Antenna beam calibration method based on feature extraction

Abstract

The invention belongs to the technical field of antenna precision correction, and particularly relates to an antenna beam calibration method based on feature extraction, which comprises the following steps of: rotating an array antenna on a turntable according to a set rule, and measuring the real beam pointing of a beam at a part of design pointing angles; the preprocessing comprises the following steps of: averaging beam pointing on an azimuth plane, and separating error data according to the clockwise measurement direction and the anticlockwise measurement direction of the rotary table; utilizing a feature extraction algorithm to respectively extract preprocessed beam pointing error features of the rotary table in different rotation directions during measurement; fitting the error features to improve dimensions, reconstructing all beam pointing errors in a required angle range based on the fitted features, and further generating real beam pointing; and based on the real beam pointing, constructing a lookup table and writing the lookup table into array antenna wave control for completing rapid calibration of the beam pointing of the array antenna.

Description

Technical Field

[0001] The invention belongs to the technical field of antenna precision correction, and in particular relates to an antenna beam calibration method based on feature extraction. Background Art

[0002] In recent years, the advantages of array antennas have become increasingly prominent and they have been widely used in military and civilian fields. The actual beam pointing of array antennas generally deviates from the designed beam pointing, which may cause serious consequences in the radar application process. The traditional beam pointing correction method requires measuring all the actual beam pointing within the required angle range and the specified beam pointing calibration granularity, thereby establishing a pointing lookup table with a one-to-one correspondence between the designed beam pointing and the actual beam pointing to perform beam pointing correction. This method has high accuracy but high cost and time-consuming. Interpolation is also often used in engineering to reduce the number of measurements, but the reduction is limited.

[0003] Feature extraction is the process of extracting useful information from raw data to reduce the complexity of the data while retaining important features so that data analysis methods or machine learning models can process data more effectively. The most widely used feature extraction algorithms include principal component analysis (PCA), independent component analysis (ICA), and wavelet transform (WT).

[0004] PCA focuses on the variance of the data, and achieves the purpose of feature extraction and dimensionality reduction by transforming the original data into another dimension and selecting a finite number of principal components with large contributions as the basis for characterizing the original data.

[0005] ICA focuses on the independence of data. It reveals the hidden structure in the data by finding transformations that make the decomposed data statistically independent of each other, and extracts statistically independent components from the original signal.

[0006] WT decomposes the data into a set of wavelet functions to analyze the data at different scales and locations. Its advantage is that WT can provide multi-scale and multi-resolution analysis of the signal, so that features at different scales can be effectively extracted.

[0007] Most of the current phased array antenna calibration technologies are aimed at amplitude and phase calibration. However, due to factors such as mutual coupling and quantization errors, even after these calibrations, the actual beam pointing will still deviate from the designed beam pointing. This deviation is unacceptable in applications that require precise angle measurement.

[0008] Despite the high actual demand for beam pointing calibration, related research is very limited. Existing beam pointing calibration methods require measuring the beam pointing at small angle intervals within the required angle range, which is very time-consuming. Interpolation techniques are often used in engineering to reduce the number of angles required to be measured, such as cubic spline interpolation. However, even if interpolation techniques are applied, the number of angles required to be measured is still large while ensuring that the accuracy meets the requirements. Moreover, the prediction accuracy of the interpolation technique does not increase monotonically with the increase of measured data, which greatly limits the degree of reduction in the number of measurements. Moreover, the distribution of beam pointing errors has certain characteristics that can be used in beam pointing calibration, but they have not been studied in depth in the existing literature. Summary of the invention

[0009] The purpose of the present invention is to provide a fast calibration method for array antenna beam pointing based on feature extraction, so as to overcome the shortcomings of the prior art and greatly improve the calibration speed and system real-time performance while ensuring the beam pointing correction accuracy.

[0010] The technical solution of the present invention is as follows:

[0011] An antenna beam calibration method based on feature extraction, the specific process is:

[0012] The array antenna rotates on the turntable according to the set rules, and the real beam pointing of the beam under some designed pointing angles is measured, and preprocessing is performed as follows: the beam pointing is averaged on the azimuth plane, and the error data is separated according to the clockwise and counterclockwise measurement directions of the turntable;

[0013] The feature extraction algorithm is used to extract the pre-processed beam pointing error characteristics at different rotation directions of the turntable during measurement.

[0014] Fitting the error features to improve the dimension, reconstructing all beam pointing errors within the required angle range based on the fitted features, and then generating the real beam pointing;

[0015] Based on the real beam pointing, a lookup table is constructed and written into the array antenna beam control to complete the rapid calibration of the array antenna beam pointing.

[0016] Furthermore, the array antenna of the present invention measures the real pointing direction of the beam at some designed pointing angles on a turntable according to a set rule, including:

[0017] (1) Set the ranges of azimuth and elevation planes to be calibrated respectively. and [-θ M ,θ M ], set the angle intervals that need to be calibrated for the azimuth and elevation planes to be and Δθ;

[0018] (2) From the initial pointing angle of the turntable Initially, the angular velocities of the turntable in the azimuth and elevation planes are set to and θ V＝ 0° / s and rotate to complete a line of scanning;

[0019] (3) Then the pitch pointing angle is stepped once by Δθ, and the angular velocity of the turntable in the azimuth and pitch planes is reset to and θ V＝ 0° / s and rotate to complete a line of scanning;

[0020] (4) Then step the elevation pointing angle by Δθ and set the angular velocity of the turntable in the azimuth and elevation planes to and θ V＝ 0° / s and rotate to complete a line scan;

[0021] (5) Repeat the above steps (3)-(4) until the pitch pointing angle is greater than θ M , end the scan.

[0022] Furthermore, the feature extraction algorithm of the present invention extracts the beam pointing error feature, including:

[0023] (1) The measured beam pointing error data are averaged in the azimuth plane, and the data are divided into clockwise measurement error data and counterclockwise measurement error data according to the different rotation directions of the turntable during measurement;

[0024] (2) The feature extraction algorithm is used to extract the data features from the two sets of data. The azimuth error and pitch angle error in each set of data are processed separately, for a total of four sets of error data.

[0025] Furthermore, the present invention describes fitting the error features to improve dimensionality, and the specific process is: fitting and improving the dimensions of the features of each group of error data by a smoothing spline fitting algorithm, and when fitting the features, traversing and selecting smoothing parameters to obtain the best fitting results of the features, specifically: traversing and selecting smoothing parameters, and performing smoothing spline fitting on the features; reconstructing based on different fitting results, and calculating three times the standard deviation of some data; the three times the standard deviation of the some data is: three times the standard deviation of the difference between the reconstructed pointing error and the measured pointing error at the measurement angle; selecting the fitting result corresponding to the minimum three times the standard deviation of the partial data as the best fitting result.

[0026] Furthermore, all beam pointing errors within the required angle range are reconstructed based on the fitted features to generate real beam pointing, including: reconstructing according to the best fitting result of the features; reconstructing the clockwise error data and the counterclockwise error data respectively; rearranging the clockwise measurement error and the error predicted based on it, and the counterclockwise measurement error and the error predicted based on it in the order of the pitch angles to obtain all beam pointing errors; and obtaining the predicted all beam pointings based on the designed beam pointing.

[0027] Furthermore, in one implementation, PCA is used for feature extraction, which has the following characteristics:

[0028] PCA is used to process the beam pointing errors in different rotation directions of the turntable separately to obtain the principal components of each group of data. Each group of error data contains azimuth error and elevation error, which are processed separately. Each principal component is fitted and dimensionally upgraded through smoothing spline fitting, and all beam pointing errors are predicted based on the reconstructed principal components after fitting.

[0029] Furthermore, the principal components are fitted and dimensionally upgraded by smoothing spline fitting, and reconstruction is performed based on the fitted principal components, including:

[0030] First, the principal component corresponding to the maximum eigenvalue is selected, and the smoothing parameters are traversed to perform smoothing spline fitting on the principal component to obtain the fitting results under various smoothing parameters;

[0031] Perform PCA reconstruction based on the fitted principal components and calculate three times the standard deviation of the difference between the reconstructed pointing error and the measured pointing error at the measured angle;

[0032] The fitting result corresponding to the minimum three-fold standard deviation is selected as the best fitting result of the principal component, and the three-fold standard deviation is recorded for subsequent operations;

[0033] Then select the principal components in turn and repeat the above operation until the difference between three adjacent standard deviations is less than 0.01°, and output the beam pointing prediction result at this time.

[0034] Furthermore, the predicting of all beam pointings of the present invention includes:

[0035] PCA reconstruction requires principal components, corresponding eigenvectors, and mean vectors;

[0036] Reconstruct the clockwise error data and the counterclockwise error data respectively;

[0037] The clockwise measurement error and the error based on its prediction, the counterclockwise measurement error and the error based on its prediction are rearranged in the order of their pitch angles (by θ M to -θ M ) and then get the total beam pointing error;

[0038] According to the designed beam pointing, the predicted overall beam pointing is obtained.

[0039] Furthermore, in one implementation, ICA is used for feature extraction, which has the following characteristics:

[0040] ICA is used to process the beam pointing errors of the turntable in different rotation directions, and the independent components and corresponding mixing matrices of each group of data are obtained. Each group of error data contains azimuth error and elevation error, which are processed separately; each group of independent components is fitted and dimensioned by smoothing spline fitting, and reconstructed based on the fitted independent components to predict all beam pointing errors;

[0041] The step of fitting and dimensionality-raising the independent components by smoothing spline fitting includes:

[0042] The independent components of each group of data are fitted and dimensioned through smoothing spline fitting;

[0043] The dimension after the fitting dimension upgrade depends on the actual needs;

[0044] When fitting independent components, smoothing parameters are selected traversally to obtain the best fitting results for each group of independent components.

[0045] Furthermore, the present invention traverses and selects smoothing parameters to obtain the best fitting result for each group of independent components, including:

[0046] Traverse and select smoothing parameters, and perform smoothing spline fitting on the group of independent components;

[0047] Perform ICA reconstruction with the mixing matrix of the data set respectively, and calculate the three times standard deviation of some data;

[0048] The fitting result corresponding to the minimum three-times standard deviation is selected as the best fitting result;

[0049] The best fitting result is taken as the fitting result of this group of independent components.

[0050] Furthermore, the predicting of all beam pointings of the present invention includes:

[0051] ICA reconstruction was performed based on the fitted independent components;

[0052] ICA reconstruction requires a mixing matrix, an independent component matrix, and a mean vector;

[0053] Reconstruct the clockwise error data and the counterclockwise error data respectively;

[0054] The clockwise measurement error and the error based on its prediction, the counterclockwise measurement error and the error based on its prediction are rearranged in the order of their pitch angles (by θ M to -θM ) and then get the total beam pointing error;

[0055] According to the designed beam pointing, the predicted overall beam pointing is obtained.

[0056] In one implementation, WT is used for feature extraction, which has the following characteristics:

[0057] Select the wavelet basis, use WT to process the beam pointing errors of different rotation directions of the turntable, and obtain the wavelet coefficients of each group of data. Each group of error data contains azimuth error and elevation error, which are processed separately; fit and upgrade the dimensions of each group of wavelet coefficients through smoothing spline fitting, and reconstruct based on the fitted wavelet coefficients to predict all beam pointing errors;

[0058] The step of fitting and dimensionalizing each group of wavelet coefficients by smoothing spline fitting includes:

[0059] The wavelet coefficients of each group of data are fitted and dimensioned by smoothing spline fitting;

[0060] The dimension after the fitting dimension upgrade depends on the actual needs;

[0061] When fitting wavelet coefficients, smoothing parameters are traversed and selected to obtain the best fitting result for each set of wavelet coefficients.

[0062] Furthermore, the traversal and selection of smoothing parameters in the present invention to obtain the best fitting result of the wavelet coefficients includes:

[0063] Traverse and select smoothing parameters, and perform smoothing spline fitting on the group of wavelet coefficients;

[0064] WT reconstruction was performed separately, and three times the standard deviation of some data was calculated;

[0065] The fitting result corresponding to the minimum three-times standard deviation is selected as the best fitting result;

[0066] The best fitting result is taken as the fitting result of the group of wavelet coefficients.

[0067] Furthermore, the present invention predicts all beam pointings, including:

[0068] WT reconstruction is performed based on the fitted wavelet coefficients;

[0069] WT reconstruction requires wavelet coefficients, wavelet basis, and decomposition levels;

[0070] Reconstruct the clockwise error data and the counterclockwise error data respectively;

[0071] The clockwise measurement error and the error based on its prediction, the counterclockwise measurement error and the error based on its prediction are rearranged in the order of their pitch angles (by θM to -θ M ) and then get the total beam pointing error;

[0072] According to the designed beam pointing, the predicted overall beam pointing is obtained.

[0073] Beneficial effects:

[0074] First, the array antenna beam pointing rapid calibration method provided by the present invention can predict all beam pointings within the required angle range and within the specified beam pointing calibration granularity while only measuring the actual beam pointing of the beam at a small part of the designed pointing angles; and if the three times the standard deviation of the deviation between the predicted error and the actual error is less than 0.25° as the indicator requirement, the method can meet the standard.

[0075] Second, the predicted beam pointing and the corresponding designed beam pointing are imported into the lookup table and written into the array antenna, so as to realize the correction of the array antenna pointing accuracy.

[0076] Third, the method provided by the present invention effectively improves the beam pointing accuracy and real-time performance of the array antenna system, and has low cost and low system complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0078] Figure 1 FIG. 1 is a flow chart of an antenna beam calibration method based on feature extraction provided by the present invention;

[0079] Figure 2 The figure shows the running trajectory diagram of the beam pointing in space of the traditional calibration method provided by the embodiment of the present invention taking a partial angle range of the pitch plane as an example;

[0080] Figure 3 The figure shows the running trajectory diagram of the beam pointing in space of the fast calibration method provided by the embodiment of the present invention taking a partial angle range of the pitch plane as an example;

[0081] Figure 4 FIG. 1 is a schematic diagram of an experimental verification process of a PCA-based beam pointing fast calibration method provided by an embodiment of the present invention;

[0082] Figure 5 The figure shows a flow chart of fitting principal components and reconstructing in a PCA-based beam pointing fast calibration method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0083] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0084] It should be noted that the following embodiments and features in the embodiments may be combined with each other in the absence of conflict; and, based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making any creative work are within the scope of protection of the present disclosure.

[0085] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0086] like Figure 1 As shown, the antenna beam calibration method based on feature extraction in the embodiment of the present application has the following specific process:

[0087] S1: The array antenna rotates on the turntable according to the set rules to measure the actual beam pointing at some designed pointing angles;

[0088] S2: Using feature extraction algorithm to extract beam pointing error features at different rotation directions of the turntable during measurement;

[0089] S3: Fitting the error feature to improve the dimension;

[0090] S4: reconstruct all beam pointing errors within the required angle range to generate the real beam pointing;

[0091] S5: Based on the real beam pointing, a lookup table is constructed and the lookup table is written into the array antenna for calibration.

[0092] After obtaining the real beam pointing of the measurement beam under some designed pointing angles in step S1, the method further includes preprocessing the beam pointing error. The preprocessing includes: averaging the beam pointing error on the azimuth plane; and separating the error data according to the clockwise and counterclockwise measurement directions of the turntable.

[0093] The schematic diagram of the trajectory of measuring the total beam pointing error using a partial angle range of the pitch plane as an example is as follows: Figure 3As shown, the intervals in the figure are only for illustration and are not the only implementation form.

[0094] 1) The beam pointing measurement in the azimuth plane is sampled by time, and the number of points depends on the speed of the turntable. The measured azimuth angle interval is much smaller than the required angle interval. The beam pointing error is averaged in the azimuth plane in the preprocessing to eliminate the random jitter generated during the measurement to a certain extent and process it into the required angle interval;

[0095] The characteristics of the measurement data under different scanning directions of the turntable are different, so the error data of different scanning directions are separated in the preprocessing;

[0096] 2) At the same time, the average processing in the above preprocessing process is a moving average processing.

[0097] like Figure 2 As shown in FIG. 1 , the trajectory of measuring all beam pointing errors in a certain angle range of the pitch plane is taken as an example. The specific operation is:

[0098] Set the ranges that need to be calibrated for the azimuth and elevation planes to be and [-θ M ,θ M ], the angle intervals that need to be calibrated are and Δθ;

[0099] The array antenna is installed on a two-dimensional turntable, and the initial pointing angle of the turntable is Initially, the angular velocities of the turntable in the azimuth and elevation planes are set to and θ V＝ 0° / s and rotate to complete a line scan;

[0100] Then the pitch pointing angle is stepped once by Δθ, and the rotation angular velocity of the turntable in the azimuth and pitch planes is reset to and θ V＝ 0° / s and rotate to complete a line scan;

[0101] Then step the elevation pointing angle by Δθ and set the rotation angular velocity of the turntable in the azimuth and elevation planes to and θ V＝ 0° / s and rotate to complete a line scan;

[0102] Repeat the above steps until the pitch pointing angle is greater than θ M , end the scan.

[0103] Another embodiment of the present application uses a feature extraction algorithm to respectively extract the beam pointing error features of the turntable in different rotation directions during measurement, specifically:

[0104] Firstly, the measured beam pointing error data are averaged in the azimuth plane, and the data are divided into clockwise measurement error data and counterclockwise measurement error data according to the different rotation directions of the turntable during measurement;

[0105] Secondly, feature extraction algorithms are used on the two groups of data to extract data features. The azimuth error and pitch angle error in each group of data are processed separately, totaling four groups of error data.

[0106] The feature extraction algorithms used in this step include but are not limited to principal component analysis (PCA), independent component analysis (ICA), wavelet transform (WT), etc.

[0107] In another embodiment of the present application, fitting the error feature to improve the dimension includes:

[0108] The smoothing spline fitting algorithm is used to fit and upgrade the characteristics of each group of error data. The dimension after fitting depends on the actual needs. When fitting the characteristics, the smoothing parameters are traversed and selected to obtain the best fitting result of the characteristics.

[0109] In this embodiment, smoothing parameters are traversed and selected to obtain the best fitting result of the feature, including: traversing and selecting smoothing parameters, performing smoothing spline fitting on the feature; reconstructing based on different fitting results respectively, and calculating three times the standard deviation of part of the data; the three times the standard deviation of the part of the data is: three times the standard deviation of the difference between the reconstructed pointing error and the measured pointing error at the measurement angle; selecting the fitting result corresponding to the minimum three times the standard deviation of the part of the data as the best fitting result.

[0110] In another embodiment of the present application, all beam pointing errors within a required angle range are reconstructed based on the fitted feature, thereby generating beam pointing, including: reconstructing according to the best fitting result of the feature; reconstructing clockwise error data and counterclockwise error data respectively; rearranging the clockwise measurement error and the error predicted based on the clockwise measurement error, and the counterclockwise measurement error and the error predicted based on the clockwise measurement error according to the pitch angle order (by θ M to -θ M ) and then the total beam pointing error is obtained; according to the designed beam pointing, the predicted total beam pointing is obtained.

[0111] Embodiment 1:

[0112] This embodiment implements rapid beam pointing calibration based on PCA.

[0113] It should be noted that in this exemplary embodiment, all beam pointings need to be measured in advance in order to subsequently verify the effectiveness of the method. The verification flow diagram of this exemplary embodiment based on MATLAB R2023a is as follows: Figure 4 As shown, including:

[0114] SS1: Pre-process all beam pointing errors within the required angle range and the specified beam pointing calibration granularity obtained by the turntable measurement to obtain the beam pointing errors under some design angles;

[0115] SS2: Perform PCA processing on the beam pointing error, extract the error characteristics, and obtain the principal component of the error data;

[0116] SS3: Use smoothing spline fitting to fit each principal component and improve the dimension. Reconstruct based on the fitted principal components and predict all beam pointing errors within the required angle range;

[0117] SS4: Compare the predicted error with the true error to verify the effectiveness of the method.

[0118] Figure 5 The schematic diagram of the process of fitting the principal components and reconstructing in this embodiment is shown, including:

[0119] First, the principal component corresponding to the largest eigenvalue is selected, and smoothing parameters are selected to perform smoothing spline fitting on the principal component to obtain the fitting results under various smoothing parameters;

[0120] Secondly, PCA reconstruction is performed based on the fitted principal components, and three times the standard deviation of the difference between the reconstructed pointing error and the measured pointing error at the measured angle is calculated;

[0121] Again, select the fitting result corresponding to the minimum three times standard deviation as the best fitting result of the principal component, and record the three times standard deviation for subsequent operations;

[0122] Finally, the principal components are selected in turn, and the above operation is repeated until the difference between three adjacent standard deviations is less than 0.01°, and the beam pointing prediction result at this time is output.

[0123] The measured beam pointing is subtracted from the designed beam pointing to obtain the beam pointing error data at all designed angles.

[0124] It is understandable that the specific steps of angle measurement are well known to those skilled in the art and will not be described in detail in this specification.

[0125] Furthermore, the preprocessing steps in step SS1 are: the beam pointing error is sampled at 1 / 2 equal intervals on the azimuth plane and averaged every 5 points; the error data is separated according to the clockwise and counterclockwise measurement directions of the turntable; the obtained clockwise measurement data and counterclockwise measurement data are sampled at 1 / 10 equal intervals on the pitch plane respectively.

[0126] The trajectory of the beam pointing error at some angles obtained after preprocessing is similar to the beam pointing measurement trajectory of the traditional method, except that the angle interval becomes Δθ and 19Δθ alternately. The schematic diagram of the trajectory of measuring the total beam pointing error using a partial angle range of the pitch plane as an example is shown in Figure 3 shown.

[0127] 1) Each embodiment is to verify the effectiveness of the fast calibration method. It is necessary to measure the actual beam pointing at all angles in advance so as to compare it with the predicted beam pointing to evaluate the prediction effect.

[0128] However, when the fast calibration method is actually used, the only data available are the true beam pointing at some angles measured according to established rules. Therefore, it is necessary to sample the true beam pointing at all angles in the pitch plane to simulate the actual usage scenario, that is, to perform equal-interval sampling in the pitch plane during the preprocessing process. In this embodiment, 1 / 10 equal-interval sampling is used.

[0129] The beam pointing measurement in the azimuth plane is sampled by time, and the number of points depends on the speed of the turntable. The measured azimuth angle interval is much smaller than the required angle interval. In the preprocessing, the beam pointing error is sampled at 1 / 2 equal intervals in the azimuth plane and averaged every 5 points (the azimuth angle interval measured here is 0.01°, and the required angle interval is 0.1°) to eliminate the random jitter generated during the measurement to a certain extent and process it into the required angle interval.

[0130] The characteristics of the measurement data under different scanning directions of the turntable are different, so the error data of different scanning directions are separated in the preprocessing;

[0131] If all the original data are directly sampled 1 / k in the pitch plane, the sampled data may be all measurement data in a certain scanning direction. Therefore, the measurement data in different scanning directions are separated before sampling.

[0132] The preprocessed error data can be regarded as Figure 5 The measurement results are obtained by measuring the measurement trace shown.

[0133] It should be noted that in step SS2, the clockwise measurement error data and the counterclockwise measurement error data need to be processed by PCA respectively. Each set of data is further divided into azimuth error data and elevation error data, which are processed separately.

[0134] PCA transforms the original data into another dimension, and selects a limited number of principal components with large contributions as the basis for characterizing the original data, so as to achieve the purpose of feature extraction and dimensionality reduction. The steps of PCA processing are as follows:

[0135] Decentralize the error data X to obtain X′;

[0136] Construct the covariance matrix V of X′;

[0137] Perform eigendecomposition on the covariance matrix V to obtain the eigenvalue α and the corresponding eigenvector W;

[0138] The extracted main component F = W T X′.

[0139] Furthermore, the obtained principal components are fitted and dimensioned in turn, and all beam pointing error data are reconstructed. The specific steps are the same as Figure 3 Same as shown.

[0140] It should be noted that the formula for PCA reconstruction is:

[0141] X″=W n F n +μ

[0142] Among them, F n is the selected principal component, W n is the corresponding eigenvector, and μ is the mean vector of samples in the data.

[0143] In this method, the principal components, eigenvectors and mean vectors after fitting can be selected for PCA reconstruction to predict the entire beam pointing error.

[0144] Based on the above verification process, this embodiment verifies 6 groups of measured data, and the verification results are shown in Table 1.

[0145] The standard is considered to be met if three times the standard deviation of the deviation between the designed beam pointing and the predicted beam pointing is less than 0.25°.

[0146] It should be noted that, in this embodiment, the deviation between the designed beam pointing and the predicted beam pointing is equivalent to the deviation between the actual beam pointing error and the predicted beam pointing error.

[0147] Table 1

[0148]

[0149] In this embodiment, the verification of the five groups of data all met the standards, and the processing time for each group of data was about 1 minute, which shows the effectiveness of the method.

[0150] Embodiment 2

[0151] This embodiment uses ICA as the feature extraction algorithm, and the specific implementation process is similar to that of the first embodiment.

[0152] It should be noted that, in this embodiment, when fitting independent components, all independent components need to be fitted and then reconstructed.

[0153] It should be noted that ICA assumes that the original data X is a linear mixture of several independent components S, that is:

[0154] X=AS

[0155] Among them, X is the original data matrix, S is the independent component matrix, and A is the mixing matrix. The ICA algorithm estimates the independent component matrix S and the mixing matrix A from the original data. During reconstruction, the original data X can be estimated by reconstructing the known independent component matrix S and the mixing matrix A using the formula.

[0156] In this method, the fitted independent components can be selected and reconstructed together with the mixing matrix through ICA to predict the entire beam pointing error.

[0157] The results of verifying 5 groups of measured data in this embodiment are shown in Table 2.

[0158] Table 2

[0159]

[0160] In this embodiment, the verification of 5 groups of data basically meets the standard, and the processing time of each group of data is about 4 minutes, which shows the effectiveness of this method.

[0161] Embodiment 3

[0162] This embodiment uses WT as the feature extraction algorithm, and the specific implementation process is similar to that of the first embodiment.

[0163] It should be noted that, in this embodiment, when fitting the wavelet coefficients, all the wavelet coefficients need to be fitted, and reconstruction is performed accordingly.

[0164] WT analyzes the signal by scaling and translating the wavelet basis functions. Scaling determines the frequency of the wavelet basis functions, translation determines the location of the wavelet basis functions, and the wavelet coefficients represent the strength of the data at different scales and locations.

[0165] During reconstruction, the original signal can be reconstructed from the wavelet coefficients, wavelet basis functions and the number of decomposition levels.

[0166] In this method, the fitted wavelet coefficients can be selected and used together with the set wavelet basis functions and decomposition levels to perform wavelet reconstruction, and the entire beam pointing error can be predicted.

[0167] The results of verifying 5 groups of measured data in this embodiment are shown in Table 3.

[0168] Table 3

[0169]

[0170] In this embodiment, the verification results of the five groups of data all meet the requirements, and the processing time for each group of data is about 3 minutes, which shows the effectiveness of this method.

[0171] Furthermore, in specific practice, the predicted beam pointing error can be used to obtain all beam pointings within the required angle range and calibration interval according to the corresponding designed beam pointing.

[0172] A lookup table can be established based on the predicted beam pointing and the corresponding designed beam pointing, and written into the array to complete the pointing calibration.

[0173] The beam pointing fast calibration method provided by the present invention can predict all beam pointing within the required angle range and at the specified beam pointing calibration granularity while only measuring the real beam pointing at a few actual angles. Compared with the prior art, this method requires fewer measurements, greatly improves efficiency, has high accuracy, low cost, and is easy to implement in engineering.

[0174] It should be noted that, in the description of the present invention, the terms "first", "second" and the like are only used for illustrative purposes and do not indicate or imply relative importance. In addition, in the description of the present application, unless otherwise explicitly stated, the term "plurality" refers to at least two.

[0175] Any process or method described in any flow chart or other manner described in the present invention can be understood as a code module, fragment or part of executable instructions including one or more steps for performing a specific logical function or process. The scope of the preferred embodiment of the present invention also includes other implementations, which may not be performed in the order shown, and will be understood by those skilled in the art to which the present invention relates.

[0176] It is understood that when describing the present invention, terms such as "one embodiment", "some embodiments", "examples", "specific examples" or "some examples" refer to at least one embodiment or example with specific features, structures, materials or characteristics. The use of these terms does not necessarily refer to the same embodiment or example, and the specific features, structures, materials or characteristics in the description can be appropriately combined in one or more embodiments or examples.

[0177] The above specific implementation further describes the purpose, technical scheme and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation of the present invention and is not intended to limit the scope of protection of the present invention. Any modification, equivalent replacement or improvement made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. An antenna beam calibration method based on feature extraction, characterized in that: The specific process is: The array antenna rotates on the turntable according to the set rules, and the real beam pointing of the beam under some designed pointing angles is measured, and preprocessing is performed as follows: the beam pointing is averaged on the azimuth plane, and the error data is separated according to the clockwise and counterclockwise measurement directions of the turntable; The feature extraction algorithm is used to extract the pre-processed beam pointing error characteristics at different rotation directions of the turntable during measurement. Fitting the error features to improve the dimension, reconstructing all beam pointing errors within the required angle range based on the fitted features, and then generating the real beam pointing; Based on the real beam pointing, a lookup table is constructed and written into the array antenna beam control to complete the rapid calibration of the array antenna beam pointing.

2. The antenna beam calibration method based on feature extraction according to claim 1, characterized in that: The array antenna measures the actual pointing direction of the beam at some designed pointing angles according to a set rule on the turntable, including: (1) Set the ranges of azimuth and elevation planes to be calibrated respectively. and [-θ M ,θ M ], set the angle intervals that need to be calibrated for the azimuth and elevation planes to be and Δθ; (2) From the initial pointing angle of the turntable Initially, the angular velocities of the turntable in the azimuth and elevation planes are set to and θ V＝ 0° / s and rotate to complete a line of scanning; (3) Then the pitch pointing angle is stepped once by Δθ, and the angular velocity of the turntable in the azimuth and pitch planes is reset to and θ V＝ 0° / s and rotate to complete a line of scanning; (4) Then step the elevation pointing angle by Δθ and set the angular velocity of the turntable in the azimuth and elevation planes to and θ V＝ 0° / s and rotate to complete a line of scanning; (5) Repeat the above steps (3)-(4) until the pitch pointing angle is greater than θ M , end the scan.

3. The antenna beam calibration method based on feature extraction according to claim 2, characterized in that: The feature extraction algorithm extracts beam pointing error features, including: (1) The measured beam pointing error data are averaged in the azimuth plane, and the data are divided into clockwise measurement error data and counterclockwise measurement error data according to the different rotation directions of the turntable during measurement; (2) The feature extraction algorithm is used to extract the data features from the two sets of data. The azimuth error and pitch angle error in each set of data are processed separately, for a total of four sets of error data.

4. The antenna beam calibration method based on feature extraction according to claim 1 or 3, characterized in that: The process of fitting the error features to improve the dimension is as follows: the features of each group of error data are fitted and improved in dimension by using a smoothing spline fitting algorithm. When fitting the features, smoothing parameters are traversed and selected to obtain the best fitting result of the features.

5. The antenna beam calibration method based on feature extraction according to claim 4, characterized in that: The specific process of obtaining the best fitting result is: traverse and select smoothing parameters, and perform smoothing spline fitting on the features; reconstruct based on different fitting results respectively, and calculate the three times standard deviation of some data; the three times standard deviation of the partial data is: three times the standard deviation of the difference between the reconstructed pointing error and the measured pointing error at the measurement angle; select the fitting result corresponding to the minimum three times standard deviation of the partial data as the best fitting result.

6. The antenna beam calibration method based on feature extraction according to claim 4, characterized in that: Reconstructing all beam pointing errors within a required angle range based on the fitted features, and then generating a true beam pointing, including: reconstructing according to the best fitting result of the features; reconstructing clockwise error data and counterclockwise error data respectively; rearranging the clockwise measurement error and the error predicted based thereon, and the counterclockwise measurement error and the error predicted thereon in the order of the pitch angles to obtain all beam pointing errors; and obtaining the predicted all beam pointings based on the designed beam pointing.

7. The antenna beam calibration method based on feature extraction according to claim 1, characterized in that: The feature extraction algorithm is principal component analysis PCA, independent component analysis or wavelet transform.

Images