscillator signal, and the sum frequency component of the mixed received signal and the local oscillator signal is filtered out through a low-pass filter to obtain the real part of the signal; In the orthogonal phase channel, the local oscillator signal is phase-shifted to a preset value, and the phase-shifted local oscillator signal is mixed with the received signal. The sum frequency component of the mixed received signal and the local oscillator signal is filtered out through a low-pass filter to obtain the imaginary part of the signal.

4. The method for intelligently identifying radar active interference types based on multi-domain information fusion according to claim 1, characterized in that: The processing matrix and the non-negative observation sample matrix U1 are fused to obtain the reconstruction matrix X 01 ,include: Splitting the processing matrix to obtain a first sub-fusion matrix; Sub-matrix x 12 Reassign the non-negative observation sample matrix U1 and determine the first target matrix; The processing matrix and the first target matrix are merged and concatenated to determine the reconstruction matrix X 01 .

5. The method for intelligently identifying radar active interference types based on multi-domain information fusion according to claim 1, characterized in that: The reconstructed matrix X 01 Fuse with the non-negative observation sample matrix U2 to obtain the reconstruction matrix, including: The reconstructed matrix X 01 Perform segmentation to obtain the second sub-fusion matrix; Through the preset hyperparameters, the reconstruction matrix X is controlled when cropping and patching 01 The distance between the sub-fusion matrix and the second sub-fusion matrix, and based on the distance, the sub-matrix x 12 Reassign the non-negative observation sample matrix U2 and determine the first target matrix; The processing matrix and the first target matrix are merged and concatenated to determine the reconstruction matrix X 02 .

6. The method for intelligently identifying the type of radar active interference based on multi-domain information fusion according to claim 1, characterized in that: The reconstructed matrix X 03 Fuse with the non-negative observation sample matrix U4 to obtain the reconstruction matrix X 04 ,include: The reconstructed matrix X 03 Perform segmentation to obtain the third sub-fusion matrix; For the third sub-fusion matrix x 22 Reassign the non-negative observation sample matrix U4 and reconstruct the matrix to obtain the reconstructed matrix X 04 .

7. The method for intelligently identifying the type of radar active interference based on multi-domain information fusion according to claim 1, characterized in that: The residual network classifier adopts ResNet18, ResNet50, WideResNet28_2 and WideResNet50_2.

Images