Electromagnetic field quantity extrapolation method and device based on finite difference and neural network fusion

Abstract

The invention relates to an electromagnetic field quantity extrapolation method and device based on finite difference and neural network fusion. The method comprises the steps that a multi-layer concentric spherical surface with the antenna phase center as the sphere center is constructed; dividing each layer of concentric spherical surface in the first direction and the second direction to obtain a plurality of neurons; according to the voltage value of each cell in the first direction and the voltage value of each cell in the second direction, carrying out iterative calculation on the voltage value of each cell between the concentric spherical surfaces of the innermost layer and the outermost layer in the first direction or the second direction, and obtaining an iterative calculation result of each cell in the first direction or an iterative calculation result of each cell in the second direction, calculating a voltage value of each cell in the outermost concentric spherical surface, and when the outermost concentric spherical surface meets a preset convergence condition, determining that the current electromagnetic field quantity extrapolation is completed. Therefore, the problems of insufficient computer memory and incapability of directly performing far-field calculation caused by the background technology are solved, and the calculation efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a method and device for extrapolating electromagnetic field quantities based on the fusion of finite differences and neural networks. Background Art

[0002] Finite difference methods are numerical computational techniques that replace differentials with differences and derivatives with difference quotients. They are widely used in mathematics, physics, and modern informatics. The application of finite difference methods in electromagnetics has made significant progress, with numerous commercial software and practical applications verifying their correctness and reliability.

[0003] However, the finite difference method also faces many challenges in its application. For example, in principle, the finite difference method can only solve problems within a limited area. In the numerical calculation of antennas, the difference method is usually only used to study the near field of the antenna (that is, the area close to the antenna and of limited size). Once the far field needs to be calculated, the amount of calculation will increase sharply with the increase in distance, resulting in insufficient computer memory, making it difficult to directly use it for far-field calculations. Therefore, existing near-field to far-field conversion technology usually requires the help of other methods to complete it, which urgently needs to be solved. Summary of the Invention

[0004] The present application provides an electromagnetic field quantity extrapolation method and device based on the fusion of finite differences and neural networks to solve the problem of insufficient computer memory and inability to directly perform far-field calculations due to the sharp increase in calculation amount when calculating the far field of the antenna in the background technology, and improve the calculation efficiency.

[0005] The first embodiment of the present application provides an electromagnetic field quantity extrapolation method based on finite difference and neural network fusion, comprising the following steps:

[0006] Construct a multi-layer concentric sphere with the phase center of the antenna as the sphere center;

[0007] Dividing each layer of concentric spherical surface in a first direction and a second direction based on a preset step size to obtain a plurality of neurons, wherein each neuron includes a plurality of cells;

[0008] Determining a voltage value of each cell in the first direction and a voltage value of each cell in the second direction;

[0009] performing, based on the voltage value of each cell in the first direction and the voltage value of each cell in the second direction, iterative calculation of the voltage value of each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface in the first direction or the second direction based on a preset iterative formula, to obtain an iterative calculation result in the first direction or the second direction for each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface;

[0010] The voltage value of each cell in the outermost concentric sphere is calculated according to the iterative calculation result in the first direction or the iterative calculation result in the second direction. When the outermost concentric sphere meets the preset convergence condition, the iterative calculation is stopped and it is determined that the extrapolation of the current electromagnetic field quantity is completed.

[0011] According to one embodiment of the present application, determining the voltage value of each cell in the first direction and the voltage value of each cell in the second direction includes:

[0012] Acquire the electric field intensity of each cell in the first direction, the inner arc length of each cell in the first direction, the electric field intensity of each cell in the second direction, and the inner arc length of each cell in the second direction;

[0013] Obtaining a voltage value of each cell in the first direction according to the product of the electric field intensity of each cell in the first direction and the inner arc length of each cell in the first direction;

[0014] The voltage value of each cell in the second direction is obtained according to the product of the electric field intensity of each cell in the second direction and the inner arc length of each cell in the second direction.

[0015] According to one embodiment of the present application, the inner arc length of each cell in the first direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the first direction, and the inner arc length of each cell in the second direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the second direction.

[0016] According to one embodiment of the present application, the inner arc radius of each cell in the first direction is obtained by the current near-field sampling radius, a preset wavelength, and the number of extrapolated circles of each cell.

[0017] According to one embodiment of the present application, the iterative calculation of the voltage value of each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface in the first direction or the second direction based on a preset iterative formula to obtain the iterative calculation result of each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface in the first direction or the iterative calculation result in the second direction includes:

[0018] calculating, according to voltage values ​​of a plurality of cells adjacent to each cell in the first direction, an iterative calculation result of each cell in the first direction;

[0019] An iterative calculation result of each cell in the second direction is obtained by calculation according to voltage values ​​of a plurality of cells adjacent to each cell in the second direction.

[0020] According to one embodiment of the present application, stopping the iterative calculation when the outermost concentric spherical surface meets a preset convergence condition includes:

[0021] Obtaining the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration;

[0022] Determine whether a difference between a maximum voltage value of a cell in the outermost concentric sphere of the current iteration and a maximum voltage value of a cell in the outermost concentric sphere of the previous iteration is less than a preset threshold, and whether a direction of the maximum voltage value of a cell in the outermost concentric sphere of the current iteration is the same as that of the maximum voltage value of a cell in the outermost concentric sphere of the previous iteration;

[0023] If the difference between the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration is less than the preset threshold, and the direction of the maximum voltage value of the cells in the outermost concentric sphere of the current iteration is the same as the direction of the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration, then it is determined that the outermost concentric sphere meets the preset convergence condition.

[0024] According to the electromagnetic field quantity extrapolation method based on the fusion of finite differences and neural networks in an embodiment of the present application, multiple layers of concentric spheres are divided in the first and second directions, respectively. Based on the voltage value of each cell in the first direction and the voltage value in the second direction, the voltage value of each cell between the innermost and outermost concentric spheres in the first or second direction is iteratively calculated. The iterative calculation result in the first direction or the iterative calculation result in the second direction for each cell is obtained. When the outermost concentric sphere meets the preset convergence condition, the current electromagnetic field quantity extrapolation is determined to be complete. This solves the problem of insufficient computer memory and the inability to directly perform far-field calculations caused by the background technology, and improves computational efficiency.

[0025] The second embodiment of the present application provides an electromagnetic field quantity extrapolation device based on the fusion of finite difference and neural network, comprising:

[0026] A construction module, used for constructing a multi-layer concentric sphere with the phase center of the antenna as the sphere center;

[0027] a partitioning module, configured to partition each layer of concentric spherical surface in a first direction and a second direction based on a preset step size to obtain a plurality of neurons, wherein each neuron includes a plurality of cells;

[0028] a determination module, configured to determine a voltage value of each cell in the first direction and a voltage value of each cell in the second direction;

[0029] an iterative calculation module, configured to iteratively calculate the voltage value of each cell in the first direction or the second direction in a concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface based on the voltage value of each cell in the first direction and the voltage value of each cell in the second direction, based on a preset iterative formula, to obtain an iterative calculation result in the first direction or the iterative calculation result in the second direction for each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface;

[0030] An extrapolation module is used to calculate the voltage value of each cell in the outermost concentric sphere according to the iterative calculation result in the first direction or the iterative calculation result in the second direction, and when the outermost concentric sphere meets the preset convergence condition, stop the iterative calculation and determine that the extrapolation of the current electromagnetic field quantity is completed.

[0031] According to one embodiment of the present application, the determining module is configured to:

[0032] Acquire the electric field intensity of each cell in the first direction, the inner arc length of each cell in the first direction, the electric field intensity of each cell in the second direction, and the inner arc length of each cell in the second direction;

[0033] Obtaining a voltage value of each cell in the first direction according to the product of the electric field intensity of each cell in the first direction and the inner arc length of each cell in the first direction;

[0034] The voltage value of each cell in the second direction is obtained according to the product of the electric field intensity of each cell in the second direction and the inner arc length of each cell in the second direction.

[0035] According to one embodiment of the present application, the inner arc length of each cell in the first direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the first direction, and the inner arc length of each cell in the second direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the second direction.

[0036] According to one embodiment of the present application, the inner arc radius of each cell in the first direction is obtained by the current near-field sampling radius, a preset wavelength, and the number of extrapolated circles of each cell.

[0037] According to one embodiment of the present application, the iterative calculation module is used to:

[0038] calculating, according to voltage values ​​of a plurality of cells adjacent to each cell in the first direction, an iterative calculation result of each cell in the first direction;

[0039] An iterative calculation result of each cell in the second direction is obtained by calculation according to voltage values ​​of a plurality of cells adjacent to each cell in the second direction.

[0040] According to one embodiment of the present application, the extrapolation module is used to:

[0041] Obtaining the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration;

[0042] Determine whether a difference between a maximum voltage value of a cell in the outermost concentric sphere of the current iteration and a maximum voltage value of a cell in the outermost concentric sphere of the previous iteration is less than a preset threshold, and whether a direction of the maximum voltage value of a cell in the outermost concentric sphere of the current iteration is the same as that of the maximum voltage value of a cell in the outermost concentric sphere of the previous iteration;

[0043] If the difference between the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration is less than the preset threshold, and the direction of the maximum voltage value of the cells in the outermost concentric sphere of the current iteration is the same as the direction of the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration, then it is determined that the outermost concentric sphere meets the preset convergence condition.

[0044] According to the electromagnetic field quantity extrapolation device based on the fusion of finite differences and neural networks in an embodiment of the present application, multiple layers of concentric spheres are divided in the first direction and the second direction, respectively. Based on the voltage value of each cell in the first direction and the voltage value in the second direction, the voltage value of each cell between the innermost and outermost concentric spheres in the first direction or the second direction is iteratively calculated, obtaining the iterative calculation result of each cell in the first direction or the iterative calculation result in the second direction. When the outermost concentric sphere meets the preset convergence condition, the current electromagnetic field quantity extrapolation is determined to be complete. This solves the problem of insufficient computer memory and the inability to directly perform far-field calculations caused by the background technology, and improves computing efficiency.

[0045] The third aspect of the present application provides an electronic device, comprising: 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 electromagnetic field quantity extrapolation method based on finite difference and neural network fusion as described in the above embodiment.

[0046] The fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the electromagnetic field quantity extrapolation method based on the fusion of finite differences and neural networks as described in the above embodiments.

[0047] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0049] Figure 1 A schematic diagram of near-far field transformation using a background technology method;

[0050] Figure 2 This is a flow chart of an electromagnetic field quantity extrapolation method based on finite difference and neural network fusion according to an embodiment of the present application;

[0051] Figure 3 A schematic diagram of constructing a multi-layer spherical surface according to one embodiment of the present application;

[0052] Figure 4 Schematic diagram of the near-field direction of a half-wave oscillator;

[0053] Figure 5 is the far-field pattern calculated using the method of this application;

[0054] Figure 6 The far-field pattern is calculated with reference to industry software;

[0055] Figure 7 This is a schematic diagram of the smoothing settings of reference industry software;

[0056] Figure 8 This is the real far-field pattern calculated by reference to industry software without smoothing.

[0057] Figure 9 is an E-plane pattern at an azimuth angle of 0 degrees extracted from the conversion result according to one embodiment of the present application;

[0058] Figure 10 is an E-plane pattern at an azimuth angle of 90 degrees extracted from the conversion result according to one embodiment of the present application;

[0059] Figure 11 is a horizontal plane direction pattern extracted from the conversion result according to one embodiment of the present application;

[0060] Figure 12 The E-plane pattern at azimuth angle 0 degrees is referenced by the software;

[0061] Figure 13 Schematic diagram of a block diagram of an electromagnetic field quantity extrapolation device based on finite difference and neural network fusion according to an embodiment of the present application;

[0062] Figure 14 Schematic diagram of the structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0063] The following describes in detail embodiments of the present application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0064] The following describes an electromagnetic field quantity extrapolation method and device based on finite difference and neural network fusion according to an embodiment of the present application with reference to the accompanying drawings.

[0065] Before introducing the electromagnetic field quantity extrapolation method based on the fusion of finite difference and neural network in the embodiment of the present application, a brief introduction to the near-far field conversion method in the related art is first given.

[0066] While there are various existing solutions for near-field to far-field conversion in related technologies, their core ideas are essentially the same: first, the field quantity at the observation point is obtained to analyze the field evolution pattern, and then the field quantity distribution of the target field is derived through numerical calculation. Among them, the finite difference method is often used to calculate the near-field distribution from a field source (such as an antenna). Its basic principle is that the electric field value at a certain point in space (or grid cell) is determined by the average value of the electric fields generated at that point by the electric fields in the six surrounding grids. This method is a numerical method that can accurately solve field quantities. After decades of development, it has been widely verified as a reliable numerical calculation method.

[0067] However, directly applying the finite difference method to the near-far field transformation will face significant computational resource challenges. Figure 1 As shown in the figure, around the antenna, the finite difference grid is distributed in the shape of a rectangular parallelepiped in six directions: front, back, left, right, and top. When the distance from the antenna doubles, the total number of grid cells increases to eight times the original, resulting in a sharp increase in computational complexity. Typically, the far-field region is many orders of magnitude farther from the antenna than the near-field region (far exceeding the case of doubling the distance), further increasing the amount of computation. Therefore, directly using the finite difference method to calculate the far-field distribution will result in insufficient computer memory, which is the fundamental reason why this method has not been widely used for far-field calculations.

[0068] In recent years, neural network algorithms, as an emerging computing technology, have made major breakthroughs in many fields. Their powerful data processing and pattern recognition capabilities have given them significant advantages in solving complex problems. They are receiving more and more attention and their application areas are becoming more and more extensive.

[0069] Therefore, based on the problem that the amount of calculation in the traditional finite difference method increases exponentially during the extrapolation process and cannot be carried out for long-distance calculations, this application solves the technical difficulty that the traditional finite difference method cannot directly calculate the far field of the antenna by adopting a method that combines the finite difference method and the artificial neural network algorithm, and realizes efficient and accurate direct calculation of the target field.

[0070] It should be noted that there are many types of neural network algorithms, and this invention primarily utilizes artificial neural network algorithms. This invention uses wireless communications as an example, specifically involving the near-to-far-field transformation of spherical waves in the electromagnetic field. The invention can be extended to encompass, but not limited to, the calculation of various field quantities in various fields of computational electromagnetics, such as computational fluid dynamics and computational thermodynamics.

[0071] The following introduces the electromagnetic field quantity extrapolation method based on the fusion of finite difference and neural network proposed in this application.

[0072] Specifically, Figure 2 A flow chart of an electromagnetic field quantity extrapolation method based on the fusion of finite difference and neural network provided in an embodiment of the present application.

[0073] It should be noted that the electromagnetic field quantity extrapolation method based on the fusion of finite differences and neural networks in the embodiment of the present application takes the near-field and far-field changes of the antenna as an example. For the calculation of other physical fields (such as magnetic fields, temperature fields, flow fields, etc.), the grid division and deduction method of the present application can also be used. The calculation of different physical fields can be described and solved using different field quantity characteristic equations according to their own physical characteristics.

[0074] like Figure 2 As shown, the electromagnetic field quantity extrapolation method based on the fusion of finite difference and neural network includes the following steps:

[0075] In step S201 , a multi-layer concentric spherical surface is constructed with the phase center of the antenna as the sphere center.

[0076] Specifically, if Figure 3 As shown, in the embodiment of the present application, a set of concentric spheres is constructed with the phase center P of the antenna as the center of the sphere, and these spheres are also concentric with the sphere of near-field sampling.

[0077] Furthermore, with the near-field sampling radius r as the starting value, the radius difference between two adjacent spherical surfaces is a preset path difference (e.g., a quarter wavelength). In addition, in addition to the spherical surface with the initial radius r, at least a plurality of additional concentric spherical surfaces are constructed, the number of which can be determined according to actual needs (e.g., extending outward by 50 wavelengths).

[0078] In step S202 , each layer of concentric spherical surface is divided in a first direction and a second direction based on a preset step size to obtain a plurality of neurons, wherein each neuron includes a plurality of cells.

[0079] Among them, the first direction is the polar angle direction in the spherical coordinate system, and the second direction is the azimuthal angle direction in the spherical coordinate system. The preset step size can be pre-set by technical personnel in this field according to actual conditions, such as a step size of 1°, which is not specifically limited here.

[0080] Specifically, if Figure 3 As shown in FIG3 , with the phase center P of the antenna as the nerve center, each concentric sphere is divided from the inside out in the first direction θ and the second direction φ, both with a certain angle step size (for example, 1°), and the area contained in the sphere is divided into many neurons, such as neuron a, neuron b, neuron c, etc. in FIG3 .

[0081] Furthermore, on the multiple spheres constructed above, these neurons are systematically divided into multiple cells. In the plane view of the polar angle direction θ, a cell A on neuron a (such as Figure 3 Taking the unit marked in red in the figure as an example, its adjacent relationship can be described as follows: there is a cell B adjacent to the neuron b adjacent to cell A; there is a cell C adjacent to the neuron c adjacent to cell A; on the same neuron a, there are two cells, namely cell A- and cell A+ adjacent to it.

[0082] Similarly, by converting the viewing angle to the azimuth angle φ plane, it can be obtained that the cells adjacent to cell A also include cell D on neuron d and cell E on neuron e.

[0083] Therefore, the embodiment of the present application combines the traditional finite difference method and the neural network algorithm to upgrade the rectangular cells to arc cells, so that when the distance of the field from the antenna increases, the calculation amount remains unchanged, instead of the calculation amount increasing sharply as in the traditional finite difference method. This solves the problem of severe computer memory shortage caused by using the traditional finite difference method to calculate the antenna far field, thereby improving computing efficiency.

[0084] In step S203 , a voltage value of each cell in the first direction and a voltage value of each cell in the second direction are determined.

[0085] Furthermore, in some embodiments, determining the voltage value of each cell in the first direction and the voltage value of each cell in the second direction includes: obtaining the electric field strength of each cell in the first direction, the inner arc length of each cell in the first direction, the electric field strength of each cell in the second direction, and the inner arc length of each cell in the second direction; obtaining the voltage value of each cell in the first direction according to the product of the electric field strength of each cell in the first direction and the inner arc length of each cell in the first direction; obtaining the voltage value of each cell in the second direction according to the product of the electric field strength of each cell in the second direction and the inner arc length of each cell in the second direction.

[0086] Specifically, the embodiment of the present application can use the finite difference method to determine the near-far field conversion parameters, and the extrapolated parameters are the cell voltages, converting the vector (electric field) into a scalar (voltage). The voltage of each cell is determined by its electric field strength and the arc length within the cell.

[0087] Illustratively, in the embodiments of the present application, the electric field intensity of each cell in the first direction and the electric field intensity of each cell in the second direction can be obtained through measurement or numerical simulation.

[0088] For example, for cell A, its electric field includes an electric field intensity component in a first direction θ and an electric field intensity component in a second direction φ, and its voltage also includes a voltage value component in the first direction θ and a voltage value component in the second direction φ.

[0089] Furthermore, the voltage value of each cell in the first direction is obtained based on the product of the electric field strength of each cell in the first direction and the inner arc length of each cell in the first direction, and the voltage value of each cell in the second direction is obtained based on the product of the electric field strength of each cell in the second direction and the inner arc length of each cell in the second direction. For ease of understanding, the voltage value of each cell in the first direction and the voltage value in the second direction of the embodiment of the present application can be expressed as:

[0090]

[0091] Among them, U θ is the voltage value of a cell in the first direction, E θ is the electric field intensity of a cell in the first direction, l θ is the length of the inner arc of a cell in the first direction, is the voltage value of a cell in the second direction, is the electric field intensity of a cell in the second direction, is the length of the inner arc of a cell in the second direction.

[0092] Among them, the voltage value u of a cell in the first direction θ , the electric field strength E of a cell in the first direction θ , the voltage value of a cell in the second direction and the electric field strength of a cell in the second direction All are phasors.

[0093] Therefore, the embodiment of the present application converts the electric field intensity of the near-field measurement (simulation) result into a scalar voltage to facilitate processing by the finite difference method. Furthermore, to leverage the strengths of C++ programming, the voltage U is calculated in phasor form, further improving computational efficiency.

[0094] Furthermore, in some embodiments, the inner arc length of each cell in the first direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the first direction, and the inner arc length of each cell in the second direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the second direction.

[0095] Specifically, the inner arc length l of a cell in the first direction θ Can be Figure 3The red arc segment shown. The inner arc length of each cell in the first direction of the embodiment of the present application is determined by the current near-field sampling radius and the inner arc radius of each cell in the first direction, and the inner arc length of each cell in the second direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the second direction. For ease of understanding, taking the inner arc length of cell A as an example, the inner arc length of cell A in the first direction and the inner arc length of the second direction of the embodiment of the present application can be expressed as:

[0096]

[0097] Among them, l θ is the length of the inner arc of a cell in the first direction, r Aθ is the inner arc radius of cell A in the first direction, is the length of the inner arc of a cell in the second direction, is the inner arc radius of cell A in the second direction.

[0098] Furthermore, in some embodiments, the inner arc radius of each cell in the first direction is obtained by the current near-field sampling radius, the preset wavelength, and the number of extrapolated circles of each cell.

[0099] Preferably, taking the preset wavelength as 1 / 4 as an example, the inner arc radius of the cell A in the first direction of the embodiment of the present application can be expressed as:

[0100]

[0101] Among them, r Aθ is the inner arc radius of cell A in the first direction, r is the near-field sampling radius, n A is the number of extrapolated circles of cell A.

[0102] Similarly, the inner arc radius of cell A in the second direction can be expressed as:

[0103]

[0104] Furthermore, in the boundary conditions, it is necessary to determine the initial calculation level, that is, a sphere with a radius of r. This sphere is a hard source, representing a known, determined field source. On the sphere with a radius of r, the above formulas (1) and (2) are applied to convert the known electric field intensity into a voltage value. The voltage values ​​of the cells in the first direction and the second direction of the 0th layer (i.e., the innermost layer) of the sphere are obtained, providing the starting field quantity for subsequent iterations.

[0105] In step S204, according to the voltage value of each cell in the first direction and the voltage value of each cell in the second direction, based on a preset iterative formula, the voltage value of each cell in the concentric sphere between the innermost concentric sphere and the outermost concentric sphere in the first direction or the second direction is iteratively calculated to obtain the iterative calculation result in the first direction or the iterative calculation result in the second direction for each cell in the concentric sphere between the innermost concentric sphere and the outermost concentric sphere.

[0106] Furthermore, in some embodiments, based on a preset iterative formula, the voltage value of each cell in the concentric sphere between the innermost concentric sphere and the outermost concentric sphere in the first direction or the second direction is iteratively calculated to obtain the iterative calculation result in the first direction or the iterative calculation result in the second direction of each cell in the concentric sphere between the innermost concentric sphere and the outermost concentric sphere, including: calculating the iterative calculation result of each cell in the first direction according to the voltage values ​​of multiple cells adjacent to each cell in the first direction; calculating the iterative calculation result of each cell in the second direction according to the voltage values ​​of multiple cells adjacent to each cell in the second direction.

[0107] It's understandable that applying a neural network algorithm to traditional finite-difference methods makes the transmission coefficients between adjacent neurons and cells very simple, allowing complex problems to be solved using simple arithmetic operations. In addition to the cells on neuron a being stimulated by neighboring neurons b and c, the cells on neuron b are also stimulated by non-adjacent neuron c. It's just that the cells on neuron c first stimulate the cells on neuron a, which then further stimulate the cells on neuron b through the cells on neuron a. This reduces complex problems to simple ones, improving computational efficiency.

[0108] Specifically, taking cell A as an example, there are 6 cells around cell A, namely cell B on neuron b adjacent to cell A, cell C on neuron c adjacent to cell A, cells A- and A+ on the same neuron a, cell D on neuron d, and cell E on neuron e. The following preset iterative formula is used to iterate in the first direction:

[0109]

[0110] Among them, U θA is the voltage value of cell A in the first direction, U θB is the voltage value of cell B in the first direction, U θC is the voltage value of cell C in the first direction, U θD is the voltage value of cell D in the first direction, U θEis the voltage value of cell E in the first direction, U θA- is the voltage value of cell A in the first direction, U θA+ is the voltage value of cell A+ in the first direction.

[0111] It should be noted that the iterative calculation in the first direction of the embodiment of the present application is only performed on the cells in the middle layer. If the voltage value of the cells in the 0th layer (i.e., the innermost layer) or the voltage of the cells in the outermost layer appears in the above formula, it only appears on the right side of the equal sign of formula (7), that is, the iteration of the embodiment of the present application is only performed on the cells in the middle layer, i.e., the 1st layer to the N-1th layer. The voltage values ​​of the 0th layer and the outermost layer do not participate in the iterative update, but are only calculated as known quantities.

[0112] Similarly, formula (8) can be used to iterate cell A in the second direction:

[0113]

[0114] in, is the voltage value of cell A in the second direction, is the voltage value of cell B in the second direction, is the voltage value of cell C in the second direction, is the voltage value of cell D in the second direction, is the voltage value of cell E in the second direction, is the voltage value of cell A in the second direction, U θA+ is the voltage value of cell A+ in the second direction.

[0115] In step S205, the voltage value of each cell in the outermost concentric sphere is calculated according to the iterative calculation result in the first direction or the iterative calculation result in the second direction. When the outermost concentric sphere meets the preset convergence condition, the iterative calculation is stopped and it is determined that the extrapolation of the current electromagnetic field quantity is completed.

[0116] Furthermore, in some embodiments, when the outermost concentric sphere meets the preset convergence condition, the iterative calculation is stopped, including: obtaining the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration; judging whether the difference between the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration is less than a preset threshold, and whether the direction of the maximum voltage value of the cells in the outermost concentric sphere of the current iteration is the same as the direction of the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration; if the difference between the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration is less than the preset threshold, and the direction of the maximum voltage value of the cells in the outermost concentric sphere of the current iteration is the same as the direction of the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration, then it is determined that the outermost concentric sphere meets the preset convergence condition.

[0117] The preset threshold value may be a smaller threshold value preset by those skilled in the art according to actual conditions, such as 0.1%, and is not specifically limited here.

[0118] Specifically, after each iteration is completed, the maximum value of the voltage values ​​of all cells on the current outermost concentric sphere and its corresponding direction are recorded. Similarly, the maximum voltage value on the outermost concentric sphere in the previous iteration and its corresponding direction are recorded, and the difference between the maximum voltage value of the current iteration and the maximum voltage value of the previous iteration is calculated. It is judged whether the difference between the maximum voltage value of the current iteration and the maximum voltage value of the previous iteration is less than a preset threshold, and whether the direction of the maximum voltage value of the current iteration is the same as the direction of the maximum voltage value of the previous iteration. If the difference between the maximum voltage value of the current iteration and the maximum voltage value of the previous iteration is less than the preset threshold, and the direction of the maximum voltage value of the current iteration (θ2, φ2) is the same as the direction of the maximum voltage value of the previous iteration (θ1, φ1), it is determined that the current iteration has met the convergence condition, that is, the near-far field extrapolation is completed, and the iteration is exited.

[0119] In order to facilitate those skilled in the art to understand more clearly and intuitively the beneficial effects of the electromagnetic field quantity extrapolation method based on the fusion of finite differences and neural networks proposed in this application, the following takes the dipole antenna (half-wave oscillator) familiar to those skilled in the art as an example to demonstrate the effect of directly performing near-to-far field conversion using the example of the present invention.

[0120] Specifically, if Figure 4 As shown, Figure 4 The near-field pattern of the half-wave oscillator shown in FIG. 1 is obtained by modeling and simulating with a certain reference software. Further, the far-field pattern calculated using the exemplary method of the present invention is shown in FIG. Figure 5As shown, the "apple diagram" well known to those skilled in the art is obtained. Figure 5 This indicates that the near-to-far field conversion is successfully completed using the exemplary method of the present invention.

[0121] Among them, the near-field data used in the embodiment of the present application can be derived after modeling and simulation in a certain industry software. The software can directly calculate the far field of the antenna, and the obtained directional pattern is as follows: Figure 6 As shown, it should be noted that Figure 6 The maximum gain value in the upper left corner is 2.07dBi. Figure 5 and Figure 6 , it can be found that the results calculated by this software are obviously smoother than those of the present invention. However, Figure 7 As shown in the figure, the result calculated by the software is obviously smoother than the result of the example of the present invention because the software used as the reference has selected the smoothing process by default. Figure 7 The default checkbox for smoothing is “√”. If it is not checked, the actual simulation result will be displayed. Figure 8 shown.

[0122] Furthermore, in order to more accurately observe the near-far field conversion results of the embodiment of the present application, the plane directional pattern is now extracted from the conversion results. Among them, the azimuth angle 0 degree E plane directional pattern of the embodiment of the present application is as follows Figure 9 As shown, the E-plane directional diagram of the embodiment of the present application at an azimuth angle of 90 degrees is as follows Figure 10 As shown, the horizontal plane direction diagram of the embodiment of the present application is as follows Figure 11 Correspondingly, the simulation results of the reference software also include the directional patterns of the above three planes. Among them, the azimuth angle 0 degree E plane directional pattern of the reference software is as follows Figure 12 shown.

[0123] Will Figure 9 and Figure 12 A comparison shows no significant difference in the calculation results between the two. However, the advantage of the present embodiment over the reference software (which does not use the finite difference method) lies in its faster calculation speed, with its computational efficiency several dozen times greater than that of the reference software. Furthermore, the reference software is typically deployed on workstations or servers with hundreds of GB of memory and CPUs with over 100 threads, while the example program of the present embodiment can complete the near-to-far field conversion on a standard PC.

[0124] Thus, by combining a neural network algorithm with the traditional finite-difference method, this embodiment of the present application overcomes the technical difficulty of the traditional finite-difference method in calculating the antenna far field due to computer memory limitations. The above examples demonstrate that the present invention has successfully resolved this technical difficulty. Furthermore, to further improve computational efficiency, the program is preferably written in a programming language suitable for scientific computing, such as C / C++ or Fortran.

[0125] According to the electromagnetic field quantity extrapolation method based on the fusion of finite differences and neural networks in an embodiment of the present application, multiple layers of concentric spheres are divided in the first and second directions, respectively. Based on the voltage value of each cell in the first direction and the voltage value in the second direction, the voltage value of each cell between the innermost and outermost concentric spheres in the first or second direction is iteratively calculated. The iterative calculation result in the first direction or the iterative calculation result in the second direction for each cell is obtained. When the outermost concentric sphere meets the preset convergence condition, the current electromagnetic field quantity extrapolation is determined to be complete. This solves the problem of insufficient computer memory and the inability to directly perform far-field calculations caused by the background technology, and improves computational efficiency.

[0126] Next, an electromagnetic field quantity extrapolation device based on finite difference and neural network fusion proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.

[0127] Figure 13 It is a block diagram of an electromagnetic field quantity extrapolation device based on the fusion of finite difference and neural network in an embodiment of the present application.

[0128] like Figure 13 As shown, the electromagnetic field quantity extrapolation device 10 based on the fusion of finite difference and neural network includes: a construction module 100, a division module 200, a determination module 300, an iterative calculation module 400 and an extrapolation module 500.

[0129] Among them, the construction module 100 is used to construct a multi-layer concentric sphere with the phase center of the antenna as the sphere center; the division module 200 is used to divide each layer of the concentric sphere in the first direction and the second direction based on a preset step size to obtain multiple neurons, wherein each neuron includes multiple cells; the determination module 300 is used to determine the voltage value of each cell in the first direction and the voltage value of each cell in the second direction; the iterative calculation module 400 is used to determine the voltage value of each cell in the first direction and the voltage value of each cell in the second direction based on a preset iterative formula. The voltage value of each cell in the concentric sphere between the concentric sphere and the outermost concentric sphere is iteratively calculated in the first direction or the second direction to obtain the iterative calculation result in the first direction or the iterative calculation result in the second direction for each cell in the concentric sphere between the innermost concentric sphere and the outermost concentric sphere; the extrapolation module 500 is used to calculate the voltage value of each cell in the outermost concentric sphere according to the iterative calculation result in the first direction or the iterative calculation result in the second direction, and when the outermost concentric sphere meets the preset convergence condition, the iterative calculation is stopped and it is determined that the extrapolation of the current electromagnetic field quantity is completed.

[0130] Furthermore, in some embodiments, the determination module 200 is used to: obtain the electric field strength of each cell in the first direction, the inner arc length of each cell in the first direction, the electric field strength of each cell in the second direction, and the inner arc length of each cell in the second direction; obtain the voltage value of each cell in the first direction according to the product of the electric field strength of each cell in the first direction and the inner arc length of each cell in the first direction; obtain the voltage value of each cell in the second direction according to the product of the electric field strength of each cell in the second direction and the inner arc length of each cell in the second direction.

[0131] Furthermore, in some embodiments, the inner arc length of each cell in the first direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the first direction, and the inner arc length of each cell in the second direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the second direction.

[0132] Furthermore, in some embodiments, the inner arc radius of each cell in the first direction is obtained by the current near-field sampling radius, the preset wavelength, and the number of extrapolated circles of each cell.

[0133] Furthermore, in some embodiments, the iterative calculation module 400 is used to: calculate the iterative calculation result of each cell in the first direction based on the voltage values ​​of multiple cells adjacent to each cell in the first direction; and calculate the iterative calculation result of each cell in the second direction based on the voltage values ​​of multiple cells adjacent to each cell in the second direction.

[0134] Furthermore, in some embodiments, the extrapolation module 500 is used to: obtain the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration; determine whether the difference between the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration is less than a preset threshold, and whether the direction of the maximum voltage value of the cells in the outermost concentric sphere of the current iteration is the same as the direction of the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration; if the difference between the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration is less than the preset threshold, and the direction of the maximum voltage value of the cells in the outermost concentric sphere of the current iteration is the same as the direction of the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration, then it is determined that the outermost concentric sphere meets the preset convergence condition.

[0135] It should be noted that the above explanation of the embodiment of the electromagnetic field quantity extrapolation method based on the fusion of finite difference and neural network is also applicable to the electromagnetic field quantity extrapolation device based on the fusion of finite difference and neural network in this embodiment, and will not be repeated here.

[0136] According to the electromagnetic field quantity extrapolation device based on the fusion of finite differences and neural networks in an embodiment of the present application, multiple layers of concentric spheres are divided in the first direction and the second direction, respectively. Based on the voltage value of each cell in the first direction and the voltage value in the second direction, the voltage value of each cell between the innermost and outermost concentric spheres in the first direction or the second direction is iteratively calculated, obtaining the iterative calculation result of each cell in the first direction or the iterative calculation result in the second direction. When the outermost concentric sphere meets the preset convergence condition, the current electromagnetic field quantity extrapolation is determined to be complete. This solves the problem of insufficient computer memory and the inability to directly perform far-field calculations caused by the background technology, and improves computing efficiency.

[0137] Figure 14 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0138] Memory 1401 , processor 1402 , and computer programs stored in the memory 1401 and executable on the processor 1402 .

[0139] When the processor 1402 executes the program, the electromagnetic field quantity extrapolation method based on the fusion of finite difference and neural network provided in the above embodiment is implemented.

[0140] Furthermore, the electronic device further includes:

[0141] The communication interface 1403 is used for communication between the memory 1401 and the processor 1402 .

[0142] The memory 1401 is used to store computer programs that can be run on the processor 1402 .

[0143] The memory 1401 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0144] If the memory 1401, processor 1402, and communication interface 1403 are implemented independently, the communication interface 1403, memory 1401, and processor 1402 can be connected to each other via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 14 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0145] Optionally, in a specific implementation, if the memory 1401, the processor 1402 and the communication interface 1403 are integrated on a chip, the memory 1401, the processor 1402 and the communication interface 1403 can communicate with each other through an internal interface.

[0146] The processor 1402 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0147] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned electromagnetic field quantity extrapolation method based on the fusion of finite differences and neural networks.

[0148] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0149] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. Throughout the description of this application, "plurality" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0150] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for extrapolating electromagnetic field quantities based on the fusion of finite difference and neural network, characterized in that: The following steps are involved: Construct a multi-layer concentric sphere with the phase center of the antenna as the sphere center; Dividing each layer of concentric spherical surface in a first direction and a second direction based on a preset step size to obtain a plurality of neurons, wherein each neuron includes a plurality of cells; Determining a voltage value of each cell in the first direction and a voltage value of each cell in the second direction; performing, based on the voltage value of each cell in the first direction and the voltage value of each cell in the second direction, iterative calculation of the voltage value of each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface in the first direction or the second direction based on a preset iterative formula, to obtain an iterative calculation result in the first direction or the second direction for each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface; The voltage value of each cell in the outermost concentric sphere is calculated according to the iterative calculation result in the first direction or the iterative calculation result in the second direction. When the outermost concentric sphere meets the preset convergence condition, the iterative calculation is stopped and it is determined that the extrapolation of the current electromagnetic field quantity is completed.

2. The method according to claim 1, characterized in that The determining of the voltage value of each cell in the first direction and the voltage value of each cell in the second direction includes: Acquire the electric field intensity of each cell in the first direction, the inner arc length of each cell in the first direction, the electric field intensity of each cell in the second direction, and the inner arc length of each cell in the second direction; Obtaining a voltage value of each cell in the first direction according to the product of the electric field intensity of each cell in the first direction and the inner arc length of each cell in the first direction; The voltage value of each cell in the second direction is obtained according to the product of the electric field intensity of each cell in the second direction and the inner arc length of each cell in the second direction.

3. The method according to claim 2, characterized in that The inner arc length of each cell in the first direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the first direction, and the inner arc length of each cell in the second direction is determined by the current near-field sampling radius and the inner arc radius of each cell in the second direction.

4. The method according to claim 3, characterized in that The inner arc radius of each cell in the first direction is obtained by the current near-field sampling radius, the preset wavelength and the number of extrapolated circles of each cell.

5. The method according to claim 1, wherein The iterative calculation of the voltage value of each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface in the first direction or the second direction based on a preset iterative formula to obtain the iterative calculation result of each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface in the first direction or the iterative calculation result in the second direction includes: calculating, according to voltage values ​​of a plurality of cells adjacent to each cell in the first direction, an iterative calculation result of each cell in the first direction; An iterative calculation result of each cell in the second direction is obtained by calculation according to voltage values ​​of a plurality of cells adjacent to each cell in the second direction.

6. The method according to claim 1, characterized in that When the outermost concentric spherical surface satisfies a preset convergence condition, stopping the iterative calculation includes: Obtaining the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration; Determine whether a difference between a maximum voltage value of a cell in the outermost concentric sphere of the current iteration and a maximum voltage value of a cell in the outermost concentric sphere of the previous iteration is less than a preset threshold, and whether a direction of the maximum voltage value of a cell in the outermost concentric sphere of the current iteration is the same as that of the maximum voltage value of a cell in the outermost concentric sphere of the previous iteration; If the difference between the maximum voltage value of the cells in the outermost concentric sphere of the current iteration and the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration is less than the preset threshold, and the direction of the maximum voltage value of the cells in the outermost concentric sphere of the current iteration is the same as the direction of the maximum voltage value of the cells in the outermost concentric sphere of the previous iteration, then it is determined that the outermost concentric sphere meets the preset convergence condition.

7. An electromagnetic field quantity extrapolation device based on the fusion of finite difference and neural network, characterized in that: include: A construction module, used for constructing a multi-layer concentric sphere with the phase center of the antenna as the sphere center; a partitioning module, configured to partition each layer of concentric spherical surface in a first direction and a second direction based on a preset step size to obtain a plurality of neurons, wherein each neuron includes a plurality of cells; a determination module, configured to determine a voltage value of each cell in the first direction and a voltage value of each cell in the second direction; an iterative calculation module, configured to iteratively calculate the voltage value of each cell in the first direction or the second direction in a concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface based on the voltage value of each cell in the first direction and the voltage value of each cell in the second direction, based on a preset iterative formula, to obtain an iterative calculation result in the first direction or the iterative calculation result in the second direction for each cell in the concentric spherical surface between the innermost concentric spherical surface and the outermost concentric spherical surface; An extrapolation module is used to calculate the voltage value of each cell in the outermost concentric sphere according to the iterative calculation result in the first direction or the iterative calculation result in the second direction, and when the outermost concentric sphere meets the preset convergence condition, stop the iterative calculation and determine that the extrapolation of the current electromagnetic field quantity is completed.

8. The device according to claim 7, characterized in that The determining module is configured to: Acquire the electric field intensity of each cell in the first direction, the inner arc length of each cell in the first direction, the electric field intensity of each cell in the second direction, and the inner arc length of each cell in the second direction; Obtaining a voltage value of each cell in the first direction according to the product of the electric field intensity of each cell in the first direction and the inner arc length of each cell in the first direction; The voltage value of each cell in the second direction is obtained according to the product of the electric field intensity of each cell in the second direction and the inner arc length of each cell in the second direction.

9. An electronic device, characterized in that: include: 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 electromagnetic field quantity extrapolation method based on finite difference and neural network fusion as described in any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program is executed by a processor to implement the electromagnetic field quantity extrapolation method based on finite difference and neural network fusion as described in any one of claims 1 to 6.

Images