A method, system, device, and medium for phase-free planar near-field data reconstruction based on convolution-deconvolution.

By combining convolution-deconvolution and conjugate gradient methods, the problem of low reconstruction accuracy in phase-free plane near-field measurements is solved, achieving efficient and accurate near-field data reconstruction and far-field pattern recovery.

CN117574031BActive Publication Date: 2026-11-17XI AN JIAOTONG UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202311464954.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-06
Publication Date
2026-11-17
Estimated Expiration
2043-11-06

AI Technical Summary

Technical Problem

Existing phase-free plane near-field measurement methods suffer from low reconstruction accuracy and high testing costs in the millimeter-wave band. In particular, the traditional iterative inverse Fourier transform method and cost function method need to be improved in terms of iterative convergence speed and reconstruction accuracy.

Method used

A convolution-deconvolution-based method is adopted, which iteratively transforms the electric fields on two near-field planes through spatial convolution-deconvolution functions, and combines the conjugate gradient method to minimize the cost function, thereby realizing complex field transformation between planes and improving accuracy.

Benefits of technology

It improves the accuracy and iterative convergence of phase-free plane near-field data reconstruction, reduces testing costs, and effectively reconstructs the complex field and corresponding far-field pattern on the target near-field plane.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117574031B_ABST
    Figure CN117574031B_ABST
Patent Text Reader

Abstract

This invention belongs to the field of microwave measurement technology and relates to a phase-free planar near-field data reconstruction method based on convolution-deconvolution. It utilizes spatial convolution-deconvolution to iteratively transform the electric fields on two near-field planes. Simultaneously, a cost function is constructed for the complex field on plane S1, and then minimized using the conjugate gradient method to obtain the optimal reconstruction result after each iteration. After multiple iterations, the reconstruction error of the complex field on plane S1 gradually converges, and the NFT far-field pattern derived from the reconstructed complex field on plane S1 shows good accuracy compared to the theoretical value. This method not only overcomes the phase loss limitation of phase-free planar near-field testing but also reduces the testing cost of complex fields and effectively reconstructs the complex field and corresponding far-field pattern on the target near-field plane.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of microwave measurement technology, specifically relating to a method, system, device, and medium for reconstructing phase-free planar near-field data based on convolution-deconvolution. Background Technology

[0002] Near-field measurement technology has been widely used in electromagnetic propagation and antenna measurement. Because near-field testing requires near-to-far-field transformation (NFT) algorithms to obtain the far-field radiation pattern of the antenna under test (AUT), planar near-field testing is a more convenient and efficient method compared to spherical and ellipsoidal near-field testing. However, in the millimeter-wave band, obtaining accurate phase information becomes more difficult, and the corresponding testing costs are higher. Furthermore, phase information at higher frequencies is more susceptible to interference from the testing environment, while amplitude information has relatively high robustness. Therefore, in millimeter-wave testing, using phase-free testing can reduce testing complexity and cost, and minimize the impact of probe measurement errors on the test results.

[0003] Currently, phase reconstruction methods for near-field measurements in phase-free planes are mainly divided into two categories: the first is the Inverse Iterative Fourier Transform (IFT) method, and the second is the cost function method. The IFT method requires acquiring amplitude information on two near-field planes, plane 1 and plane 2. Then, an initial phase is given to plane 1 to form an initial two-dimensional complex field. Subsequently, multiple iterative operations are performed between the two planes using forward and inverse Fourier transforms to obtain the iteratively improved two-dimensional complex field of plane 1. However, the convergence and data reconstruction accuracy of this method are significantly affected by the choice of the initial phase.

[0004] The second type of method constructs a corresponding cost function for the two-dimensional electric field amplitude, equivalent magnetic current, or plane-wave spectrum (PWS) of planar near-field data. In this way, the reconstruction accuracy of the two-dimensional complex field is improved by reducing the cost function during multiple iterations of reconstruction. However, this type of method only performs near-field and far-field data extrapolation and reconstruction on one near-field plane or two near-field planes, without realizing information exchange and data reconstruction between planes. Therefore, its iterative convergence speed and reconstruction accuracy need to be further improved. Summary of the Invention

[0005] The purpose of this invention is to provide a phase-free planar near-field data reconstruction method, system, device and medium based on convolution-deconvolution, which solves the problem of low accuracy in existing reconstruction methods.

[0006] This invention is achieved through the following technical solution:

[0007] A phase-free planar near-field data reconstruction method based on convolution-deconvolution includes the following steps:

[0008] Step 1: Obtain the planar electric field amplitudes of the antenna under test on planes S1 and S2 using a planar near-field testing system;

[0009] Step 2: Assign an initial phase to plane S1, and combine the initial phase with the amplitude of the plane electric field that has been collected on plane S1 to obtain the initial complex field on plane S1;

[0010] Step 3: Substitute the initial complex field on plane S1 into the spatial convolution function to obtain a two-dimensional complex field on plane S2.

[0011] Subsequently, the amplitude of the planar electric field on plane S2 obtained in step 1 is used to replace the amplitude of the two-dimensional complex field on plane S2 obtained by convolution, and the reconstructed complex field on plane S2 is obtained.

[0012] Step 4: Substitute the complex field on plane S2 into the spatial deconvolution function, and reconstruct the reconstructed complex field on plane S1 in reverse.

[0013] At this point, the complex field on the plane S1 after the spatial convolution-deconvolution processing as described in steps 3-4 is denoted as matrix U. n And calculate the corresponding cost function, where n represents the number of iterations to complete steps 3-4;

[0014] Step 5: Find the descending gradient of the cost function and apply it to matrix U. n Calculate the updated U based on n+1 ;

[0015] Step 6: Calculate U n+1 The error between the known electric field amplitude on plane S1;

[0016] If the amplitude error is small and tends to stabilize, it indicates that the iterative process of steps 2-5 is gradually converging. Stop the iteration and reconstruct the U value. n+1 As the final value of the complex field on plane S1; otherwise, U n+1 Substitute the initial complex field from step 3 onto plane S1, and repeat the iterative process from step 3 to step 6 until the error converges.

[0017] Furthermore, in step 1, the coordinates on the near-field scanning plane S1 and plane S2 are represented as r′=(x′,y′,z′) and r=(x,y,z), respectively, and the measured electric field vectors corresponding to plane S1 and plane S2 are represented as E1(r′) and E2(r), respectively.

[0018] Furthermore, in step 2, the measurement amplitudes along the x and y directions on the scanning plane S1 are known, denoted as |E|. 1,x′ |and|E 1,y′ | where the phase is an unknown quantity, and an initial phase, denoted as E, needs to be given. 1,x′,pha and E 1,y′,pha ;

[0019] Combining the known amplitude with the given initial phase, we obtain the initial complex field E on plane S1. 1,n (r′).

[0020] Furthermore, in steps 3 and 4, the construction process of the spatial convolution function and the spatial deconvolution function is as follows:

[0021] The electric field in a plane in free space can be represented in the form of an equivalent magnetic current:

[0022] J 1,m (r′)=-2n×E1(r′) (10)

[0023] Where E1(r′) is the electric field vector on plane S1, J 1,m (r′) is the equivalent magnetic current, and its corresponding vector potential A 1,m for:

[0024]

[0025] Where G(rr′) represents the Green's function, the electric field E2(r) on plane S2 is expressed as:

[0026]

[0027] From equations (1)-(3), we obtain that E1(r′) and E2(r) satisfy a convolution relationship, which is:

[0028] F(x,y)=∫∫f(x′,y′)g(xx′,yy′)dx′dy′ (13)

[0029] Where f(x,y) represents the electric field component on plane S1, F(x,y) represents the electric field component on plane S2, and g(x,y) is the spatial Green's function;

[0030] Therefore, the electric fields on planes S1 and S2 can achieve the mutual conversion between spatial convolution and deconvolution, resulting in:

[0031]

[0032]

[0033] Equations (5) and (6) are defined as spatial convolution function and spatial deconvolution function, respectively.

[0034] Furthermore, in step 4, the complex field U that has undergone the above convolution-deconvolution process for the nth time is... n The corresponding cost function is expressed as:

[0035]

[0036] Where G is the Green's function matrix, and |E2| is the magnitude matrix of the measured electric field vector E2(r) on plane S2. Describes the 2-norm of a vector. It is the normalization coefficient.

[0037] Furthermore, in step 5, the cost function is minimized using the cost function and the conjugate gradient of CG, and U is... n Updated to U n+1 The iteratively updated complex field matrix U n+1 The expression is:

[0038] U n+1 =U n +V n =U n +g n +β n V n-1 (17)

[0039] Where g n It is C n (U n The gradient of β n It is the weighting coefficient, V n-1 It is the gradient value from the previous iteration, with an initial value of V0 = 0.

[0040] Furthermore, in step 6, after obtaining U n+1 Then, calculate U n+1 The magnitude error ε between the magnitude matrix |E1| of the measured electric field vector E1(r′) on plane S1 and the magnitude of the electric field vector E1(r′) is... n The expression is:

[0041]

[0042] This invention also discloses a phase-free planar near-field data reconstruction system based on convolution-deconvolution, comprising:

[0043] The data acquisition module is used to acquire the planar electric field amplitude of the antenna under test on planes S1 and S2, which is collected using the planar near-field testing system.

[0044] The initial complex field module is used to assign an initial phase to plane S1 and combine the initial phase with the plane electric field amplitude that has been collected on plane S1 to obtain the initial complex field on plane S1.

[0045] The convolution module is used to substitute the initial complex field on plane S1 into the spatial convolution function, thereby convolving to obtain a two-dimensional complex field on plane S2.

[0046] Replace the amplitude of the two-dimensional complex field on plane S2 obtained by convolution with the amplitude of the planar electric field on plane S2, and obtain the reconstructed complex field on plane S2.

[0047] The deconvolution module is used to substitute the complex field on plane S2 into the spatial deconvolution function and reconstruct the reconstructed complex field on plane S1 in reverse.

[0048] The cost module is used to denote the complex field on the plane S1 after spatial convolution-deconvolution as matrix U. n And calculate the corresponding cost function, where n represents the number of iterations to complete steps 3-4;

[0049] The update module is used to find the descending gradient of the cost function and in matrix U. n Calculate the updated U based on n+1 ;

[0050] The error calculation and judgment module is used to calculate U. n+1 The error between the known electric field amplitude on plane S1;

[0051] If the amplitude error is small and tends to stabilize, it indicates that the iterative process is gradually converging. Stop the iteration and reconstruct the U value. n+1 As the final value of the complex field on plane S1; otherwise, U n+1 Substitute it into the spatial convolution function as the initial complex field on the plane S1, and re-execute the iteration process until the error converges.

[0052] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the phase-free planar near-field data reconstruction method based on convolution-deconvolution.

[0053] The present invention also discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the phase-free planar near-field data reconstruction method based on convolution-deconvolution.

[0054] Compared with the prior art, the present invention has the following beneficial technical effects:

[0055] This invention proposes a phase-free planar near-field data reconstruction method based on convolution-deconvolution. This method utilizes spatial convolution-deconvolution to iteratively transform the electric fields on two near-field planes. Simultaneously, a cost function is constructed for the complex field on plane S1, and then minimized using the conjugate gradient (CG) method to obtain the optimal reconstruction result after each iteration. After multiple iterations, the reconstruction error of the complex field on plane S1 gradually converges, and the NFT far-field pattern derived from the reconstructed complex field on plane S1 shows good accuracy compared to the theoretical value.

[0056] This method effectively achieves the mutual conversion of complex fields between two near-field planes. Since convolution in the spatial domain corresponds to multiplication in the wavenumber domain, convolution-deconvolution can be implemented through element-wise multiplication and division of matrices, simplifying the overall computation. Furthermore, compared to the traditional first-type IFT method, this invention combines the spatial convolution-deconvolution method with a cost function and the CG method, thereby improving reconstruction accuracy and error convergence. Therefore, the resulting far-field radiation pattern of the AUT is closer to the true value. This method not only overcomes the phase loss limitation of near-field testing on phaseless planes but also reduces the testing cost of complex fields and effectively reconstructs the complex field on the target near-field plane and the corresponding far-field radiation pattern. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the planar near-field testing environment structure used in an embodiment of the present invention;

[0058] Figure 2 This is a schematic diagram of the dual-plane near-field scanning parameters and structural principle used in an embodiment of the present invention;

[0059] Among them, 101 is the probe; 102 is the robotic arm; 103 is the absorbing material; 104 is the antenna under test; 105 is plane S2; and 106 is plane S1.

[0060] Figure 3 This refers to the complex electric field information combined on plane S1 in this embodiment of the invention; wherein, Figure 3 a is the electric field amplitude collected on plane S1 in this embodiment of the invention; Figure 3b is the electric field amplitude of the given initial phase;

[0061] Figure 4 These are the reconstruction error curves corresponding to multiple iterations in the embodiments of the present invention;

[0062] Figure 5 It is the phase information on plane S1; where, Figure 5 a is the reconstructed phase on plane S1 in this embodiment of the invention. Figure 5 b is the theoretical phase on plane S1 according to the embodiment of the present invention;

[0063] Figure 6 These are the far-field radiation patterns corresponding to each method; among them, Figure 6 a is the calculated theoretical far-field radiation pattern; Figure 6 b is the far-field radiation pattern reconstructed using the reconstruction method of this invention; Figure 6 c is the reconstructed far-field pattern under the traditional IFT method;

[0064] Figure 7 This is a comparison of far-field radiation patterns under three methods; among them, Figure 7 a is a comparison of the far-field radiation pattern on the E-plane. Figure 7 b is a comparison of the far-field radiation pattern on the H plane;

[0065] Figure 8 This describes the overall system architecture of the present invention and the process correspondence between each module. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of the present invention clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention; that is, the described embodiments are only a part of the embodiments of the present invention, and not all of them.

[0067] The components described and illustrated in the accompanying drawings and embodiments of this invention can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the invention provided in the following drawings is not intended to limit the scope of the claimed invention, but merely to illustrate one selected embodiment of the invention. All other embodiments obtained by those skilled in the art based on the accompanying drawings and embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0068] It should be noted that the terms “comprising,” “including,” or any other variations are intended to cover non-exclusive inclusion, such that a process, element, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to the process, element, method, article, or apparatus.

[0069] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0070] The overall test environment of this invention embodiment is as follows: Figure 1 As shown, the robotic arm 102 is controlled to perform planar near-field testing of the antenna 104 under test by the probe 101. Since the method requires switching between two near-field planes, two near-field planes, S1106 and S2105, are selected, as follows: Figure 2 As shown in the diagram. The antenna under test (TUT) 104 operates at a frequency of 30 GHz with a gain of 18 dBi. The distance between plane S1106 and the aperture of TUT 104 is d1 = λ, and the distance between plane S2105 and the aperture of TUT 104 is d2 = 5λ. The sampling interval on both near-field planes is 0.47λ, and the scanning area of ​​both planes S1106 and S2105 is 16 cm × 16 cm. Therefore, the two near-field sampling planes S1106 and S2105 will yield a 35 × 35 sample matrix.

[0071] The following section explains the convolution-deconvolution method and the cost function, i.e., the CG method, and verifies the effectiveness of the reconstruction method.

[0072] According to the equivalence principle, if the radiation field in space is radiated by an ideal electrical conductor, then the forward spatial radiation field of the antenna 104 under test can be represented by the equivalent magnetic current, i.e.

[0073] J 1,m (r′)=-2n×E1(r′) (1)

[0074] Where E1(r′) is the electric field vector on the plane S1106, J 1,m (r′) is the equivalent magnetic current on plane S1106, r′=(x′,y′,z′) are the coordinates on plane S1106, and n is the normal vector perpendicular to plane S1. Correspondingly, the equivalent magnetic current J 1,m Vector potential A of (r′) 1,m It can be represented as:

[0075]

[0076] Where r = (x, y, z) represents the coordinates on plane S2105, G(rr′) is the scalar Green's function, and the wave number k satisfies k = 2πf / c. Subsequently, the electric field E2(r) on plane S2105 is calculated using the vector potential from equation (2).

[0077]

[0078] To further reveal the correspondence between the electric fields on the two near-field planes, the vector integral of equation (3) is decomposed into scalar integrals on the x and y axes, resulting in:

[0079]

[0080]

[0081] Among them, E 1,x′ and E 1,y′ It is the scalar component of E1(r′), E 2,x and E 2,y It is a scalar component of E2(r), and the Euclidean distance between plane S1106 and plane S2105 is R=((xx′)). 2 +(yy′) 2 +(zz′) 2 ) 1 / 2 Its spatial structure and parameter relationships are as follows: Figure 2 As shown.

[0082] By observing equations (4) and (5), it can be seen that the electric field E2(r) on plane S2105 can be obtained by convolving the electric field E1(r′) on plane S1106. Therefore, equations (4) and (5) can be simplified as follows:

[0083] F(x,y)=∫∫f(x′,y′)g(xx′,yy′)dx′dy′ (6)

[0084] Where f(x,y) represents the electric field component E on plane S1106. 1,x′ and E 1,y′ F(x,y) represents the electric field component E on plane S2105. 2,x and E 2,y g(x,y) represents the Green's function. Therefore, using equation (6), the electric field on plane S2105 can be derived from the electric field on plane S1106, and equation (6) is named spatial convolution, with the geometric correspondence as follows: Figure 2 As shown.

[0085] After obtaining the spatial convolution correspondences described above, it is necessary to find the inverse correspondences between the two planes, i.e., spatial deconvolution. Since convolution in the spatial domain can be represented as scalar multiplication in the wavenumber domain (i.e., multiplying corresponding elements of two matrices using Hadamard multiplication), deconvolution in the spatial domain can also be represented as division in the wavenumber domain (i.e., dividing corresponding elements of two matrices).

[0086]

[0087]

[0088] Among them, F and F -1 represents the forward Fourier transform and the inverse Fourier transform, respectively; ⊙ is the Hadamard multiplication operation; and % represents the element-wise division of the two matrices. Therefore, the convolution-deconvolution process between the electric field E1(r′) on plane S1106 and the electric field E2(r) on plane S2105 can be achieved by equations (7) and (8).

[0089] It is important to note that since this method employs a phase-free plane near-field test, the phase information of E1(r′) and E2(r) is unknown. Therefore, an assumed initial phase E needs to be given to E1(r′). 1,x′,pha and E 1,y′,pha The magnitude information of E1(r′) and E2(r) is known as |E 1,x′ |,|E 1,y′ |,|E 2,x |and|E 2,y Therefore, the initial phase information (E) on plane S1106 can be obtained. 1,x′,pha and E 1,y′,pha ,like Figure 3 (b) shown) and known amplitude information (|E) 1,x′ |and|E 1,y′ |, such as Figure 3 (a) is combined to obtain the initial complex field E on the plane S1106. 1,n (r′), n=1,2,...k, where n represents the number of iterations of the method.

[0090] At this time, E 1,n Substituting (r′) into equation (7), we can obtain the complex field E on the plane S2105. 2,n (r). E 2,n The magnitude of (r) is replaced with the known |E 2,x |and|E 2,y | can be used to obtain the reconstructed complex field E′ on plane S2105. 2,n (r). Then, E2′ ,n Substituting (r) into equation (8) yields the reconstructed complex field E′ on plane S1106. 1,n (r′). The complex field E′ 1,n (r′) is the reconstructed complex field after a complete convolution-deconvolution process.

[0091] At this point, the principle of convolution-deconvolution has been explained in detail. The cost function and CG method need to be elaborated below to form the complete reconstruction method.

[0092] Typically, repeatedly executing the above convolution-deconvolution process cannot guarantee accurate reconstruction of phase information on the plane S1106, and the number of convergence iterations is often large. Therefore, it is necessary to introduce a cost function and gradient descent to improve iterative convergence and increase reconstruction accuracy. Since the reconstructed phase information lies on the plane S1106, a cost function needs to be constructed for the plane S1106.

[0093] First, equations (4) and (5) can be simplified to

[0094] E 2,x =∫∫E 1,x′ G x-x′ ds′ (9)

[0095] E 2,y =∫∫E 1,y′ G y-y′ ds′ (10)

[0096] In this regard, assuming that the sample dimensions on plane S1106 and plane S2105 are both N×M, then equations (9) and (10) can be expressed as matrix multiplication:

[0097]

[0098] Among them, G x-x′ and G y-y′ It is a (2N×2M, 2N×2M) Green's function matrix, E 1,x′ and E 1,y′ E1(r′) is a complex field column vector (N×M,1) on the plane S1106. 2,x and E 2,y E2(r) is a complex field column vector (N×M,1) on the plane S2105, and equation (11) can be simplified to:

[0099] GU=E (12)

[0100] Assuming that the measured amplitudes of the known planar electric fields E1(r′) and E2(r) are rearranged to form matrices |E1] and |E2|, i.e. (N×M,1) column vectors, then equation (12) can be used to calculate the cost function of the plane S1106.

[0101]

[0102] in, It is the normalization coefficient. It is the 2-norm of the vector, U n This is the complex field reconstruction result on the plane S1106 after the nth iteration. To this end, the cost function of equation (13) can be minimized using the CG conjugate gradient method, and U... nUpdated to U n+1 U n+1 The expression is:

[0103] U n+1 =U n +V n =U n +g n +β n V n-1 (14)

[0104] Where, β n It is the weighting coefficient, V n The initial value is V0 = 0, V n-1 It is the gradient value from the previous iteration, g n It is C n (U n The gradient of ) is expressed as:

[0105]

[0106] Meanwhile, the Polak-Ribiere weighting coefficient β n Calculation as follows

[0107]

[0108] in, It is C n (U n The conjugate gradient of ) g n-1 C is the value at the (n-1)th iteration. n-1 (U n-1 The gradient value of ) It is the 2-norm of the vector.

[0109] At this point, the improved U is obtained through equations (14)-(16). n+1 And in order to verify U n+1 Whether the iteration stopping condition is met and the expected reconstruction result is achieved requires calculating U. n+1 The error between the known electric field amplitude |E1| on the plane S1106 and the known electric field amplitude |E1| is...

[0110]

[0111] U under the nth iteration n+1 The error ε between the measured amplitude |E1| and the actual amplitude n Curves Figure 4 As shown, if the error ε n If the value is small and tends to stabilize, it indicates that the reconstruction error is gradually converging and U... n+1 The value is close to the expected theoretical value, therefore the iterative process can be stopped, and U... n+1As a reconstructed complex field on plane S1106. Otherwise, it is necessary to continue repeating the convolution-deconvolution process of equations (7)-(8) and the cost function minimization process of equations (13)-(17) until the error ε n convergence.

[0112] Finally, after 200 iterations of the method described above, the following result was obtained: Figure 5 The reconstructed phase on plane S1106 shown in figure a, and the true phase on plane S1106 as shown in figure a. Figure 5 As shown in b, a consistency between the reconstruction result and the actual value is observed, which demonstrates the accuracy of the reconstruction result.

[0113] Accordingly, the far field of the NFT derived from the measured near field of the complex plane is as follows: Figure 6 As shown in figure a, the complex near-field U on the reconstructed plane S1106 n+1 The derived NFT far field is as follows Figure 6 As shown in b, to verify the effectiveness and advantages of the method, an NFT far field reconstructed using the traditional IFT method was added, such as... Figure 6 As shown in c. The tangential components of the three NFT far-field patterns on the E-plane and H-plane are respectively as follows. Figure 7 As shown in 7a and 7b.

[0114] Meanwhile, the relative errors on the E-plane and the H-plane are calculated as follows:

[0115]

[0116] Among them, E theo E represents the measured far-field tangential pattern of the NFT. recon This represents the far-field tangential pattern of the NFT reconstructed using the method described in this invention or the IFT method, where θ1 = -82° and θ2 = 82° are the confidence angles on the E-plane and H-plane. Calculations show that the relative error between the NFT far-field reconstructed by the method described in this invention and the theoretical far-field on the E-plane and H-plane is D. E,1 =3.64%, D H,1 =4.87%; while the relative error between the NFT far field reconstructed by the IFT method and the theoretical far field on the E-plane and H-plane is D. E,2 =7.22%, D H,2 =8.75%.

[0117] This demonstrates that the NFT far-field accuracy reconstructed by the proposed method is higher than that reconstructed by the traditional IFT method, thus illustrating the effectiveness and accuracy of the proposed method.

[0118] The method described in this invention utilizes spatial convolution-deconvolution to achieve biplane complex field transformation. Simultaneously, it reduces the complex field reconstruction error under multiple iterations through cost functions and conjugate gradients, thereby improving the iteration convergence speed and reconstruction accuracy. This method can reconstruct phase-free data in the millimeter-wave band, effectively recovering the phase information of near-field data while reducing testing overhead, and obtaining a relatively accurate far-field radiation pattern.

[0119] The process relationships between the functional modules corresponding to the method described in this invention and the overall system architecture are as follows: Figure 8 As shown, the present invention also discloses a phase-free planar near-field data reconstruction system based on convolution-deconvolution, comprising:

[0120] The data acquisition module is used to acquire the planar electric field amplitude of the antenna under test 104 on planes S1106 and S2105, which is collected by the planar near-field test system.

[0121] The initial complex field module is used to assign an initial phase to the plane S1106 and combine the initial phase with the plane electric field amplitude that has been collected on the plane S1106 to obtain the initial complex field on the plane S1106.

[0122] The convolution module is used to substitute the initial complex field on plane S1106 into the spatial convolution function, thereby convolving to obtain a two-dimensional complex field on plane S2105.

[0123] Subsequently, the amplitude of the planar electric field on the obtained plane S2105 is used to replace the amplitude of the two-dimensional complex field on the plane S2105 obtained by convolution, and the reconstructed complex field on the plane S2105 is obtained.

[0124] The deconvolution module is used to substitute the complex field on plane S2105 into the spatial deconvolution function and reconstruct the reconstructed complex field on plane S1106 in reverse.

[0125] The cost module is used to denote the complex field on the plane S1106 after spatial convolution-deconvolution as matrix U. n And calculate the corresponding cost function, where n represents the number of iterations to complete steps 3-4;

[0126] The update module is used to find the descending gradient of the cost function and in matrix U. n Calculate the updated U based on n+1 ;

[0127] The error calculation and judgment module is used to calculate U. n+1 The error between the known electric field amplitude on plane S1106;

[0128] If the amplitude error is small and tends to stabilize, it indicates that the iterative process is gradually converging. Stop the iteration and reconstruct the U value. n+1 As the final value of the complex field on plane S1106; otherwise, U n+1 Substitute it into the spatial convolution function as the initial complex field on the plane S1106, and re-execute the iteration process until the error converges.

[0129] This invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the phase-free planar near-field data reconstruction method based on convolution-deconvolution. The memory may include main memory, such as high-speed random access memory, or it may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry-standard architecture bus, a peripheral component interconnection standard bus, an extended industry-standard architecture bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory stores the program; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.

[0130] This invention also discloses a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the steps of the phase-free planar near-field data reconstruction method based on convolution-deconvolution. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A phase-free planar near-field data reconstruction method based on convolution-deconvolution, characterized in that, Includes the following steps: Step 1: Obtain the planar electric field amplitude of the antenna under test (104) on plane S1 (106) and plane S2 (105) using the planar near-field testing system; Step 2: Assign an initial phase to plane S1(106) and combine the initial phase with the amplitude of the plane electric field that has been collected on plane S1(106) to obtain the initial complex field on plane S1(106); Step 3: Substitute the initial complex field on plane S1 (106) into the spatial convolution function to obtain a two-dimensional complex field on plane S2 (105); Subsequently, the amplitude of the planar electric field on the plane S2(105) obtained in step 1 is used to replace the amplitude of the two-dimensional complex field on the plane S2(105) obtained by convolution, and the reconstructed complex field on the plane S2(105) is obtained. Step 4: Substitute the complex field on plane S2(105) into the spatial deconvolution function, and reconstruct the reconstructed complex field on plane S1(106) in reverse. At this point, the complex field on the plane S1(106) after the spatial convolution-deconvolution processing as described in steps 3-4 is denoted as matrix U. n And calculate the corresponding cost function, where n represents the number of iterations to complete steps 3-4; Step 5: Find the descending gradient of the cost function and apply it to matrix U. n Calculate the updated U based on n+1 ; Step 6: Calculate U n+1 The error between the known electric field amplitude on plane S1(106); If the amplitude error is small and tends to stabilize, it indicates that the iterative process of steps 2-5 is gradually converging. Stop the iteration and reconstruct the U value. n+1 As the final value of the complex field on plane S1(106); Otherwise, U n+1 Substitute the initial complex field from step 3 onto plane S1(106), and repeat the iterative process from step 3 to step 6 until the error converges.

2. The method for reconstructing phase-free planar near-field data based on convolution-deconvolution according to claim 1, characterized in that, In step 1, the coordinates on the near-field scanning plane S1 (106) and plane S2 (105) are represented as r′=(x′,y′,z′) and r=(x,y,z), respectively, and the measured electric field vectors corresponding to plane S1 (106) and plane S2 (105) are represented as E1(r′) and E2(r), respectively.

3. The method for reconstructing phase-free planar near-field data based on convolution-deconvolution according to claim 1, characterized in that, In step 2, the measurement amplitudes along the x and y directions on the scanning plane S1(106) are known and denoted as |E 1,x′ |and|E 1,y′ | where the phase is an unknown quantity, and an initial phase needs to be given, denoted as E. 1,x′,pha and E 1,y′,pha ; Combining the known amplitude with the given initial phase, we obtain the initial complex field E on the plane S1(106). 1,n (r′).

4. The phase-free planar near-field data reconstruction method based on convolution-deconvolution according to claim 1, characterized in that, In steps 3 and 4, the construction process of the spatial convolution function and the spatial deconvolution function is as follows: The electric field in a plane in free space can be represented in the form of an equivalent magnetic current: J 1,m (r′)=-2n×E1(r′) (1) Where E1(r′) is the electric field vector on the plane S1(106), J 1,m (r′) is the equivalent magnetic current, and its corresponding vector potential A 1,m for: Where G(rr′) represents the Green's function, the electric field E2(r) on the plane S2(105) can be expressed as: From equations (1)-(3), we obtain that E1(r′) and E2(r) satisfy a convolution relationship, which is: F(x,y)=∫∫f(x′,y′)g(xx′,yy′)dx′dy′ (4) Where f(x,y) represents the electric field component on plane S1(106), F(x,y) represents the electric field component on plane S2(105), and g(x,y) is the spatial Green's function; Therefore, the electric fields on planes S1(106) and S2(105) can achieve the mutual conversion between spatial convolution and deconvolution, resulting in: Equations (5) and (6) are defined as spatial convolution function and spatial deconvolution function, respectively.

5. The method for reconstructing phase-free planar near-field data based on convolution-deconvolution according to claim 2, characterized in that, In step 4, the complex field U that has undergone the above convolution-deconvolution process for the nth time is... n The corresponding cost function is expressed as: Where G is the Green's function matrix, and |E2| is the magnitude matrix of the measured electric field vector E2(r) on the plane S2(105). Describes the 2-norm of a vector. It is the normalization coefficient.

6. The method for reconstructing phase-free planar near-field data based on convolution-deconvolution according to claim 1, characterized in that, In step 5, the cost function is minimized using the cost function and the conjugate gradient of CG, and U is... n Updated to U n+1 The iteratively updated complex field matrix U n+1 The expression is: U n+1 =U n +V n =U n +g n +β n V n-1 (8) Where g n It is C n (U n The gradient of β n It is the weighting coefficient, V n-1 It is the gradient value from the previous iteration, with an initial value of V0 = 0.

7. The method for reconstructing phase-free planar near-field data based on convolution-deconvolution according to claim 2, characterized in that, In step 6, after obtaining U n+1 Then, calculate U n+1 The magnitude error ε between the magnitude matrix |E1| of the measured electric field vector E1(r′) on plane S1(106) and the magnitude of the electric field vector E1(r′) is... n The expression is:

8. A phase-free planar near-field data reconstruction system based on convolution-deconvolution, characterized in that, include: The data acquisition module is used to acquire the planar electric field amplitude of the antenna under test (104) on plane S1 (106) and plane S2 (105) using the planar near-field test system; The initial complex field module is used to assign an initial phase to the plane S1(106) and combine the initial phase with the plane electric field amplitude that has been collected on the plane S1(106) to obtain the initial complex field on the plane S1(106); The convolution module is used to substitute the initial complex field on plane S1 (106) into the spatial convolution function, thereby convolving to obtain a two-dimensional complex field on plane S2 (105); Replace the amplitude of the two-dimensional complex field on the plane S2(105) obtained by convolution with the amplitude of the planar electric field on the obtained plane S2(105), and obtain the reconstructed complex field on the plane S2(105); The deconvolution module is used to substitute the complex field on plane S2(105) into the spatial deconvolution function and reconstruct the reconstructed complex field on plane S1(106) in reverse. The cost module is used to denote the complex field on the plane S1(106) after spatial convolution-deconvolution as matrix U. n And calculate the corresponding cost function, where n represents the number of iterations to complete steps 3-4; The update module is used to find the descending gradient of the cost function and in matrix U. n Calculate the updated U based on n+1 ; The error calculation and judgment module is used to calculate U. n+1 The error between the known electric field amplitude on plane S1(106); If the amplitude error is small and tends to stabilize, it indicates that the iterative process is gradually converging. Stop the iteration and reconstruct the U. n+1 As the final value of the complex field on plane S1(106); Otherwise, U n+1 Substitute it into the spatial convolution function as the initial complex field on the plane S1(106), and re-execute the iteration process until the error converges.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the phase-free planar near-field data reconstruction method based on convolution-deconvolution as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the phase-free planar near-field data reconstruction method based on convolution-deconvolution as described in any one of claims 1 to 7.

Citation Information

Patent Citations

  • Large-view-field holographic projection method and system based on deep learning accelerated calculation

    CN114002931A

  • Antenna aperture diffraction field reconstruction method based on similarity principle

    CN115238227A