A method for calibrating amplitude and phase errors of an ultra-high frequency radar array based on MSCO-PSO

Abstract

The application discloses a kind of based on MSCO-PSO's ultra-high frequency radar array amplitude phase error calibration method, including obtaining radar echo data, the covariance matrix of array data is constructed, to obtain the coarse measurement of angle of arrival, then utilize MSCO-PSO algorithm to estimate the phase error and amplitude error of each channel, to calibrate steering vector, estimate the angle of arrival after calibration, until meet the requirement, output the estimated value of amplitude and phase error.The MSCO-PSO algorithm of the scheme is based on the framework of traditional PSO algorithm, by introducing Latin hypercube sampling for population initialization, combined with adaptive particle classification strategy, improved speed update mechanism and Gaussian random disturbance, enhance the global search ability and convergence stability of algorithm in high-dimensional channel amplitude phase error parameter space, can effectively estimate array channel amplitude phase error, restore array directional diagram characteristics, significantly improve the angle measurement precision and radial flow field space consistency, provide reliable array angle measurement basis for subsequent section surface flow velocity inversion.

Description

Technical Field

[0001] This invention belongs to the field of antenna array channel calibration, specifically relating to a method for calibrating the amplitude and phase errors of an ultra-high frequency radar array based on MSCO-PSO. Background Technology

[0002] Ultra-high frequency (UHF) radar offers advantages such as non-contact operation, all-weather capability, and wide coverage. Because its operating frequency band matches the roughness scale of the river surface, it can effectively acquire information from the scattered echoes from the river surface, gradually becoming an important means of detecting river surface velocity. Radar measures flow by inverting the surface velocity of the river cross-section through the acquisition of electromagnetic scattered echo information from the river surface, avoiding the safety risks and operational limitations associated with direct contact between the instrument and the water body in traditional measurement methods. It is particularly suitable for long-term continuous observation under conditions of wide channels, high flow velocities, and floods.

[0003] In practical multi-channel radar systems, each receiving channel often exhibits inconsistent amplitude and phase responses due to factors such as device differences and temperature drift. Array amplitude and phase errors disrupt ideal radiation pattern characteristics, causing a decrease in main lobe gain, main lobe broadening, and an increase in side lobes, thus weakening the system's directivity. More critically, channel inconsistencies lead to spatial spectrum distortion in the MUSIC algorithm, producing false or shifted peaks. This can cause errors in angle-of-arrival estimation under multi-target conditions, which in turn propagate to the subsequent velocity inversion process, reducing the accuracy and reliability of cross-sectional surface velocity estimation. Therefore, calibrating the array channel amplitude and phase errors is crucial for ensuring flow measurement performance.

[0004] Traditional array calibration often relies on active reference sources or prior information, such as estimating error parameters through active calibration signals or least-squares fitting. These methods are effective in controlled environments, but in field deployments, they are often limited by the difficulty of setting up reference sources and the destructive effect of clutter noise on the stability of the calibration signal, making it difficult to guarantee accuracy. To reduce dependence on reference sources, passive calibration strategies have been developed, such as utilizing the covariance structure of received data. However, passive methods typically rely on certain signal assumptions, and their performance is sensitive to the number of signal sources, signal-to-noise ratio, and number of snapshots, lacking robustness in complex interference or multi-source scenarios.

[0005] Essentially, amplitude and phase error joint calibration is a nonlinear, non-convex high-dimensional optimization problem. Traditional gradient-based optimization algorithms are susceptible to local optima due to initial conditions. Therefore, recent work has increasingly transformed it into a parameter optimization problem, employing intelligent optimization algorithms to search for optimal solutions in a high-dimensional parameter space. Among these, the PSO algorithm, due to its simple structure, few parameters, and ease of integration with angle measurement algorithms such as MUSIC, is widely used for array calibration and signal angle of arrival estimation. However, traditional PSO may still suffer from slow convergence, insufficient population diversity, and premature local optima under high-dimensional, multi-source, and low signal-to-noise ratio conditions. To address these issues, improved algorithms such as IWO-PSO enhance global search capabilities by introducing diffusion mechanisms, improving calibration accuracy and robustness to some extent. However, they may still face shortcomings such as slower convergence in later stages, insufficiently adaptive perturbation strategies, and parameter sensitivity. Therefore, it is necessary to further design more efficient and stable optimization methods to improve the convergence speed, global search capability, and result consistency of array amplitude and phase calibration in complex scenarios. Summary of the Invention

[0006] To address the aforementioned problems, the present invention aims to provide a method for calibrating the amplitude and phase errors of ultra-high frequency radar arrays based on Multi-Strategy Channel-Optimized Particle Swarm Optimization (MSCO-PSO). Building upon the traditional particle swarm optimization algorithm framework, this method introduces Latin hypercube sampling for population initialization and combines an adaptive particle classification strategy, an improved velocity update mechanism, and Gaussian random perturbation to enhance the algorithm's global search capability and convergence stability in the high-dimensional channel amplitude and phase error parameter space, thereby achieving the calibration of array amplitude and phase errors.

[0007] The specific technical solution for achieving the objective of this invention is as follows:

[0008] A method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO includes the following steps:

[0009] Step 1: Acquire multi-channel echo data received by the UHF radar array, perform signal processing on the echo data, and obtain the range Doppler spectrum;

[0010] Step 2: Obtain the effective spectral points based on the distance-Doppler spectrum and construct the covariance matrix of the array data;

[0011] Step 3: Use the MUSIC algorithm to process the covariance matrix of the array data to obtain a coarse measurement of the angle of arrival. ;

[0012] Step 4: Coarse measurement based on angle of arrival The phase error of each channel is estimated based on the MSCO-PSO algorithm;

[0013] Step 5: Coarse measurement based on angle of arrival Based on the estimated phase error, the amplitude error of each channel is estimated using the MSCO-PSO algorithm;

[0014] Step 6: Calibrate the steering vector using the estimated amplitude and phase errors, and apply the calibrated steering vector to the MUSIC algorithm to estimate the calibrated angle of arrival. ;

[0015] Step 7: Calculate the difference between the estimated angle of arrival before and after calibration. If the difference is greater than the threshold... If the number of iterations is less than or equal to the maximum number of iterations, return to step 4; if the difference is less than the threshold... If the number of iterations exceeds the maximum number of iterations, the algorithm terminates and outputs estimated values ​​of amplitude and phase errors to calibrate the array amplitude and phase errors.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0017] (1) Low initial requirements and no need for external calibration source: This invention adopts a passive calibration approach, which only uses the radar echo data itself to estimate the channel error. It does not rely on any external reference signal source or pre-acquired calibration field data. It can be directly applied in the field environment, which overcomes the problem of difficulty in obtaining calibration source in traditional active calibration and has higher scene adaptability.

[0018] (2) Strong global search capability and high calibration accuracy: The MSCO-PSO algorithm introduced in this invention has been enhanced by multiple strategies on the basis of traditional PSO, including Latin hypercube population initialization, adaptive subgroup partitioning, improved velocity update formula and Gaussian perturbation mechanism. These improvements have greatly improved the global search capability and convergence stability of the algorithm, and effectively prevented particles from getting trapped in local optima;

[0019] (3) Improved direction finding and velocity inversion performance: Through calibration, the amplitude and phase responses of each channel are restored to consistency, and the array pattern distortion is corrected. After calibration using the method of this invention, the peak of the MUSIC spatial spectrum is more prominent and its position is more accurate, and the angle of arrival estimation error is reduced. Accurate DOA estimation makes subsequent velocity inversion calculations more reliable. After calibration, the radial velocity distribution is smoother and more continuous, and the correlation with the measurement results of the acoustic Doppler current meter is significantly improved, with a significant reduction in both mean absolute error and root mean square error;

[0020] (4) Robustness and applicability to complex environments: The method of this invention maintains robust performance even in multi-target, low signal-to-noise ratio, and strong interference scenarios. After calibration, more effective Doppler frequencies are utilized, greatly improving data quality, and the method can still extract reliable information in complex backgrounds. Long-term observation and analysis also prove that the stability and consistency of radar flow velocity inversion results are significantly improved after array calibration using this method. Therefore, this invention is particularly suitable for environments with no reference source and unstable channel conditions in the field, and can stably improve the measurement accuracy of UHF radar systems over a long period of time, possessing significant engineering application value.

[0021] The present invention will be further described below with reference to specific embodiments. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the process for the MSCO-PSO-based ultra-high frequency radar array amplitude and phase error calibration method of the present invention.

[0023] Figure 2 This is a schematic diagram of the MSCO-PSO algorithm flow of the present invention.

[0024] Figure 3 This is a schematic diagram comparing the antenna radiation patterns before and after calibration in an embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram showing the comparison of the MUSIC spatial spectrum before and after calibration in an embodiment of the present invention.

[0026] Figure 5 This is a flow field diagram before channel calibration in an embodiment of the present invention.

[0027] Figure 6 This is a flow field diagram after channel calibration in an embodiment of the present invention. Detailed Implementation

[0028] Example

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0031] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0032] Combination Figure 1 A method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO includes the following steps:

[0033] Step 1: Acquire multi-channel echo data received by the UHF radar array, perform signal processing on the echo data, and obtain the range Doppler spectrum:

[0034] The radar echo data is first subjected to Fourier transform in the fast time dimension to achieve equivalent pulse compression and obtain the range spectrum.

[0035] Subsequently, coherent accumulation and spectral analysis of the echoes from each distance cell are performed in the slow time dimension to obtain the corresponding Doppler spectrum distribution.

[0036] Through the two-stage processing described above, the target's distance and velocity information can be obtained simultaneously. Mathematically, this process of determining distance and velocity is equivalent to performing two Fourier transforms on the echo signal.

[0037] Step 2: Set the signal-to-noise ratio threshold based on the distance-Doppler spectrum, separate the first-order spectral region, obtain the effective spectral points, and construct the covariance matrix of the array data:

[0038] Step 2-1: Smooth the original echo Doppler spectrum to suppress spikes and glitches caused by random noise. Introduce a noise threshold to filter spectral points. For example, consider frequency points with spectral power 15dB above the noise floor as effective echo regions to eliminate spectral points with low signal-to-noise ratios within the positive and negative first-order Bragg peak regions. This improves the signal-to-noise ratio of effective spectral points while ensuring the integrity of the spectral region.

[0039] Step 2-2: Calculate the power change rate between adjacent spectral points to obtain the corresponding difference spectrum distribution. In the difference spectrum, the frequency point corresponding to the position where the echo power change is most significant is taken as the outer boundary of the positive and negative first-order Bragg peak regions, thereby realizing the division of the effective spectral region.

[0040] Steps 2-3: Extract complex echo data of each receiving channel at the corresponding frequency points within the effective spectrum region, and construct array snapshot vectors according to the array channel order; perform statistical averaging on multiple array snapshot vectors, that is, accumulate the product of each array snapshot vector and its conjugate transpose and take the average, thereby obtaining the covariance matrix of the array data.

[0041] Step 3: Use the MUSIC algorithm to process the covariance matrix of the array data to obtain a coarse measurement of the angle of arrival. :

[0042] The covariance matrix is ​​decomposed into eigenvalues ​​to separate the signal subspace and the noise subspace. Based on the orthogonality between the array steering vector and the noise subspace, the MUSIC spatial spectrum function is constructed.

[0043] Perform a spectral peak search within a preset angle search range, and obtain the angle corresponding to the spectral peak as the coarse angle of arrival of the signal at that frequency point. .

[0044] Step 4, Combining Figure 2 Coarse measurement based on angle of arrival The phase error of each channel is estimated based on the MSCO-PSO algorithm:

[0045] Step 4-1: Initialize the population using the Latin hypercube sampling method. Latin hypercube sampling is a multi-dimensional stratified sampling method. Compared with simple random sampling, this method can generate more uniformly distributed random samples, thereby improving the individual diversity of the initial population.

[0046] In the phase error estimation stage, the particle position vector is represented as: Where M is the number of array elements. Let be the phase error of the m-th channel. The first channel is the reference channel, and its phase error is fixed at 0. Therefore, the actual optimization dimension is The basic steps for extracting N samples in a d-dimensional space are as follows:

[0047] (1) Set the range of values ​​for each dimension Divide it into N segments of equal length that do not overlap;

[0048] (2) Randomly select a data point from each segment interval of each dimension;

[0049] (3) Randomly select one data point from the data points selected in (2) for each dimension to form a d-dimensional data vector, thereby forming N d-dimensional vector samples;

[0050] Step 4-2: Begin population iteration:

[0051] To expand the search area and improve the global search capability of the algorithm, this invention employs a multi-population information interaction strategy. An adaptive classification coefficient divides the population into an elite subpopulation and a general subpopulation. These two subpopulations evolve according to different speed and position update formulas. Finally, the subpopulations are merged for information interaction, thereby improving the algorithm's optimization performance. Specifically, before each iteration, an adaptive classification coefficient based on the norm is used to classify the particle population levels.

[0052] Before each iteration, the fitness value of each particle is calculated based on its position vector. Then, the particles are sorted from highest to lowest fitness value. Finally, the sorted population is divided according to the adaptive classification coefficient. Divided into two subpopulations, when the particle ordinal number At that time, particles with better fitness values ​​were classified into the elite subpopulation, while the remaining particles were classified into the ordinary subpopulation. The calculation formula is:

[0053]

[0054] in, The number of particles in the population. This represents the current iteration number. It is the current population fitness vector. The infinite norm represents the extreme value of fitness. Let L1 be the norm, and let represent the sum of fitness scores.

[0055] Step 4-3: The improved population velocity update formula guides particle convergence. The position of the particle population is updated based on the position iteration formula, and the optimal particle is determined according to the fitness function.

[0056] The position update formula for the particle population is:

[0057]

[0058] in, Let i be the updated position of the i-th particle population. This represents the position of the i-th particle population before the update. The updated velocity of the i-th particle population;

[0059] In this study, the elite subpopulation has higher particle fitness values, containing more effective information. To maintain its superior information, this invention employs a PSO velocity update formula with improved inertia weights. The ordinary subpopulation has disadvantaged particle positions. This invention introduces a random perturbation factor to ensure that particles explore a larger area in the early stages of the algorithm and converge quickly in the later stages.

[0060] Accordingly, this invention proposes a non-linearly decreasing inertia weight based on the Sigmoid function, and the velocity of the elite particle population. The updated formula is:

[0061]

[0062]

[0063] in, This represents the current iteration number. and These are the learning factors for the local optimal particle and the global optimal particle, respectively. and All are random number vectors in the range [-1, 1]; the first In the next iteration, the local optimal individual particle position is used The globally optimal particle position is represented by... express; The inertial weights are non-linearly decreasing. and These are the upper and lower limits of the inertia weight, respectively. This represents the current iteration number. This represents the maximum number of iterations.

[0064] For the ordinary subpopulation, to improve the algorithm's convergence accuracy and individual diversity, and to avoid getting trapped in local optima, a random perturbation factor is introduced to the particle's current velocity. The updated formula is:

[0065]

[0066] in, It is a vector of random numbers between [-1, 1]; the formula contains the velocities of individuals from the original population. and random disruptive factors The algorithm consists of two parts, with the random disruption factor decreasing as the number of iterations increases. This strategy has a wide exploration range in the early stages of the algorithm and can converge quickly in the later stages.

[0067] Step 4-4: Incorporate a Gaussian perturbation strategy into the current particle population position to allow particles to escape local optima. As the population iterates, particles may become stuck in local optima, which can affect the final optimization efficiency. To avoid the algorithm getting stuck in local optima, this paper introduces a Gaussian perturbation strategy, enabling individuals to escape local optima and expand the search space.

[0068] After merging subpopulations, if a particle's current fitness value is equal to its individual optimal fitness value, then the particle is perturbed to prevent it from getting trapped in a local optimum. The calculation formula for incorporating the Gaussian perturbation strategy is as follows:

[0069]

[0070] In the formula, and The first The positions of the particles before and after the perturbation. The mean is 0 and the variance is If the normal distribution is followed, and the fitness is optimized, then new individuals are used. Replace the original individual Conversely, no replacement is needed;

[0071] Step 4-5: Repeat steps 4-2 to 4-4 until the maximum number of iterations is reached. At the end of the iteration, the globally optimal particle position is determined. This is the estimated phase error of each channel. .

[0072] Step 5, Combining Figure 2 Coarse measurement based on angle of arrival Based on the estimated phase error, the amplitude error of each channel is estimated using the MSCO-PSO algorithm:

[0073] Step 5-1: Initialize the population using the Latin hypercube sampling method:

[0074] Coarse Angle of Arrival for Fixed Signal and phase error The amplitude error particle position vector can be expressed as: ,in, The number of array elements; For the first The amplitude error of each channel is calculated, with the first channel serving as the reference channel, whose amplitude error is fixed at 1. The actual optimization dimensions are The basic steps for extracting N samples in a d-dimensional space are as follows:

[0075] (1) Set the range of values ​​for each dimension Divide it into N segments of equal length that do not overlap;

[0076] (2) Randomly select a data point from each segment interval of each dimension;

[0077] (3) Randomly select one data point from the data points selected in (2) for each dimension to form a d-dimensional data vector, thereby forming N d-dimensional vector samples;

[0078] Step 5-2: Begin population iteration:

[0079] Before each iteration, the particle population is classified into different levels based on the norm-adaptive classification coefficients:

[0080] Before each iteration, the fitness value of each particle is calculated based on its position vector. Then, the particles are sorted from highest to lowest fitness value. Finally, the sorted population is divided according to the adaptive classification coefficient. Divided into two subpopulations, when the particle ordinal number At that time, particles with better fitness values ​​were classified into the elite subpopulation, while the remaining particles were classified into the ordinary subpopulation. The calculation formula is:

[0081]

[0082] in, The number of particles in the population. This represents the current iteration number. It is the current population fitness vector. The infinite norm represents the extreme value of fitness. Let L1 be the norm, and let represent the sum of fitness scores.

[0083] Step 5-3: Update the position of the particle population based on the position iteration formula of the particle population, and determine the optimal particle according to the fitness function. The particle with the largest fitness value is the optimal particle.

[0084] The position update formula for the particle population is:

[0085]

[0086] in, Let i be the updated position of the i-th particle population. This represents the position of the i-th particle population before the update. The updated velocity of the i-th particle population;

[0087] Among them, the speed of the elite particle population The updated formula is:

[0088]

[0089]

[0090] in, This represents the current iteration number. and These are the learning factors for the local optimal particle and the global optimal particle, respectively. and All are random number vectors in the range [-1, 1]; the first In the next iteration, the local optimal individual particle position is used The globally optimal particle position is represented by... express; The inertial weights are non-linearly decreasing. and These are the upper and lower limits of the inertia weight, respectively. This represents the current iteration number. This represents the maximum number of iterations.

[0091] The speed of ordinary particle population The updated formula is:

[0092]

[0093] in, It is a vector of random numbers between [-1, 1].

[0094] Step 5-4: Incorporate a Gaussian perturbation strategy into the current particle population position to make the particles escape local extrema:

[0095]

[0096] In the formula, and The first The positions of the particles before and after the perturbation. The mean is 0 and the variance is If the normal distribution is followed, and the fitness is optimized, then new individuals are used. Replace the original individual Conversely, no replacement is needed;

[0097] Step 5-5: Repeat steps 5-2 to 5-4 until the maximum number of iterations is reached. At the end of the iteration, the globally optimal particle position is determined. This is the estimated phase error of each channel. .

[0098] Furthermore, the construction process of the particle fitness function in this embodiment is as follows:

[0099] Radar echo signals can be represented as:

[0100]

[0101] in, For the observation vector, The number of array elements. For the number of snapshots, Indicates the transpose operation; Let be the complex envelope of the signals, assuming that the signals are independent of each other; It is a complex Gaussian white noise vector, and is related to... Irrelevant; This indicates the index of the current snapshot. and Let the amplitude and phase error matrices of the channel and the array manifold matrix be represented as follows:

[0102]

[0103]

[0104] in, It is the ideal guiding vector; Indicates the number of incident signals; and They represent the first Amplitude and phase errors of each channel;

[0105] Construct the fitness function:

[0106]

[0107] Pedometer Maximum number of iterations , The number of signal sources.

[0108] Step 6: Calibrate the steering vector using the estimated amplitude and phase errors, and apply the calibrated steering vector to the MUSIC algorithm to estimate the calibrated angle of arrival. :

[0109] Based on the estimated amplitude and phase errors, a corresponding amplitude and phase calibration matrix is ​​constructed to correct the ideal array steering vector, resulting in a calibrated array steering vector. This calibrated steering vector is then substituted into the MUSIC spatial spectrum function to reconstruct the calibrated spatial spectrum. A peak search is performed within a preset angle search range to obtain the angle corresponding to the peak as the calibrated signal arrival angle. .

[0110] Step 7: Calculate the difference between the estimated angle of arrival before and after calibration. If the difference is greater than the threshold... If the number of iterations is less than or equal to the maximum number of iterations, return to step 4; if the difference is less than the threshold... If the number of iterations exceeds the maximum number of iterations, the algorithm terminates, outputting estimated values ​​for amplitude and phase errors to calibrate the array's amplitude and phase errors.

[0111] In this embodiment, the calculation of the first... The estimated angle of arrival obtained in the next iteration With the The estimated angle of arrival obtained in the next iteration The absolute value of the difference between them. When the following conditions are met... and If the condition is met, return to step 5 to continue the optimization process; when the condition is satisfied... or When the time is up, the algorithm terminates and outputs estimated values ​​of amplitude and phase errors, thus achieving calibration of array amplitude and phase errors.

[0112] Figure 3 This is a comparison of the antenna radiation patterns before and after calibration. Figure 3 It can be preliminarily observed that MSCO-PSO performs best in terms of array pattern recovery, and its calibrated pattern is closest to the ideal array response.

[0113] Figure 4 The image shows a comparison of the MUSIC spatial spectrum before and after calibration. After calibration using the MSCO-PSO method, the peak value of the MUSIC spatial spectrum increased by more than 25 dB compared to the uncalibrated case, and the main lobe became significantly narrower and more concentrated, reaching an advanced level among similar methods.

[0114] Figure 5 and Figure 6 A comparison of flow field diagrams before and after channel calibration is provided. After calibration using the MSCO-PSO method, the radial flow field distribution obtained by inversion exhibits a more ordered spatial morphology that conforms to actual hydrodynamic characteristics, showing the most consistent and stable flow field structure, providing a more reliable basis for subsequent flow field inversion.

[0115] Overall, the MSCO-PSO algorithm exhibits significant advantages in accuracy, robustness, and generalizability. Based on the traditional particle swarm optimization algorithm framework, this method enhances the algorithm's global search capability and convergence stability in the high-dimensional channel amplitude and phase error parameter space by introducing multi-strategy improvements, thereby achieving array amplitude and phase error calibration. Verification through this embodiment confirms the effectiveness and advantages of the method of this invention in antenna array amplitude and phase error calibration.

[0116] In addition, this application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to perform the following steps:

[0117] Step 1: Acquire multi-channel echo data received by the UHF radar array, perform signal processing on the echo data, and obtain the range Doppler spectrum;

[0118] Step 2: Obtain the effective spectral points based on the distance-Doppler spectrum and construct the covariance matrix of the array data;

[0119] Step 3: Use the MUSIC algorithm to process the covariance matrix of the array data to obtain a coarse measurement of the angle of arrival. ;

[0120] Step 4: Coarse measurement based on angle of arrival The phase error of each channel is estimated based on the MSCO-PSO algorithm;

[0121] Step 5: Coarse measurement based on angle of arrival Based on the estimated phase error, the amplitude error of each channel is estimated using the MSCO-PSO algorithm;

[0122] Step 6: Calibrate the steering vector using the estimated amplitude and phase errors, and apply the calibrated steering vector to the MUSIC algorithm to estimate the calibrated angle of arrival. ;

[0123] Step 7: Calculate the difference between the estimated angle of arrival before and after calibration. If the difference is greater than the threshold... If the number of iterations is less than or equal to the maximum number of iterations, return to step 4; if the difference is less than the threshold... If the number of iterations exceeds the maximum number of iterations, the algorithm terminates and outputs estimated values ​​of amplitude and phase errors to calibrate the array amplitude and phase errors.

[0124] This application also provides a computer-storeable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, performs the following steps:

[0125] Step 1: Acquire multi-channel echo data received by the UHF radar array, perform signal processing on the echo data, and obtain the range Doppler spectrum;

[0126] Step 2: Obtain the effective spectral points based on the distance-Doppler spectrum and construct the covariance matrix of the array data;

[0127] Step 3: Use the MUSIC algorithm to process the covariance matrix of the array data to obtain a coarse measurement of the angle of arrival. ;

[0128] Step 4: Coarse measurement based on angle of arrival The phase error of each channel is estimated based on the MSCO-PSO algorithm;

[0129] Step 5: Coarse measurement based on angle of arrival Based on the estimated phase error, the amplitude error of each channel is estimated using the MSCO-PSO algorithm;

[0130] Step 6: Calibrate the steering vector using the estimated amplitude and phase errors, and apply the calibrated steering vector to the MUSIC algorithm to estimate the calibrated angle of arrival. ;

[0131] Step 7: Calculate the difference between the estimated angle of arrival before and after calibration. If the difference is greater than the threshold... If the number of iterations is less than or equal to the maximum number of iterations, return to step 4; if the difference is less than the threshold... If the number of iterations exceeds the maximum number of iterations, the algorithm terminates and outputs estimated values ​​of amplitude and phase errors to calibrate the array amplitude and phase errors.

[0132] The embodiments described above are merely one implementation method of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO, characterized in that, Includes the following steps: Step 1: Acquire multi-channel echo data received by the UHF radar array, perform signal processing on the echo data, and obtain the range Doppler spectrum; Step 2: Obtain the effective spectral points based on the distance-Doppler spectrum and construct the covariance matrix of the array data; Step 3: Use the MUSIC algorithm to process the covariance matrix of the array data to obtain a coarse measurement of the angle of arrival. ; Step 4: Coarse measurement based on angle of arrival The phase error of each channel is estimated based on the MSCO-PSO algorithm; Step 5: Coarse measurement based on angle of arrival Based on the estimated phase error, the amplitude error of each channel is estimated using the MSCO-PSO algorithm; Step 6: Calibrate the steering vector using the estimated amplitude and phase errors, and apply the calibrated steering vector to the MUSIC algorithm to estimate the calibrated angle of arrival. ; Step 7: Calculate the difference between the estimated angle of arrival before and after calibration. If the difference is greater than the threshold... If the number of iterations is less than or equal to the maximum number of iterations, then return to step 4; If the difference is less than the threshold If the number of iterations exceeds the maximum number of iterations, the algorithm terminates and outputs estimated values ​​of amplitude and phase errors to calibrate the array amplitude and phase errors.

2. The method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO according to claim 1, characterized in that, The signal processing of the echo data in step 1 to obtain the range Doppler spectrum specifically involves: The radar echo data is first subjected to Fourier transform in the fast time dimension to achieve equivalent pulse compression and obtain the range spectrum. Subsequently, coherent accumulation and spectral analysis of the echoes from each distance cell are performed in the slow time dimension to obtain the corresponding Doppler spectrum distribution.

3. The method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO according to claim 1, characterized in that, The construction of the covariance matrix of the array data in step 2 is specifically as follows: Step 2-1: Introduce a noise threshold to filter spectral points and smooth the original echo Doppler spectrum to suppress spikes and punctures caused by random noise while ensuring the integrity of the spectral region. Step 2-2: Calculate the power change rate between adjacent spectral points to obtain the corresponding difference spectrum distribution. In the difference spectrum, the frequency point corresponding to the position where the echo power change is most significant is taken as the outer boundary of the positive and negative first-order Bragg peak regions, thereby realizing the division of the effective spectral region. Steps 2-3: Extract complex echo data of each receiving channel at the corresponding frequency points within the effective spectrum region, and construct array snapshot vectors according to the array channel order; perform statistical averaging on multiple array snapshot vectors, that is, accumulate the product of each array snapshot vector and its conjugate transpose and take the average, thereby obtaining the covariance matrix of the array data.

4. The method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO according to claim 1, characterized in that, The coarse measurement of the angle of arrival in step 3 is specifically as follows: The covariance matrix is ​​decomposed into eigenvalues ​​to separate the signal subspace and the noise subspace. Based on the orthogonality between the array steering vector and the noise subspace, the MUSIC spatial spectrum function is constructed. Perform a spectral peak search within a preset angle search range, and obtain the angle corresponding to the spectral peak as the coarse angle of arrival of the signal at that frequency point. .

5. The method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO according to claim 1, characterized in that, The step 4, which estimates the phase error of each channel based on the MSCO-PSO algorithm, specifically involves: Step 4-1: Initialize the population using the Latin hypercube sampling method: In the phase error estimation stage, the particle position vector is represented as: Where M is the number of array elements. To determine the phase error of the m-th channel, the basic steps for extracting N samples in the d-dimensional space are as follows: (1) Set the range of values ​​for each dimension Divide it into N segments of equal length that do not overlap; (2) Randomly select a data point from each segment interval of each dimension; (3) Randomly select one data point from the data points selected in (2) for each dimension to form a d-dimensional data vector, thereby forming N d-dimensional vector samples; Step 4-2: Begin population iteration: Before each iteration, the particle population is classified into different levels based on the norm-adaptive classification coefficients: Before each iteration, the fitness value of each particle is calculated based on its position vector. Then, the particles are sorted from highest to lowest fitness value. Finally, the sorted population is divided according to the adaptive classification coefficient. Divided into two subpopulations, when the particle ordinal number At that time, particles with better fitness values ​​were classified into the elite subpopulation, while the remaining particles were classified into the ordinary subpopulation. The calculation formula is: ； in, The number of particles in the population. This represents the current iteration number. It is the current population fitness vector. The infinite norm represents the extreme value of fitness. Let L1 be the norm, and let represent the sum of fitness scores. Step 4-3: Update the position of the particle population based on the position iteration formula of the particle population, and determine the optimal particle according to the fitness function; The position update formula for the particle population is: ； in, Let i be the updated position of the i-th particle population. This represents the position of the i-th particle population before the update. The updated velocity of the i-th particle population; Among them, the speed of the elite particle population The updated formula is: ； ； in, This represents the current iteration number. and These are the learning factors for the local optimal particle and the global optimal particle, respectively. and All are random number vectors in the range [-1, 1]; the first In the next iteration, the local optimal individual particle position is used The globally optimal particle position is represented by... express; The inertial weights are non-linearly decreasing. and These are the upper and lower limits of the inertia weight, respectively. This represents the current iteration number. This represents the maximum number of iterations. The speed of ordinary particle population The updated formula is: ； in, It is a vector of random numbers between [-1, 1]. Step 4-4: Incorporate a Gaussian perturbation strategy into the current particle population position to make the particles escape local extrema. ； In the formula, and The first The positions of the particles before and after the perturbation. The mean is 0 and the variance is If the normal distribution is followed, and the fitness is optimized, then new individuals are used. Replace the original individual Conversely, no replacement is needed; Step 4-5: Repeat steps 4-2 to 4-4 until the maximum number of iterations is reached. At the end of the iteration, the globally optimal particle position is determined. This is the estimated phase error of each channel. .

6. The method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO according to claim 1, characterized in that, The estimation of amplitude error for each channel based on the MSCO-PSO algorithm in step 5 is specifically as follows: Step 5-1: Initialize the population using the Latin hypercube sampling method: Coarse Angle of Arrival for Fixed Signal and phase error The amplitude error particle position vector can be expressed as: ,in, The number of array elements; For the first The basic steps for extracting N samples from the amplitude error of each channel in a d-dimensional space are as follows: (1) Set the range of values ​​for each dimension Divide it into N segments of equal length that do not overlap; (2) Randomly select a data point from each segment interval of each dimension; (3) Randomly select one data point from the data points selected in (2) for each dimension to form a d-dimensional data vector, thereby forming N d-dimensional vector samples; Step 5-2: Begin population iteration: Before each iteration, the particle population is classified into different levels based on the norm-adaptive classification coefficients: Before each iteration, the fitness value of each particle is calculated based on its position vector. Then, the particles are sorted from highest to lowest fitness value. Finally, the sorted population is divided according to the adaptive classification coefficient. Divided into two subpopulations, when the particle ordinal number At that time, particles with better fitness values ​​were classified into the elite subpopulation, while the remaining particles were classified into the ordinary subpopulation. The calculation formula is: ； in, The number of particles in the population. This represents the current iteration number. It is the current population fitness vector. The infinite norm represents the extreme value of fitness. Let L1 be the norm, and let represent the sum of fitness scores. Step 5-3: Update the position of the particle population based on the position iteration formula of the particle population, and determine the optimal particle according to the fitness function; The position update formula for the particle population is: ； in, Let i be the updated position of the i-th particle population. This represents the position of the i-th particle population before the update. The updated velocity of the i-th particle population; Among them, the speed of the elite particle population The updated formula is: ； ； in, This represents the current iteration number. and These are the learning factors for the local optimal particle and the global optimal particle, respectively. and All are random number vectors in the range [-1, 1]; the first In the next iteration, the local optimal individual particle position is used The globally optimal particle position is represented by... express; The inertial weights are non-linearly decreasing. and These are the upper and lower limits of the inertia weight, respectively. This represents the current iteration number. This represents the maximum number of iterations. The speed of ordinary particle population The updated formula is: ； in, It is a vector of random numbers between [-1, 1]. Step 5-4: Incorporate a Gaussian perturbation strategy into the current particle population position to make the particles escape local extrema: ； In the formula, and The first The positions of the particles before and after the perturbation. The mean is 0 and the variance is If the normal distribution is followed, and the fitness is optimized, then new individuals are used. Replace the original individual Conversely, no replacement is needed; Step 5-5: Repeat steps 5-2 to 5-4 until the maximum number of iterations is reached. At the end of the iteration, the globally optimal particle position is determined. This is the estimated phase error of each channel. .

7. The method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO according to claim 5 or 6, characterized in that, The fitness function of the particle is: ； ； ； in, It is the ideal guiding vector; Indicates the number of incident signals; and They represent the first Amplitude and phase errors of each channel, The number of signal sources This represents the noise subspace, which consists of eigenvectors corresponding to small eigenvalues.

8. The method for calibrating the amplitude and phase error of an ultra-high frequency radar array based on MSCO-PSO according to claim 1, characterized in that, The estimated and calibrated angle of arrival in step 6 Specifically: Based on the estimated amplitude and phase errors, a corresponding amplitude and phase calibration matrix is ​​constructed, and the ideal array steering vector is corrected to obtain the calibrated array steering vector. The calibrated steering vector is substituted into the MUSIC spatial spectrum function to reconstruct the calibrated spatial spectrum. A peak search is then performed within a preset angle search range to obtain the angle corresponding to the peak, which is taken as the calibrated signal arrival angle. .

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-8.

10. A computer-storable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-8.

Images