Phased array radar beam forming method in shielding environment

Abstract

The invention discloses a phased array radar beam forming method in a shielding environment. The method comprises the following steps: S1, establishing a forming array covariance matrix of a phased array radar beam; s2, forming output distribution of radar beams is generated based on a generative adversarial network (GAN) generative model, and a guide target of a forming array covariance matrix is optimized in combination with an output probability value of a discriminant model of a GAN generative adversarial network, so that the generated radar beams are closer to real radar beams; and S3, the influence of the earth curvature and atmospheric refraction is considered, the formed target shielding interval of the phased array radar beam is compensated, and automatic forming of the radar beam is realized. According to the method, the problems of beam main lobe distortion and insufficient shielding compensation are effectively solved by establishing the array covariance matrix, optimizing the guiding target by using the GAN and combining earth curvature and atmospheric refraction compensation, and the method has the advantages of improving the beam forming quality, optimizing the array covariance matrix guiding target and automatically compensating the shielding interval.

Description

Technical Field

[0001] This application relates to the field of radar signal processing technology, and more specifically, to a method for beamforming of a phased array radar under obstructed conditions. Background Technology

[0002] Phased array radar beamforming technology is one of the key technologies in modern radar systems. It forms a beam in a specific direction by electronically controlling the phase and amplitude of the radiating elements in the array, enabling rapid scanning of the target. However, in practical applications, the radar beam propagation process is often affected by geographical obstacles such as mountains, hills, vegetation, and buildings, as well as environmental factors such as rain, fog, and electromagnetic interference, leading to beam main lobe energy attenuation and shape distortion, which seriously affects radar detection performance.

[0003] Various beamforming methods have been proposed in the prior art to address these problems. While holographic metasurface antenna-based methods offer flexible configuration of element states, their beam scanning range is limited by array aperture and element spacing. Polarization-phase modulation-based methods suffer from cross-polarization coupling issues under orthogonal electric field polarization conditions. Methods based on the multidimensional Newton orthogonal matched pursuit algorithm are heavily reliant on the target sparsity assumption. Methods based on decellularized millimeter-wave MIMO exhibit poor scalability in the high-frequency band. These methods generally suffer from unstable main lobe gain, large sidelobe level fluctuations, and abnormal variations in the number of array elements, making it difficult to guarantee radar detection quality in complex obstructed environments.

[0004] Especially when considering the effects of Earth's curvature and atmospheric refraction, traditional methods have limited ability to compensate for beam shielding angles, leading to a further decline in radar detection performance in shielded areas. How to effectively optimize the covariance matrix of beamforming arrays and improve beamforming quality in shielded environments has become a pressing technical challenge.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] The purpose of this invention is to provide a phased array radar beamforming method under obstructed environments to solve the problems existing in the prior art.

[0007] The above-mentioned technical objective of the present invention is achieved through the following technical solution:

[0008] A method for beamforming of a phased array radar in an obstructed environment includes the following steps:

[0009] S1. Establish the shaping array covariance matrix of the phased array radar beam;

[0010] S2. Based on the generative model of Generative Adversarial Network (GAN), the radar beam shaping output distribution is generated, and combined with the output probability value of the discriminative model of GAN, the guiding target of the shaping array covariance matrix is ​​optimized, so that the generated radar beam is closer to the real radar beam.

[0011] S3. Considering the effects of Earth's curvature and atmospheric refraction, compensation is made for the target obstruction range of the phased array radar beam to achieve automatic radar beam shaping.

[0012] Furthermore, the specific method for step S1 is as follows:

[0013] Array antenna by Composed of omnidirectional array elements, assuming the signal is a narrowband far-field signal, the array combiner output is obtained by adjusting the signal amplitude and phase within the channel through weight vectors, as shown in the following formula:

[0014]

[0015] In formulas (1) to (3), The signal amplitude within the channel; The phase of the signal within the channel; For weight vectors; For input signals; Radar radius; The signal period; The array output power in the direction of the main lobe; The input signal received by the antenna array element; This is the output signal for array combining; For the first The input signal received by each antenna element;

[0016] exist Based on this, the shaping array covariance matrix of the phased array radar beam is established, as shown in the following formula:

[0017]

[0018] In formula (4), The shaping array covariance matrix of the phased array radar beam; Main lobe energy; The signal amplitude within the main lobe beamwidth channel; It is white noise; This represents the array output power in the sidelobe direction.

[0019] Furthermore, the specific method for step S2 is as follows:

[0020] Assuming the radar beamforming data distribution in the obstructed environment is as follows: ;

[0021] Generative models in Generative Adversarial Networks (GANs) In this process, a set of random noise is input, and a radar beamforming image that meets the requirements of the obstructed environment is generated, thus obtaining the radar beamforming output distribution. ;

[0022] In the discriminative model of Generative Adversarial Networks (GANs) In this process, it is used to determine whether the input vector comes from a real occlusion environment, and to measure... and The distance between them outputs a probability value. Distinguish Does it closely resemble a real radar beam?

[0023] The objective function of the Generative Adversarial Network (GAN) for radar beamforming is established, as shown in the following formula:

[0024]

[0025] In formula (5), The target is guided by the generative adversarial network (GAN) for automatic radar beamforming. For discriminative models Predicted values ​​for actual beamforming; for For generated samples The predicted value;

[0026] Set the generative model No change, only training the discriminant model ,generate ;

[0027] This indicates the distribution of the generated radar beamforming data. Distribution of radar beamforming data in real-world environments The distance between them;

[0028] Then train the generative model. ,let Minimum, shrink and The distance between them is used to find the optimal solution for the Generative Adversarial Network (GAN), as shown in the following formula:

[0029]

[0030] In formula (6), The optimal output distribution for radar beamforming; The optimal output distribution for true radar beamforming; , Substituting into the objective function, we get:

[0031]

[0032] In formula (7), The target is guided by the covariance matrix of the phased array radar beamforming array; when Approaching At that time, the radar beam generated by the Generative Adversarial Network (GAN) is closer to the real radar beam.

[0033] Furthermore, the specific method for step S3 is as follows:

[0034] Based on the effects of Earth's curvature and atmospheric refraction, the shielding angle of the radar beam is calculated using the following formula:

[0035]

[0036] In formula (8), The shielding angle of the radar beam; This represents the height of the beam center axis above the ground at the slant distance. Radar altitude; Slope distance; Let be the equivalent Earth radius under atmospheric refraction; compensation is applied for the portion below the radar echo shielding angle, using the following formula:

[0037]

[0038] In formula (9), The reflectivity factor value after occlusion compensation; The actual reflectivity factor value observed by the radar; The compensation factor for the shading angle; when > At that time, the shading angle was effectively compensated. ≈0, the azimuth beam can still continue to form; according to ,Adjustment This enables the automatic shaping of phased array radar beams.

[0039] In summary, the present invention has the following beneficial effects:

[0040] In existing technologies, portable phased array radar beamforming methods face challenges such as main lobe energy attenuation and beam distortion under high obstruction environments. Traditional methods optimize beam directivity using techniques such as holographic metasurface antennas and polarization phase modulation, but these methods suffer from limitations such as limited beam scanning range and insufficient compensation for cross-polarization coupling. Especially under complex terrain and atmospheric refraction conditions, fixed-parameter models struggle to dynamically adapt to changes in the obstruction angle, leading to increased main lobe gain fluctuations and abnormally elevated sidelobe levels, severely impacting radar detection accuracy.

[0041] To address these issues, the inventors discovered that beam distortion is essentially the result of the combined effects of environmental interference and physical propagation. Traditional static compensation methods cannot adapt to dynamic occlusion scenarios, while Generative Adversarial Networks (GANs) possess the dual characteristics of dynamic generation and discriminative optimization. By combining the physical propagation model with a deep learning framework, an adaptive adjustment mechanism for beamforming parameters is constructed, simultaneously considering quantitative compensation for Earth's curvature and atmospheric refraction, forming a closed-loop optimization system. This approach breaks through the separation between environmental modeling and beam optimization in traditional methods, achieving coordinated control of environmental perception and beamforming.

[0042] This invention effectively solves the problems of beam main lobe distortion and insufficient occlusion compensation by establishing an array covariance matrix, using generative adversarial networks (GANs) to optimize the guiding target, and combining Earth curvature and atmospheric refraction compensation. It has the advantages of improving beamforming quality, optimizing array covariance matrix for guiding targets, and automatically compensating for occlusion intervals. Attached Figure Description

[0043] Figure 1 This is a flowchart of the phased array radar beamforming method under obstructed conditions as described in this invention.

[0044] Figure 2 This is a schematic diagram of the gridded radar polar coordinate data described in this invention.

[0045] Figure 3 This is a simulation diagram of each station of the radar network described in this invention.

[0046] Figure 4 This is the radar beamforming guide diagram described in this invention.

[0047] Figure 5 This is a diagram showing the occlusion compensation result described in this invention.

[0048] Figure 6 This is a diagram showing the radar beamforming performance results described in this invention. Detailed Implementation

[0049] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below with reference to the figures and specific embodiments.

[0050] like Figure 1 As shown, this application proposes a beamforming method for phased array radar under obstructed conditions, including the following steps:

[0051] S1. Establish the shaping array covariance matrix of the phased array radar beam;

[0052] S2. Based on the generative model of Generative Adversarial Network (GAN), the radar beam shaping output distribution is generated, and combined with the output probability value of the discriminative model of GAN, the guiding target of the shaping array covariance matrix is ​​optimized, so that the generated radar beam is closer to the real radar beam.

[0053] S3. Considering the effects of Earth's curvature and atmospheric refraction, compensation is made for the target obstruction range of the phased array radar beam to achieve automatic radar beam shaping.

[0054] The shaped array covariance matrix is ​​a mathematical representation describing the signal correlation among the array antenna elements. Specifically, it can be calculated by determining the ratio of main lobe energy to side lobe power, and is used to establish the basic model of beam energy distribution. The generative adversarial network (GAN) model generates beam distributions that conform to the characteristics of obstructed environments through adversarial training. This can be implemented using a deep convolutional neural network architecture to simulate beam attenuation patterns in real-world environments. The output probability value of the discriminative model is a quantitative indicator evaluating the similarity between the generated beam and the real beam. This can be achieved using the 0-1 range of the sigmoid function output, and is used to guide the dynamic adjustment of the covariance matrix parameters. Obstruction angle compensation corrects the beam obstruction angle deviation caused by the Earth's curvature and atmospheric refraction. This can be achieved by calculating the beam height at the slant range using an equivalent Earth radius model, and is used to eliminate the impact of terrain obstruction on the beam propagation path.

[0055] Specifically, the process begins by constructing a covariance matrix using signals received by an array antenna to quantify the energy distribution relationship between the main lobe and side lobes. Then, within a Generative Adversarial Network (GAN) framework, the generator produces candidate beam distributions based on random noise, while the discriminator compares real-world data with generated data to output probability values, driving iterative optimization of the guidance target parameters of the covariance matrix. Finally, the shielding angle is calculated based on an equivalent Earth radius model, and the beam coverage area is corrected using a reflectivity factor. These three steps form a closed-loop optimization system, where the physical model provides environmental constraints, and the GAN enables intelligent search of the parameter space, collectively ensuring the directionality and anti-interference capabilities of beamforming.

[0056] Through the above technical solutions, this application effectively suppresses the main lobe energy attenuation under obstructed environments, reduces beam shape distortion, and improves the spatial resolution of radar detection information. By dynamically adjusting the covariance matrix to guide the target, the adaptability of the beam pattern to complex terrain is enhanced. Combined with the obstruction angle compensation mechanism of the physical propagation model, the signal blind zone caused by terrain obstruction is significantly reduced, ensuring the detection reliability of the radar system in harsh environments.

[0057] The specific method for step S1 is as follows:

[0058] Array antenna by Composed of omnidirectional array elements, assuming the signal is a narrowband far-field signal, the array combiner output is obtained by adjusting the signal amplitude and phase within the channel through weight vectors, as shown in the following formula:

[0059]

[0060] In formulas (1) to (3), The signal amplitude within the channel; The phase of the signal within the channel; For weight vectors; For input signals; Radar radius; The signal period; The array output power in the direction of the main lobe; The input signal received by the antenna array element; This is the output signal for array combining; For the first The input signal received by each antenna element;

[0061] exist Based on this, the shaping array covariance matrix of the phased array radar beam is established, as shown in the following formula:

[0062]

[0063] In formula (4), The shaping array covariance matrix of the phased array radar beam; Main lobe energy; The signal amplitude within the main lobe beamwidth channel; It is white noise; This represents the array output power in the sidelobe direction.

[0064] Among them, omnidirectional array elements refer to array units capable of uniformly radiating or receiving electromagnetic waves within a 360-degree horizontal plane. Their isotropic characteristics allow beam pointing adjustment to be unrestricted by the array element pattern. Narrowband far-field signals refer to electromagnetic waves with a bandwidth much smaller than the carrier frequency and satisfying the Fraunhofer diffraction condition; their propagation model can be simplified to the plane wavefront assumption. Weight vectors are complex vectors used to control the amplitude and phase distribution of signals in each channel of the array. By adjusting this vector, the spatial response characteristics of the beam can be dynamically changed. The shaped array covariance matrix is ​​a mathematical structure reflecting the statistical correlation of the array's output signals. Specifically, it can be calculated from the expected value of the outer product of the received signal samples, and its element values ​​characterize the degree of signal correlation between different array elements.

[0065] Specifically, a narrowband far-field signal is received by an array antenna composed of omnidirectional elements. Weight vectors are used to weight the amplitude and phase of the signals in each channel, forming a spatially selective synthetic beam. The array combiner output signal is obtained by superimposing the weighted received signals from each element; its power expression can be decomposed into output components in the main lobe direction and side lobe directions. The established covariance matrix uses the main lobe energy parameter to maintain the gain in the target direction, the main lobe beamwidth parameter to control the beamwidth, the white noise parameter to quantify the level of environmental interference, and the side lobe output power parameter to suppress signal leakage in non-target directions. This matrix comprehensively characterizes the key performance indicators of beamforming through mathematical modeling, providing a quantifiable objective function for subsequent optimization algorithms.

[0066] Compared to existing technologies, traditional methods employ fixed array weights or single-parameter optimization strategies, making it difficult to simultaneously achieve both main lobe focusing and side lobe suppression capabilities. For example, holographic metasurface-based schemes are limited by element spacing constraints, resulting in a restricted beam scanning range; polarization modulation-based schemes cannot effectively compensate for attenuation due to cross-polarization coupling. This scheme constructs a multi-dimensional covariance matrix to jointly model parameters such as main lobe energy, beam broadening, noise suppression, and side lobe leakage, forming a systematic beamforming optimization framework. Compared to the isolated treatment of performance indicators in existing technologies, this matrix can more comprehensively reflect the overall performance requirements of beamforming.

[0067] Through the above technical solution, this application effectively solves the performance fluctuation problem caused by unreasonable array covariance matrix construction. The assumption of omnidirectional array elements and narrowband far-field signals reduces model complexity, dynamic adjustment of weight vectors enables flexible control of beam shape, and multi-parameter joint modeling of the covariance matrix provides a quantitative evaluation benchmark for beamforming quality. This scheme can stabilize the main lobe gain level, suppress abnormal fluctuations in side lobe levels, and optimize the signal correlation characteristics between array elements, thereby improving the overall quality of radar detection information.

[0068] The specific method for step S2 is as follows:

[0069] Assuming the radar beamforming data distribution in the obstructed environment is as follows: ;

[0070] Generative models in Generative Adversarial Networks (GANs) In this process, a set of random noise is input, and a radar beamforming image that meets the requirements of the obstructed environment is generated, thus obtaining the radar beamforming output distribution. ;

[0071] In the discriminative model of Generative Adversarial Networks (GANs) In this process, it is used to determine whether the input vector comes from a real occlusion environment, and to measure... and The distance between them outputs a probability value. Distinguish Does it closely resemble a real radar beam?

[0072] The objective function of the Generative Adversarial Network (GAN) for radar beamforming is established, as shown in the following formula:

[0073]

[0074] In formula (5), The target is guided by the generative adversarial network (GAN) for automatic radar beamforming. For discriminative models Predicted values ​​for actual beamforming; for For generated samples The predicted value;

[0075] Set the generative model No change, only training the discriminant model ,generate ;

[0076] This indicates the distribution of the generated radar beamforming data. Distribution of radar beamforming data in real-world environments The distance between them;

[0077] Then train the generative model. ,let Minimum, shrink and The distance between them is used to find the optimal solution for the Generative Adversarial Network (GAN), as shown in the following formula:

[0078]

[0079] In formula (6), The optimal output distribution for radar beamforming; The optimal output distribution for true radar beamforming; , Substituting into the objective function, we get:

[0080]

[0081] In formula (7), The target is guided by the covariance matrix of the phased array radar beamforming array; when Approaching At that time, the radar beam generated by the Generative Adversarial Network (GAN) is closer to the real radar beam.

[0082] In Generative Adversarial Networks (GANs), the generative model refers to a network structure that generates a simulated beamforming distribution from random noise input. This can be implemented using a multi-layer convolutional neural network, simulating signal attenuation characteristics under occlusion to generate physically plausible beam shapes. The discriminative model is a network structure used to evaluate the similarity between generated and real samples. This can be implemented using a combination of fully connected layers and activation functions, quantifying distribution differences by calculating probability values ​​to provide optimization feedback. The objective function is a mathematical expression that measures the difference between the generated and real distributions. This can be constructed using cross-entropy or Wasserstein distance, unifying the optimization direction of the generation and discrimination processes. The optimal output distribution refers to the equilibrium state reached when the GAN training converges. This can be solved iteratively using the gradient descent algorithm, making the statistical characteristics of the generated beams approximate real-world data.

[0083] Specifically, this technical solution dynamically optimizes the beamforming process through the adversarial training mechanism of Generative Adversarial Networks (GANs). After receiving random noise input, the generative model uses neural network parameters to generate beamforming images with occlusion adaptability, simulating signal propagation characteristics in complex environments. The discriminative model performs binary classification on the input data, outputting probability values ​​that represent the distribution distance between generated and real samples. During alternating training, the generative model parameters are first fixed, and the discriminative model is optimized by maximizing the discrimination accuracy to precisely measure distribution differences. Subsequently, the discriminative model parameters are fixed again, and the generative model is optimized by minimizing the differences between generated and real samples, gradually reducing the distribution distance. When training reaches Nash equilibrium, the beamforming distribution output by the generative model achieves optimal matching with the distribution of real-world data. At this point, the guiding target of the covariance matrix is ​​dynamically adjusted to the optimal state for adapting to the occlusion environment.

[0084] Compared to existing technologies, traditional beamforming methods typically employ fixed compensation strategies or static modeling to address occlusion issues, making it difficult to accurately match the complex and ever-changing distribution of real-world environments. Our proposed solution, however, utilizes the dynamic adversarial mechanism of Generative Adversarial Networks (GANs) to autonomously learn signal attenuation patterns under occlusion, achieving adaptive adjustment of the beamforming distribution without the need for manual modeling. Compared to methods based on orthogonal projection or sparse reconstruction, this technique effectively avoids the error accumulation problem caused by pre-defined model assumptions, significantly improving the spatial matching accuracy between beamform and the real environment.

[0085] Through the above technical solutions, this application solves the problems of abnormal main lobe gain, side lobe level fluctuations, and unstable array element number caused by the mismatch between beamforming output and the actual distribution under obstructed environments. The adversarial training mechanism of Generative Adversarial Networks (GANs) enables the statistical characteristics of the generated beam to gradually approximate real environmental data. By dynamically optimizing the covariance matrix to guide the target, it ensures that the beamforming results can maintain stable main lobe energy concentration and side lobe suppression capability under complex obstructed conditions, thereby improving the quality and reliability of radar detection information.

[0086] The specific method for step S3 is as follows:

[0087] The scanning data of portable phased array radar beams is a data set stored in polar coordinates. The radar data is transformed into a Cartesian coordinate system with uniform spatial distribution to describe the data, enabling the comparison of gridded radar data within the common scanning coverage area.

[0088] like Figure 2 As shown, during the conversion of radar data to Cartesian coordinates, the coordinates of the two grid points deviate due to the influence of the obstructing environment, affecting the beamforming quality. The radar's obstruction angle is defined as the radar beam angle. Above the obstruction angle, the portion is unobstructed; below the obstruction angle, the portion is obstructed.

[0089] Based on the effects of Earth's curvature and atmospheric refraction, the shielding angle of the radar beam is calculated using the following formula:

[0090]

[0091] In formula (8), The shielding angle of the radar beam; This represents the height of the beam center axis above the ground at the slant distance. Radar altitude; Slope distance; Let be the equivalent Earth radius under atmospheric refraction; compensation is applied for the portion below the radar echo shielding angle, using the following formula:

[0092]

[0093] In formula (9), The reflectivity factor value after occlusion compensation; The actual reflectivity factor value observed by the radar; The compensation factor for the shading angle; when > At that time, the shading angle was effectively compensated. ≈0, the azimuth beam can still continue to form; according to ,Adjustment This enables the automatic shaping of phased array radar beams.

[0094] The shielding angle refers to the critical angle at which the radar beam propagation path is blocked by terrain or obstacles. It can be calculated using the geometric relationship between radar altitude, slant range, and equivalent Earth radius, and is used to quantify the shielding effect of Earth's curvature and atmospheric refraction on the beam propagation path. The compensation factor is a dynamic adjustment coefficient used to eliminate signal attenuation below the shielding angle. It can be implemented using the proportional relationship between the reflectivity factor and the shielding angle threshold, and is used to maintain the stability of the reflectivity factor when the beam propagation path is obstructed. The equivalent Earth radius refers to the radius of curvature of the Earth after correcting for atmospheric refraction effects. It can be implemented using empirical parameters from the standard atmospheric refraction model, and is used to accurately simulate the propagation trajectory of the radar beam under actual atmospheric conditions.

[0095] Specifically, firstly, a geometric model is used to establish a functional relationship between the beam center axis height above the ground and the radar altitude, slant range, and equivalent Earth radius. The actual shielding angle under the current propagation path is calculated, transforming the beam path offset caused by Earth's curvature and atmospheric refraction into a quantifiable shielding threshold. Then, based on the comparison between the actual observed reflectivity factor and the shielding angle, a compensation factor is dynamically generated to correct the reflectivity factor. When the observed value exceeds the shielding angle, the compensation factor adjusts the reflectivity factor to near zero, eliminating abnormal signal attenuation caused by terrain obstruction and ensuring the continuity of beamforming in that azimuth. Finally, the compensated reflectivity factor is fed back to the beamforming control system, forming a closed-loop adjustment mechanism to achieve adaptive beamforming in complex terrain and atmospheric environments.

[0096] Compared to existing technologies, current methods typically ignore the combined effects of Earth's curvature and atmospheric refraction when dealing with terrain shading, compensating for the shading area only through a fixed threshold. This leads to inaccurate shading angle calculations and insufficient compensation. For example, methods based on holographic metasurface antennas lack a correlation model between physical environmental parameters and the shading angle, making them unable to adapt to beam propagation variations under different terrain and atmospheric conditions. Methods based on polarization-phase modulation rely on orthogonal subspace projection, making it difficult to maintain stable compensation effects in complex refractive environments. This proposed solution establishes a joint geometric-physical model of the beam propagation path, accurately quantifies the impact of environmental factors on the shading angle, and combines a dynamic compensation mechanism to achieve real-time correction of the shading area, thus solving the problems of inaccurate shading angle calculations and compensation lags in traditional methods.

[0097] Through the above technical solution, this application can effectively suppress the abnormal increase of the shielding angle caused by the curvature of the earth and atmospheric refraction, maintain the stability of the reflectivity factor through a dynamic compensation mechanism, ensure the continuous shaping capability of the radar beam in complex terrain and atmospheric environment, and improve the target detection accuracy and signal quality in the shielded area.

[0098] Example

[0099] Considering a uniform array with 20 array elements, a desired target direction of -10°, and a signal bandwidth of 50 μs, the boundary flow field in region A is simulated at a sampling frequency of 0.95 GHz to verify the effectiveness of the portable phased array radar beamforming method for obstructed environments optimized by the GAN algorithm. The radar network observation platform consists of four X-band radars, including one radar signal with a klystron transmitting system and three single-array radar signals with solid-state transmitting systems.

[0100] Simulation results of each site in the radar network are as follows: Figure 3 As shown, A is the center point of the radar station; B, C, and D are the stations where the radar signal needs to reach. Continuous radar signals are sent to stations A, B, C, and D to simulate radar signals in an obstructed environment, replicating a large number of similar signals. The simulation parameters are shown in Table 1 below.

[0101] Table 1 Interference Simulation Parameters

[0102]

[0103] As shown in Table 1, a large number of continuous wave signals with the same center frequency and similar frequency domain width as the radar signal are used to simulate the portable phased array radar signal in the obstructed environment, ensuring the radar beam simulation effect.

[0104] Automatic Radar Beamforming and Guiding Analysis

[0105] Narrowband radar signals with different beam directions are transmitted toward point A. The signal wavelength is 1, the main polarization direction is φ, and the cross polarization direction is θ, forming a phased array radar beam array.

[0106] Beamforming guidance situation as follows Figure 4 As shown, ① is the radar beam steering pattern with a transmission angle of {(φ,θ)=(20°,40°)}; ② is the radar beam steering pattern optimized from ①; ③ is the radar beam steering pattern with a transmission angle of {(φ,θ)=(30°,120°)}; ④ is the radar beam steering pattern optimized from ③; ⑤ is the radar beam steering pattern with a transmission angle of {(φ,θ)=(45°,230°)}; and ⑥ is the radar beam steering pattern optimized from ⑤. Compared with ①, ③, and ⑤, ②, ④, and ⑥ have significantly higher main lobe gain. During beamforming, they can more effectively concentrate energy in the target direction, have stronger blocking loss capability in the obstructed area, and have better automatic beamforming effect.

[0107] The reflectivity factor at the elevation angle of the radar network is obtained, and the azimuth of the partially blocked radar beam is compensated for, resulting in more accurate radar detection results.

[0108] Obstruction compensation details are as follows Figure 5 As shown, (1), (3), and (5) are the mean radar beam reflectivity factor (PPI) maps for stations B, C, and D, respectively; (2), (4), and (6) are the mean radar beam reflectivity factor (PPI) maps for (1), (3), and (5) after occlusion compensation, respectively. The red dashed box area represents the occlusion area. After occlusion compensation, the radar data in (2), (4), and (6) are clearer, and the information from the automatic radar beamforming feedback is more accurate.

[0109] In the simulation of automatic radar beamforming, the goal is to minimize the number of array elements used, while ensuring the stability of the main gain and peak sidelobe level. This results in the robust performance of automatic radar beamforming.

[0110] Radar beamforming performance results are as follows Figure 6As shown, a) presents the performance results of the portable phased array radar beamforming method based on a holographic metasurface antenna; b) presents the performance results of the portable phased array radar beamforming method based on polarization-phase modulation; and c) presents the performance results of the portable phased array radar beamforming method for obstructed environments optimized by the GAN algorithm designed in this paper. This paper utilizes the GAN algorithm to simulate obstructed targets of different attitudes, sizes, and materials, generating data on Doppler effects and scattering characteristic changes to cover extreme cases. The generator predicts the optimal beam parameters, and the discriminator evaluates the beam steering map quality, effectively improving the quality of automatic radar beamforming. The final beamforming results show that the main lobe gain fluctuates around 50dB, the peak sidelobe level fluctuates around -5dB, and the number of array elements fluctuates around 500. It can form phased array radar beams under conditions of high directivity, high interference immunity, and high degrees of freedom, demonstrating good automatic radar beamforming performance. Therefore, the method designed in this paper can simulate different terrains and weather conditions according to the actual environment. During the generation of adversarial training, it can simulate interference signals, adjust the beam direction in real time, avoid interference signals, and form a radar beam with better detection effect, which plays an important role in the detection of obscured areas.

[0111] The portable phased array radar beamforming method for obstructed environments of this invention reduces the impact of errors on beamforming performance by reconstructing the covariance matrix of the received signal. A GAN algorithm is used to optimize the steering vector of the matrix, better adapting to complex obstructed environments. By using a reflectivity factor to compensate for the obstruction effect in beamforming, and with echo compensation, the automatic radar beamforming effect is effectively improved, providing quality assurance for radar information in obstructed environments.

[0112] In this document, the terms "upper," "lower," "front," "back," "left," "right," "top," "bottom," "inner," "outer," "vertical," and "horizontal," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only used for the clarity of expressing the technical solution and for the convenience of description, and therefore should not be construed as limiting the present invention.

[0113] In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, which includes not only the elements listed but also other elements not expressly listed.

[0114] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for beamforming of a phased array radar under obstructed conditions, characterized in that, Includes the following steps: S1. Establish the shaping array covariance matrix of the phased array radar beam; S2. Based on the generative model of Generative Adversarial Network (GAN), the radar beam shaping output distribution is generated, and combined with the output probability value of the discriminative model of GAN, the guiding target of the shaping array covariance matrix is ​​optimized, so that the generated radar beam is closer to the real radar beam. S3. Considering the effects of Earth's curvature and atmospheric refraction, compensation is made for the target obstruction range of the phased array radar beam to achieve automatic radar beam shaping.

2. The phased array radar beamforming method under obstructed conditions according to claim 1, characterized in that, The specific method for step S1 is as follows: Array antenna by Composed of omnidirectional array elements, assuming the signal is a narrowband far-field signal, the array combiner output is obtained by adjusting the signal amplitude and phase within the channel through weight vectors, as shown in the following formula: In formulas (1) to (3), The signal amplitude within the channel; The phase of the signal within the channel; For weight vectors; For input signals; Radar radius; The signal period; The array output power in the direction of the main lobe; The input signal received by the antenna array element; This is the output signal for array combining; For the first The input signal received by each antenna element; exist Based on this, the shaping array covariance matrix of the phased array radar beam is established, as shown in the following formula: In formula (4), The shaping array covariance matrix of the phased array radar beam; Main lobe energy; The signal amplitude within the main lobe beamwidth channel; It is white noise; This represents the array output power in the sidelobe direction.

3. The phased array radar beamforming method under obstructed conditions according to claim 1, characterized in that, The specific method for step S2 is as follows: Assuming the radar beamforming data distribution in the obstructed environment is as follows: ; Generative models in Generative Adversarial Networks (GANs) In this process, a set of random noise is input, and a radar beamforming image that meets the requirements of the obstructed environment is generated, thus obtaining the radar beamforming output distribution. ; In the discriminative model of Generative Adversarial Networks (GANs) In this process, it is used to determine whether the input vector comes from a real occlusion environment, and to measure... and The distance between them outputs a probability value. Distinguish Does it closely resemble a real radar beam? The objective function of the Generative Adversarial Network (GAN) for radar beamforming is established, as shown in the following formula: In formula (5), The target is guided by the generative adversarial network (GAN) for automatic radar beamforming. For discriminative models Predicted values ​​for actual beamforming; for For generated samples The predicted value; Set the generative model No change, only training the discriminant model ,generate ; This indicates the distribution of the generated radar beamforming data. Distribution of radar beamforming data in real-world environments The distance between them; Then train the generative model. ,let Minimum, shrink and The distance between them is used to find the optimal solution for the Generative Adversarial Network (GAN), as shown in the following formula: In formula (6), The optimal output distribution for radar beamforming; The optimal output distribution for true radar beamforming; , Substituting into the objective function, we get: In formula (7), The target is guided by the covariance matrix of the phased array radar beamforming array; when Approaching At that time, the radar beam generated by the Generative Adversarial Network (GAN) is closer to the real radar beam.

4. The automatic beamforming method for phased array radar under obstructed conditions according to claim 1, characterized in that, The specific method for step S3 is as follows: Based on the effects of Earth's curvature and atmospheric refraction, the shielding angle of the radar beam is calculated using the following formula: In formula (8), The shielding angle of the radar beam; This represents the height of the beam center axis above the ground at the slant distance. Radar altitude; Slope distance; Let be the equivalent Earth radius under atmospheric refraction; compensation is applied for the portion below the radar echo shielding angle, using the following formula: In formula (9), The reflectivity factor value after occlusion compensation; The actual reflectivity factor value observed by the radar; The compensation factor for the shading angle; when > At that time, the shading angle was effectively compensated. ≈0, the azimuth beam can still continue to form; according to ,Adjustment This enables the automatic shaping of phased array radar beams.

Images