Deep generative adversarial radar signal enhancement method and system oriented to low signal-to-noise ratio

Abstract

The invention provides a low signal-to-noise ratio-oriented deep generative adversarial radar signal enhancement method and system. The method comprises the following steps of: extracting target scattering statistical characteristics from a preset historical database; generating an emission polarization state configuration instruction set according to the target scattering statistical characteristics; according to the emission polarization state configuration instruction set, obtaining an emission signal with a set polarization state parameter corresponding to the optimal emission polarization state, and emitting the emission signal in a low signal-to-noise ratio environment; receiving an echo signal generated after the signal is transmitted, and generating a noise polarization state parameter; obtaining a polarization filtering signal based on the noise polarization state parameter and a set polarization state parameter; based on target scattering statistical characteristics, obtaining a reconstructed polarization filtering signal, and obtaining an enhanced radar signal; according to the technical scheme provided by the invention, deep generative adversarial network reconstruction fusing optimal polarization emission, adaptive polarization filtering and target scattering characteristics is realized, and the radar echo signal quality and detection performance are improved in a low signal-to-noise ratio environment.

Description

Technical Field

[0001] This application relates to the field of radar signal processing technology, and in particular to a method and system for enhancing deep generative adversarial radar signals for low signal-to-noise ratio. Background Technology

[0002] In low signal-to-noise ratio environments with strong clutter interference, radar detection faces the challenge of target echo signals being overwhelmed by intense noise and clutter. This is especially true when the target's scattering characteristics and background interference highly overlap in the time and frequency domains, making effective differentiation difficult. Therefore, there is an urgent need for a technique that can extract and enhance the inherent distinguishable features of targets from complex electromagnetic environments. This would improve the radar system's target detection, identification, and tracking capabilities under harsh conditions, ensuring the robustness and reliability of the sensing system.

[0003] To address these needs, current mainstream solutions propose radar echo enhancement methods based on deep neural networks. These methods construct training datasets containing echo samples from various typical targets and utilize convolutional neural networks to perform end-to-end feature learning and noise suppression on the original received signal, attempting to improve signal quality at the receiving end without altering the radar's transmitted waveform. However, existing solutions have significant drawbacks. For example, they do not fully utilize prior physical knowledge of the interaction between electromagnetic waves and targets, resulting in the network bearing an excessive burden of nonlinear reconstruction; the separability between targets and noise in the input signal to the convolutional neural network is low, making it difficult for the network to focus on refined reconstruction of target features. Summary of the Invention

[0004] This application provides a method and system for enhancing deep generative adversarial radar signals with low signal-to-noise ratio, in order to solve problems in the prior art such as insufficient network burden on nonlinear reconstruction, low separability between target and noise, and difficulty in fine-grained network reconstruction.

[0005] In a first aspect, this application provides a method for enhancing deep generative adversarial radar signals with low signal-to-noise ratio, including:

[0006] Extract target scattering statistical features from a pre-set historical database;

[0007] Based on the target scattering statistical characteristics, the optimal emission polarization state is calculated to generate an emission polarization state configuration instruction set;

[0008] According to the transmission polarization state configuration instruction set, the radar signal is subjected to carrier modulation processing to obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state and transmitted in a low signal-to-noise ratio environment;

[0009] The system receives the echo signal generated after transmitting the transmitted signal, and separates the horizontal polarization component and the vertical polarization component of the echo signal to generate noise polarization state parameters.

[0010] Based on the noise polarization state parameters and the set polarization state parameters, adaptive polarization filtering is performed on the echo horizontal polarization component and the echo vertical polarization component to obtain a polarization filtered signal.

[0011] Based on the target scattering statistical characteristics, the polarization-filtered signal is reconstructed using a deep generative adversarial network to obtain a reconstructed polarization-filtered signal. The reconstructed polarization-filtered signal is then verified, and the verified reconstructed polarization-filtered signal is used as an enhanced radar signal.

[0012] Optionally, based on the target scattering statistical characteristics, an optimal emission polarization state is calculated to generate an emission polarization state configuration instruction set, including:

[0013] The target scattering statistical characteristics are processed by polarization state to generate a polarization state distribution map containing multiple polarization states;

[0014] Calculate the matching degree between each candidate polarization state in the polarization state distribution map and the target scattering statistical features, and aggregate all matching degrees to obtain a set of polarization state matching degrees;

[0015] From the set of polarization state matching degrees, select the candidate polarization state corresponding to the matching degree with the largest value, and take the candidate polarization state as the optimal emission polarization state;

[0016] The amplitude values ​​of the horizontal component, the amplitude values ​​of the vertical component, the phase values ​​of the horizontal component, and the phase values ​​of the vertical component are resolved from the optimal emission polarization state to calculate the amplitude ratio parameter and the phase difference parameter.

[0017] Based on the preset instruction format requirements, the amplitude ratio parameter and the phase difference parameter are phase encoded to generate a transmit polarization state configuration instruction set containing polarization modulation parameters.

[0018] Optionally, the target scattering statistical characteristics are subjected to polarization state processing to generate a polarization state distribution map containing multiple polarization states, including:

[0019] The target scattering statistical characteristics are decomposed to obtain multiple eigenvalues, and the eigenvector corresponding to the eigenvalue with the largest value is taken as the target principal polarization scattering component.

[0020] The polarization ellipticity angle parameter and polarization orientation angle parameter of the target principal polarization scattering component are calculated to determine the polarization state space boundary range of the target scattering statistical characteristics;

[0021] According to the preset coordinate point generation rules, multiple coordinate points are generated within the boundary of the polarization state space, and all coordinate points are vector-transformed based on the preset transformation rules to obtain multiple polarization state description vectors.

[0022] According to the preset partitioning rules, all polarization state description vectors are divided into spatial grids to obtain a set of polarization state spatial grid cells;

[0023] Calculate the cell density of each cell in the polarization state space grid cell set, combine all cell densities to obtain polarization state density distribution data, and perform pseudo-color encoding on the polarization state density distribution data to obtain a polarization state distribution map containing multiple polarization states.

[0024] Optionally, according to the transmission polarization state configuration instruction set, the radar signal is subjected to carrier modulation processing to obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state, and transmitted in a low signal-to-noise ratio environment, including:

[0025] The transmit polarization state configuration instruction set is parsed to obtain amplitude ratio code and phase difference code, and the amplitude ratio code and phase difference code are decoded to obtain amplitude ratio setting value and phase difference setting value.

[0026] Carrier modulation is performed on the radar signal to obtain an initial carrier modulation signal, and orthogonal polarization separation processing is performed on the initial carrier modulation signal to obtain the horizontal polarization component and the vertical polarization component of the initial carrier modulation signal.

[0027] Based on the amplitude ratio setting value, the horizontal polarization component and the vertical polarization component are subjected to amplitude processing to obtain the scaled horizontal component and the magnified vertical component.

[0028] Based on the phase difference setting value, the phase offset adjustment is performed on the amplified vertical component to obtain a phase-matched vertical component. The scaled horizontal component and the phase-matched vertical component are then vector-synthesized to obtain a modulation signal with the optimal transmit polarization state.

[0029] The modulation signal and the preset pulse gating signal are time-domain superimposed to obtain a pulse modulation signal. The pulse modulation signal is then bandpass filtered to obtain a radar pulse transmission signal.

[0030] The radar pulse transmission signal is amplified to generate a transmission signal that meets the transmission power requirements and corresponds to the optimal transmission polarization state with set polarization parameters, and then transmitted in a low signal-to-noise ratio environment.

[0031] Optionally, the horizontal polarization component and the vertical polarization component of the echo signal are separated from the echo signal to generate noise polarization state parameters, including:

[0032] The echo signal is subjected to orthogonal polarization separation processing to obtain the horizontal polarization component and the vertical polarization component of the echo.

[0033] Using the set polarization state parameters corresponding to the optimal transmit polarization state, a reference signal component consistent with the polarization characteristics of the transmit signal is constructed. The echo horizontal polarization component and the echo vertical polarization component are coherently compared with the reference signal component to adjust the phase of the reference signal component and generate a matched horizontal reference component and a matched vertical reference component.

[0034] Remove the component in the horizontal polarization component of the echo that is consistent with the matched horizontal reference component to obtain the horizontal residual component; remove the component in the vertical polarization component of the echo that is consistent with the matched vertical reference component to obtain the vertical residual component.

[0035] Calculate the autocorrelation value of the horizontal residual component, the autocorrelation value of the vertical residual component, and the horizontal-vertical cross-correlation value between the horizontal residual component and the vertical residual component to construct the noise polarization characteristic matrix.

[0036] Based on the noise polarization characteristic matrix, the eigenvector corresponding to the largest eigenvalue is determined as the main noise polarization state, so as to extract the noise polarization state parameters from the main noise polarization state.

[0037] Optionally, based on the noise polarization state parameters and the set polarization state parameters, adaptive polarization filtering is performed on the echo horizontal polarization component and the echo vertical polarization component to obtain a polarization-filtered signal, including:

[0038] Based on the noise polarization state parameters and the set polarization state parameters, the noise location point and the target location point in the polarization state space are determined, and the spatial distance between the noise location point and the target location point is calculated. The spatial distance is used as the polarization state difference value.

[0039] Based on a preset intensity conversion rule, the polarization state difference value is quantized and converted to obtain the noise suppression intensity coefficient;

[0040] Based on the noise suppression intensity coefficient, the horizontal channel weight coefficient and the vertical channel weight coefficient are calculated to scale the echo horizontal polarization component and the echo vertical polarization component, so as to obtain the scaled echo horizontal component and the scaled echo vertical component.

[0041] The scaled horizontal and vertical components of the echo are vector-superimposed to obtain the polarized filtered signal.

[0042] Optionally, based on the target scattering statistical characteristics, the polarization-filtered signal is reconstructed using a deep generative adversarial network to obtain a reconstructed polarization-filtered signal. The reconstructed polarization-filtered signal is then verified, and the verified reconstructed polarization-filtered signal is used as an enhanced radar signal, including:

[0043] The target scattering statistical features are decomposed into a feature matrix to obtain the principal feature components. The principal feature components are then normalized to obtain the target feature constraint vector.

[0044] Based on the target feature constraint vector, the polarization filter signal is reconstructed by the generator of a deep generative adversarial network to obtain the reconstructed polarization filter signal.

[0045] The physical features of the reconstructed signal are extracted from the reconstructed polarization-filtered signal, and the feature distribution distance between the physical features of the reconstructed signal and the target scattering statistical features is calculated using a discriminator of a deep generative adversarial network.

[0046] When the feature distribution distance value is less than or equal to the preset verification threshold, the reconstructed polarization filter signal is determined to have passed verification, and the reconstructed polarization filter signal that has passed verification is used as the enhanced radar signal.

[0047] Alternatively, when the feature distribution distance value is greater than a preset verification threshold, the feature distribution distance value is used as a feedback error. Based on the feedback error, a target reconstructed polarized filter signal with a feature distribution distance value less than or equal to the preset verification threshold is reconstructed through a deep adversarial network. The target reconstructed polarized filter signal is then used as an enhanced radar signal.

[0048] Secondly, this application provides a deep generative adversarial radar signal enhancement system for low signal-to-noise ratio radar, comprising:

[0049] The extraction module is used to extract target scattering statistical features from a preset historical database;

[0050] The calculation module is used to calculate the optimal emission polarization state based on the target scattering statistical characteristics, so as to generate an emission polarization state configuration instruction set;

[0051] The modulation module is used to configure the instruction set according to the transmission polarization state, perform carrier modulation processing on the radar signal, obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state, and transmit it in a low signal-to-noise ratio environment;

[0052] The separation module is used to receive the echo signal generated after the transmitted signal is transmitted, and to separate the horizontal polarization component and the vertical polarization component of the echo signal from the echo signal to generate noise polarization state parameters.

[0053] The filtering module is used to perform adaptive polarization filtering on the echo horizontal polarization component and the echo vertical polarization component based on the noise polarization state parameters and the set polarization state parameters to obtain a polarization filtered signal.

[0054] The verification module is used to reconstruct the polarization-filtered signal based on the target scattering statistical characteristics using a deep generative adversarial network to obtain a reconstructed polarization-filtered signal, and to verify the reconstructed polarization-filtered signal, using the verified reconstructed polarization-filtered signal as an enhanced radar signal.

[0055] Thirdly, this application provides a computing device, including 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 method for enhancing deep generative adversarial radar signals with low signal-to-noise ratio as described in the first aspect above.

[0056] Fourthly, this application provides a computer storage medium storing a computer program, which, when executed by a computer, implements a method for enhancing deep generative adversarial radar signals with low signal-to-noise ratio as described in the first aspect.

[0057] In this application, target scattering statistical features are extracted from a preset historical database; based on the target scattering statistical features, an optimal transmission polarization state is calculated to generate a transmission polarization state configuration instruction set; based on the transmission polarization state configuration instruction set, the radar signal is subjected to carrier modulation processing to obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state, and transmitted in a low signal-to-noise ratio environment; the echo signal generated after the transmission signal is received, and the echo horizontal polarization component and echo vertical polarization component are separated from the echo signal to generate noise polarization state parameters; based on the noise polarization state parameters and the set polarization state parameters, the echo horizontal polarization component and echo vertical polarization component are subjected to adaptive polarization filtering processing to obtain a polarization filtered signal; based on the target scattering statistical features, the polarization filtered signal is reconstructed using a deep generative adversarial network to obtain a reconstructed polarization filtered signal, and the reconstructed polarization filtered signal is verified, with the verified reconstructed polarization filtered signal used as an enhanced radar signal. The technical solution provided in this application acquires the electromagnetic scattering behavior of targets under different observation conditions, providing physically meaningful prior knowledge support for subsequent signal processing. It achieves active matching between the polarization state of the transmitted signal and the target characteristics, enhancing the target echo energy and suppressing unwanted scattering responses. This enhances the distinguishability of the target from the background at the signal source. Real-time perception of the polarization characteristics of the interference environment is achieved, providing key parameter basis for subsequent interference suppression. A polarization domain filter can be dynamically constructed to suppress clutter and noise that do not match the transmitted polarization while retaining the target-related signal components, thus improving the signal-to-clutter ratio of the echo signal. Integrating physical prior knowledge into the deep learning process allows the network to focus more on the fine recovery of target features, improving the accuracy and robustness of signal reconstruction; ensuring the reliability of the enhanced signal, and ultimately obtaining more resolving radar observation data. Furthermore, this application generates a distribution map covering multiple polarization modes, quantifies the matching degree between each candidate polarization state and the target characteristics, and selects the optimal transmission polarization state from it; it further analyzes its key parameters such as amplitude ratio and phase difference, and encodes them according to the standard instruction format to form an executable polarization modulation instruction set, thereby realizing accurate mapping and automated configuration from target characteristics to transmission waveform parameters.

[0058] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 A flowchart of a deep generative adversarial radar signal enhancement method for low signal-to-noise ratio provided in this application is shown;

[0061] Figure 2 A schematic diagram of the structure of a deep generative adversarial radar signal enhancement system for low signal-to-noise ratio provided in this application is shown;

[0062] Figure 3 A schematic diagram of the structure of a computing device provided in this application is shown. Detailed Implementation

[0063] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0064] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0065] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0066] To address the problem of difficulty in distinguishing target features in low signal-to-noise ratio radar signals under strong clutter interference, this application proposes a signal enhancement method that integrates physical priors and deep learning. This method overcomes the limitations of existing pure receiver-side deep neural network enhancement schemes, which suffer from poor target and interference separability and heavy network reconstruction burden due to neglecting transmit control and physical interaction priors. It separates and suppresses mismatched interference, improves the signal quality of the input deep network, and ensures output reliability through a verification mechanism. A closed-loop processing chain is constructed to enhance the signal enhancement effect and robustness in complex electromagnetic environments.

[0067] Figure 1This application provides a flowchart of a deep generative adversarial radar signal enhancement method for low signal-to-noise ratio embodiments, as shown below. Figure 1 As shown, the method includes:

[0068] Step 101: Extract target scattering statistical features from the preset historical database.

[0069] In this step, the preset historical database refers to a structured repository that stores historical radar observation data. Target scattering statistical characteristics refer to the statistically quantified scattering characteristics extracted from the historical database.

[0070] In this embodiment of the application, a historically stored polarization scattering matrix dataset is invoked, and a statistical averaging operation is performed on the scattering characteristics of all samples in the dataset to generate a three-dimensional coherence matrix containing the average scattering characteristics of the target as the target scattering statistical feature.

[0071] Step 102: Calculate the optimal emission polarization state based on the target scattering statistical characteristics to generate an emission polarization state configuration instruction set.

[0072] In this step, the optimal transmit polarization state refers to the electromagnetic wave polarization state that maximizes the signal-to-noise ratio of the target echo. The transmit polarization state configuration instruction set refers to the machine-executable instructions that control the polarization state of the radar transmitter.

[0073] Step 103: According to the transmission polarization state configuration instruction set, perform carrier modulation processing on the radar signal to obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state and transmit it in a low signal-to-noise ratio environment.

[0074] In this step, the radar signal refers to the frequency-modulated continuous wave baseband signal. Setting the polarization state parameters refers to the combination of physical parameters corresponding to the optimal transmit polarization state. The transmitted signal refers to the radio frequency electromagnetic wave that has undergone polarization modulation and pulse shaping.

[0075] Step 104: Receive the echo signal generated after transmitting the transmitted signal, and separate the horizontal polarization component and the vertical polarization component of the echo signal to generate noise polarization state parameters.

[0076] In this step, the echo signal refers to the received signal generated after the transmitted signal is reflected by the target. The horizontal polarization component of the echo refers to the time-domain waveform data of the echo signal in the horizontal polarization channel. The vertical polarization component of the echo refers to the time-domain waveform data of the echo signal in the vertical polarization channel. The noise polarization state parameter refers to the descriptive quantity of the polarization characteristics of the ambient noise.

[0077] Step 105: Based on the noise polarization state parameters and the set polarization state parameters, perform adaptive polarization filtering on the echo horizontal polarization component and the echo vertical polarization component to obtain the polarization filtered signal.

[0078] In this step, the polarization-filtered signal refers to the output signal after adaptive polarization suppression processing, in which its noise components are selectively attenuated.

[0079] Step 106: Based on the target scattering statistical characteristics, the polarization filter signal is reconstructed using a deep generative adversarial network to obtain a reconstructed polarization filter signal. The reconstructed polarization filter signal is then verified, and the verified reconstructed polarization filter signal is used as an enhanced radar signal.

[0080] In this step, a deep generative adversarial network (GAN) refers to a neural network architecture that includes a conditional generator and a feature discriminator. The reconstructed polarization-filtered signal refers to the time-domain signal reconstructed through the generator network, which preserves the target scattering characteristics and suppresses noise. The enhanced radar signal refers to the output signal that has passed the final verification.

[0081] This application's embodiments improve target energy focusing by dynamically optimizing the emission polarization state based on historical scattering characteristics; accurately suppress environmental noise polarization components through dual-channel adaptive filtering; reconstruct target scattering characteristics using a conditional constraint generation network; and finally ensure the physical consistency of the output signal through statistical feature verification.

[0082] This application provides a specific embodiment. Step 102 involves calculating the optimal emission polarization state based on the target scattering statistical characteristics to generate an emission polarization state configuration instruction set, specifically including the following steps:

[0083] Step 201: Perform polarization state processing on the target scattering statistical characteristics to generate a polarization state distribution map containing multiple polarization states.

[0084] In this step, the polarization state distribution map refers to a two-dimensional heatmap that visualizes the probability of occurrence of each state point in the polarization state space. The horizontal axis is the polarization ellipticity angle ranging from -45 degrees to +45 degrees, the vertical axis is the polarization orientation angle ranging from 0 degrees to 180 degrees, and the color depth reflects the matching probability with the target scattering characteristics.

[0085] In this embodiment, eigenvalue decomposition is performed on the three-dimensional coherence matrix of the target scattering statistical characteristics to extract the eigenvector corresponding to the largest eigenvalue as the target principal polarization scattering component. The polarization ellipticity angle parameter and polarization orientation angle parameter of this component are calculated. Based on these two parameters, the polarization state space boundary range is determined. Within this boundary range, a set of candidate polarization state coordinates is generated at 0.5 degree intervals. Each coordinate point is converted into a polarization state description vector in Stokes vector form. Finally, kernel density estimation is performed on all description vectors to generate a polarization state distribution map.

[0086] Step 202: Calculate the matching degree between each candidate polarization state and the target scattering statistical features in the polarization state distribution map, and aggregate all matching degrees to obtain a set of polarization state matching degrees.

[0087] In this step, candidate polarization states refer to discretely sampled state points in the polarization state distribution map. Each point is uniquely determined by its ellipticity angle and orientation angle coordinates, representing a possible emission polarization state. The matching degree refers to the quantified similarity value between a single candidate polarization state and the target's scattering statistical characteristics. The polarization state matching degree set refers to the data structure that stores the matching degrees of all candidate polarization states.

[0088] In this embodiment of the application, the matching degree between each candidate polarization state and the target scattering statistical feature in the polarization state distribution map is calculated by the polarization matching algorithm. The calculation formula is: matching degree = correlation coefficient between candidate polarization state and target scattering statistical feature ÷ product of their magnitudes. All the calculated matching degrees are then integrated to obtain a set of polarization state matching degrees.

[0089] Step 203: Select the candidate polarization state with the largest matching degree from the set of polarization state matching degrees, and take the candidate polarization state as the optimal emission polarization state.

[0090] In this step, the optimal emission polarization state refers to the polarization state corresponding to the highest matching degree selected from the matching degree set.

[0091] In this embodiment, the values ​​in the matching degree set are traversed, the index number corresponding to the maximum value is identified, the candidate polarization state point in the polarization state distribution map is located according to the index number, and the mathematical coordinates of the candidate polarization state point are converted into physical polarization state parameters, namely the combination of ellipticity angle and orientation angle, as the optimal emission polarization state.

[0092] Step 204: Extract the horizontal component amplitude value, vertical component amplitude value, horizontal component phase value, and vertical component phase value from the optimal emission polarization state to calculate the amplitude ratio parameter and phase difference parameter.

[0093] In this step, the horizontal component amplitude value refers to the amplitude of the horizontal channel in the complex representation of the optimal emitter polarization state. The vertical component amplitude value refers to the amplitude of the vertical channel in the complex representation of the optimal emitter polarization state. The horizontal component phase value refers to the phase angle of the horizontal channel in the optimal emitter polarization state. The vertical component phase value refers to the phase angle of the vertical channel in the optimal emitter polarization state. The amplitude ratio parameter refers to the ratio of the horizontal component amplitude value to the vertical component amplitude value. The phase difference parameter refers to the angular difference between the horizontal component phase value and the vertical component phase value.

[0094] In this embodiment, the horizontal component amplitude value, vertical component amplitude value, horizontal component phase value, and vertical component phase value are extracted from the optimal emission polarization state using signal analysis technology. Then, the amplitude ratio parameter is calculated using the formula: Amplitude ratio parameter = Horizontal component amplitude value ÷ Vertical component amplitude value. Finally, the phase difference parameter is calculated using the formula: Phase difference parameter = Horizontal component phase value - Vertical component phase value.

[0095] Step 205: Based on the preset instruction format requirements, perform phase encoding on the amplitude ratio parameter and the phase difference parameter to generate a transmit polarization state configuration instruction set containing polarization modulation parameters.

[0096] In this step, the preset command format requires that the command specifications be parseable by the radar transmitter.

[0097] In this embodiment, based on the instruction format requirements, the amplitude ratio parameter is quantized into a 12-bit binary code, the phase difference parameter is quantized into an 8-bit binary code, an 8-bit instruction header identifier and an 8-bit checksum are added, and combined into a 36-bit instruction frame as a transmit polarization state configuration instruction set containing polarization modulation parameters.

[0098] This application's embodiments visualize the feature matching probability distribution through polarization state distribution maps; quantitative evaluation of matching metrics enables scientific optimization of the emission state; parameter parsing ensures physical executability; and standardized instruction coding ensures device compatibility. The entire process forms a lossless conversion chain from feature analysis to hardware control.

[0099] This application provides a specific embodiment. Step 201 involves polarization state processing of the target scattering statistical characteristics to generate a polarization state distribution map containing multiple polarization states. This specifically includes the following steps:

[0100] Step 211: Decompose the target scattering statistical features to obtain multiple feature values, and take the feature vector corresponding to the feature value with the largest value as the target principal polarization scattering component.

[0101] In this step, eigenvalues ​​refer to the scalar eigenvalues ​​obtained after eigenvalue decomposition of the coherence matrix, reflecting the primary and secondary distribution relationships of scattered energy. The largest eigenvalue corresponds to the primary scattered energy. The target principal polarization scattering component refers to the eigenvector corresponding to the largest eigenvalue.

[0102] In this embodiment, the Jacobian eigenvalue decomposition algorithm is performed on the matrix of the target scattering statistical characteristics to calculate three eigenvalues. The eigenvalue with the largest value is selected, and its corresponding eigenvector is identified as the target principal polarization scattering component.

[0103] Step 212: Calculate the polarization ellipticity angle parameter and polarization orientation angle parameter of the target principal polarization scattering component to determine the polarization state space boundary range of the target scattering statistical characteristics.

[0104] In this step, the polarization ellipticity angle parameter refers to the angular parameter describing the flattening of the polarization ellipse. The polarization orientation angle parameter refers to the angular parameter describing the direction of the principal axis of the polarization ellipse. The polarization state space boundary range refers to the effective region used to generate candidate polarization states.

[0105] In this embodiment, the real and imaginary parts of the horizontal and vertical components of the feature vector are extracted, and the polarization ellipticity angle parameter and polarization orientation angle parameter of the target principal polarization scattering component are calculated to determine the polarization state space boundary range.

[0106] Step 213: Generate multiple coordinate points within the polarization state space boundary according to the preset coordinate point generation rules, and perform vector transformation on all coordinate points based on the preset transformation rules to obtain multiple polarization state description vectors.

[0107] In this step, the preset coordinate point generation rule refers to the rule of sampling at fixed angular intervals within the boundary range. The polarization state description vector refers to a three-dimensional vector in Stokes parameter form, used to quantitatively characterize the polarization state.

[0108] In this embodiment of the application, a coordinate point grid is generated according to a preset coordinate point generation rule, and a Stokes vector transformation operation is performed on each coordinate point based on a preset transformation rule to obtain multiple polarization state description vectors.

[0109] Step 214: According to the preset partitioning rules, divide all polarization state description vectors into spatial grids to obtain a set of polarization state spatial grid cells.

[0110] In this step, the preset partitioning rule refers to the rule for dividing the polarization state space. The set of polarization state space grid cells refers to the set of grid containers covering the entire polarization state space, and each cell has a unique spatial location identifier.

[0111] In this embodiment of the application, the polarization state space is divided into multiple grid cells according to a preset partitioning rule. Each grid cell covers a certain rectangular area, resulting in a set of polarization state space grid cells containing all grid cells.

[0112] Step 215: Calculate the cell density of each cell in the polarization state space grid cell set, combine all cell densities to obtain polarization state density distribution data, and perform pseudo-color encoding on the polarization state density distribution data to obtain a polarization state distribution map containing multiple polarization states.

[0113] In this step, the cell density of a grid cell refers to the number of polarization state description vectors falling within a single grid cell, reflecting the frequency of polarization states occurring at that spatial location. The polarization state density distribution data refers to a two-dimensional matrix data structure that stores the density values ​​of all grid cells.

[0114] In this embodiment, the number of polarization state description vectors falling into each grid cell is counted as the cell density. All cell densities are combined into polarization state density distribution data in the form of a two-dimensional density matrix. A preset pseudo-color mapping rule is applied to convert the density values ​​into red, green, and blue color values ​​to generate a pseudo-color coded image as a polarization state distribution map.

[0115] This application's embodiments accurately extract the main scattering mode through feature decomposition; establish a bridge between physical properties and mathematical representation through elliptic parameter calculation; quantify the state distribution law through gridded density statistics; and provide an intuitive analytical view through pseudo-color encoding. The entire process forms a complete transformation chain from abstract features to spatial distribution.

[0116] This application provides a specific embodiment. Step 103 involves performing carrier modulation processing on the radar signal according to the transmission polarization state configuration instruction set to obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state, and transmitting it in a low signal-to-noise ratio environment. This specifically includes the following steps:

[0117] Step 301: Parse the transmit polarization state configuration instruction set to obtain the amplitude ratio code and phase difference code, and perform phase decoding operation on the amplitude ratio code and phase difference code to obtain the amplitude ratio setting value and phase difference setting value.

[0118] In this step, the amplitude ratio code refers to the radix of the horizontal-to-vertical amplitude ratio in the instruction set. The phase difference code refers to the eight-bit binary number representing the phase difference between the two channels in the instruction set. The amplitude ratio setting value refers to the physical amplitude ratio obtained after dequantization. The phase difference setting value refers to the physical phase difference value obtained after decoding.

[0119] In this embodiment of the application, the transmit polarization state configuration instruction set is parsed by an instruction parser to extract the amplitude ratio code and phase difference code. Then, the phase decoding algorithm is used to decode these two codes to obtain the amplitude ratio setting value and phase difference setting value for subsequent processing.

[0120] Step 302: Carrier modulation is performed on the radar signal to obtain an initial carrier modulation signal, and orthogonal polarization separation processing is performed on the initial carrier modulation signal to obtain the horizontal polarization component and the vertical polarization component of the initial carrier modulation signal.

[0121] In this step, the initial carrier modulation signal refers to the radio frequency continuous wave generated by up-conversion of the baseband signal. The horizontal polarization component refers to the time-domain waveform of the initial carrier modulation signal in the horizontal polarization channel, with the electric field vector parallel to the horizontal dipole of the antenna. The vertical polarization component refers to the time-domain waveform of the initial carrier modulation signal in the vertical polarization channel.

[0122] In this embodiment, the radar signal is input to the quadrature upconverter and mixed with the local oscillator to generate a radio frequency carrier signal as the initial carrier modulation signal. The initial carrier modulation signal is decomposed into horizontal polarization components and vertical polarization components by a hybrid coupler.

[0123] Step 303: According to the amplitude ratio setting value, perform amplitude processing on the horizontal polarization component and the vertical polarization component to obtain the scaled horizontal component and the magnified vertical component.

[0124] In this step, the scaled horizontal component refers to the horizontally polarized signal processed by the digital attenuator. The amplified vertical component refers to the vertically polarized signal processed by the programmable amplifier.

[0125] In this embodiment, the amplitude ratio setting value is input to the horizontal channel digital attenuator to perform signal attenuation operation on the horizontal polarization component to generate a scaled horizontal component, and the reciprocal of the amplitude ratio setting value is input to the vertical channel programmable gain amplifier to perform signal amplification operation on the vertical polarization component to generate an amplified vertical component.

[0126] Step 304: According to the phase difference setting value, the phase offset adjustment is performed on the amplified vertical component to obtain the phase-matched vertical component. The scaled horizontal component and the phase-matched vertical component are vector synthesized to obtain the modulation signal with the optimal transmit polarization state.

[0127] In this step, the phase-matched vertical component refers to the vertically polarized signal adjusted by the digital phase shifter. The modulated signal refers to the composite signal of the scaled horizontal component and the phase-matched vertical component, whose instantaneous polarization state is equal to the optimal emitter polarization state.

[0128] In this embodiment, a phase difference setting value is loaded into a digital phase shifter to perform a phase shift operation on the amplified vertical component to generate a phase-matched vertical component. The scaled horizontal component and the phase-matched vertical component are then vector-superimposed by a dual-channel power combiner to generate a modulation signal with the optimal transmit polarization state.

[0129] Step 305: Perform time-domain superposition processing on the modulation signal and the preset pulse gating signal to obtain a pulse modulation signal, and perform bandpass filtering processing on the pulse modulation signal to obtain a radar pulse transmission signal.

[0130] In this step, the preset pulse gating signal refers to a periodic rectangular pulse sequence. The pulse modulation signal refers to the time-domain envelope signal generated by multiplying the modulation signal and the pulse gating signal, which has continuous waveform characteristics with pulse interruptions. The radar pulse transmission signal refers to a compliant radio frequency signal processed by bandpass filtering.

[0131] In this embodiment, a pulse modulation signal is generated by performing a time-domain multiplication operation between the modulation signal and a preset pulse gating signal using an analog multiplier, and a surface acoustic wave filter is used to perform bandpass filtering on the pulse modulation signal to generate a radar pulse transmission signal.

[0132] Step 306: Amplify the power of the radar pulse transmission signal to generate a transmission signal that meets the transmission power requirements and corresponds to the set polarization parameters of the optimal transmission polarization state, and transmit it in a low signal-to-noise ratio environment.

[0133] In this step, meeting the transmit power requirement means that the output signal of the power amplifier reaches the minimum radiated power threshold of the radar system.

[0134] In this embodiment, the radar pulse transmission signal is input into a gallium nitride power amplifier to perform a power boosting operation, and the output radio frequency signal is used as a transmission signal with set polarization parameters. The signal is then transmitted through a dual-polarized microstrip antenna array in a low signal-to-noise ratio environment.

[0135] This application's embodiments ensure zero-error parameter conversion through instruction parsing; optimize polarization purity through dual-channel independent modulation; precisely control pulse waveforms through time-domain gating; and overcome path loss due to low signal-to-noise ratio through power enhancement. The entire process achieves high-fidelity conversion from digital instructions to physical radiation.

[0136] This application provides a specific embodiment. Step 104 involves separating the horizontal polarization component and the vertical polarization component of the echo signal to generate noise polarization state parameters. This specifically includes the following steps:

[0137] Step 401: Perform orthogonal polarization separation processing on the echo signal to obtain the horizontal polarization component and the vertical polarization component of the echo.

[0138] In this embodiment, an orthogonal polarization separator is used to decompose the echo signal into a horizontal polarization component and a vertical polarization component by means of filtering and phase separation techniques, utilizing the orthogonal characteristics of the polarization signal in the horizontal and vertical directions.

[0139] Step 402: Using the set polarization state parameters corresponding to the optimal transmit polarization state, construct a reference signal component that is consistent with the polarization characteristics of the transmit signal, and perform a coherent comparison between the echo horizontal polarization component and the echo vertical polarization component and the reference signal component to adjust the phase of the reference signal component, thereby generating a matched horizontal reference component and a matched vertical reference component.

[0140] In this step, the transmitted signal polarization characteristics refer to the inherent properties of the transmitted signal in the polarization dimension, used to reflect the polarization state of the transmitted signal. The reference signal component refers to a signal component constructed based on the set polarization state parameters of the optimal transmitted polarization state, consistent with the transmitted signal polarization characteristics, used as a benchmark for comparison with the echo polarization component to distinguish the target signal from noise signals. The matched horizontal reference component refers to the horizontal reference component that matches the echo horizontal polarization component in phase after coherent comparison and phase adjustment, used for subsequent removal of the target signal portion from the echo horizontal polarization component. The matched vertical reference component refers to the vertical reference component that matches the echo vertical polarization component in phase after coherent comparison and phase adjustment, used for subsequent removal of the target signal portion from the echo vertical polarization component.

[0141] In this embodiment, using the set polarization state parameters corresponding to the optimal transmit polarization state, a reference signal component consistent with the polarization characteristics of the transmit signal is constructed through signal synthesis technology. Then, the echo horizontal polarization component and the echo vertical polarization component are coherently compared with the reference signal component through a cross-correlation algorithm. The phase of the reference signal component is adjusted according to the comparison results to generate a matched horizontal reference component that matches the phase of the echo horizontal polarization component and a matched vertical reference component that matches the phase of the echo vertical polarization component.

[0142] Step 403: Remove the component of the echo horizontal polarization component that is consistent with the matched horizontal reference component to obtain the horizontal residual component; remove the component of the echo vertical polarization component that is consistent with the matched vertical reference component to obtain the vertical residual component.

[0143] In this step, the horizontal residual component refers to the component remaining after removing the component in the horizontal polarization component of the echo that matches the matched horizontal reference component. The vertical residual component refers to the component remaining after removing the component in the vertical polarization component of the echo that matches the matched vertical reference component.

[0144] In this embodiment, signal subtraction is used to remove the component in the horizontal polarization component of the echo that is consistent with the matched horizontal reference component, so as to obtain the horizontal residual component after removing the horizontal polarization part of the target signal; similarly, the component in the vertical polarization component of the echo that is consistent with the matched vertical reference component is removed, so as to obtain the vertical residual component after removing the vertical polarization part of the target signal.

[0145] Step 404: Calculate the autocorrelation value of the horizontal residual component, the autocorrelation value of the vertical residual component, and the horizontal-vertical cross-correlation value between the horizontal residual component and the vertical residual component to construct the noise polarization characteristic matrix.

[0146] In this step, the horizontal component autocorrelation value refers to the time integral of the horizontal residual component with its own delayed signal, reflecting the correlation and stability of the horizontal residual component itself, and is an important element in constructing the noise polarization characteristic matrix. The vertical component autocorrelation value refers to the time integral of the vertical residual component with its own delayed signal, reflecting the correlation and stability of the vertical residual component itself, and is an important element in constructing the noise polarization characteristic matrix. The horizontal-vertical cross-correlation value refers to the time integral of the horizontal residual component with the delayed signal of the vertical residual component, reflecting the correlation between the horizontal and vertical residual components, and is an important element in constructing the noise polarization characteristic matrix. The noise polarization characteristic matrix is ​​a matrix formed by arranging the horizontal component autocorrelation values, vertical component autocorrelation values, and horizontal-vertical cross-correlation values ​​according to a preset structure, used to comprehensively describe the characteristics of noise in the polarization dimension.

[0147] In this embodiment of the application, the autocorrelation value of the horizontal component of the horizontal residual component, the autocorrelation value of the vertical component of the vertical residual component, and the horizontal and vertical cross-correlation value between the horizontal and vertical residual components are calculated by autocorrelation and cross-correlation calculation methods. Then, these three values ​​are arranged according to a preset matrix structure to construct a noise polarization characteristic matrix.

[0148] Step 405: Based on the noise polarization characteristic matrix, determine the eigenvector corresponding to the largest eigenvalue as the main noise polarization state, so as to extract the noise polarization state parameters from the main noise polarization state.

[0149] In this step, the main noise polarization state refers to the eigenvector corresponding to the largest eigenvalue in the noise polarization characteristic matrix. This vector best reflects the main polarization state of the noise and is the dominant manifestation of noise in the polarization space.

[0150] In this embodiment, the noise polarization characteristic matrix is ​​decomposed using an eigenvalue decomposition algorithm to obtain multiple eigenvalues ​​and corresponding eigenvectors. The eigenvector corresponding to the largest eigenvalue is determined as the main noise polarization state, and then the noise polarization state parameters reflecting the noise polarization characteristics are extracted from the main noise polarization state.

[0151] This application embodiment achieves accurate characterization of noise polarization state by performing orthogonal polarization separation on the echo signal, constructing a reference signal component and performing phase matching, removing the target signal component to obtain the residual component, calculating the correlation value to construct the noise polarization characteristic matrix, and finally extracting the main noise polarization state parameters. This provides a reliable noise characteristic basis for subsequent adaptive polarization filtering and effectively improves the accuracy of noise separation in radar signals under low signal-to-noise ratio environments.

[0152] This application provides a specific embodiment. Step 105 involves performing adaptive polarization filtering on the echo horizontal polarization component and the echo vertical polarization component based on the noise polarization state parameters and the set polarization state parameters to obtain a polarization-filtered signal. This specifically includes the following steps:

[0153] Step 501: Based on the noise polarization state parameters and the set polarization state parameters, determine the noise location point and the target location point in the polarization state space, calculate the spatial distance between the noise location point and the target location point, and use the spatial distance as the polarization state difference value.

[0154] In this step, the noise location point refers to the specific spatial location in the polarization state space used to characterize the polarization state of the noise, reflecting the polarization characteristics of the noise. The target location point refers to the specific spatial location in the polarization state space used to characterize the polarization state of the target signal, reflecting the polarization characteristics of the target signal. Spatial distance refers to the straight-line distance between the noise location point and the target location point in the polarization state space, used to quantify the physical separation between them in the polarization space. The polarization state difference value refers to using the spatial distance between the noise location point and the target location point as a numerical value to characterize the degree of difference in their polarization characteristics.

[0155] In this embodiment, based on the noise polarization state parameters and the set polarization state parameters, the noise polarization state is mapped to a noise location point in the polarization state space using polarization state space modeling technology, and the set polarization state parameters are mapped to a target location point. Then, the spatial distance between the noise location point and the target location point is calculated using the Euclidean distance algorithm, and this spatial distance is used as the polarization state difference value that characterizes the difference in polarization characteristics between the two.

[0156] Step 502: Based on the preset intensity conversion rule, the polarization state difference value is quantized and converted to obtain the noise suppression intensity coefficient.

[0157] In this step, the preset intensity conversion rule refers to the pre-defined mapping relationship between polarization state difference values ​​and noise suppression intensity coefficients, used to quantitatively convert noise suppression intensity indicators based on the degree of polarization difference. The noise suppression intensity coefficient refers to the coefficient used to control the degree of noise suppression, obtained by converting polarization state difference values ​​based on the preset intensity conversion rule.

[0158] In this embodiment of the application, based on a preset intensity conversion rule, the polarization state difference value is quantized by a numerical quantization conversion algorithm to obtain a noise suppression intensity coefficient used to control the degree of noise suppression.

[0159] Step 503: Calculate the horizontal channel weight coefficient and the vertical channel weight coefficient based on the noise suppression intensity coefficient, and scale the echo horizontal polarization component and the echo vertical polarization component to obtain the scaled echo horizontal component and the scaled echo vertical component.

[0160] In this step, the horizontal channel weight coefficients refer to weight parameters calculated based on the noise suppression intensity coefficient and the amplitude proportion of the echo horizontal polarization component. These parameters are used to scale the echo horizontal polarization component to adjust the signal preservation and noise suppression levels in the horizontal polarization direction. The vertical channel weight coefficients refer to weight parameters calculated based on the noise suppression intensity coefficient and the amplitude proportion of the echo vertical polarization component. These parameters are used to scale the echo vertical polarization component to adjust the signal preservation and noise suppression levels in the vertical polarization direction. The scaled echo horizontal component is the signal component obtained by multiplying the echo horizontal polarization component by the horizontal channel weight coefficients. After scaling, the horizontal polarization portion of the target signal is enhanced, while the noise horizontal polarization portion is suppressed. The scaled echo vertical component is the signal component obtained by multiplying the echo vertical polarization component by the vertical channel weight coefficients. After scaling, the vertical polarization portion of the target signal is enhanced, while the noise vertical polarization portion is suppressed.

[0161] In this embodiment, based on the noise suppression intensity coefficient and the polarization characteristics of the echo horizontal polarization component and the echo vertical polarization component, the horizontal channel weight coefficient is calculated using a weight allocation algorithm. The calculation formula is: Horizontal channel weight coefficient = Noise suppression intensity coefficient × Amplitude ratio of the horizontal polarization component. The vertical channel weight coefficient is calculated as follows: Vertical channel weight coefficient = Noise suppression intensity coefficient × Amplitude ratio of the vertical polarization component. Then, the echo horizontal polarization component is multiplied by the horizontal channel weight coefficient, and the echo vertical polarization component is multiplied by the vertical channel weight coefficient to obtain the scaled echo horizontal component and the scaled echo vertical component.

[0162] Step 504: Vector superposition of the scaled echo horizontal component and the scaled echo vertical component to obtain the polarized filtered signal.

[0163] In this embodiment of the application, the scaled echo horizontal component and the scaled echo vertical component are superimposed in amplitude and synthesized in phase on the same time dimension by vector synthesis technology to obtain a polarized filtered signal after adaptive polarization filtering.

[0164] This application embodiment determines the spatial difference based on the polarization state parameters of noise and target, converts it to obtain the suppression intensity coefficient, calculates the weight coefficient to scale the echo component and synthesizes the filtered signal, thereby realizing adaptive polarization filtering of the echo signal, suppressing noise while retaining the target signal, and improving the signal-to-noise ratio and quality of radar signal in low signal-to-noise ratio environments.

[0165] This application provides a specific embodiment. Step 106 involves reconstructing the polarization-filtered signal based on the target scattering statistical characteristics using a deep generative adversarial network (GAN) to obtain a reconstructed polarization-filtered signal. The reconstructed polarization-filtered signal is then verified, and the verified reconstructed polarization-filtered signal is used as an enhanced radar signal. The specific steps include:

[0166] Step 601: Perform feature matrix decomposition on the target scattering statistical features to obtain the principal feature components, and normalize the principal feature components to obtain the target feature constraint vector.

[0167] In this step, the principal feature component refers to the feature component with the largest variance obtained after principal component analysis decomposition of the feature matrix of the target scattering statistical features. It can reflect the main information of the target scattering statistical features and is used to generate the target feature constraint vector in the subsequent process. The target feature constraint vector refers to the vector obtained after normalizing the principal feature component. It is used to guide the generator of the deep generative adversarial network to reconstruct the signal and ensure that the reconstructed signal conforms to the target features.

[0168] In this embodiment of the application, the target scattering statistical features are decomposed into a feature matrix. Specifically, the feature matrix corresponding to the target scattering statistical features is decomposed into multiple feature components by principal component analysis algorithm. The feature component with the largest variance is selected as the principal feature component. Then, the principal feature component is normalized by standardization processing to obtain the target feature constraint vector used for constraint signal reconstruction.

[0169] Step 602: Based on the target feature constraint vector, the polarization-filtered signal is reconstructed using a generator from a deep generative adversarial network to obtain the reconstructed polarization-filtered signal.

[0170] In this step, the generator refers to the neural network module in the deep generative adversarial network used for signal reconstruction, which is used to enhance the signal under low signal-to-noise ratio.

[0171] In this embodiment, based on the target feature constraint vector, a generator of a deep generative adversarial network uses the target feature constraint vector as a priori condition to perform feature mapping and signal reconstruction processing on the polarization filter signal, generating a reconstructed polarization filter signal that matches the target scattering statistical characteristics.

[0172] Step 603: Extract the physical features of the reconstructed signal from the reconstructed polarization-filtered signal, and calculate the feature distribution distance between the physical features of the reconstructed signal and the target scattering statistical features using the discriminator of the deep generative adversarial network.

[0173] In this step, the reconstructed signal physical features refer to the features reflecting the physical properties of the signal extracted from the reconstructed polarized filtered signal, used for comparison and verification with the target scattering statistical features. The discriminator refers to the neural network module in the deep generative adversarial network used for feature verification, calculating the difference between the reconstructed signal physical features and the target scattering statistical features, outputting a feature distribution distance value, and determining the validity of the reconstructed signal. The feature distribution distance value refers to the distribution difference metric between the reconstructed signal physical features and the target scattering statistical features calculated by the discriminator, used to quantitatively evaluate the degree of agreement between the reconstructed signal and the target features.

[0174] In this embodiment, the reconstructed polarized filtered signal is extracted using a signal feature extraction algorithm to extract physical features of the reconstructed signal, including information such as amplitude, phase, and spectral distribution. Then, the discriminator of a deep generative adversarial network is used to calculate the feature distribution distance between the physical features of the reconstructed signal and the target scattering statistical features.

[0175] Step 604: When the feature distribution distance value is less than or equal to the preset verification threshold, it is determined that the reconstructed polarization filter signal has passed the verification, and the reconstructed polarization filter signal that has passed the verification is used as the enhanced radar signal.

[0176] In this step, the preset verification threshold refers to a pre-set critical value used to determine whether the reconstructed polarization filter signal meets the requirements. When the characteristic distribution distance value is less than or equal to the threshold, it indicates that the reconstructed signal meets the enhancement requirements.

[0177] In this embodiment of the application, when the feature distribution distance value is less than or equal to a preset verification threshold, the reconstructed polarization filter signal is determined to have passed verification through the judgment logic, and the reconstructed polarization filter signal that has passed verification is directly used as the enhanced radar signal.

[0178] Step 605: Alternatively, when the feature distribution distance value is greater than a preset verification threshold, the feature distribution distance value is used as a feedback error. Based on the feedback error, a target reconstructed polarized filter signal with a feature distribution distance value less than or equal to the preset verification threshold is reconstructed through a deep adversarial network. The target reconstructed polarized filter signal is used as an enhanced radar signal.

[0179] In this step, the feedback error refers to the error value that reflects the degree of difference between the reconstructed signal and the target feature when the feature distribution distance value is greater than the preset verification threshold.

[0180] In this embodiment, when the feature distribution distance value is greater than the preset verification threshold, the feature distribution distance value is used as a feedback error. The backpropagation algorithm of the deep adversarial network is used to adjust the network parameters of the generator based on the feedback error, and the polarization filter signal is reconstructed until the target reconstructed polarization filter signal with a feature distribution distance value less than or equal to the preset verification threshold is obtained. The target reconstructed polarization filter signal is then used as an enhanced radar signal.

[0181] This application embodiment obtains a constraint vector by processing the statistical features of target scattering, reconstructs the signal using a generator from a deep generative adversarial network, verifies the signal using a discriminator, and adjusts the reconstruction process based on the results. This achieves accurate reconstruction and optimization of the polarization-filtered signal, effectively improving the quality and recognizability of radar signals in low signal-to-noise ratio environments.

[0182] Figure 2 This application provides a schematic diagram of the structure of a deep generative adversarial radar signal enhancement system for low signal-to-noise ratio, as shown in the embodiment of this application. Figure 2 As shown, the system includes:

[0183] Extraction module 21 is used to extract target scattering statistical features from a preset historical database.

[0184] The calculation module 22 is used to calculate the optimal emission polarization state based on the target scattering statistical characteristics, so as to generate an emission polarization state configuration instruction set.

[0185] The modulation module 23 is used to perform carrier modulation processing on the radar signal according to the transmission polarization state configuration instruction set, to obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state, and to transmit it in a low signal-to-noise ratio environment.

[0186] The separation module 24 is used to receive the echo signal generated after the transmitted signal is transmitted, and to separate the echo horizontal polarization component and the echo vertical polarization component from the echo signal to generate noise polarization state parameters.

[0187] The filtering module 25 is used to perform adaptive polarization filtering on the echo horizontal polarization component and the echo vertical polarization component based on the noise polarization state parameters and the set polarization state parameters to obtain a polarization filtered signal.

[0188] The verification module 26 is used to reconstruct the polarization filter signal based on the target scattering statistical characteristics using a deep generative adversarial network to obtain a reconstructed polarization filter signal, and to verify the reconstructed polarization filter signal, using the verified reconstructed polarization filter signal as an enhanced radar signal.

[0189] Figure 2The aforementioned deep generative adversarial radar signal enhancement system for low signal-to-noise ratio radar can perform... Figure 1 The implementation principle and technical effects of the deep generative adversarial radar signal enhancement method for low signal-to-noise ratio (SNR) described in the illustrated embodiment will not be repeated here. The specific operation methods of each module and unit in the deep generative adversarial radar signal enhancement system for low SNR described in the above embodiments have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0190] In one possible design, Figure 2 The deep generative adversarial radar signal enhancement system for low signal-to-noise ratio shown in the embodiment can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32.

[0191] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0192] The processing component 32 is used for the above Figure 1 The above embodiment describes a method for enhancing deep generative adversarial radar signals with low signal-to-noise ratio.

[0193] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as 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 components to perform the above-described method.

[0194] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as Random Access Memory (RAM), Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0195] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0196] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0197] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0198] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0199] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The illustrated embodiment is a deep generative adversarial radar signal enhancement method for low signal-to-noise ratio.

[0200] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0201] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0202] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0203] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for enhancing deep generative adversarial radar signals with low signal-to-noise ratio, characterized in that, include: Extract target scattering statistical features from a pre-set historical database; Based on the target scattering statistical characteristics, the optimal emission polarization state is calculated to generate an emission polarization state configuration instruction set; According to the transmission polarization state configuration instruction set, the radar signal is subjected to carrier modulation processing to obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state and transmitted in a low signal-to-noise ratio environment; The system receives the echo signal generated after transmitting the transmitted signal, and separates the horizontal polarization component and the vertical polarization component of the echo signal to generate noise polarization state parameters. Based on the noise polarization state parameters and the set polarization state parameters, adaptive polarization filtering is performed on the echo horizontal polarization component and the echo vertical polarization component to obtain a polarization filtered signal. Based on the target scattering statistical characteristics, the polarization-filtered signal is reconstructed using a deep generative adversarial network to obtain a reconstructed polarization-filtered signal. The reconstructed polarization-filtered signal is then verified, and the verified reconstructed polarization-filtered signal is used as an enhanced radar signal.

2. The method according to claim 1, characterized in that, Based on the target scattering statistical characteristics, the optimal emission polarization state is calculated to generate an emission polarization state configuration instruction set, including: The target scattering statistical characteristics are processed by polarization state to generate a polarization state distribution map containing multiple polarization states; Calculate the matching degree between each candidate polarization state in the polarization state distribution map and the target scattering statistical features, and aggregate all matching degrees to obtain a set of polarization state matching degrees; From the set of polarization state matching degrees, select the candidate polarization state corresponding to the matching degree with the largest value, and take the candidate polarization state as the optimal emission polarization state; The amplitude values ​​of the horizontal component, the amplitude values ​​of the vertical component, the phase values ​​of the horizontal component, and the phase values ​​of the vertical component are resolved from the optimal emission polarization state to calculate the amplitude ratio parameter and the phase difference parameter. Based on the preset instruction format requirements, the amplitude ratio parameter and the phase difference parameter are phase encoded to generate a transmit polarization state configuration instruction set containing polarization modulation parameters.

3. The method according to claim 2, characterized in that, The target scattering statistical characteristics are processed by polarization state analysis to generate a polarization state distribution map containing multiple polarization states, including: The target scattering statistical characteristics are decomposed to obtain multiple eigenvalues, and the eigenvector corresponding to the eigenvalue with the largest value is taken as the target principal polarization scattering component. The polarization ellipticity angle parameter and polarization orientation angle parameter of the target principal polarization scattering component are calculated to determine the polarization state space boundary range of the target scattering statistical characteristics; According to the preset coordinate point generation rules, multiple coordinate points are generated within the boundary of the polarization state space, and all coordinate points are vector-transformed based on the preset transformation rules to obtain multiple polarization state description vectors. According to the preset partitioning rules, all polarization state description vectors are divided into spatial grids to obtain a set of polarization state spatial grid cells; Calculate the cell density of each cell in the polarization state space grid cell set, combine all cell densities to obtain polarization state density distribution data, and perform pseudo-color encoding on the polarization state density distribution data to obtain a polarization state distribution map containing multiple polarization states.

4. The method according to claim 1, characterized in that, According to the transmission polarization state configuration instruction set, the radar signal is subjected to carrier modulation processing to obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state, and transmitted in a low signal-to-noise ratio environment, including: The transmit polarization state configuration instruction set is parsed to obtain amplitude ratio code and phase difference code, and the amplitude ratio code and phase difference code are decoded to obtain amplitude ratio setting value and phase difference setting value. Carrier modulation is performed on the radar signal to obtain an initial carrier modulation signal, and orthogonal polarization separation processing is performed on the initial carrier modulation signal to obtain the horizontal polarization component and the vertical polarization component of the initial carrier modulation signal. Based on the amplitude ratio setting value, the horizontal polarization component and the vertical polarization component are subjected to amplitude processing to obtain the scaled horizontal component and the magnified vertical component. Based on the phase difference setting value, the phase offset adjustment is performed on the amplified vertical component to obtain a phase-matched vertical component. The scaled horizontal component and the phase-matched vertical component are then vector-synthesized to obtain a modulation signal with the optimal transmit polarization state. The modulation signal and the preset pulse gating signal are time-domain superimposed to obtain a pulse modulation signal. The pulse modulation signal is then bandpass filtered to obtain a radar pulse transmission signal. The radar pulse transmission signal is amplified to generate a transmission signal that meets the transmission power requirements and corresponds to the optimal transmission polarization state with set polarization parameters, and then transmitted in a low signal-to-noise ratio environment.

5. The method according to claim 1, characterized in that, Separating the horizontal and vertical polarization components of the echo signal to generate noise polarization state parameters includes: The echo signal is subjected to orthogonal polarization separation processing to obtain the horizontal polarization component and the vertical polarization component of the echo. Using the set polarization state parameters corresponding to the optimal transmit polarization state, a reference signal component consistent with the polarization characteristics of the transmit signal is constructed. The echo horizontal polarization component and the echo vertical polarization component are coherently compared with the reference signal component to adjust the phase of the reference signal component and generate a matched horizontal reference component and a matched vertical reference component. Remove the component in the horizontal polarization component of the echo that is consistent with the matched horizontal reference component to obtain the horizontal residual component; remove the component in the vertical polarization component of the echo that is consistent with the matched vertical reference component to obtain the vertical residual component. Calculate the autocorrelation value of the horizontal residual component, the autocorrelation value of the vertical residual component, and the horizontal-vertical cross-correlation value between the horizontal residual component and the vertical residual component to construct the noise polarization characteristic matrix. Based on the noise polarization characteristic matrix, the eigenvector corresponding to the largest eigenvalue is determined as the main noise polarization state, so as to extract the noise polarization state parameters from the main noise polarization state.

6. The method according to claim 1, characterized in that, Based on the noise polarization state parameters and the set polarization state parameters, adaptive polarization filtering is performed on the echo horizontal polarization component and the echo vertical polarization component to obtain a polarization-filtered signal, including: Based on the noise polarization state parameters and the set polarization state parameters, the noise location point and the target location point in the polarization state space are determined, and the spatial distance between the noise location point and the target location point is calculated. The spatial distance is used as the polarization state difference value. Based on a preset intensity conversion rule, the polarization state difference value is quantized and converted to obtain the noise suppression intensity coefficient; Based on the noise suppression intensity coefficient, the horizontal channel weight coefficient and the vertical channel weight coefficient are calculated to scale the echo horizontal polarization component and the echo vertical polarization component, so as to obtain the scaled echo horizontal component and the scaled echo vertical component. The scaled horizontal and vertical components of the echo are vector-superimposed to obtain the polarized filtered signal.

7. The method according to claim 1, characterized in that, Based on the target scattering statistical characteristics, the polarization-filtered signal is reconstructed using a deep generative adversarial network (GAN) to obtain a reconstructed polarization-filtered signal. The reconstructed polarization-filtered signal is then verified, and the verified signal is used as an enhanced radar signal, including: The target scattering statistical features are decomposed into a feature matrix to obtain the principal feature components. The principal feature components are then normalized to obtain the target feature constraint vector. Based on the target feature constraint vector, the polarization filter signal is reconstructed by the generator of a deep generative adversarial network to obtain the reconstructed polarization filter signal. The physical features of the reconstructed signal are extracted from the reconstructed polarization-filtered signal, and the feature distribution distance between the physical features of the reconstructed signal and the target scattering statistical features is calculated using a discriminator of a deep generative adversarial network. When the feature distribution distance value is less than or equal to the preset verification threshold, the reconstructed polarization filter signal is determined to have passed verification, and the reconstructed polarization filter signal that has passed verification is used as the enhanced radar signal. Alternatively, when the feature distribution distance value is greater than a preset verification threshold, the feature distribution distance value is used as a feedback error. Based on the feedback error, a target reconstructed polarized filter signal with a feature distribution distance value less than or equal to the preset verification threshold is reconstructed through a deep adversarial network. The target reconstructed polarized filter signal is then used as an enhanced radar signal.

8. A deep generative adversarial radar signal enhancement system for low signal-to-noise ratio radar, characterized in that, include: The extraction module is used to extract target scattering statistical features from a preset historical database; The calculation module is used to calculate the optimal emission polarization state based on the target scattering statistical characteristics, so as to generate an emission polarization state configuration instruction set; The modulation module is used to configure the instruction set according to the transmission polarization state, perform carrier modulation processing on the radar signal, obtain a transmission signal with set polarization state parameters corresponding to the optimal transmission polarization state, and transmit it in a low signal-to-noise ratio environment; The separation module is used to receive the echo signal generated after the transmitted signal is transmitted, and to separate the horizontal polarization component and the vertical polarization component of the echo signal from the echo signal to generate noise polarization state parameters. The filtering module is used to perform adaptive polarization filtering on the echo horizontal polarization component and the echo vertical polarization component based on the noise polarization state parameters and the set polarization state parameters to obtain a polarization filtered signal. The verification module is used to reconstruct the polarization-filtered signal based on the target scattering statistical characteristics using a deep generative adversarial network to obtain a reconstructed polarization-filtered signal, and to verify the reconstructed polarization-filtered signal, using the verified reconstructed polarization-filtered signal as an enhanced radar signal.

9. 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 method for enhancing deep generative adversarial radar signals with low signal-to-noise ratio as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, The system contains a computer program that, when executed by a computer, implements a method for enhancing deep generative adversarial radar signals with low signal-to-noise ratio as described in any one of claims 1 to 7.

Images