Array radar angle measurement method, device, equipment and storage medium

Abstract

The present invention provides an array radar angle measurement method, which can be applied to the field of radar technology. The array radar angle measurement method includes: obtaining an initial sample, which is a multi-channel echo signal; converting the initial sample into a beam space using a coherent synthesis technology to obtain a feature sample; inputting the feature sample into an optimization model, and outputting an angle value, wherein the optimization model is constructed based on a popular regularization constraint and a support vector regression algorithm. The present invention also provides an array radar angle measurement device, equipment, and storage medium.

Description

Technical Field

[0001] The present invention relates to the field of radar technology, and more specifically to an array radar angle measurement method, device, equipment and storage medium. Background Art

[0002] Among the angle positioning methods of array radar, the model-based angle estimation method is greatly affected by the array amplitude and phase errors, and the angle measurement performance is significantly reduced. This makes the robust array radar angle measurement technology the focus of scholars at home and abroad.

[0003] The support vector regression technology based on manifold regularization can adaptively learn the amplitude and phase errors of the array and realize semi-supervised small sample learning. However, this type of method performs learning operations in the array element space, and the computational complexity is high. The accuracy of angle estimation is highly dependent on the training set. For arrays with different errors, a large amount of data needs to be obtained for retraining. Summary of the invention

[0004] In view of the above problems, the present invention provides an array radar angle measurement method, device, equipment, medium and program product.

[0005] According to a first aspect of the present invention, there is provided an array radar angle measurement method, comprising: obtaining an initial sample, the initial sample being a multi-channel echo signal; converting the initial sample into a beam space using a coherent synthesis technique to obtain a feature sample; inputting the feature sample into an optimization model to output an angle value, wherein the optimization model is constructed based on a popular regularization constraint and a support vector regression algorithm.

[0006] According to an embodiment of the present invention, obtaining the initial sample includes: using any transmitting antenna of the array radar to transmit a radar signal to surrounding objects to obtain a target reflected wave signal, where the target reflected wave signal is obtained by reflecting the radar signal; using multiple receiving antennas of the array radar to receive the target reflected wave signal to obtain the initial sample.

[0007] According to an embodiment of the present invention, the initial samples are converted to beam space using coherent synthesis technology to obtain feature samples; the initial samples are weighted summed using coherent synthesis technology to obtain first processed samples; and feature selection is performed on the first processed samples to obtain feature samples.

[0008] According to an embodiment of the present invention, feature selection of the first processed sample to obtain a feature sample includes: mapping the first processed sample to a correlation matrix; determining the upper triangular elements of the correlation matrix as the second processed sample; normalizing multiple second processed samples to obtain multiple candidate feature samples; and determining the candidate feature samples that meet preset conditions among the multiple candidate feature samples as feature samples.

[0009] According to an embodiment of the present invention, feature samples include labeled samples and unlabeled samples, a first moment of obtaining the labeled samples is earlier than a second moment of obtaining the unlabeled samples, a target angle of the labeled samples is known, and a target angle of the unlabeled samples is unknown; wherein, the feature samples are input into the optimization model, and the output angle value comprises: constructing an initial model based on popular regularization constraints and a support vector regression algorithm; processing the initial model according to a representation theorem to obtain a test model; simplifying the initial model to obtain an optimization objective function; solving the optimization objective function to obtain a training model; inputting labeled samples and unlabeled samples into the training model to obtain a combination coefficient; inputting the combination coefficient into the test model to obtain an optimization model; inputting the unlabeled samples of the angle to be estimated into the optimization model to obtain the angle value.

[0010] According to an embodiment of the present invention, an initial model is constructed based on an initial norm, initial constraints and an initial loss function; wherein, simplifying the initial model to obtain an optimization objective function comprises: performing a regenerative transformation on the initial norm to obtain a target norm; simplifying the initial constraints according to a graph Laplace matrix to obtain a manifold constraint; simplifying the initial loss function according to the manifold constraint and the target norm to obtain a loss function; and updating the initial model with the target norm and the manifold constraint loss function to obtain an optimization objective function.

[0011] According to an embodiment of the present invention, the feature sample is input into the optimization model, and the output angle value also includes: recording the number of executions of the angle value obtained according to the optimization model, and when the number of executions is less than or equal to a preset threshold, the optimization model is adopted; when the number of executions is greater than the preset threshold, the initial model is retrained.

[0012] The second aspect of the present invention provides an array radar angle measurement device, including: an acquisition module, used to obtain initial samples, the initial samples are echo signals obtained after overlapping scanning of the beams emitted by the array radar; a processing module, used to use coherent synthesis technology to reduce the dimension of the initial samples to obtain feature samples; and an output module, used to input the feature samples into an optimization model and output angle values, the optimization model is constructed based on popular regularization constraints and support vector regression algorithm.

[0013] The third aspect of the present invention provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the above-mentioned array radar angle measurement method.

[0014] The fourth aspect of the present invention also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the above-mentioned array radar angle measurement method.

[0015] The fifth aspect of the present invention also provides a computer program product, including a computer program, which implements the above-mentioned array radar angle measurement method when executed by a processor.

[0016] According to an embodiment of the present invention, the coherent synthesis technology is used to convert the acquired initial samples into the beam space to obtain feature samples; then the feature samples are input into the optimization model to output the angle value, and the optimization model is constructed based on the popular regularization constraint and the support vector regression algorithm. By using the coherent synthesis method to perform dimensionality reduction processing on the initial samples, and using the optimization model constructed based on the popular regularization constraint and the support vector regression algorithm, semi-supervised training is performed on the labeled samples and the unlabeled samples, thereby avoiding the high dependence on the training set data, resulting in the need to obtain a large amount of labeled data for multiple trainings, and the problem of large errors in the output angle values, and realizing the technical effect of reducing the sensitivity of the phase error and improving the angle measurement accuracy while expanding the amount of training set data while participating in the training with unlabeled samples. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above contents and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0018] Figure 1 An application scenario diagram of an array radar angle measurement method, device, equipment, medium, and program product according to an embodiment of the present invention is shown;

[0019] Figure 2 A flow chart of an array radar angle measurement method according to an embodiment of the present invention is shown;

[0020] Figure 3 A schematic diagram of beam overlap scanning of an array radar according to an embodiment of the present invention is shown;

[0021] Figure 4 A flow chart of acquiring feature samples according to an array radar angle measurement method according to an embodiment of the present invention is shown;

[0022] Figure 5 A flow chart of obtaining angle values ​​according to an array radar angle measurement method according to an embodiment of the present invention is shown;

[0023] Figure 6 A flow chart of simplifying an initial model according to an array radar angle measurement method according to an embodiment of the present invention is shown;

[0024] Figure 7 A schematic diagram showing angle value test results of a spectrum method according to an embodiment of the present invention is shown;

[0025] Figure 8A schematic diagram showing error test results of a spectrum method according to an embodiment of the present invention is shown;

[0026] Fig. 9 A schematic diagram showing angle value test results of the SVR method according to an embodiment of the present invention is shown;

[0027] Fig.10 A schematic diagram showing error test results of an SVR method according to an embodiment of the present invention is shown;

[0028] Fig.11 A schematic diagram showing angle value test results of an array radar angle measurement method according to an embodiment of the present invention is shown;

[0029] Fig.12 A schematic diagram showing error test results of an array radar angle measurement method according to an embodiment of the present invention is shown;

[0030] Fig.13 A schematic diagram showing relative errors of multiple methods according to an embodiment of the present invention;

[0031] Fig.14 A structural block diagram of an array radar angle measuring device according to an embodiment of the present invention is shown; and

[0032] Fig.15 A block diagram of an electronic device suitable for implementing an array radar angle measurement method according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0033] Below, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of embodiments of the present invention. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of concepts of the present invention.

[0034] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0035] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0036] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0037] In forward-looking imaging, antenna pattern deconvolution technology is usually used to improve resolution. However, due to the limitation of single antenna aperture on resolution, the resolution is still not high at the current stage, and samples are scarce, resulting in reduced angle measurement performance. At the same time, when the platform to be measured is in motion, the antenna pattern is inaccurate, and the model mismatch caused by the platform movement reduces the accuracy.

[0038] Based on the above problems, the inventors found that overlapping scanning of a large-aperture multi-channel array can break through the limitation of a single antenna aperture. At the same time, an optimization model constructed based on a semi-supervised learning method with manifold regularization is used to process the array signal to eliminate the influence of errors in related technologies on the angle measurement accuracy.

[0039] Figure 1 An application scenario diagram of an array radar angle measurement method, device, equipment, medium, and program product according to an embodiment of the present invention is shown.

[0040] like Figure 1 As shown, the application scenario 100 according to this embodiment may include an array radar 101, a terminal device 102, a network 103, and a server 104. The network 103 is used to provide a medium for a communication link between the array radar 101, the terminal device 102, and the server 104. The network 103 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0041] The array radar 101 may be a device for transmitting radar signals, acquiring multi-channel echo signals, and sending the multi-channel echo signals to the server 104 via the network 103 .

[0042] The user can use the terminal device 102 to interact with the server 104 through the network 103 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 102, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0043] The terminal device 102 may be any electronic device having a display screen and supporting web browsing, including but not limited to a smart phone, a tablet computer, a laptop computer, a desktop computer, and the like.

[0044] The server 104 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the terminal device 102. The background management server may analyze and process received data such as user requests, and feed back processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0045] It should be noted that the array radar angle measurement method provided in the embodiment of the present invention can generally be performed by the server 104. Accordingly, the array radar angle measurement device provided in the embodiment of the present invention can generally be set in the server 104. The array radar angle measurement method provided in the embodiment of the present invention can also be performed by a server or server cluster that is different from the server 104 and can communicate with the terminal device 102 and / or the server 104. Correspondingly, the array radar angle measurement device provided in the embodiment of the present invention can also be set in a server or server cluster that is different from the server 104 and can communicate with the terminal device 102 and / or the server 104.

[0046] It should be understood that Figure 1 The number of array radars, terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0047] The following will be based on Figure 1 The scene described by Figure 2~Figure 13 The array radar angle measurement method according to an embodiment of the present invention is described in detail.

[0048] Figure 2 A flow chart of an array radar angle measurement method according to an embodiment of the present invention is shown.

[0049] like Figure 2 As shown, the array radar angle measurement method of this embodiment includes operations S210 to S230.

[0050] In operation S210 , an initial sample is acquired, where the initial sample is a multi-channel echo signal.

[0051] According to an embodiment of the present invention, the initial sample includes a tag echo signal and a tag-free echo signal.

[0052] In operation S220, the initial samples are converted into a beam space using a coherent synthesis technique to obtain feature samples.

[0053] According to an embodiment of the present invention, the coherent synthesis technology divides a target with a wide spectral range into several channels for independent modulation of amplitude and phase, and finally coherently superimposes the optical and electrical fields of several channels to synthesize a signal of the expected waveform.

[0054] According to an embodiment of the present invention, the coherent synthesis technology is based on two basic conditions, including that when an airborne radar performs air-to-ground mode detection, the distance between the radar and the target is usually much larger than the radar aperture; and a small aperture radar is used for deconvolution, that is, the distance between each two array elements is very small.

[0055] According to the embodiment of the present invention, since the far-field condition is met, the target echo can be regarded as a plane wave; and the distance between radar array elements is very close, the angle and distance of different array elements to the same target are very close, that is, the Doppler information of two array elements is also the same. Therefore, the coherent synthesis technology is used to process the echo signal, and the Doppler information contained in the echo is almost not destroyed.

[0056] In operation S230, the feature sample is input into an optimization model, and an angle value is output. The optimization model is constructed based on a popular regularization constraint and a support vector regression algorithm.

[0057] According to an embodiment of the present invention, the main task of the support vector regression (SVR) algorithm is to train the functional relationship between input values ​​and output values, and it has a strong nonlinear fitting capability.

[0058] According to an embodiment of the present invention, manifold regularization is a semi-supervised regularization method based on the "manifold hypothesis". The "manifold hypothesis" assumes that all labeled data and unlabeled data are located in a low-dimensional manifold space. If two input samples are similar in this structure, then the output results should also be similar. The degree of proximity between samples can often be expressed by the Euclidean distance defined by two samples in this space. The above assumptions must require that the decision function must be locally smooth at the sample points to ensure that adjacent inputs are represented by adjacent outputs.

[0059] According to an embodiment of the present invention, the coherent synthesis technology is used to convert the acquired initial samples into the beam space to obtain feature samples; then the feature samples are input into the optimization model to output the angle value, and the optimization model is constructed based on the popular regularization constraint and the support vector regression algorithm. By using the coherent synthesis method to perform dimensionality reduction processing on the initial samples, and using the optimization model constructed based on the popular regularization constraint and the support vector regression algorithm, semi-supervised training is performed on the labeled samples and the unlabeled samples, thereby avoiding the high dependence on the training set data, resulting in the need to obtain a large amount of labeled data for multiple trainings, and the problem of large errors in the output angle values, and realizing the technical effect of reducing the sensitivity of the phase error and improving the angle measurement accuracy while expanding the amount of training set data while participating in the training with unlabeled samples.

[0060] According to an embodiment of the present invention, obtaining the initial sample includes: using any transmitting antenna of the array radar to transmit a radar signal to surrounding objects to obtain a target reflected wave signal, where the target reflected wave signal is obtained by reflecting the radar signal; using multiple receiving antennas of the array radar to receive the target reflected wave signal to obtain the initial sample.

[0061] According to an embodiment of the present invention, any transmitting antenna of the array radar transmits a beam for scanning in the first pulse cycle. After the scanning is completed, the main lobe of the beam transmitted in the next pulse cycle of the array radar is offset by one super-resolution unit and then scanned; wherein one pulse cycle includes three super-resolution units.

[0062] Figure 3 A schematic diagram of beam overlap scanning of an array radar according to an embodiment of the present invention is shown.

[0063] like Figure 3 As shown in Figure 2, several super-resolution units are divided. Assume that each beam width is divided into three super-resolution units. The main lobe of the first pulse cycle beam points to the first three super-resolution units. Figure 3 The main lobe direction of the second pulse period beam rotates by an angle of a super-resolution unit. Figure 3 In the figure, it is marked with the number 2; and so on, the main lobe of the beam shifts by one super-resolution unit for data acquisition every pulse cycle. The finer the division of the super-resolution unit, the more accurate the result should be.

[0064] According to an embodiment of the present invention, multiple receiving antennas of an array radar can collect reflected wave signals of multiple targets at the same time to achieve fast overlapping scanning, thereby avoiding the problem of taking a long time. At the same time, the use of an array radar can flexibly schedule beams without the need to scan in sequence, and a reasonable design of the scanning wave position can effectively increase the Doppler bandwidth.

[0065] According to an embodiment of the present invention, the initial samples are converted to beam space using coherent synthesis technology to obtain feature samples, including: performing weighted summation on the initial samples using coherent synthesis technology to obtain first processed samples; and performing feature selection on the first processed samples to obtain feature samples.

[0066] According to an embodiment of the present invention, assuming that a planar array is used for scanning, a certain array element is selected as a reference array element, and its coordinates are set to (0,0). The relative position of the i-th array element of the array radar relative to the reference array element is set to , the wavelength of the radar signal is In the kth pulse repetition period, the radar transmit beam angle value is , as shown in formula (1), the initial samples are converted to beam space using coherent synthesis technology, where represents the initial sample of the i-th beam space channel, Then it is represented as the kth first processed sample, and j is an imaginary unit. After L pulse repetition cycles, the first processed sample Y after coherent synthesis can be expressed as formula (2), where the first processed sample Y includes The first treatment subsample.

[0067] (1);

[0068] (2);

[0069] According to an embodiment of the present invention, the array radar has multiple receiving channels, which will cause the dimension of the initial sample to increase sharply. In order not to increase the dimension of data processing, the coherent synthesis technology is used to reduce the dimension of the initial sample, which effectively avoids the degradation of the angle measurement performance caused by the excessively high dimension of the initial sample. At the same time, when the transmission power is the same, the first processed sample has a larger signal-to-noise ratio.

[0070] Figure 4 A flow chart of acquiring feature samples of an array radar angle measurement method according to an embodiment of the present invention is shown.

[0071] like Figure 4 As shown, performing feature selection on the first processed sample to obtain a feature sample includes operations S410 to S440.

[0072] In operation S410, a first processed sample is mapped to a correlation matrix.

[0073] According to an embodiment of the present invention, the coherently synthesized first processed samples are mapped to a higher dimensional correlation matrix.

[0074] According to an embodiment of the present invention, due to the influence of noise and error, mapping the coherently synthesized first processed samples to a higher-dimensional correlation matrix can improve the stability of the first processed samples.

[0075] In operation S420, upper triangular elements of the correlation matrix are determined as second processing samples.

[0076] According to the embodiment of the present invention, since radar data usually has amplitude and phase, the above correlation matrix is ​​a complex symmetric matrix, and the upper triangular part or the lower triangular part (excluding diagonal elements) of the correlation matrix is ​​taken as the input element. If the number of samples scanned by the radar signal is L, the correlation matrix has L 2 elements, excluding the diagonal elements, are L 2 -L elements, and then take the upper triangle or lower triangle as the target diagonal elements, the number of which is L', as shown in formula (3). The second processed sample is expressed as shown in formula (4), where Y' is expressed as the matrix form of the second processed sample, which includes A second treatment subsample.

[0077] (3);

[0078] (4);

[0079] In operation S430, a normalization process is performed on the plurality of second processed samples to obtain a plurality of candidate feature samples.

[0080] According to an embodiment of the present invention, the second processed sample is a complex vector at this time, so it is necessary to perform normalization processing to map the second processed sample to a specified range to obtain multiple candidate feature samples. , as shown in formula (5), where Y' is the second processed sample obtained above, real represents the real part of the second processed sample, and imag represents the imaginary part of the second processed sample.

[0081] (5);

[0082] In operation S440, a candidate feature sample that meets a preset condition among the plurality of candidate feature samples is determined as a feature sample.

[0083] According to an embodiment of the present invention, the preset condition may be to sort the sample values ​​of a plurality of candidate feature samples from large to small, and select the candidate feature samples with the top five sample values.

[0084] According to an embodiment of the present invention, mapping the first processed sample to the correlation matrix improves the stability of the sample data while also increasing the dimension of the first processed sample, determining a candidate feature sample with a larger value among multiple candidate feature samples as a feature sample, and performing normalization processing on the first processed sample to ensure that no complex numbers appear in the angle measurement results.

[0085] Figure 5 A flow chart of obtaining angle values ​​of an array radar angle measurement method according to an embodiment of the present invention is shown.

[0086] like Figure 5 As shown, the feature samples include labeled samples and unlabeled samples, the first moment of obtaining the labeled samples is earlier than the second moment of obtaining the unlabeled samples, the target angle of the labeled samples is known, and the target angle of the unlabeled samples is unknown; wherein, the feature samples are input into the optimization model, and the output angle value includes operation S510~operation S570.

[0087] In operation S510, an initial model is constructed based on the popular regularization constraint and the support vector regression algorithm.

[0088] According to an embodiment of the present invention, by testing the function Transform the feature samples so that the transformed test function is close to the angle value of the target reflected wave signal. , under the support vector regression algorithm, the corresponding optimization problem As shown in formula (6), is the loss function, where is the angle value of the labeled sample, are labeled samples. is the regularization term, where is the regularization parameter. The regularization term is used to add regular restrictions to the above loss function to prevent the model from overfitting. From this formula, it can be seen that the result of the optimization problem is closely related to the given labeled samples, and the structural information between the unlabeled samples and the labeled samples is not utilized.

[0089] (6);

[0090] According to the embodiment of the present invention, due to the influence of factors such as array configuration, pattern error, motion error, noise, etc., the feature samples may not be uniform in the original space and should be constrained in a certain manifold space. In. In , if the similarity distance between labeled samples and unlabeled samples is small enough, then the angle values ​​of labeled samples and unlabeled samples should also be close. Therefore, the above optimization problem is subjected to manifold regularization constraints, as shown in formula (7), , is the regularization parameter, controls the smoothness in the reproducing kernel Hilbert space, Controls the strength of the manifold regularization constraint. is a weight parameter. If the value is too large, the loss function cannot work effectively. If the value is too small, the similarity relationship between samples cannot be fully mined.

[0091] (7);

[0092] in, is the loss function, which represents the empirical risk. After the action, the degree of closeness between the angle estimation result of the labeled sample and the preset label angle value.

[0093] According to an embodiment of the present invention, yes The norm defined in the reproducing Hilbert space is used to prevent the model from overfitting.

[0094] According to an embodiment of the present invention, Representing manifold space The popular structure of the model is the constraints of the model, where Indicates labeled samples and unlabeled samples In the manifold The similarity in is as shown in formula (8), with labeled samples and unlabeled samples The smaller the distance between The larger the To ensure that the estimated angle values ​​are highly close. Only a finite number of neighboring points are taken, that is, only a sufficiently large finite number of ,the remaining All are reset to zero. is the Gaussian bandwidth parameter, The value of determines the similarity at a certain Euclidean distance. If the value is too large, the similarity between labeled samples and unlabeled samples will decrease, and each sample will become an isolated individual without connection. If the value is too large, the similarity between labeled samples and unlabeled samples will increase, and samples with a relatively long distance may also be considered similar samples.

[0095] (8);

[0096] According to an embodiment of the present invention, in general, the loss function can be defined as a square loss function, as shown in formula (9), so the initial model can be constructed by formula (10).

[0097] (9);

[0098] (10);

[0099] In operation S520, the initial model is processed according to the representation theorem to obtain a test model.

[0100] According to an embodiment of the present invention, the representation theorem is a theorem in statistical learning, which can be used to solve the optimization problem represented by formula (10). As shown in formula (11), Z is an unlabeled sample of the angle value to be estimated, is the i-th sub-coefficient in the combined coefficient, Target supervision value output for the test model.

[0101] (11);

[0102] According to an embodiment of the present invention, suppose there are m label samples, and suppose the combination coefficient is B, including Sub-coefficients, that is . The solution can be a kernel function A linear combination of It is converted into formula (12), where is the bandwidth parameter of the kernel function, The value of determines the similarity under the Euclidean distance. If the value is too large, the similarity between labeled samples and unlabeled samples will be reduced, and each sample will be an independent individual without connection. If the value is too large, the similarity between labeled samples and unlabeled samples will increase, and samples with a relatively long distance may also be considered similar samples.

[0103] (12);

[0104] In operation S530, the initial model is simplified to obtain an optimization objective function.

[0105] Figure 6 A flow chart of simplifying an initial model of an array radar angle measurement method according to an embodiment of the present invention is shown.

[0106] like Figure 6 As shown, the initial model is constructed based on the initial norm, initial constraints and initial loss function; wherein, simplifying the initial model to obtain the second optimization objective function includes operation S610 to operation S640.

[0107] In operation S610, the initial norm is regeneratively transformed to obtain a target norm.

[0108] According to an embodiment of the present invention, yes The norm defined in the reproducing Hilbert space is expressed according to the reproducibility as formula (13), where represents the jth combination coefficient.

[0109] According to an embodiment of the present invention, K is a kernel function The square matrix is ​​composed of, then formula (13) can be expressed in matrix form, and the norm to be solved is shown in formula (14), where is the transpose of the combination coefficient B.

[0110] (13);

[0111] (14);

[0112] In operation S620, as shown in formula (15), the initial constraint condition is simplified according to the graph Laplacian matrix to obtain the manifold constraint condition. Where L is the graph Laplacian matrix, Represents the manifold regularization parameter of the feature sample, with a value of 1.

[0113] (15);

[0114] In operation S630, as shown in formula (16), the initial loss function is simplified according to the manifold constraint and the target norm to obtain a loss function.

[0115] (16);

[0116] According to an embodiment of the present invention, when there are m labeled samples and n unlabeled samples, the above is a column vector, representing the feature sample matrix. It is a diagonal matrix, which represents the selection matrix of labeled samples.

[0117] In operation S640, the target norm, manifold constraint and loss function are used to update the initial model to obtain an optimized target function. The operation of obtaining the optimized target function is shown in formula (17).

[0118] (17);

[0119] In operation S540, the optimization objective function is solved to obtain a training model. The solving process is shown in formula (18), where I represents a unit matrix.

[0120] (18);

[0121] In operation S550, the labeled samples and the unlabeled samples are input into the training model to obtain the combination coefficients.

[0122] According to an embodiment of the present invention, since the feature samples include labeled samples and unlabeled samples, semi-supervised training and learning can be implemented when input into the training model.

[0123] In operation S560, the combined coefficients are input into the test model to obtain an optimized model.

[0124] According to an embodiment of the present invention, the combination coefficients obtained above are used to replace the combination coefficients in the above test model to obtain an updated test model, that is, an optimized model.

[0125] In operation S570, an unlabeled sample of the angle to be estimated is input into the optimization model to obtain an angle value.

[0126] According to an embodiment of the present invention, a sample whose angle needs to be estimated is extracted from the unlabeled samples, and the sample is input into the optimization model obtained above. The optimization model performs calculation processing on the sample to obtain an angle value corresponding to the unlabeled sample whose angle needs to be estimated.

[0127] According to an embodiment of the present invention, the feature sample is input into the optimization model, and the output angle value also includes: recording the number of executions of the angle value obtained according to the optimization model, and when the number of executions is less than or equal to a preset threshold, the optimization model is adopted; when the number of executions is greater than the preset threshold, the initial model is retrained.

[0128] According to an embodiment of the present invention, as the number of executions accumulates, the dimension of the data becomes higher and higher, which in turn brings an excessively large computational burden. Therefore, the number of executions is limited by a preset threshold to effectively avoid the problem of excessively high dimensions.

[0129] According to an embodiment of the present invention, an array radar first obtains an initial sample by overlapping scanning of a beam; performs dimension reduction processing on the initial sample by using coherent synthesis technology to obtain a first processed sample, maps the first processed sample to a high-dimensional correlation matrix, removes the upper triangular elements as a second processed sample, and then performs normalization processing on the second processed sample to obtain a feature sample; constructs an initial model based on a popular regularization constraint and a support vector regression algorithm, and processes the initial model according to a representation theorem to obtain a training model; simplifies the initial model to obtain an optimization objective function and solves it to obtain a training model; inputs the feature sample into the training model, and inputs the obtained combination coefficient into a test model to obtain an optimized model; inputs an unlabeled sample of an angle to be estimated into the optimization model to obtain an angle value.

[0130] According to an embodiment of the present invention, in order to test the array radar angle measurement method of the present invention, a computer is used to perform simulation based on a simulation tool. It is assumed that the flight speed of the carrier aircraft is 60m / s, the minimum beam width of the array is 3.38°, the scanning range is ±10°, a total of 71 overlapping wave positions are uniformly scanned, and the pulse repetition frequency is 10KHz.

[0131] Figure 7 A schematic diagram showing angle value test results of a spectrum method according to an embodiment of the present invention is shown; Figure 8 A schematic diagram showing error test results of a spectrum method according to an embodiment of the present invention is shown; Fig. 9 A schematic diagram showing angle value test results of the SVR method according to an embodiment of the present invention is shown; Fig.10 A schematic diagram showing error test results of an SVR method according to an embodiment of the present invention is shown; Fig.11 A schematic diagram showing angle value test results of an array radar angle measurement method according to an embodiment of the present invention is shown; Fig.12 A schematic diagram showing error test results of an array radar angle measurement method according to an embodiment of the present invention is shown; Fig.13 A schematic diagram showing relative errors of multiple methods according to an embodiment of the present invention is shown.

[0132] like Figures 7 to 13As shown, the average error measured based on the spectral estimation method is 5.49%, the average error measured based on the SVR method is 3.65%, and the average error measured based on the array radar angle measurement method of the present invention is 3.48%. Compared with the traditional method, the array radar angle measurement method of the present invention further explores the popular structure of feature samples and enables unlabeled samples to be applied to the optimization model. Therefore, compared with the traditional method, the angle measured is more accurate and the relative error is smaller.

[0133] Based on the above array radar angle measurement method, the present invention also provides an array radar angle measurement device. Fig.14 The device is described in detail.

[0134] Fig.14 A structural block diagram of an array radar angle measuring device according to an embodiment of the present invention is shown.

[0135] like Fig.14 As shown, the array radar angle measuring device 1400 of this embodiment includes an acquisition module 1410, a processing module 1420 and an output module 1430.

[0136] The acquisition module 1410 is used to acquire an initial sample, which is a multi-channel echo signal. In one embodiment, the acquisition module 1410 can be used to perform the operation S210 described above, which will not be described in detail here.

[0137] The processing module 1420 is used to convert the initial samples into the beam space by using the coherent synthesis technology to obtain the characteristic samples. In one embodiment, the processing module 1420 can be used to perform the operation S220 described above, which will not be described in detail here.

[0138] The output module 1430 is used to input the feature sample into the optimization model and output the angle value. The optimization model is constructed based on the popular regularization constraint and the support vector regression algorithm. In one embodiment, the output module 1430 can be used to perform the operation S230 described above, which will not be repeated here.

[0139] According to an embodiment of the present invention, the coherent synthesis technology is used to convert the acquired initial samples into the beam space to obtain feature samples; then the feature samples are input into the optimization model to output the angle value, and the optimization model is constructed based on the popular regularization constraint and the support vector regression algorithm. By using the coherent synthesis method to perform dimensionality reduction processing on the initial samples, and using the optimization model constructed based on the popular regularization constraint and the support vector regression algorithm, semi-supervised training is performed on the labeled samples and the unlabeled samples, thereby avoiding the high dependence of the training set data, resulting in the need to obtain a large amount of labeled data for multiple training, and the problem of large error in the output angle value, so as to achieve the technical effect of reducing the sensitivity of the phase error and improving the angle measurement accuracy while expanding the amount of training set data while participating in the training of unlabeled samples. According to an embodiment of the present invention, the acquisition module 1410 includes a transmitting unit and a receiving unit.

[0140] The transmitting unit is used to transmit radar signals to surrounding objects using any transmitting antenna of the array radar to obtain target reflected wave signals, which are obtained by reflecting the radar signals.

[0141] The receiving unit is used to receive the target reflected wave signal using multiple receiving antennas of the array radar to obtain initial samples.

[0142] According to an embodiment of the present invention, the processing module 1420 includes a first processing unit and a selection unit.

[0143] The first processing unit is used to perform weighted summation on the initial samples by using the coherent synthesis technology to obtain the first processed samples.

[0144] The selection unit is used to perform feature selection on the first processed sample to obtain a feature sample.

[0145] According to an embodiment of the present invention, the selection unit includes a mapping subunit, a first determination subunit, a processing subunit and a second determination subunit.

[0146] The mapping subunit is used to map the first processed sample to a correlation matrix.

[0147] The first determining subunit is used to determine the upper triangular elements of the correlation matrix as second processing samples.

[0148] The processing subunit is used to perform normalization processing on the multiple second processed samples to obtain multiple candidate feature samples.

[0149] The second determining subunit determines a candidate feature sample with a larger value among the plurality of candidate feature samples as a feature sample.

[0150] According to an embodiment of the present invention, the output module 1430 includes a construction unit, a second processing unit, a simplification unit, a solving unit, a first obtaining unit, a second obtaining unit and a third obtaining unit.

[0151] The construction unit is used to construct an initial model based on popular regularization constraints and support vector regression algorithm.

[0152] The second processing unit is used to process the initial model according to the representation theorem to obtain a test model.

[0153] The simplified element is used to simplify the initial model and obtain the optimized objective function.

[0154] The solving unit is used to solve the optimization objective function to obtain a training model.

[0155] The first obtaining unit is used to input the labeled samples and the unlabeled samples into the training model to obtain the combination coefficients.

[0156] The second obtaining unit is used to input the combination coefficients into the test model to obtain the optimization model.

[0157] The third obtaining unit is used to input the unlabeled sample of the angle to be estimated into the optimization model to obtain the angle value.

[0158] According to an embodiment of the present invention, the simplification unit includes a conversion subunit, a first simplification subunit, a second simplification subunit and an update subunit.

[0159] The conversion subunit is used to perform regenerative transformation on the initial norm to obtain the target norm.

[0160] The first simplification subunit is used to simplify the initial constraint conditions according to the graph Laplacian matrix to obtain the manifold constraint conditions.

[0161] The second simplification subunit is used to simplify the initial loss function according to the manifold constraint and the target norm to obtain a loss function.

[0162] The updating subunit is used to update the initial model with the target norm, manifold constraints and loss function to obtain the optimized target function.

[0163] According to an embodiment of the present invention, the output module 1430 further includes a recording unit and a training unit.

[0164] The recording unit is used to record the number of executions of the new feature sample input into the optimization model. When the number of executions is less than or equal to a preset threshold, the current optimization model is still used.

[0165] The training unit is used to retrain the initial model when the number of executions is greater than a preset threshold.

[0166] According to an embodiment of the present invention, any multiple modules among the acquisition module 1410, the processing module 1420 and the output module 1430 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present invention, at least one of the acquisition module 1410, the processing module 1420 and the output module 1430 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in any appropriate combination of any of them. Alternatively, at least one of the acquisition module 1410, the processing module 1420 and the output module 1430 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be executed.

[0167] Fig.15 A block diagram of an electronic device suitable for implementing an array radar angle measurement method according to an embodiment of the present invention is shown.

[0168] like Fig.15 As shown, the electronic device 1500 according to an embodiment of the present invention includes a processor 1501, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 1502 or the program loaded from the storage part 1508 to the random access memory (RAM) 1503. The processor 1501 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 1501 may also include an onboard memory for caching purposes. The processor 1501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0169] In RAM 1503, various programs and data required for the operation of electronic device 1500 are stored. Processor 1501, ROM 1502 and RAM 1503 are connected to each other via bus 1504. Processor 1501 performs various operations of the method flow according to the embodiment of the present invention by executing the programs in ROM 1502 and / or RAM 1503. It should be noted that the program can also be stored in one or more memories other than ROM 1502 and RAM 1503. Processor 1501 can also perform various operations of the method flow according to the embodiment of the present invention by executing the programs stored in the one or more memories.

[0170] According to an embodiment of the present invention, the electronic device 1500 may further include an input / output (I / O) interface 1505, which is also connected to the bus 1504. The electronic device 1500 may further include one or more of the following components connected to the I / O interface 1505: an input portion 1506 including a keyboard, a mouse, etc.; an output portion 1507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 1508 including a hard disk, etc.; and a communication portion 1509 including a network interface card such as a LAN card, a modem, etc. The communication portion 1509 performs communication processing via a network such as the Internet. A drive 1510 is also connected to the I / O interface 1505 as needed. A removable medium 1511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1510 as needed, so that a computer program read therefrom is installed into the storage portion 1508 as needed.

[0171] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiment; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present invention is implemented.

[0172] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present invention, the computer-readable storage medium may include the ROM 1502 and / or RAM 1503 described above and / or one or more memories other than ROM 1502 and RAM 1503.

[0173] The embodiment of the present invention also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the above method provided by the embodiment of the present invention.

[0174] The computer program executes the above functions defined in the system / device of the embodiment of the present invention when it is executed by the processor 1501. According to the embodiment of the present invention, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0175] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 1509, and / or installed from the removable medium 1511. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0176] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1509, and / or installed from the removable medium 1511. When the computer program is executed by the processor 1501, the above functions defined in the system of the embodiment of the present invention are performed. According to the embodiment of the present invention, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.

[0177] According to an embodiment of the present invention, the program code for executing the computer program provided by the embodiment of the present invention can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0178] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0179] It will be appreciated by those skilled in the art that the features described in the various embodiments and / or claims of the present invention may be combined and / or combined in various ways, even if such combinations and / or combinations are not explicitly described in the present invention. In particular, the features described in the various embodiments and / or claims of the present invention may be combined and / or combined in various ways without departing from the spirit and teachings of the present invention. All of these combinations and / or combinations fall within the scope of the present invention.

[0180] The embodiments of the present invention are described above. However, these embodiments are only for the purpose of illustration, and are not intended to limit the scope of the present invention. Although each embodiment is described above, it does not mean that the measures in each embodiment cannot be used in combination. The scope of the present invention is defined by the attached claims and their equivalents. Without departing from the scope of the present invention, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present invention.

Claims

1. A method for measuring angles by an array radar, characterized in that: The method comprises: Acquire an initial sample, where the initial sample is a multi-channel echo signal; The initial samples are converted into beam space by using coherent synthesis technology to obtain characteristic samples; The feature samples are input into an optimization model, and an angle value is output. The optimization model is constructed based on a popular regularization constraint and a support vector regression algorithm. The feature samples include labeled samples and unlabeled samples. The first moment of obtaining the labeled samples is earlier than the second moment of obtaining the unlabeled samples. The target angle of the labeled samples is known, while the target angle of the unlabeled samples is unknown. The feature sample is input into the optimization model, and the angle value is output, including: An initial model is constructed based on the popular regularization constraint and the support vector regression algorithm; Processing the initial model according to the representation theorem to obtain a test model; Simplifying the initial model to obtain an optimization objective function; Solving the optimization objective function to obtain a training model; Inputting the labeled samples and the unlabeled samples into the training model to obtain a combination coefficient; Inputting the combination coefficients into the test model to obtain the optimization model; The unlabeled sample of the angle to be estimated is input into the optimization model to obtain the angle value.

2. The array radar angle measurement method according to claim 1, characterized in that: The obtaining of the initial sample comprises: Using any transmitting antenna of the array radar to transmit a radar signal to a surrounding object to obtain a target reflected wave signal, wherein the target reflected wave signal is obtained by reflecting the radar signal; The target reflected wave signal is received by using multiple receiving antennas of the array radar to obtain the initial sample.

3. The array radar angle measurement method according to claim 1, characterized in that: The coherent synthesis technology is used to convert the initial sample into the beam space to obtain the characteristic sample, including: Using the coherent synthesis technology to perform weighted summation on the initial samples to obtain a first processed sample; Feature selection is performed on the first processed sample to obtain a feature sample.

4. The array radar angle measurement method according to claim 3, characterized in that: The step of performing feature selection on the first processed sample to obtain a feature sample includes: mapping the first processed samples to a correlation matrix; determining the upper triangular elements of the correlation matrix as second processing samples; performing normalization processing on the plurality of the second processed samples to obtain a plurality of candidate feature samples; A candidate feature sample that meets a preset condition among the plurality of candidate feature samples is determined as the feature sample.

5. The array radar angle measurement method according to claim 1, characterized in that: The initial model is constructed based on the initial norm, initial constraint conditions and initial loss function; The simplification of the initial model to obtain an optimization objective function includes: Performing a regenerative transformation on the initial norm to obtain a target norm; Simplifying the initial constraint condition according to the graph Laplace matrix to obtain a manifold constraint condition; Simplifying the initial loss function according to the manifold constraint and the target norm to obtain a loss function; The target norm, the manifold constraint and the loss function are used to update the initial model to obtain the optimized target function.

6. The array radar angle measurement method according to claim 1, characterized in that: Inputting the feature sample into the optimization model and outputting the angle value also includes: Recording the number of executions of obtaining the angle value according to the optimization model, and adopting the optimization model when the number of executions is less than or equal to a preset threshold; When the execution times are greater than the preset threshold, the initial model is retrained.

7. An array radar angle measuring device, characterized in that: The device comprises: An acquisition module, used for acquiring an initial sample, wherein the initial sample is a multi-channel echo signal; A processing module, used for converting the initial samples into beam space by using coherent synthesis technology to obtain characteristic samples; and An output module is used to input the feature samples into an optimization model and output an angle value, wherein the optimization model is constructed based on a popular regularization constraint and a support vector regression algorithm, wherein the feature samples include labeled samples and unlabeled samples, and the first moment of obtaining the labeled samples is earlier than the second moment of obtaining the unlabeled samples; the target angle of the labeled samples is known; and the target angle of the unlabeled samples is unknown; The feature sample is input into the optimization model, and the angle value is output, including: An initial model is constructed based on the popular regularization constraint and the support vector regression algorithm; Processing the initial model according to the representation theorem to obtain a test model; Simplifying the initial model to obtain an optimization objective function; Solving the optimization objective function to obtain a training model; Inputting the labeled samples and the unlabeled samples into the training model to obtain a combination coefficient; Inputting the combination coefficients into the test model to obtain the optimization model; The unlabeled sample of the angle to be estimated is input into the optimization model to obtain the angle value.

8. An electronic device, characterized in that: The electronic device comprises: one or more processors; a memory for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: Executable instructions are stored thereon, and when the instructions are executed by a processor, the processor executes the method according to any one of claims 1 to 6.

Images