Rapid testing method of radar scattering based on polarization decomposition

Abstract

The present invention belongs to the field of radar signal processing technology and discloses a rapid radar scattering test method based on polarization decomposition. The method comprises the following steps: scanning a 3D target model, simulating scattering field distributions under different incident angles and polarization modes, calculating polarization contrast, and identifying polarization-sensitive areas; establishing a dynamic coordinate system conversion model, fusing RTK data, IMU data, and laser tracking data, and correcting the vehicle position and model by combining a Doppler frequency shift model and a least squares method; using a nature-inspired optimization algorithm to calculate the vehicle's optimal test path based on sensitive areas; establishing a geometric environment model, obtaining backscattered signals through beamforming, analyzing a polarization coherence matrix, obtaining high-entropy areas, and decomposing scattering components. The method forms a closed-loop optimization from the data acquisition source to the scattering component analysis end, significantly improving test efficiency and real-time performance.

Description

Technical Field

[0001] The present invention relates to the technical field of radar signal processing, and more particularly to a rapid radar scattering testing method based on polarization decomposition. Background Art

[0002] Current radar cross-section (RCS) testing primarily relies on a static rotating platform within a microwave anechoic chamber, enabling multi-angle measurements through the target's rotation. However, this method is difficult to apply to large, fixed targets such as aircraft and radar vehicles, and anechoic chamber space and cost constraints limit test flexibility. In real-world outdoor scenarios, large, immobile targets make traditional methods difficult to apply directly. Furthermore, multipath reflections, ground clutter, and motion errors in complex environments can reduce measurement accuracy.

[0003] Chinese patent application publication number CN116224271A discloses a method for detecting weak targets on the sea surface based on polarization scattering characteristics: Step 1: Receive radar echo data from four polarization channels of the sea surface respectively to form a polarization scattering matrix; Step 2: Perform polarization feature decomposition of signal echo samples according to the characteristics of the polarization scattering matrix through Cameron decomposition. For any polarization scattering matrix, perform a similarity operation with the test scattering matrix to determine whether they belong to the same scattering type; Step 3: Based on the difference in scattering types between clutter and target units, propose scattering weights for Cameron polarization decomposition; Step 4: Calculate the detection rate and false alarm rate based on the scattering weights of the Cameron polarization decomposition.

[0004] Although the above methods can meet most scenarios, research and practical application of the above methods and existing technologies have revealed that the above methods and existing technologies have at least the following defects:

[0005] Traditional polarization decomposition methods use mathematical modeling to decompose the scattering mechanism into different physical components. This requires extensive matrix operations and eigenvalue decomposition on the radar echo data. This restricts the effectiveness of traditional polarization decomposition methods in detecting faint targets on the sea surface, affecting both detection rate and real-time performance.

[0006] In view of this, the present invention proposes a rapid radar scattering test method based on polarization decomposition to solve the above problems. Summary of the Invention

[0007] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a rapid radar scattering test method based on polarization decomposition, comprising the following steps:

[0008] Scan and acquire the 3D model of the target, simulate the scattered field distribution of the 3D model surface of the target under different incident angles and polarization modes, and identify the polarization-sensitive area on the target surface;

[0009] Establish a dynamic coordinate system conversion model between the vehicle position coordinate system and the target coordinate system, obtain RTK data, IMU data and laser tracking data, and combine the Doppler frequency shift model and the least squares method to obtain and correct the vehicle position and dynamic coordinate system conversion model;

[0010] A nature-inspired optimization algorithm is used to calculate the optimal test path for the mobile car based on the polarization-sensitive areas of the target surface.

[0011] A geometric environment model of ground reflections and surrounding objects is established. When the car moves along the optimal test path, the backscattered signal of the target is acquired through beamforming technology. The polarization coherence matrix is analyzed and the scattering entropy and anisotropy of the polarization coherence matrix are calculated to obtain the high entropy area. The high entropy area is then dynamically decomposed into scattering components to obtain the decomposition results.

[0012] Furthermore, the method for identifying the polarization-sensitive area of the target surface includes:

[0013] Preset radar operating frequency band, polarization mode, incident angle range and wavelength parameters;

[0014] Obtain the target center point position and discretize the target's 3D model surface into W surface grids according to the wavelength constraint;

[0015] Calculate the single-station RCS value of each surface grid under the four polarization combinations of HH / HV / VH / VV;

[0016] Scanning within a preset angle range with a preset step size, recording the polarization scattering matrix at each angle;

[0017] Polarization contrast ratio was calculated based on the ratio of the maximum RCS value of a single station to the minimum RCS value of a single station for the four polarization combinations;

[0018] The areas with polarization contrast greater than a preset contrast threshold are marked as polarization-sensitive areas.

[0019] Furthermore, in the method for identifying polarization-sensitive areas on the target surface, the curvature of the 3D model surface of the target is also calculated, and local surface mesh encryption is performed on areas where the curvature change rate is greater than a preset change rate threshold; wherein the side length of a single surface mesh and the locally encrypted surface mesh both meet the corresponding preset wavelength constraints.

[0020] Furthermore, the method of obtaining the high entropy region includes:

[0021] Obtain the backscattered signal of the target, calculate the covariance of the signal vector of K independent observations, and obtain the polarization coherence matrix;

[0022] Perform eigenvalue decomposition on the polarization coherence matrix and obtain the three non-negative real numbers The diagonal matrix formed by And the corresponding eigenvector , and The unitary matrix formed by polarization coherence matrix, diagonal matrix and unitary matrix meet the matrix constraint conditions;

[0023] The scattering entropy is obtained based on the normalized non-negative real number calculation; and The difference and sum of the two are calculated, and the ratio of the difference and the sum is calculated to obtain the anisotropy; the area where the scattering entropy is not lower than the preset low entropy threshold is marked as a high entropy area.

[0024] Furthermore, the method for obtaining the decomposition result includes:

[0025] Classify the scattering components whose normalized non-negative real number is greater than a preset eigenvalue threshold as surface scattering;

[0026] If the eigenvalue If the value of the scattered light component is greater than the preset significance threshold and the anisotropy is greater than the preset anisotropy threshold, the scattered light component is classified as secondary scattering;

[0027] The scattering entropy is greater than the preset scattering entropy threshold and the eigenvalue Scattering components larger than a preset characteristic threshold are classified as volume scattering.

[0028] Furthermore, the method for establishing a dynamic coordinate system conversion model includes:

[0029] Obtain the target coordinate system and the vehicle coordinate system; the target coordinate system takes the target geometric center as its origin and is aligned with the principal inertial axes; the vehicle coordinate system takes the radar phase center as its origin, the vehicle's motion direction is the X-axis, the horizontal direction perpendicular to the motion direction is the Y-axis, and the vertical direction is the Z-axis; transform the point coordinates in the vehicle coordinate system to the target coordinate system through the translation vector and rotation matrix.

[0030] Furthermore, the method for obtaining and correcting the vehicle position and the dynamic coordinate system transformation model includes:

[0031] The laser tracking data is updated by combining the Doppler frequency shift model and the least square method to obtain updated laser tracking data;

[0032] The state vector is defined to include position deviation and attitude deviation. The position deviation is the translation deviation of the radar phase center relative to the target coordinate system; the attitude deviation is the attitude angle deviation of the vehicle.

[0033] The state transfer matrix is constructed through the identity matrix, and the process noise is obtained according to the covariance matrix of the process noise. The covariance matrix of the process noise is set according to the sensor stability, and the predicted state vector is obtained according to the state transfer matrix and the process noise.

[0034] The initial deviation is calibrated by the laser tracker, and the covariance matrix of the initial state estimation is set;

[0035] Predict the state at the next moment based on the state transfer matrix;

[0036] Obtaining the covariance of the predicted updated state estimate based on the state transfer matrix, the covariance matrix of the current state estimate, and the covariance matrix of the process noise;

[0037] Acquire sensor observations consisting of RTK data, IMU data, and updated laser tracking data;

[0038] The Kalman gain is calculated based on the covariance of the predicted updated state estimate, the sensor observation matrix composed of the sensor observation values, and the covariance matrix of the process noise;

[0039] Update the state estimate based on the covariance matrix of the process noise, the Kalman gain, the sensor observations, and the sensor observation matrix;

[0040] Update the covariance matrix of the state estimate based on the Kalman gain, the sensor observation matrix, and the covariance of the predicted updated state estimate;

[0041] Determine whether the value of the updated covariance matrix is greater than the preset variance threshold. If so, re-update to obtain a new state estimate and, until the value of the covariance matrix of the updated state estimate is less than the preset variance threshold, obtain the corresponding updated state estimate, otherwise directly obtain the updated state estimate;

[0042] According to the updated state estimation, the radar phase center position and attitude angle are corrected, the rotation matrix is updated according to the corrected attitude angle, and the vehicle position correction is converted to the target coordinate system according to the updated rotation matrix.

[0043] Furthermore, the method for obtaining updated laser tracking data includes:

[0044] Obtain the vibration acceleration vector of the IMU data, project the vibration acceleration vector to the radar line of sight direction through the radar line of sight direction vector, and obtain the effective acceleration;

[0045] Integrate the effective acceleration to obtain the velocity, and combine it with the Doppler frequency shift formula to obtain the frequency shift;

[0046] The effective acceleration is integrated twice to obtain the vibration displacement; combined with the radar wavelength, the theoretical phase error is calculated;

[0047] In a static scenario, N sets of radar echo phase and theoretical phase error values are collected, and a matrix equation is obtained by constructing a time matrix based on the acquisition time, a polynomial coefficient matrix based on polynomial coefficients used to model the theoretical phase error as a polynomial form, and a difference matrix obtained from the difference between the radar echo phase and the theoretical phase error.

[0048] The matrix equation is solved to obtain the polynomial coefficient matrix, the actual phase error is calculated in real time based on the polynomial coefficient matrix, and the actual phase error is deducted from the radar echo phase to obtain the true phase; the laser tracking data is updated according to the true phase to obtain updated laser tracking data.

[0049] Furthermore, the method for calculating the optimal test path of the mobile vehicle includes:

[0050] The test area is discretely divided into P regional grids, and corresponding regional weights are assigned to different regions according to the polarization sensitive areas;

[0051] According to the regional weight and path length, the corresponding fitness is calculated to maximize the sum of the weights of the coverage area of the regional grid and minimize the path length as the optimization goal;

[0052] Initialize the initial pheromone concentration and pheromone volatilization rate of each regional grid;

[0053] A heuristic function is constructed based on the area weight and the length between path nodes. The state transition probability is calculated based on the heuristic function and pheromone concentration. Each ant starts from the starting point and selects the next path node according to the probability until it reaches the end point.

[0054] After each ant completes the path, the pheromone is updated according to the path quality. The path quality is calculated based on the ratio of the fitness value to the global optimal fitness and the pheromone volatility rate.

[0055] After each round, the pheromone of the optimal path is enhanced by the ratio of fitness to path length;

[0056] When the preset termination condition is reached, the optimal path with the highest fitness is selected as the optimal test path.

[0057] Furthermore, the method for obtaining the backscattered signal of the target includes:

[0058] Sampling and processing the received signal containing only interference and noise, and calculating the covariance matrix of interference plus noise;

[0059] According to the incident direction of the desired signal, the desired direction steering vector is calculated using the array manifold formula;

[0060] A weight vector is obtained by calculating the covariance matrix of the interference plus noise and the desired direction steering vector;

[0061] The received signal containing only multipath interference is sampled and processed to obtain a covariance matrix of the received signal. The covariance matrix is eigen-decomposed to obtain eigenvalues and corresponding eigenvectors. After sorting the eigenvalues, the eigenvectors corresponding to the eigenvalues below a preset characteristic threshold are spanned into a noise subspace. Within a preset angle range of interest, a spatial spectrum is calculated for each angle based on the desired direction steering vector and the noise subspace, where the peak of the spatial spectrum corresponds to the direction of the interference wave.

[0062] For the nth sampling, the a priori error is calculated based on the received signal, the expected signal and the corresponding weight vector; the gain vector is calculated based on the error covariance matrix and the received signal; the weight vector is updated based on the weight vector, the a priori error and the gain vector corresponding to the received signal of the last sampling; the error covariance matrix is updated based on the error covariance matrix, the gain vector and the received signal corresponding to the received signal of the last sampling;

[0063] A covariance matrix loading is obtained, and the covariance matrix loading is added to the diagonal of the covariance matrix of interference plus noise to obtain an adjusted covariance matrix. Beamforming weights are calculated by adjusting the covariance matrix, and a backscattered signal of the target is formed according to the beamforming weights.

[0064] The technical effects and advantages of the radar scattering rapid testing method based on polarization decomposition of the present invention are as follows:

[0065] The present invention uses 3D modeling and electromagnetic simulation to identify polarization-sensitive areas, and combines nature-inspired algorithms to plan the optimal path, so that the radar focuses on high-value areas to collect signals, reducing invalid scanning and redundant data; secondly, through multi-source data fusion and Kalman filtering to dynamically correct the coordinate system, millimeter-level positioning accuracy is ensured, and signal dislocation and repeated verification caused by platform vibration or coordinate deviation are avoided; thirdly, a geometric environment model is constructed and combined with beamforming to suppress multipath interference, improve the signal-to-noise ratio, and reduce the noise interference and iterative correction number of subsequent polarization decomposition; finally, a fast classification method based on scattering entropy and eigenvalue threshold is used to skip the full matrix decomposition of low-contribution areas, reducing computational complexity; the present invention forms a closed-loop optimization from the data acquisition source to the end of scattering component analysis, significantly improving test efficiency and real-time performance, while ensuring the accuracy of polarization scattering characteristic analysis, providing an efficient and reliable technical path for rapid detection and identification of targets in dynamic scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 Schematic diagram of the rapid radar scattering test method based on polarization decomposition of the present invention;

[0067] Figure 2 This is a data flow diagram of Example 1 of the present invention;

[0068] Figure 3 Schematic diagram of the vehicle detour path feedback optimization method of the present invention;

[0069] Figure 4 This is a data flow diagram of Example 2 of the present invention. DETAILED DESCRIPTION

[0070] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0071] Example 1

[0072] See also Figure 1 、 Figure 2 As shown, this embodiment provides a rapid radar scattering test method based on polarization decomposition, including:

[0073] Scan and acquire the 3D model of the target, simulate the scattered field distribution of the 3D model surface of the target under different incident angles and polarization modes, and identify the polarization-sensitive area on the target surface;

[0074] Methods for identifying polarization-sensitive areas on a target surface include:

[0075] Preset radar operating frequency band, polarization mode, incident angle range and wavelength parameters;

[0076] Full-wave electromagnetic simulation using the finite-difference time-domain method;

[0077] Get the target center point position, discretize the target 3D model surface according to the preset dimension, and form W surface grids. The side length of a single surface grid meets the wavelength constraint, such as the side length of a single surface grid ;in, is the radar working wavelength; calculate the curvature of the target's 3D model surface, and perform local surface mesh encryption on areas with sudden curvature changes (such as the curvature change rate is greater than the preset change rate threshold). The corresponding surface mesh side length of the encrypted area is .

[0078] For each surface grid of the target's 3D model, the single-station RCS values under the four polarization combinations of HH / HV / VH / VV are calculated respectively. Among them, HH polarization means transmitting horizontally polarized waves and receiving horizontally polarized echoes; HV polarization means transmitting horizontally polarized waves and receiving vertically polarized echoes; VH polarization means transmitting vertically polarized waves and receiving horizontally polarized echoes; VV polarization means transmitting vertically polarized waves and receiving vertically polarized echoes.

[0079] Perform angle scanning within a preset angle range (e.g., 0° to 360°) with a step size of ≤1°, and record the polarization scattering matrix at each angle; such as the Jones matrix or Mueller matrix.

[0080] Polarization contrast ratio was calculated based on the ratio of the maximum RCS value of a single station to the minimum RCS value of a single station for the four polarization combinations;

[0081] The areas with polarization contrast greater than a preset contrast threshold are marked as polarization-sensitive areas.

[0082] By scanning and acquiring a 3D model of the target and simulating the scattering field distribution under different conditions to identify polarization-sensitive areas, rapid radar scattering testing based on polarization decomposition can clearly identify areas on the target surface that respond significantly to polarization signals, laying the foundation for subsequent targeted collection of scattered signals and analysis of target polarization characteristics, thereby improving test efficiency and accuracy. In addressing the high computational complexity and lack of real-time performance of traditional polarization decomposition methods, accurately identifying polarization-sensitive areas can reduce unnecessary computational effort, focus on key areas for analysis, and avoid redundant calculations of the entire target model. At the same time, identifying sensitive areas in advance helps optimize the test process, making signal acquisition and processing more targeted, reducing data processing, and thus improving real-time performance.

[0083] Establish a dynamic coordinate system conversion model between the vehicle position coordinate system and the target coordinate system, obtain RTK data, IMU data and laser tracking data, and combine the Doppler frequency shift model and the least squares method to obtain and correct the vehicle position and dynamic coordinate system conversion model;

[0084] A dynamic coordinate conversion model is established between the vehicle's position coordinate system and the target coordinate system. This model combines RTK data (using a dual-frequency RTK module, such as the u-blox ZED-F9P, to acquire the vehicle's absolute position), IMU data (using an integrated six-axis IMU, such as the ADIS16470, to measure the vehicle's vibration acceleration or angular velocity in real time), and laser tracking data (using a laser tracker, such as the Leica AT960, to scan reflection points on the target surface in real time and acquire the target's attitude). Using a Doppler shift model and least-squares correction, the vehicle's position and attitude relative to the target can be accurately determined. This is crucial for rapid radar scattering testing based on polarization decomposition. It provides accurate position information for radar signal acquisition, enabling precise antenna alignment of the target's polarization-sensitive area, improving the accuracy and effectiveness of scattered signal acquisition. Furthermore, addressing the computational complexity and poor real-time performance of traditional polarization decomposition methods, accurate position information optimizes the data acquisition range and angle, reduces unnecessary data processing, avoids redundant calculations, and improves processing speed, facilitating rapid testing.

[0085] Methods for establishing a dynamic coordinate system transformation model include:

[0086] Obtain the target coordinate system and the vehicle coordinate system; for example, the target coordinate system takes the target geometric center as its origin and is aligned with the principal inertial axes (for example, the aircraft's longitudinal axis is the X-axis, the transverse axis is the Y-axis, and the vertical axis is the Z-axis). The vehicle coordinate system takes the radar phase center as its origin, the vehicle's motion direction is the X-axis, the horizontal direction perpendicular to the motion direction is the Y-axis, and the vertical direction is the Z-axis; transform the point coordinates in the vehicle coordinate system to the target coordinate system using the translation vector and rotation matrix.

[0087] A dynamic coordinate system conversion model is established to clarify the origin and axial definition of the target and vehicle coordinate systems, and coordinate conversion is realized through translation vectors and rotation matrices, providing a unified and accurate spatial reference for rapid radar scattering testing based on polarization decomposition. The radar carried by the vehicle can accurately locate the target based on this reference, effectively collect scattered signals from polarization-sensitive areas on the target surface, and improve test efficiency and data accuracy. To address the problems of complex calculations and poor real-time performance of traditional polarization decomposition methods, this model eliminates redundant calculations and data deviations caused by inconsistent coordinate systems, reduces unnecessary coordinate conversion and error correction steps, reduces computational complexity, and ensures the efficiency of data acquisition and processing, thereby improving the real-time performance and reliability of the overall test.

[0088] The methods for obtaining and correcting the vehicle position and dynamic coordinate system transformation model include:

[0089] The laser tracking data is updated by combining the Doppler frequency shift model and the least square method to obtain updated laser tracking data;

[0090] Methods for obtaining updated laser tracking data include:

[0091] Obtain the vibration acceleration vector of the IMU data, project the vibration acceleration vector to the radar line of sight direction through the radar line of sight direction vector, and obtain the effective acceleration;

[0092] Integrate the effective acceleration to obtain the velocity, and combine it with the Doppler frequency shift formula to obtain the frequency shift;

[0093] The effective acceleration is integrated twice to obtain the vibration displacement; combined with the radar wavelength, the theoretical phase error is calculated;

[0094] In a static scenario, N sets of radar echo phase and theoretical phase error values are collected, and a matrix equation is obtained by constructing a time matrix based on the acquisition time, a polynomial coefficient matrix based on polynomial coefficients used to model the theoretical phase error as a polynomial form, and a difference matrix obtained from the difference between the radar echo phase and the theoretical phase error.

[0095] The matrix equation is solved to obtain the polynomial coefficient matrix, the actual phase error is calculated in real time based on the polynomial coefficient matrix, and the actual phase error is deducted from the radar echo phase to obtain the true phase; the laser tracking data is updated according to the true phase to obtain updated laser tracking data.

[0096] The state vector is defined to include position deviation and attitude deviation. The position deviation is the translation deviation of the radar phase center relative to the target coordinate system; the attitude deviation is the attitude angle deviation of the vehicle (roll, pitch, yaw).

[0097] The state transfer matrix is constructed through the identity matrix, and the process noise is obtained according to the covariance matrix of the process noise. The covariance matrix of the process noise is set according to the sensor stability, and the predicted state vector is obtained according to the state transfer matrix and the process noise.

[0098] The initial deviation is calibrated by the laser tracker, and the covariance matrix of the initial state estimation is set;

[0099] Predict the state at the next moment based on the state transfer matrix;

[0100] Obtaining the covariance of the predicted updated state estimate based on the state transfer matrix, the covariance matrix of the current state estimate, and the covariance matrix of the process noise;

[0101] Acquire sensor observations consisting of RTK data, IMU data, and updated laser tracking data;

[0102] The Kalman gain is calculated based on the covariance of the predicted updated state estimate, the sensor observation matrix composed of the sensor observation values, and the covariance matrix of the process noise;

[0103] Update the state estimate based on the covariance matrix of the process noise, the Kalman gain, the sensor observations, and the sensor observation matrix;

[0104] Update the covariance matrix of the state estimate based on the Kalman gain, the sensor observation matrix, and the covariance of the predicted updated state estimate;

[0105] Determine whether the value of the updated covariance matrix is greater than the preset variance threshold. If so, re-update to obtain a new state estimate and, until the value of the covariance matrix of the updated state estimate is less than the preset variance threshold, obtain the corresponding updated state estimate, otherwise directly obtain the updated state estimate;

[0106] According to the updated state estimation, the radar phase center position and attitude angle are corrected, the rotation matrix is updated according to the corrected attitude angle, and the vehicle position correction is converted to the target coordinate system according to the updated rotation matrix.

[0107] By processing the vibration data in the IMU data to obtain the true phase to update the laser tracking data, and combining it with Kalman filtering to fuse RTK data, IMU data, and laser data, the vehicle's position and attitude deviations are accurately estimated and the dynamic coordinate system conversion model is corrected. This provides millimeter-level precision position and attitude information for rapid radar scattering testing based on polarization decomposition. This ensures the spatiotemporal alignment of the radar phase center with the target coordinate system, ensuring that the scattered signal acquisition strictly corresponds to the actual orientation of the target's polarization-sensitive area, and avoiding signal acquisition misalignment and data redundancy caused by platform vibration or coordinate deviation. To address the computational complexity and lack of real-time performance of traditional polarization decomposition methods, this correction method eliminates the cumulative impact of sensor noise and model deviations through multi-source data fusion and dynamic error compensation. This reduces invalid calculations and repeated verifications caused by position errors in subsequent polarization decomposition, improving data processing efficiency. Furthermore, the precise coordinate system conversion provides a reliable benchmark for path planning, allowing the radar to only collect directional data in high-value sensitive areas, significantly reducing data volume and computational load, and enabling low-complexity, high-real-time rapid analysis of scattering characteristics.

[0108] A nature-inspired optimization algorithm is used to calculate the optimal test path for the mobile car based on the polarization-sensitive areas of the target surface.

[0109] Methods for calculating the optimal test path of the mobile car include:

[0110] The test area is discretely divided into P regional grids, and corresponding regional weights are assigned to different regions according to the polarization sensitive areas;

[0111] According to the regional weight and path length, the corresponding fitness is calculated to maximize the sum of the weights of the coverage area of the regional grid and minimize the path length as the optimization goal;

[0112] Initialize the initial pheromone concentration and pheromone volatilization rate of each regional grid;

[0113] A heuristic function is constructed based on the area weight and the length between path nodes. The state transition probability is calculated based on the heuristic function and pheromone concentration. Each ant starts from the starting point and selects the next path node according to the probability until it reaches the end point.

[0114] After each ant completes the path, the pheromone is updated according to the path quality. The path quality is calculated based on the ratio of the fitness value to the global optimal fitness and the pheromone volatility rate.

[0115] After each round, the pheromone of the optimal path is additionally enhanced by the ratio of fitness to path length;

[0116] When the preset termination condition is reached, the optimal path with the highest fitness is selected as the optimal test path.

[0117] Using a nature-inspired optimization algorithm, the method generates an optimal test path that prioritizes high-value areas by discretizing test areas, assigning weights to polarization-sensitive areas, and constructing a fitness function that combines weights and path length. This provides efficient trajectory planning for rapid radar scattering testing based on polarization decomposition, ensuring that the radar focuses on collecting signals from polarization-sensitive areas on the target surface, reducing invalid paths and redundant data collection, and significantly improving test efficiency and coverage integrity of sensitive areas. To address the computational complexity and lack of real-time performance of traditional polarization decomposition methods, this path optimization method uses a weight-guided heuristic search to avoid indiscriminate scanning of low-value areas, significantly reducing the amount of data to be processed. Combined with a pheromone update strategy, the method strengthens the optimal path and reduces the algorithm search complexity, enabling the radar to complete high-quality data collection in key areas in a short period of time. This provides streamlined and effective input data for subsequent polarization decomposition, alleviating the computational burden at the source of data collection and improving the real-time and efficiency of the overall test process.

[0118] A geometric environment model of ground reflections and surrounding objects is established. When the vehicle moves along the optimal test path, beamforming technology is used to suppress multipath interference. The target's backscattered signal is acquired and analyzed to obtain the polarization coherence matrix. The scattering entropy and anisotropy of the polarization coherence matrix are calculated to obtain high-entropy areas. The high-entropy areas are then dynamically decomposed into scattering components to obtain the decomposition results.

[0119] A geometric environment model of ground reflection and surrounding objects is established and combined with beamforming to suppress multipath interference, providing a pure backscattered signal for rapid radar scattering testing based on polarization decomposition. By accurately constructing the polarization coherence matrix, calculating the scattering entropy, and locating the high-entropy area using anisotropy, the complex scattering components of the target are directional decomposed to ensure the accuracy and pertinence of the scattering characteristic analysis. To address the computational complexity and lack of real-time performance of traditional polarization decomposition methods, this link improves the signal-to-noise ratio through environmental interference preprocessing, avoids the contamination of the polarization matrix calculation by multipath clutter, and reduces iterative corrections and redundant calculations caused by noise in subsequent decomposition. At the same time, it focuses on high-entropy areas for dynamic component decomposition, avoiding indiscriminate processing of the entire area, significantly reducing feature extraction and matrix operations, and improving efficiency in both signal acquisition and data processing, helping to achieve low-complexity, high-real-time rapid radar scattering testing.

[0120] Methods for acquiring the backscattered signal of a target include:

[0121] Sampling and processing the received signal containing only interference and noise, and calculating the covariance matrix of interference plus noise;

[0122] According to the incident direction of the desired signal, the desired direction steering vector is calculated using the array manifold formula;

[0123] A weight vector is obtained by calculating the covariance matrix of the interference plus noise and the desired direction steering vector;

[0124] The received signal containing only multipath interference is sampled and processed to calculate the covariance matrix of the received signal. The covariance matrix is then eigen-decomposed to obtain eigenvalues and corresponding eigenvectors. After sorting the eigenvalues, the eigenvectors corresponding to eigenvalues below a preset eigenthreshhold are spanned into a noise subspace. Within a preset angle range of interest (e.g., 0°-360°), a spatial spectrum is calculated for each angle based on the desired direction-steering vector and the noise subspace. The peak of the spatial spectrum corresponds to the direction of the interference wave.

[0125] For the nth sampling, the a priori error is calculated based on the received signal, the expected signal and the corresponding weight vector; the gain vector is calculated based on the error covariance matrix and the received signal; the weight vector is updated based on the weight vector, the a priori error and the gain vector corresponding to the received signal of the last sampling; the error covariance matrix is updated based on the error covariance matrix, the gain vector and the received signal corresponding to the received signal of the last sampling;

[0126] A covariance matrix loading is obtained, and the covariance matrix loading is added to the diagonal of the covariance matrix of interference plus noise to obtain an adjusted covariance matrix. Beamforming weights are calculated by adjusting the covariance matrix, and a backscattered signal of the target is formed according to the beamforming weights.

[0127] By calculating the covariance matrix of interference plus noise, the steering vector, and iteratively updating the weight vector, combined with spatial spectrum estimation to locate the direction of the interference wave and adjust the beamforming weights, a high signal-to-noise ratio backscattered signal is provided for rapid radar scattering testing based on polarization decomposition, ensuring that the target component in the received signal is dominant, avoiding multipath interference from contaminating the polarization coherence matrix calculation, and improving the accuracy of parameter estimation such as scattering entropy and anisotropy. To address the problems of traditional polarization decomposition methods with complex calculations and insufficient real-time performance, this signal acquisition method suppresses multipath interference through preprocessing, reduces repeated iterations and error corrections caused by noise in subsequent polarization decomposition, and reduces the complexity of eigendecomposition and matrix operations. At the same time, precise interference suppression enables the radar to effectively collect the target's real scattered signal, avoids invalid data processing, improves data quality and processing efficiency at the signal source, and helps achieve low-computation, high-real-time rapid decomposition of dynamic scattering components.

[0128] Methods for obtaining high entropy regions include:

[0129] The radar transmits and receives four orthogonal polarization signals to obtain the target's backscattered signal. For a single observation, the received signal can be expressed as a three-dimensional polarization vector. The covariance of the signal vectors of K independent observations is calculated to obtain the polarization coherence matrix. ;

[0130] Perform eigenvalue decomposition on the polarization coherence matrix and obtain the three non-negative real numbers The diagonal matrix , and the corresponding eigenvectors , and The unitary matrix , ;in, is the transpose of the vector; is the unit matrix; polarization coherence matrix , diagonal matrix and unitary matrix Satisfy the matrix constraints; such as ;The sum of the eigenvalues is equal to the trace of the polarization coherence matrix;

[0131] The scattering entropy is obtained based on the normalized non-negative real number calculation; and The difference and sum of the two are calculated, and the ratio of the difference and the sum is calculated to obtain the anisotropy; the area where the scattering entropy is not lower than the preset low entropy threshold is marked as a high entropy area.

[0132] By transmitting and receiving four orthogonal polarization signals to construct a polarization coherence matrix, eigendecomposition is used to calculate scattering entropy and anisotropy and mark high-entropy areas. This provides precise positioning of target complex scattering areas for rapid radar scattering testing based on polarization decomposition, allowing subsequent dynamic component decomposition to focus on key areas with rich scattering mechanisms, avoiding indiscriminate treatment of the entire area and improving analysis efficiency and pertinence. To address the computational complexity and lack of real-time performance of traditional polarization decomposition methods, this method uses entropy to screen high-value areas, significantly reducing the scope of the area to be decomposed and reducing the amount of eigendecomposition and matrix operations. At the same time, covariance calculation and normalization processing based on statistical observations ensure the robustness of high-entropy area identification, avoid redundant data interference, reduce the computational burden at the data processing level, and help achieve efficient and real-time rapid analysis of scattering characteristics.

[0133] Methods for obtaining decomposition results include:

[0134] Classify the scattering components whose normalized non-negative real number is greater than a preset eigenvalue threshold as surface scattering;

[0135] If the eigenvalue If the value of the scattered light component is greater than the preset significance threshold and the anisotropy is greater than the preset anisotropy threshold, the scattered light component is classified as secondary scattering;

[0136] The scattering entropy is greater than the preset scattering entropy threshold and the eigenvalue Scattering components larger than a preset characteristic threshold are classified as volume scattering.

[0137] By classifying scattering components (surface, secondary, and volume scattering) using preset eigenvalue, anisotropy, and scattering entropy thresholds, this method provides a simple and efficient scattering mechanism identification method for rapid radar scattering testing based on polarization decomposition. This method ensures that the decomposition results directly correspond to the target's actual physical scattering characteristics (such as surface scattering from smooth surfaces, secondary scattering from angular structures, and volume scattering from vegetation), avoiding the redundant calculations required for complex mathematical modeling. To address the computational complexity and lack of real-time performance of traditional polarization decomposition methods, this classification method uses threshold screening to focus on significant scattering components, skipping ineffective decomposition of low-contribution components and significantly reducing the computational complexity of eigenvalue comparisons and conditional judgments. Furthermore, by performing rapid classification based on thresholds with clear physical meaning (such as eigenvalues reflecting energy contribution and scattering entropy representing randomness), this method avoids the high-complexity computations required by traditional methods that rely on full matrix decomposition. This reduces the computational burden at the end of scattering mechanism analysis, enabling rapid decomposition and identification of scattering characteristics with low time consumption and high real-time performance.

[0138] Example 2

[0139] See also Figure 3 、 Figure 4 This embodiment provides a feedback optimization method for a vehicle detour path applied to embodiment 1, comprising the following steps:

[0140] The decomposition results, RCS values and the target 3D model are superimposed to generate a four-dimensional scattering distribution map (such as a time-space-frequency-polarization four-dimensional scattering distribution map);

[0141] Mark the scattering components (e.g., red for secondary scattering, green for volume scattering) and RCS amplitude on the surface mesh of the target 3D model;

[0142] Hierarchical data storage based on scanning time series and radar operating frequency bands (such as C-band and X-band), supporting dynamic playback and multi-band comparison;

[0143] The RCS values of the four polarization channels HH / HV / VH / VV are visualized separately to form a polarization response heat map.

[0144] Finally, an interactive four-dimensional scattering distribution map is generated, which displays the dynamic changes of the scattering characteristics of each area on the target surface in real time.

[0145] Based on the real-time four-dimensional scattering distribution map, the vehicle's detour path is dynamically optimized through reinforcement learning algorithm.

[0146] Define the state space: including the current vehicle pose (position and attitude in the target coordinate system), the set of covered polarization-sensitive areas, and key features of the real-time four-dimensional scattering distribution map (such as the location of uncovered high-entropy areas and missed areas with low RCS values);

[0147] Define the action space: define the car's movement direction (forward / backward / left / right / rotation) and the next jump target grid (select uncovered or low-accuracy grids from the polarization-sensitive area);

[0148] Design reward function:

[0149] Design positive rewards based on covering high entropy areas, collecting high RCS value areas, and short path length;

[0150] Negative penalties are designed based on repeated sampling of low-weight areas and entering obstacle areas.

[0151] The dynamically optimized car detour path is obtained through iteration through the deep Q network DQN or policy gradient algorithm.

[0152] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

[0153] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A rapid radar scattering test method based on polarization decomposition, characterized in that: include: Scan and acquire the 3D model of the target, simulate the scattered field distribution of the 3D model surface of the target under different incident angles and polarization modes, and identify the polarization-sensitive area on the target surface; Establish a dynamic coordinate system conversion model between the vehicle position coordinate system and the target coordinate system, obtain RTK data, IMU data and laser tracking data, and combine the Doppler frequency shift model and the least squares method to obtain and correct the vehicle position and dynamic coordinate system conversion model; A nature-inspired optimization algorithm is used to calculate the optimal test path for the mobile car based on the polarization-sensitive areas of the target surface. A geometric environment model of ground reflections and surrounding objects is established. When the car moves along the optimal test path, the backscattered signal of the target is acquired through beamforming technology. The polarization coherence matrix is analyzed and the scattering entropy and anisotropy of the polarization coherence matrix are calculated to obtain the high entropy area. The high entropy area is then dynamically decomposed into scattering components to obtain the decomposition results.

2. The rapid radar scattering test method based on polarization decomposition according to claim 1 is characterized in that: Methods for identifying polarization-sensitive areas on a target surface include: Preset radar operating frequency band, polarization mode, incident angle range and wavelength parameters; Obtain the target center point position and discretize the target's 3D model surface into W surface grids according to the wavelength constraint; Calculate the single-station RCS value of each surface grid under the four polarization combinations of HH / HV / VH / VV; Scanning within a preset angle range with a preset step size, recording the polarization scattering matrix at each angle; Polarization contrast ratio was calculated based on the ratio of the maximum RCS value of a single station to the minimum RCS value of a single station for the four polarization combinations; The areas with polarization contrast greater than a preset contrast threshold are marked as polarization-sensitive areas.

3. The rapid radar scattering test method based on polarization decomposition according to claim 2, characterized in that: In the method for identifying polarization-sensitive areas on the target surface, the curvature of the target's 3D model surface is also calculated, and local surface mesh encryption is performed on areas where the curvature change rate is greater than a preset change rate threshold; wherein the side length of a single surface mesh and the locally encrypted surface mesh both meet corresponding preset wavelength constraints.

4. The rapid radar scattering test method based on polarization decomposition according to claim 1 is characterized in that: Methods for obtaining high entropy regions include: Obtain the backscattered signal of the target, calculate the covariance of the signal vector of K independent observations, and obtain the polarization coherence matrix; Perform eigenvalue decomposition on the polarization coherence matrix and obtain the three non-negative real numbers The diagonal matrix formed by , and the corresponding eigenvectors , and The unitary matrix formed by polarization coherence matrix, diagonal matrix and unitary matrix meet the matrix constraint conditions; The scattering entropy is obtained based on the normalized non-negative real number calculation; and The difference and sum of the two are calculated, and the ratio of the difference and the sum is calculated to obtain the anisotropy; the area where the scattering entropy is not lower than the preset low entropy threshold is marked as a high entropy area.

5. The rapid radar scattering test method based on polarization decomposition according to claim 4 is characterized in that: Methods for obtaining decomposition results include: Classify the scattering components whose normalized non-negative real number is greater than a preset eigenvalue threshold as surface scattering; If the eigenvalue If the value of the scattered light component is greater than the preset significance threshold and the anisotropy is greater than the preset anisotropy threshold, the scattered light component is classified as secondary scattering; The scattering entropy is greater than the preset scattering entropy threshold and the eigenvalue Scattering components larger than a preset characteristic threshold are classified as volume scattering.

6. The method for rapid radar scattering testing based on polarization decomposition according to claim 1, characterized in that: Methods for establishing a dynamic coordinate system transformation model include: Obtain the target coordinate system and the vehicle coordinate system; the target coordinate system takes the target geometric center as its origin and is aligned with the principal inertial axes; the vehicle coordinate system takes the radar phase center as its origin, the vehicle's motion direction is the X-axis, the horizontal direction perpendicular to the motion direction is the Y-axis, and the vertical direction is the Z-axis; transform the point coordinates in the vehicle coordinate system to the target coordinate system through the translation vector and rotation matrix.

7. The method for rapid radar scattering testing based on polarization decomposition according to claim 6, characterized in that: The methods for obtaining and correcting the vehicle position and dynamic coordinate system transformation model include: The laser tracking data is updated by combining the Doppler frequency shift model and the least square method to obtain updated laser tracking data; The state vector is defined to include position deviation and attitude deviation. The position deviation is the translation deviation of the radar phase center relative to the target coordinate system; the attitude deviation is the attitude angle deviation of the vehicle. The state transfer matrix is constructed through the identity matrix, and the process noise is obtained according to the covariance matrix of the process noise. The covariance matrix of the process noise is set according to the sensor stability, and the predicted state vector is obtained according to the state transfer matrix and the process noise. The initial deviation is calibrated by the laser tracker, and the covariance matrix of the initial state estimation is set; Predict the state at the next moment based on the state transfer matrix; Obtaining the covariance of the predicted updated state estimate based on the state transfer matrix, the covariance matrix of the current state estimate, and the covariance matrix of the process noise; Acquire sensor observations consisting of RTK data, IMU data, and updated laser tracking data; The Kalman gain is calculated based on the covariance of the predicted updated state estimate, the sensor observation matrix composed of the sensor observation values, and the covariance matrix of the process noise; Update the state estimate based on the covariance matrix of the process noise, the Kalman gain, the sensor observations, and the sensor observation matrix; Update the covariance matrix of the state estimate based on the Kalman gain, the sensor observation matrix, and the covariance of the predicted updated state estimate; Determine whether the value of the updated covariance matrix is greater than the preset variance threshold. If so, re-update to obtain a new state estimate and, until the value of the covariance matrix of the updated state estimate is less than the preset variance threshold, obtain the corresponding updated state estimate, otherwise directly obtain the updated state estimate; According to the updated state estimation, the radar phase center position and attitude angle are corrected, the rotation matrix is updated according to the corrected attitude angle, and the vehicle position correction is converted to the target coordinate system according to the updated rotation matrix.

8. The method for rapid radar scattering testing based on polarization decomposition according to claim 7, characterized in that: Methods for obtaining updated laser tracking data include: Obtain the vibration acceleration vector of the IMU data, project the vibration acceleration vector to the radar line of sight direction through the radar line of sight direction vector, and obtain the effective acceleration; Integrate the effective acceleration to obtain the velocity, and combine it with the Doppler frequency shift formula to obtain the frequency shift; The effective acceleration is integrated twice to obtain the vibration displacement; combined with the radar wavelength, the theoretical phase error is calculated; In a static scenario, N sets of radar echo phase and theoretical phase error values are collected, and a matrix equation is obtained by constructing a time matrix based on the acquisition time, a polynomial coefficient matrix based on polynomial coefficients used to model the theoretical phase error as a polynomial form, and a difference matrix obtained from the difference between the radar echo phase and the theoretical phase error. The matrix equation is solved to obtain the polynomial coefficient matrix, the actual phase error is calculated in real time based on the polynomial coefficient matrix, and the actual phase error is deducted from the radar echo phase to obtain the true phase; the laser tracking data is updated according to the true phase to obtain updated laser tracking data.

9. The method for rapid radar scattering testing based on polarization decomposition according to claim 1, characterized in that: Methods for calculating the optimal test path of the mobile car include: The test area is discretely divided into P regional grids, and corresponding regional weights are assigned to different regions according to the polarization sensitive areas; According to the regional weight and path length, the corresponding fitness is calculated to maximize the sum of the weights of the coverage area of the regional grid and minimize the path length as the optimization goal; Initialize the initial pheromone concentration and pheromone volatilization rate of each regional grid; A heuristic function is constructed based on the area weight and the length between path nodes. The state transition probability is calculated based on the heuristic function and pheromone concentration. Each ant starts from the starting point and selects the next path node according to the probability until it reaches the end point. After each ant completes the path, the pheromone is updated according to the path quality. The path quality is calculated based on the ratio of the fitness value to the global optimal fitness and the pheromone volatility rate. After each round, the pheromone of the optimal path is enhanced by the ratio of fitness to path length; When the preset termination condition is reached, the optimal path with the highest fitness is selected as the optimal test path.

10. The rapid radar scattering test method based on polarization decomposition according to claim 1, characterized in that: Methods for acquiring the backscattered signal of a target include: Sampling and processing the received signal containing only interference and noise, and calculating the covariance matrix of interference plus noise; According to the incident direction of the desired signal, the desired direction steering vector is calculated using the array manifold formula; A weight vector is obtained by calculating the covariance matrix of the interference plus noise and the desired direction steering vector; The received signal containing only multipath interference is sampled and processed to obtain a covariance matrix of the received signal. The covariance matrix is eigen-decomposed to obtain eigenvalues and corresponding eigenvectors. After sorting the eigenvalues, the eigenvectors corresponding to the eigenvalues below a preset characteristic threshold are spanned into a noise subspace. Within a preset angle range of interest, a spatial spectrum is calculated for each angle based on the desired direction steering vector and the noise subspace, where the peak of the spatial spectrum corresponds to the direction of the interference wave. For the nth sampling, the a priori error is calculated based on the received signal, the expected signal and the corresponding weight vector; the gain vector is calculated based on the error covariance matrix and the received signal; the weight vector is updated based on the weight vector, the a priori error and the gain vector corresponding to the received signal of the last sampling; the error covariance matrix is updated based on the error covariance matrix, the gain vector and the received signal corresponding to the received signal of the last sampling; A covariance matrix loading is obtained, and the covariance matrix loading is added to the diagonal of the covariance matrix of interference plus noise to obtain an adjusted covariance matrix. Beamforming weights are calculated by adjusting the covariance matrix, and a backscattered signal of the target is formed according to the beamforming weights.

Images