CN113051836B - Rapid modeling method for antenna under machine learning-assisted array environment - Google Patents
Rapid modeling method for antenna under machine learning-assisted array environment Download PDFInfo
- Publication number
- CN113051836B CN113051836B CN202110417710.4A CN202110417710A CN113051836B CN 113051836 B CN113051836 B CN 113051836B CN 202110417710 A CN202110417710 A CN 202110417710A CN 113051836 B CN113051836 B CN 113051836B
- Authority
- CN
- China
- Prior art keywords
- antenna
- array
- machine learning
- antenna unit
- position information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000010801 machine learning Methods 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000013461 design Methods 0.000 claims abstract description 24
- 238000005457 optimization Methods 0.000 claims abstract description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 16
- 238000005070 sampling Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000005684 electric field Effects 0.000 claims description 4
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 238000003786 synthesis reaction Methods 0.000 claims description 3
- 230000008878 coupling Effects 0.000 abstract description 7
- 238000010168 coupling process Methods 0.000 abstract description 7
- 238000005859 coupling reaction Methods 0.000 abstract description 7
- 238000010586 diagram Methods 0.000 abstract description 7
- 238000003491 array Methods 0.000 abstract description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000001788 irregular Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000001808 coupling effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005672 electromagnetic field Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000001308 synthesis method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/20—Design reuse, reusability analysis or reusability optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Variable-Direction Aerials And Aerial Arrays (AREA)
Abstract
The invention discloses a quick modeling method of an antenna in an array environment assisted by machine learning, which learns the influence of the position information of an antenna unit in an array on the antenna performance such as a directional diagram, return loss and the like in the array of the unit by introducing the machine learning method, so that the performance of the antenna unit in any array arrangement can be quickly predicted. The method of the invention models the antenna units by taking the mutual coupling, the electromagnetic environment of the array, the platform influence and the like into consideration, and can be used for carrying out rapid optimization design on the directional diagram of the antenna array in the actual electromagnetic environment. The method can be used in the fields of beam forming design, low side lobe design and multi-beam design of antennas of different types and antenna arrays.
Description
Technical Field
The invention belongs to the technical field of antenna design, and relates to a quick antenna modeling method under an array environment assisted by machine learning.
Background
As a research problem focused on the professions of electromagnetic field, microwave technology, signal processing and the like, the rapid optimization and design of an antenna array are hot spots and difficulties which are attempted to be solved in academia. The conventional signal processing method aims at ideal antenna units, performs array synthesis by developing different mathematical methods, and can obtain good effects in a short time, but often fails to consider the problems of mutual coupling among the antenna units, electromagnetic influence of equipment platforms where the antennas are positioned and the like in practical application, so that the designed performance cannot be achieved in full-wave simulation and practical use. In recent years, algorithms considering mutual coupling and platform electromagnetic influence appear, but are generally only suitable for optimizing the amplitude phase or rotation angle of an antenna unit when fixing the position of the antenna unit, and the position of the antenna unit can be rarely considered as an optimization parameter, so that the optimization performance is greatly limited.
Over the past decade, machine learning methods have been widely introduced into the design field of electronic devices such as antennas, passive devices, and circuit designs, and have achieved very good results. Currently, most of the antenna designs assisted by machine learning only consider the design of antenna units, but cannot solve the problem of designing more complex antenna arrays. In recent years, some documents propose an optimization and design method of an antenna array assisted by machine learning, but all have the problems of overlarge required data set and the like, and are difficult to apply practically.
Therefore, how to adopt machine learning and obtain accurate prediction results of the antenna array by using a smaller data set, namely how to quickly and accurately model the performance of the antenna unit in an array environment, so that the design and optimization process of the array are the key points for solving the design of the actual antenna array.
Disclosure of Invention
The invention aims to: aiming at the problems, the invention provides a quick modeling method for an antenna under an array environment assisted by machine learning, so as to accelerate the optimal design of the antenna array.
The technical scheme is as follows: in order to achieve the above purpose, the present invention adopts the following technical scheme:
The fast modeling method of the antenna under the machine learning-assisted array environment takes the position information of the antenna unit as the input of a machine learning algorithm, takes the antenna pattern and return loss performance corresponding to the antenna unit as the output of the machine learning algorithm, learns a corresponding agent model with low calculation complexity, predicts the characteristics of the antenna unit under any antenna array arrangement, and then carries out fast design on the antenna array.
Further, the inputs to the machine learning algorithm include absolute position information of the antenna element in the device platform and relative position information between antenna elements adjacent to the antenna element.
Further, the absolute position information of the antenna unit in the equipment platform and the relative position information between the antenna units adjacent to the antenna unit refer to taking the inverse of the distance value as the input of a machine learning algorithm, so that the relative position information of the antenna unit at the edge of the antenna array is set to 0.
Further, the output of the machine learning algorithm includes amplitude and phase information of the electric field values of the antenna unit at different angles and return loss antenna characteristics.
Furthermore, the design and optimization process of the antenna array adopts any conventional optimization method of array synthesis based on the use of a proxy model, including a convex optimization algorithm, an intelligent optimization algorithm, a deterministic optimization algorithm and other common array optimization methods
A quick modeling method for an antenna in an array environment assisted by machine learning comprises the following steps:
(1) Modeling: the absolute and relative positions of the antenna elements are taken as inputs, the antenna patterns in the array are taken as outputs, and a machine learning method is utilized to build a proxy model. The specific steps of the method can be divided into:
(101) Sampling: and (3) sampling N groups of antenna array position parameter combinations by using a random sampling method such as Latin hypercube sampling and other algorithms, and simulating the array by using full-wave simulation software such as HFSS to obtain an in-array directional diagram (Embedded ELEMENT PATTERN, EEP) of the antenna units.
(102) Learning: the inverse of the absolute position (l 0) of all antenna units on the equipment platform and the inverse and the angles of the relative positions (l 1～lM) of the adjacent antenna units are used as input parameters of machine learning, EEP is used as an output parameter, and a machine learning method such as Gaussian Process machine learning (Gaussian Process MACHINE LEARNING, GPML) is used for learning, so that a proxy model R with low calculation complexity is obtained.
(103) And (3) predicting: the pattern of the antenna unit at any position in the antenna array can be predicted by using the constructed proxy model R.
The beneficial effects are that: compared with the prior art, the invention has the following beneficial effects:
The method provided by the invention models the antenna units by taking mutual coupling, electromagnetic environment of the array, platform influence and the like into consideration, and can be used for carrying out rapid optimal design on the directional diagram of the antenna array in the actual electromagnetic environment. The method can be used in the fields of beam forming design, low side lobe design and multi-beam design of antennas of different types and antenna arrays.
Drawings
Fig. 1 is a schematic diagram of an antenna array according to an embodiment of the present invention;
fig. 2 is a schematic diagram of an example of an antenna array used to validate the present invention;
FIG. 3 is a graph of actual values of the amplitude of an antenna element versus predicted values at different M for a typical set of array arrangements;
fig. 4 is a graph showing the actual phase values of an antenna element in a typical array arrangement and the predicted phase values of different M.
Detailed Description
The technical scheme of the invention is further described below with reference to the accompanying drawings and examples.
The embodiment of the invention discloses a quick modeling method for an antenna in an array environment assisted by machine learning, which considers the condition of a linear array, and has the design task of designing and optimizing the position, amplitude and phase characteristics of P antenna units so as to achieve the required antenna array pattern performance.
Fig. 1 shows a position diagram of an antenna array on a shaped device platform. For any antenna element a, the distance l 0 between its absolute position and the initially set origin of coordinates O is defined, and the relative position between it and other antenna elements is defined by the distance l m (m=1 to M) between it and other antenna elements, where: l m includes l l,m and l r,m,ll,m and l r,m, respectively, denote the distances from the M-th antenna element on the left and from the M-th antenna element on the right, M representing the maximum value of the number of adjacent antenna elements considered on one side of the antenna element. Considering that the mutual coupling effect between adjacent cells in the antenna array is strongest and weakens as the cell spacing increases, the M value may be set to 1 or 2 in practical applications, that is, only the mutual coupling between the antenna cell and the adjacent left and right 2 cells or the mutual coupling between the antenna cell and the adjacent left and right 4 cells is considered. Here, m=1 is taken.
Considering that for antenna elements at the edge of the array, only the left or right side has adjacent elements, and the inter-element distance between the sides of the adjacent elements is difficult to define, the reciprocal of the relative position and the absolute position is uniformly set as the input parameter of the proxy model, and for the above case, the reciprocal of the inter-element distance is set to 0.
In summary, for the antenna units in the antenna array, the input parameters of the corresponding proxy model are the inverse of the absolute position, the inverse of the relative position, and the angle values, and the output parameters are the electric field values of the EEPs at the corresponding angles. For the linear array under consideration, the electric field values in the plane of interest for the upper half plane are generally considered.
Under given unit spacing limitation and other limitation conditions, firstly sampling to obtain N groups of antenna array position parameter combinations, and simulating the array by using full-wave simulation software such as HFSS to obtain EEP of the antenna units; and then taking the reciprocal of the absolute position (l 0) of all antenna units on the equipment platform and the reciprocal and the angle of the relative position (l 1～lM) of the adjacent antenna units as input parameters of machine learning, setting the term value to 0 in the condition of no adjacent units, taking EEP of the corresponding angle as output parameters, and learning the EEP by using a machine learning method such as GPML to obtain a proxy model R with low computational complexity.
By using the obtained agent model, the position and amplitude of the antenna array can be optimized by adopting a traditional array synthesis method or an intelligent optimization algorithm. After the result is obtained by optimization, the result of the accurate antenna array pattern can be obtained by using simulation software such as HFSS and the like, so as to judge whether the design index is met. If the data set is not satisfied, the obtained result can be added into the data set to relearn the agent model, so that a more accurate agent model is obtained, and the next step of optimization design is performed.
Considering the microstrip antenna array shown in fig. 2 and located on the surface of the circuit board with an irregular shape, the circuit board of the microstrip antenna array contains irregular openings and via structures, the microstrip antenna units work at 10GHz, the number of the antenna units is P, and the antenna array is arranged along the X-axis direction. Taking p=6 to 16, n=11×5=55, namely sampling 5 times for each unit number, and m=0 to 4, taking 4 times of the data sets of each unit number, adding the training set, and adding the rest 1 times into the test set. The training set is randomly sampled by 1/20, learning is carried out by GPML, a proxy model between the position information of the antenna unit and the performance of the antenna unit is established, the obtained proxy model is used for predicting the test set, the prediction result is compared with the actual simulation result to calculate Root Mean Square Error (RMSE), and the obtained result is shown in table 1. Fig. 3 and 4 show the actual values of the amplitude and phase of an antenna element and the predicted values for different M for a typical set of array arrangements. It can be seen that when m=0, that is, the relative position information of the antenna units is not used, the prediction bias of the proxy model is large, and when m=1 to 4, that is, the relative position information of the antenna units in the array is considered, the prediction bias of the proxy model is greatly reduced; in addition, it can be seen that when m=1, that is, only the mutual coupling between the antenna unit and the adjacent left and right 2 units is considered, the prediction accuracy of the proxy model can reach a quite high level. The foregoing choice of m=1 is quite reasonable in view of the reduction in machine learning time that can be brought about by the reduction in the data amount.
Table 1 shows the Root-mean-square error (RMSE) calculated by predicting the test set using the obtained proxy model and comparing the predicted result with the actual simulation result in the present invention.
TABLE 1
| RMSE | Pattern amplitude | Pattern phase | Return loss of |
| M＝0 | 1.0690 | 0.1222 | 0.2450 |
| M＝1 | 0.2592 | 0.0409 | 0.1040 |
| M＝2 | 0.2413 | 0.0418 | 0.1022 |
| M＝3 | 0.2418 | 0.0405 | 0.0828 |
| M＝4 | 0.2494 | 0.0397 | 0.0828 |
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (3)
1. A quick modeling method of an antenna under an array environment assisted by machine learning is characterized by comprising the following steps of: the method comprises the steps of taking position information of an antenna unit as input of a machine learning algorithm, taking an antenna pattern and return loss performance corresponding to the antenna unit as output of the machine learning algorithm, learning a corresponding agent model with low calculation complexity, predicting characteristics of the antenna unit under any antenna array arrangement, and designing an antenna array;
the input of the machine learning algorithm comprises absolute position information of an antenna unit in the equipment platform and relative position information between antenna units adjacent to the antenna unit;
The absolute position information of the antenna unit in the equipment platform and the relative position information between the antenna units adjacent to the antenna unit are obtained by taking the inverse of the distance value as the input of a machine learning algorithm, so that the relative position information of the antenna unit at the edge of the antenna array is set to be 0;
the output of the machine learning algorithm contains amplitude and phase information of the electric field values of the antenna elements at different angles and return loss antenna characteristics.
2. The method for rapid modeling of antennas in a machine-learning aided array environment of claim 1, wherein: the design and optimization process of the antenna array adopts any conventional optimization method of array synthesis on the basis of using a proxy model.
3. The method for rapid modeling of antennas in a machine learning aided array environment of claim 1, comprising the steps of:
S1, sampling: sampling N groups of antenna array position parameter combinations by using a random sampling method, and simulating the array by using full-wave simulation software to obtain a pattern EEP in the array of the antenna units;
S2, learning: taking the reciprocal of the absolute position l 0 of all antenna units on the equipment platform and the reciprocal and angle of the relative position l 1～lM of the antenna units adjacent to the reciprocal as input parameters of machine learning, taking the pattern EEP in the array of the antenna units as output parameters, and learning by using a machine learning method to obtain a proxy model R with low calculation complexity;
s3, predicting: and predicting the pattern of the antenna unit at any position in the antenna array by using the proxy model R established in the step S2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110417710.4A CN113051836B (en) | 2021-04-19 | 2021-04-19 | Rapid modeling method for antenna under machine learning-assisted array environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110417710.4A CN113051836B (en) | 2021-04-19 | 2021-04-19 | Rapid modeling method for antenna under machine learning-assisted array environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113051836A CN113051836A (en) | 2021-06-29 |
| CN113051836B true CN113051836B (en) | 2024-04-26 |
Family
ID=76519827
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110417710.4A Active CN113051836B (en) | 2021-04-19 | 2021-04-19 | Rapid modeling method for antenna under machine learning-assisted array environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113051836B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086531A (en) * | 2018-08-07 | 2018-12-25 | 中南大学 | Antenna design method neural network based |
| CN111985150A (en) * | 2020-07-06 | 2020-11-24 | 东南大学 | Multilayer electronic device robustness optimization design method using machine learning auxiliary optimization |
-
2021
- 2021-04-19 CN CN202110417710.4A patent/CN113051836B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086531A (en) * | 2018-08-07 | 2018-12-25 | 中南大学 | Antenna design method neural network based |
| CN111985150A (en) * | 2020-07-06 | 2020-11-24 | 东南大学 | Multilayer electronic device robustness optimization design method using machine learning auxiliary optimization |
Non-Patent Citations (2)
| Title |
|---|
| 合成孔径雷达抗干扰技术综述;黄岩 等;雷达学报;20200305;86-106 * |
| 毫米波无线通信平面圆极化天线与阵列关键技术研究;无奇;全文博士论文全文数据库;20190227;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113051836A (en) | 2021-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104992001B (en) | The accurate quick calculation method of extensive MIMO array Antenna Far Field radiation field | |
| CN110728034B (en) | Antenna rapid multi-target modeling method using multistage cooperative machine learning | |
| Yuan et al. | Multibranch artificial neural network modeling for inverse estimation of antenna array directivity | |
| CN112329303B (en) | Array antenna electromagnetic characteristic solving method based on finite element region decomposition | |
| CN110059422B (en) | Electromagnetic scattering characteristic simulation method for frequency selective surface curved surface radome | |
| CN110737873B (en) | Rapid analysis method for scattering of large-scale array antenna | |
| CN107302140A (en) | Planar antenna array Sparse methods based on quantum spider group's mechanism of Evolution | |
| Fu et al. | An efficient surrogate assisted particle swarm optimization for antenna synthesis | |
| CN105718662B (en) | Characteristics of conformal array antenna structure tolerance fast determination method based on mechanical-electric coupling | |
| CN107918696B (en) | Multi-field coupling analysis method for phased array antenna and computer program | |
| US9568624B2 (en) | Method for predicting electromagnetic radiation characteristics, computer-readable recording medium and simulator | |
| CN113051836B (en) | Rapid modeling method for antenna under machine learning-assisted array environment | |
| Feng et al. | Parallel decomposition approach to gradient-based EM optimization | |
| Ma et al. | Transfer learning for the behavior prediction of microwave structures | |
| CN116562143A (en) | Antenna topology and parameter mixing optimization method based on normalized Gaussian network | |
| CN115146544A (en) | Array antenna design method adopting knowledge and data hybrid driving | |
| CN114491992B (en) | Efficient electromagnetic scattering method based on equivalent dipole moment and physical optical method | |
| CN107305536A (en) | Mix the discontinuous Jia Lvejin methods of rank time domain | |
| Dang et al. | A fast modeling algorithm for quasi-periodic array | |
| CN111709192B (en) | Planar inverted F antenna resonant frequency prediction method based on semi-supervised learning | |
| Zhao et al. | An improved full-wave multilevel Green’s function interpolation method with RBF-QR technique for fast field evaluation | |
| Simsek et al. | Space mapping with inverse difference technique for reconfigurable antenna design problem | |
| Cheng et al. | Thinning and weighting of planar/conformal arrays considering mutual coupling effects | |
| Rahul et al. | Electromagnetic modeling of microstrip patch antennas: a neural network approach | |
| Li et al. | Parallelized Multilevel Characteristic Basis function Method (MLCBFM) combined with Adaptive Cross Approximation (ACA) for the analysis of the scattering from electrically large rough surfaces |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |