Rapid solving method for large finite periodic antenna array based on regional decomposition combined element pole

Abstract

The invention discloses a method for rapidly solving a large finite periodic antenna array based on a region decomposition synthetic pole, and relates to the technical field of array antennas. The method comprises the following steps: partitioning a to-be-solved large finite period antenna array target and extracting representative antenna units; performing mesh generation on the representative antenna unit; periodic parameters are set, and a solution grid of the whole large finite periodic structure is implicitly rebuilt by performing copy translation operation on a representative antenna unit grid; establishing a sub-region internal matrix and an inter-sub-region coupling matrix of the region decomposition synpole system; constructing a hybrid global-local preprocessor on the basis of the synpole system matrix; and iteratively solving the hybrid global-local preprocessor to obtain an unknown coefficient to be solved, and obtaining near-field and far-field electromagnetic characteristics. According to the method, the large finite period antenna array can be solved quickly and efficiently, the modeling subdivision time is greatly shortened, and the memory and calculation time consumption are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of array antenna, and particularly relates to a large finite periodic antenna array fast solving method based on region decomposition combined element pole. BACKGROUND

[0002] In recent years, with the rapid development of wireless communication technology, the performance requirements of antenna systems are also getting higher and higher. Ultra-large scale array antenna has become one of the core technologies in the fields of 5G / 6G mobile communication, satellite communication, radar detection, etc. due to its high gain, high resolution, flexible beamforming and other advantages. However, the electromagnetic simulation of ultra-large scale array antenna faces huge computing challenges. Traditional numerical methods, such as method of moments (MoM) and finite element method (FEM), often consume a large amount of computing resources and storage space when dealing with such problems, which is difficult to meet the actual engineering needs.

[0003] There are two main difficulties in the electromagnetic simulation of large-scale finite periodic array. One is that large-scale finite periodic array usually includes several thousand or even tens of thousands of basic units, and its modeling usually consumes huge hardware resources and is very slow. The other is that in the process of practical application, the large-scale finite periodic array has a huge number of unknowns, which requires a huge memory space and high hardware requirements for the computing platform, making it difficult to solve the electromagnetic characteristics.

[0004] Region decomposition technology can quickly solve the electromagnetic characteristics of large targets by decomposing complex problems into multiple sub-problems and solving them in parallel. Specifically for the electromagnetic simulation of large-scale finite periodic structure, region decomposition technology divides the entire calculation region into several non-overlapping sub-regions and independently solves the electromagnetic field equation in each sub-region. The information exchange between sub-regions is realized through boundary conditions or iterative algorithms, and the global solution is finally obtained. Although the region decomposition finite element method can also solve the finite periodic array, it needs to add an air box outside the model to set the truncation boundary, which lacks flexibility and precision. The region decomposition combined element pole method uses boundary integral method to avoid the problem of setting truncation boundary, so it does not need to add truncation boundary such as absorbing boundary, perfect matched layer, etc. outside the whole finite periodic structure. Region decomposition combined element pole is often used to solve the electromagnetic characteristics of large complex targets.

[0005] In the related art, a time-domain discontinuous Galerkin electromagnetic simulation method and system based on Robin transmission conditions are disclosed in a patent document (application number CN202510062925.7, publication number CN119849201A). The method mainly includes the following steps: geometric modeling of the simulation model, setting boundary conditions and excitation sources, obtaining the to-be-simulated geometric model; tetrahedral mesh partitioning of the to-be-simulated geometric model, obtaining the partitioned model; region division of the partitioned model, obtaining a plurality of sub-regions; selecting a time step and determining the sampling point position in the sub-region; calculating the time-domain results of all time steps by using the time-domain discontinuous Galerkin electromagnetic field simulation method, and completing the electromagnetic simulation. However, the invention still needs to perform overall geometric modeling and partitioning on the target, and a large amount of resources is required for modeling and partitioning related operations when solving large finite periodic structures, and the solution cannot be quickly and efficiently obtained.

[0006] In addition, a non-conformal mesh electromagnetic periodic structure solving method based on finite elements is disclosed in a patent document (application number CN202410921877.8, publication number CN118468676A). The method mainly includes the following steps: constructing a periodic unit model; mesh partitioning the periodic unit model into tetrahedral units; setting unknown coefficients to be solved for the periodic unit model; establishing a finite element system self-region matrix, a finite element system coupling matrix, and an excitation matrix, constructing a matrix equation about the unknown coefficients to be solved, and solving to obtain the known coefficients corresponding to the unknown coefficients to be solved, calculating the total electric field of the periodic unit model, extracting the tangential total electric field at the first excitation port and the tangential total electric field at the second excitation port; obtaining the reflection coefficient of the first excitation port according to the incident electric field and the tangential total electric field at the first excitation port; obtaining the transmission coefficient of the first excitation port and the second excitation port according to the incident electric field and the tangential total electric field at the second excitation port. Thus, the calculation efficiency of the transmission characteristic coefficient of the non-conformal mesh electromagnetic periodic structure is improved. However, the invention uses periodic boundary conditions to simulate an infinite electromagnetic periodic structure, and cannot quickly and efficiently solve finite periodic structures.

[0007] The above-mentioned "a time-domain discontinuous Galerkin electromagnetic simulation method and system based on Robin transmission conditions" needs to perform overall geometric modeling and partitioning on the target, and a large amount of hardware and time resources are required for modeling and partitioning related operations when solving large finite periodic structures, and the solution cannot be quickly and efficiently obtained. Although the used region decomposition finite element method can also solve finite periodic arrays, an air box needs to be added outside the model to set a truncation boundary, and the flexibility and precision are insufficient.

[0008] The above-mentioned existing "a method for solving electromagnetic periodic structures based on non-conformal grids of finite elements" uses periodic boundary conditions to simulate infinite electromagnetic periodic structures. Although it avoids the overall modeling of the target, it cannot quickly solve finite periodic structures. Summary of the Invention

[0009] The technical problem to be solved by the present invention is how to provide a method for quickly and efficiently solving large finite period antenna arrays, greatly reducing the time for modeling and segmentation, and reducing memory and computing time consumption.

[0010] To solve the above technical problems, the technical solution adopted by the present invention is: a fast solution method for a large finite period antenna array based on regional decomposition and combined poles, comprising the following steps:

[0011] To solve the target partition of large finite period antenna array and extract representative antenna elements;

[0012] The representative antenna unit is meshed. According to the requirements of the regional decomposition combined element method, the internal finite element area is meshed with a tetrahedral mesh, and the external boundary integral equation area is meshed with a triangular mesh.

[0013] Set periodic parameters and implicitly rebuild the solution grid of the entire large finite periodic structure by copying and translating the representative antenna unit grid;

[0014] Establish the sub-region internal matrix and inter-sub-region coupling matrix of the regional decomposition meta-polar system, then add all sub-region matrices to construct the meta-polar system matrix for the unknown number to be solved;

[0015] A hybrid global-local preconditioner is constructed based on the matrix of the combined element system. The integral equation part uses the global preconditioner matrix of near interaction, and the finite element part reuses the local preconditioner matrix.

[0016] The hybrid global-local preconditioner is solved iteratively to obtain the unknown coefficients to be solved, and the near-field and far-field electromagnetic characteristics of the large finite period antenna array target to be solved are obtained.

[0017] A further technical solution is that the method for solving the target partition of a large finite period antenna array and extracting representative antenna elements includes the following steps:

[0018] A finite periodic antenna array consisting of closely arranged identical periodic unit models is divided into a plurality of identical sub-regions according to design requirements. Each sub-region includes an antenna unit, and an antenna unit in one of the sub-regions is selected as a representative antenna unit.

[0019] Preferably, the periodic parameters include periodic number, periodic distance, translation and rotation parameters.

[0020] A further technical solution is that a method for obtaining near-field and far-field electromagnetic characteristics of a target to be solved for a large finite period antenna array includes the following steps:

[0021] The multi-wave frontier solver MUMP with multi-threaded block low-rank decomposition is used to solve the preconditioning matrix given in the preconditioning matrix. Finally, by iteratively solving the hybrid global-local preconditioner, the electromagnetic field in each sub-region can be obtained, and the near-field and far-field electromagnetic characteristics can be obtained.

[0022] The beneficial effects of adopting the above technical solution are as follows: the method described in this application partitions large finite periodic array antennas and extracts representative antenna elements. Only the representative antenna elements need to be meshed, implicitly reconstructing the mesh of the entire finite periodic array antenna, avoiding the hardware resource and time consumption associated with modeling and meshing the entire array. During the solution process, a hybrid global-local preprocessor is constructed, leveraging periodic characteristics to reuse the preprocessing matrix, reducing computational memory and time consumption, and enabling fast and efficient solution of finite periodic arrays. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] Figure 1 is a flow chart of the method according to an embodiment of the present invention;

[0025] Figure 2 is a schematic diagram of a 10x10 patch antenna array and representative antenna units in the method according to an embodiment of the present invention;

[0026] Figure 3 Schematic diagram of representative antenna unit dimensions in the method according to an embodiment of the present invention;

[0027] Figure 4 is a schematic diagram of a representative antenna unit grid decomposition in the method described in an embodiment of the present invention;

[0028] Figure 5 Schematic diagram of periodic grid generation in the method according to an embodiment of the present invention;

[0029] Figure 6 It is a curve diagram of the simulation results of the embodiment pattern in the method described in the embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following is a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts are within the scope of protection of the present invention.

[0031] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0032] like Figure 1 As shown, the present invention discloses a method for quickly solving a large finite period antenna array based on regional decomposition and combined elements, the method comprising the following steps:

[0033] Step 1: Partition the large finite periodic antenna array to be solved and extract representative antenna elements. Specifically, the finite periodic antenna array, consisting of closely spaced antenna element models with identical periods, is divided into multiple identical sub-regions according to the design requirements. Each sub-region contains one antenna element, and an antenna element in one of the sub-regions is selected as the representative antenna element.

[0034] Step 2: Mesh the representative antenna unit. According to the requirements of the regional decomposition and combined polar method, the inner finite element area is meshed with tetrahedral meshes, and the outer boundary integral equation area is meshed with triangular meshes.

[0035] Step 3: Set the periodic parameters, including the number of periodicities, periodic distance, translation, and rotation parameters. By copying and translating the representative antenna unit grid, the solution grid of the entire large finite periodic structure is implicitly rebuilt. At this time, the grids between different sub-regions are non-conformal.

[0036] Step 4: Establish the sub-region internal matrix and inter-sub-region coupling matrix of the regional decomposition meta-polar system, then add all sub-region matrices to construct the meta-polar system matrix for the unknown number to be solved.

[0037] Specifically, for an antenna array using coaxial line excitation, the boundary conditions satisfied by its ports are:

[0038]

[0039]

[0040]

[0041] The sub-region matrix of the domain decomposition composite polar system includes the internal finite element and boundary integral equation parts. The electromagnetic field of the internal finite element part of the m-th sub-region satisfies the variational equation:

[0042]

[0043] Where Ω represents the internal non-uniform medium target area, S represents the outer surface of the entire calculation target, To calculate the normal of the target outer surface, the last term represents the contribution of the coaxial line excitation to the matrix system. Discretizing the above equation using the finite element edge element basis function, we can obtain the equation:

[0044]

[0045]

[0046]

[0047]

[0048]

[0049] Finite element sub-region interface Γ m,n On the other hand, the full second-order Robin transport condition is used to connect them to improve the convergence of the domain decomposition and meta-polarization program. The equations satisfied are as follows:

[0050]

[0051] in and are the surface current and surface magnetic current on the interface, ρ Γ A set of auxiliary scalars introduced for the convenience of writing, defined as:

[0052]

[0053] β and γ are two crucial variable parameters. The convergence of the domain decomposition combined-element method is related to them. Numerical experiments show that their values ​​are determined as follows:

[0054]

[0055]

[0056] The electromagnetic field S on the outer boundary of any subregion satisfies the mixed field integral equation:

[0057]

[0058]

[0059]

[0060] Using the RWG basis function as the test function to discretize the boundary integral equation (14) through the Galerkin matching method, we can obtain:

[0061]

[0062] Here we define the inner product factor:

[0063] <a,b> V =∫ V (a·b T )dV (18)

[0064] <a,b> Γ =∫ Γ (a·b T )dS (19)

[0065] In this embodiment, the matrix in formula (17) can be expressed as:

[0066]

[0067]

[0068]

[0069]

[0070] The electromagnetic fields at the interface between the internal finite element part and the external integral equation part in each subregion are coupled using the first-order Robin transmission condition:

[0071]

[0072] At this time, the matrix equation in each sub-region can be expressed as:

[0073]

[0074] In the matrix equation, I represents the internal unknown quantity, and Γ represents the unknown quantity on the interface of the internal finite element part. The contribution of antenna excitation to the matrix is ​​represented by X m Marked, the wave port excitation only exists within the subregion or at the interface between the interior and the exterior surface. There are no excitation-related unknowns at the interface between different subregions. [K], [B], [D], [F], and [G] are sparse finite element matrices. Although [X] is a full matrix, its dimension is much smaller than that of the [K] matrix, so [K+X] can still be treated as a sparse matrix. [P] and [Q] are dense integral equation matrices.

[0075] By superimposing all the sub-region matrices (25) and rearranging the unknowns, we can obtain the final domain decomposition combined polar equation matrix:

[0076]

[0077] Among them A N and C mn (m, n = 1, 2, ..., N) represent the matrix of each finite element sub-region and the coupling matrix between adjacent finite element sub-regions respectively.

[0078] Step 5: Construct a hybrid global-local preconditioner. The integral equation part uses a nearly interactive global preconditioning matrix, and the finite element part reduces memory and time usage by reusing the local preconditioning matrix.

[0079] As the size of the computational target increases, the tetrahedral mesh of the finite element part increases rapidly, which leads to slow or even non-convergence of the matrix iterative solution. As mentioned above, although [X] in Equation (25) is a full matrix, its dimension is much smaller than the [K] matrix. Therefore, [K+X] can still be treated as a sparse matrix. In order to improve the convergence of the proposed method and enhance the computational efficiency, the matrix on the diagonal is extracted and the Schwarz type preconditioning matrix is ​​further constructed. Its form is as follows:

[0080]

[0081] Among them, A -1 is the sub-region local preconditioning matrix based on the finite element-absorbing boundary (FEM-ABC), is a global preconditioning matrix of near interactions based on sparse approximate inverse (SAI).

[0082]

[0083]

[0084]

[0085] From the finite period structure rapid modeling method in step 3, we know that the grid in each sub-region is the same. The sub-regions are divided into three categories according to their positions, which are recorded as internal regions, side regions, and corner regions. For the convenience of writing, the inverse of the matrix of different categories of regions is recorded as For the internal region, the interfaces between different sub-regions are exactly the same, so the precondition matrix (27) The same, combined with the hierarchical parallel region decomposition technology, after the parallel load is divided, the minimum set of local unit types can be established in each process. At this time, the required memory space can be greatly reduced by utilizing the local preprocessor reuse technology. At this time, it is only necessary to solve Once, the preconditioning matrix can be obtained. For the edge region, the interface of the sub-regions on different sides is different, but the interface of the regions on the same side is the same. The inverse A of the matrix of the sub-regions on the same side can be obtained. e -1 For the corner area, the interface between different areas is completely different and needs to be calculated separately. Then the preconditioning matrix can be expressed as:

[0086]

[0087] Since the finite element matrix A can be considered as a sparse matrix as a whole, it is very convenient to construct the preconditioning matrix of formula (31), and the matrix A can be directly solved -1 The inverse of is sufficient. Using the preconditioning matrix constructed by equation (31) to perform right preconditioning on equation (26), we can obtain a hybrid global-local preconditioner:

[0088]

[0089] Step 6: Iteratively solve the matrix equation to obtain the unknown coefficients to be solved, thereby solving the near-field and far-field electromagnetic characteristics.

[0090] The multi-threaded version of the block-based low-rank decomposition multi-wave frontier solver MUMP is used to solve the preconditioning matrix given in Equation (31). Finally, by iteratively solving Equation (32), the electromagnetic field in each sub-region can be obtained, and the near-field and far-field electromagnetic characteristics of the antenna array can be obtained.

[0091] Example 2

[0092] In an optional embodiment of the present invention, the effect of the fast solution method of the large finite period antenna array based on regional decomposition and combined poles provided above is verified through simulation experiments.

[0093] 1. Simulation conditions:

[0094] The hardware platform for the simulation experiment in this embodiment is: Liuhui III computer. This supercomputer server has 8 Intel Xeon Platinum 8276 CPUs 2.20GHz, each CPU has 28 cores, and the total memory size is 1TByte.

[0095] 2. Simulation content and result analysis:

[0096] Step 1: Partition the large finite periodic array target and extract representative antenna elements.

[0097] right Figure 2 The 10x10 patch antenna array shown is first divided into 100 identical sub-areas, each area contains 1 antenna element, and one of the antenna elements is selected as the representative antenna element.

[0098] The dimensions of a typical antenna unit are 60x60x0.8mm, and the relative dielectric constant of the medium is 2.44. Figure 3 .

[0099] Step 2: Mesh the representative antenna elements.

[0100] According to the requirements of the regional decomposition combined element method, the internal finite element area is meshed with tetrahedral meshes, and the boundary integral equation area is meshed with triangular meshes, such as Figure 4 shown.

[0101] Step 3: Set the periodic parameters and implicitly reconstruct the solution grid of the entire large finite periodic structure using the representative antenna unit grid.

[0102] In this embodiment, the periodic number in both the x and y directions is set to 10, and the periodic distance in both the x and y directions is set to 60 mm. In this way, the antenna units are arranged compactly without overlapping each other. By copying and translating the representative antenna unit grid, the entire 10x10 patch antenna array grid is reconstructed, as shown in FIG. Figure 5 shown.

[0103] Step 4: Construct the matrix of the combined polar system of the unknowns to be solved.

[0104] The sub-region matrix of the domain decomposition combined element system contains the internal finite element and boundary integral equation parts. The electromagnetic field of the internal finite element part of the m-th sub-region satisfies the variational equation:

[0105]

[0106] Where Ω represents the internal non-uniform medium target area, S represents the outer surface of the entire calculation target, To calculate the normal of the target outer surface, the last term represents the contribution of the coaxial line excitation to the matrix system. Discretizing the above equation using the edge element basis function, we can obtain the equation:

[0107]

[0108] Finite element sub-region interface Γ m,nOn the other hand, the full second-order Robin transport condition is used to connect them to improve the convergence of the domain decomposition and meta-polarization program. The equations satisfied are as follows:

[0109]

[0110] and are the surface current and surface magnetic current on the interface, ρ Γ A set of auxiliary scalars introduced for the convenience of writing, defined as:

[0111]

[0112] β and γ are two crucial variable parameters. The convergence of the domain decomposition combined method is related to them. Numerical experiments show that their values ​​are determined as follows:

[0113]

[0114]

[0115] The electromagnetic field S on the outer boundary of any subregion satisfies the mixed field integral equation:

[0116]

[0117] Using the RWG basis function as the test function to discretize the boundary integral equation through the Galerkin matching method, we can obtain:

[0118]

[0119] The electromagnetic fields at the interface between the internal finite element part and the external integral equation part in each subregion are coupled using the first-order Robin transmission condition:

[0120]

[0121] At this time, the matrix equation in each sub-region can be expressed as:

[0122]

[0123] In the matrix equation, I represents the internal unknown quantity, and Γ represents the unknown quantity on the interface of the internal finite element part. The contribution of antenna excitation to the matrix is ​​represented by X mMarked out, the wave port excitation only exists within the subregion or at the interface between the interior and the exterior surface. There are no excitation-related unknowns at the interface between different subregions. [K], [B], [D], [F], and [G] are sparse finite element matrices. Although [X] is a full matrix, its dimension is much smaller than that of the [K] matrix, so [K+X] can still be treated as a sparse matrix. [P] and [Q] are dense integral equation matrices.

[0124] By superimposing all the sub-region matrices and rearranging the unknowns, we can obtain the final regional decomposition combined element system matrix:

[0125]

[0126] Among them A N and C mn (m, n = 1, 2, ..., N) represent the matrix of each finite element sub-region and the coupling matrix between adjacent finite element sub-regions respectively.

[0127] Step 5: Construct a hybrid global-local preconditioner. The integral equation part uses a nearly interactive global preconditioning matrix, and the finite element part reduces memory and time usage by reusing the preconditioning matrix.

[0128] As the computational target size increases, the number of tetrahedral meshes in the finite element part increases rapidly, which results in slow or even non-convergence of the matrix iterative solution. As mentioned earlier, although [X] is a full matrix, its dimension is much smaller than the [K] matrix. Therefore, [K+X] can still be treated as a sparse matrix. In order to improve the convergence of the proposed method and enhance computational efficiency, the matrix on the diagonal is extracted and the Schwarz-type preconditioning matrix is ​​further constructed. Its form is as follows:

[0129]

[0130] Among them, A -1 is the sub-region local preconditioning matrix based on the finite element-absorbing boundary (FEM-ABC), is a global preconditioning matrix of near interactions based on sparse approximate inverse (SAI).

[0131]

[0132]

[0133]

[0134] From the finite period structure rapid modeling method in step 3, we know that the grid in each sub-region is the same. The sub-regions are divided into three categories according to their positions, which are recorded as internal regions, side regions, and corner regions. For the convenience of writing, the inverse of the matrix of different categories of regions is recorded as For the internal area, the interfaces between different sub-areas are exactly the same, so the precondition matrix The same, combined with the hierarchical parallel region decomposition technology, after the parallel load is divided, the minimum set of local unit types can be established in each process. At this time, the required memory space can be greatly reduced by utilizing the local preprocessor reuse technology. At this time, it is only necessary to solve Once, the preconditioning matrix can be obtained. For edge regions, the interfaces of sub-regions on different edges are different, but the interfaces of regions on the same edge are the same. The inverse of the matrix of the region on the same edge can be obtained. For the corner area, the interface between different areas is completely different and needs to be calculated separately. Then the preconditioning matrix can be expressed as:

[0135]

[0136] Since the finite element matrix A can be regarded as a sparse matrix as a whole, it is very convenient to construct a preconditioning matrix and directly solve the matrix A -1 The inverse of is sufficient. By using the constructed preconditioning matrix and the meta-polar system matrix to perform right preconditioning, we can obtain a hybrid global-local preconditioner:

[0137]

[0138] Step 6: Iteratively solve the matrix equation to obtain the unknown coefficients to be solved, thereby solving the near-field and far-field electromagnetic characteristics.

[0139] The multi-threaded version of the block-based low-rank decomposition multi-wave frontier solver MUMP is used to solve the preconditioning matrix. Finally, by iteratively solving the regional meta-polar system equation, the electromagnetic field in each sub-region can be obtained, and finally the radiation characteristics of the antenna array can be obtained, such as Figure 6 shown.

[0140] Analysis of experimental results:

[0141] The directional pattern calculation results of the present invention are as follows Figure 6 As shown, the calculation frequency is set to 4.0GHz. Figure 6 It can be seen that the result curve of the method of the present invention at the sampling point is very consistent with the result curve of the FEKO software, indicating that the accuracy of the method proposed in this application is good enough.

[0142] Table 1 - Performance comparison table

[0143]

[0144] Table 1 shows that the proposed method has a significant computational efficiency advantage over FEKO. For a comparable number of unknowns, the proposed method requires 48.1 GB of peak memory, while FEKO requires 72.4 GB. The proposed method takes 3 minutes and 54 seconds to compute, while FEKO takes 19 minutes and 32 seconds. The proposed preprocessor reuse technique reduces the computational time by 28.7 GB of memory and 1 minute and 26 seconds.

[0145] In summary, the method described in this application is based on numerical calculations. The examples demonstrate the directional pattern calculation results of this method and FEKO software, demonstrating the accuracy and adaptability of the method of the present invention. In summary, the above is only one embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A fast solution method for large finite period antenna arrays based on regional decomposition and combined elements, characterized by The steps include: To solve the target partition of large finite period antenna array and extract representative antenna elements; The representative antenna unit is meshed. According to the requirements of the regional decomposition combined element method, the internal finite element area is meshed with a tetrahedral mesh, and the external boundary integral equation area is meshed with a triangular mesh. Set periodic parameters and implicitly rebuild the solution grid of the entire large finite periodic structure by copying and translating the representative antenna unit grid; Establish the sub-region internal matrix and inter-sub-region coupling matrix of the regional decomposition meta-polar system, then add all sub-region matrices to construct the meta-polar system matrix for the unknown number to be solved; A hybrid global-local preconditioner is constructed based on the matrix of the combined element system. The integral equation part uses the global preconditioner matrix of near interaction, and the finite element part reuses the local preconditioner matrix. The hybrid global-local preconditioner is solved iteratively to obtain the unknown coefficients to be solved, and the near-field and far-field electromagnetic characteristics of the large finite period antenna array target to be solved are obtained.

2. The fast solution method for large finite period antenna array based on regional decomposition and combined poles according to claim 1 is characterized in that: The method for partitioning a large finite period antenna array target and extracting representative antenna elements includes the following steps: A finite periodic antenna array consisting of closely arranged identical periodic unit models is divided into a plurality of identical sub-regions according to design requirements. Each sub-region includes an antenna unit, and an antenna unit in one of the sub-regions is selected as a representative antenna unit.

3. The method for rapidly solving a large finite period antenna array based on regional decomposition and combined poles according to claim 1, characterized in that: The periodic parameters include periodic number, periodic distance, translation and rotation parameters.

4. The fast solution method for a large finite period antenna array based on regional decomposition and combined poles according to claim 1 is characterized in that: The method for constructing the Heyuanji system matrix includes the following steps: For an antenna array using coaxial line excitation, the boundary conditions satisfied by its ports are: The sub-region matrix of the regional decomposition combined element system includes the internal finite element and boundary integral equation parts. The electromagnetic field of the internal finite element part of the m-th sub-region satisfies the variational equation: Where Ω represents the internal non-uniform medium target area, S represents the outer surface of the entire calculation target, To calculate the normal of the target outer surface, the last term represents the contribution of the coaxial line excitation to the matrix system. Discretizing the above equation using the finite element edge element basis function, the equation is obtained: Finite element sub-region interface Γ m,n On the other hand, the full second-order Robin transmission condition is used for connection, and the equations satisfied are as follows: in, and are the surface current and surface magnetic current on the interface, ρ Γ A set of auxiliary scalars are introduced, defined as: β and γ are two variable parameters, and the value rules are as follows: The electromagnetic field S on the outer boundary of any subregion satisfies the mixed field integral equation: Using the RWG basis function as the test function, the boundary integral equation (14) is discretized by the Galerkin matching method, and the following is obtained: Define the inner product factor: a,b> V =∫ V (a·b T )dV(18) <a,b> Γ =∫ Γ (a·b T )dS(19) The matrix in equation (17) is expressed as: The electromagnetic fields at the interface between the internal finite element part and the external integral equation part in each subregion are coupled using the first-order Robin transmission condition: At this time, the internal matrix equation of each sub-region is expressed as: In the matrix equation, I represents the internal unknown quantity, Γ represents the unknown quantity on the interface of the internal finite element part; X m is the contribution of antenna excitation to the matrix; [K], [B], [D], [F], and [G] are sparse finite element matrices, [X] is a full matrix whose dimension is much smaller than the [K] matrix, and [K+X] is treated as a sparse matrix. [P] and [Q] are dense integral equation matrices; All sub-region internal matrices (25) are superimposed together and the unknowns are rearranged to obtain the final regional decomposition combined element system matrix: Among them A N and C mn (m, n = 1, 2, ..., N) represent the matrix of each finite element sub-region and the coupling matrix between adjacent finite element sub-regions respectively.

5. The fast solution method for a large finite period antenna array based on regional decomposition and combined poles according to claim 4 is characterized in that: The method for constructing a hybrid global-local preprocessor comprises the following steps: The matrix on the diagonal of the combined-element system matrix is ​​extracted to construct a Schwarz-type preconditioning matrix, which has the following form: Among them, A -1 is the sub-region local preconditioning matrix based on the finite element-absorbing boundary, is a near-interaction global preconditioning matrix based on sparse approximate inverse; The grid in each sub-region is the same. The sub-regions are divided into three categories according to their positions, namely internal regions, side regions, and corner regions. The inverse of the matrix of different categories of regions is recorded as For the internal area, the interfaces between different sub-areas are exactly the same, so the precondition matrix formula (27) The same, combined with the hierarchical parallel region decomposition technology, after the parallel load is divided, the minimum set of local unit types is established in each process, and the local preprocessor reuse technology is used to reduce the required memory space. At this time, it is only necessary to solve Once, the preconditioning matrix can be obtained; for the edge region, the interface of the sub-regions on different sides is different, but the interface of the region on the same side is the same, and the inverse of the matrix of the sub-regions on the same side is For reuse; for corner areas, the interfaces of different areas are completely different and need to be calculated separately; then the preconditioning matrix is ​​expressed as: Using the preconditioning matrix constructed by equation (31), equation (26) is right-preconditioned to obtain a hybrid global-local preconditioner:

6. The fast solution method for a large finite period antenna array based on regional decomposition and combined poles according to claim 5 is characterized in that: The method for obtaining the near-field and far-field electromagnetic characteristics of a large finite period antenna array target to be solved includes the following steps: The multi-wave frontier solver MUMP with multi-threaded block low-rank decomposition is used to solve the preconditioning matrix given in the preconditioning matrix. Finally, by iteratively solving the hybrid global-local preconditioner, the electromagnetic field in each sub-region can be obtained, and the near-field and far-field electromagnetic characteristics can be obtained.

Images