CN116341360A - Antenna topology pixel optimization design method adopting machine learning auxiliary optimization - Google Patents
Antenna topology pixel optimization design method adopting machine learning auxiliary optimization Download PDFInfo
- Publication number
- CN116341360A CN116341360A CN202310044507.6A CN202310044507A CN116341360A CN 116341360 A CN116341360 A CN 116341360A CN 202310044507 A CN202310044507 A CN 202310044507A CN 116341360 A CN116341360 A CN 116341360A
- Authority
- CN
- China
- Prior art keywords
- antenna
- optimization
- algorithm
- topological
- machine learning
- 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.)
- Pending
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 57
- 238000013461 design Methods 0.000 title claims abstract description 32
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000010801 machine learning Methods 0.000 title claims abstract description 19
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 35
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 12
- 238000012795 verification Methods 0.000 claims abstract description 4
- 238000013528 artificial neural network Methods 0.000 claims abstract 2
- 239000011159 matrix material Substances 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 6
- 239000002184 metal Substances 0.000 claims description 6
- 238000004088 simulation Methods 0.000 claims description 5
- 230000004044 response Effects 0.000 claims description 4
- 239000002245 particle Substances 0.000 claims description 2
- 230000005855 radiation Effects 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000005672 electromagnetic field Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Medical Informatics (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Variable-Direction Aerials And Aerial Arrays (AREA)
Abstract
The invention discloses an antenna topology pixel optimization design method adopting machine learning auxiliary optimization. The method is based on an iterative machine learning aided optimization algorithm architecture, a convolutional neural network is introduced in each iteration of an algorithm, a proxy model between pixelated antenna topology and antenna performance is established, a binary evolution algorithm is introduced to optimize the proxy model, then an optimization result is verified, whether the algorithm is terminated or not is judged, if not, the optimization and verification result is added into a data set to retrain the neural network, and the next algorithm iteration is carried out. Compared with the traditional antenna topological pixel optimization method only based on the evolutionary algorithm, the method greatly improves the convergence rate of optimization and realizes the great improvement of the optimization result in the same time. The method can be used in the field of topology optimization of antennas of different types.
Description
Technical Field
The invention belongs to the technical field of antenna design, and relates to an antenna topology pixel optimization design method adopting machine learning auxiliary optimization.
Background
As a research problem focused on the professions of electromagnetic field and microwave technology, communication and information systems and the like, the topological optimization design of the antenna is always a hot spot and a difficult point of academic research. The traditional antenna topology optimization method can be divided into an antenna structure topology optimization design based on an evolutionary algorithm and an antenna structure topology optimization design based on a gradient algorithm. The method has the advantages that the method is easy to integrate with commercial software, has strong expandability, does not need sensitivity information, can search the globally optimal topological structure as far as possible, and has long time consumption and high requirement on computing resources; the method can effectively improve the solving efficiency of the antenna topological structure optimization problem, but is easy to sink into local optimization, and is difficult to integrate with commercial software due to the fact that sensitivity information is needed.
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. At present, most of antenna designs assisted by machine learning only consider parameter designs of antennas after antenna topology fixing, but cannot be applied to topology optimization designs of antennas.
Disclosure of Invention
The invention aims to: aiming at the problems, the invention provides an antenna topological pixel optimization design method adopting machine learning auxiliary optimization, which realizes great convergence acceleration compared with the traditional antenna topological pixel optimization design method based on an evolutionary algorithm.
The technical scheme is as follows: in order to achieve the above purpose, the present invention adopts the following technical scheme:
an antenna topological pixel optimization design method adopting machine learning auxiliary optimization comprises the following steps:
(1) Initial sampling: the topological area range for antenna design is first determined and divided into m x n sub-blocks, each corresponding to an element in a binary m x n matrix. The following mapping relation is adopted: the element of the matrix with 0 is mapped to air and 1 is mapped to metal. Thus, the topology of the planar antenna can be represented by constructing a binary m×n matrix. By means of random sampling, uniform sampling or mixed sampling, k binary m×n matrixes are obtained and defined as input parameters X of the data set.
(2) Full wave simulation builds a dataset: converting the sample X in the step (1) into an antenna topological structure model, and calculating by using full-wave simulation software such as HFSS to obtain a corresponding antenna performance response set, wherein the corresponding antenna performance response set is defined as an output parameter Y of a data set. Thereby constructing a data set M consisting of the input parameter X and the output parameter Y together.
(3) Convolutional neural network training: and training the data set M by introducing a convolutional neural network and a learning method thereof, and learning the relation between input parameters and output parameters to obtain a proxy model R. For the common antenna topology optimization design problem, a convolutional neural network structure with a small number of layers should be selected so as to reduce the consumption of training time. A typical convolutional neural network structure is shown in fig. 2 and includes two convolutional layers, two activation function layers, two pooling layers, and a fully-connected layer.
(4) Binary evolution algorithm optimization: binary evolutionary algorithms, such as binary particle swarm algorithm (BinaryParticle Swarm Optimization, BPSO), are introduced to optimize elements in a binary m n matrix.
The fitness function may be set to reflect, radiate, etc. or a combination of properties of the antenna. The antenna being at different inputs X i Performance under Y i And (3) predicting the agent model R obtained through training in the step (3).
Because the agent model is adopted to optimize the antenna topology, the calculation time required by the traditional optimization process based on the evolutionary algorithm is greatly reduced. The optimized matrix is defined as X 1 The prediction performance of the corresponding agent model is Y 1 。
(5) Full-wave simulation verification: for matrix X obtained by optimization 1 Converting the antenna topology structure into an antenna topology structure model, and calculating by using full-wave simulation to obtain the real performance Y of the verified antenna 2 。
(6) Judging whether a termination condition is satisfied: the termination condition may be set to achieve a maximum iteration number limit or the true performance Y of the antenna 2 The optimization objective has been met; if the termination condition is met, the algorithm is ended, and if the termination condition is not met, the antenna topology X obtained by the iteration is obtained 1 And the true performance Y of the corresponding antenna 2 And (3) adding the data set M, namely updating the data set, returning to the step (3), and retraining the convolutional neural network.
Compared with the prior art, the invention has the following beneficial effects:
the method provided by the invention can greatly improve the convergence rate of the topological pixel optimization design of the traditional antenna structure based on the evolutionary algorithm, and improve the performance of the finally optimized antenna structure. The method can be used in the field of topological optimization design of different types of antenna structures.
Drawings
FIG. 1 is a block diagram of an algorithm of an antenna topology pixel optimization design method using machine learning aided optimization as described in the present invention;
FIG. 2 is a block diagram of an exemplary convolutional neural network used in the present invention;
FIG. 3 is a schematic diagram of a side-fed microstrip patch antenna optimized for an embodiment of the present invention;
FIG. 4 is a graph comparing the convergence speed of the optimization method described in the present invention with that of the conventional BPSO algorithm in the implementation example;
fig. 5 is a schematic diagram showing the comparison of the antenna performance obtained by optimizing the optimization method of the present invention in the implementation example with the initial value.
Detailed Description
The technical scheme of the invention is further described below with reference to the accompanying drawings and examples.
The invention discloses an antenna adopting machine learning to assist in optimizationA topological pixel optimization design method. Consider a typical three-frequency side-fed microstrip patch antenna design having a structure as shown in fig. 3, comprising a layer of height h=1.2 mm and width w 1 Length l =18 mm 1 Dielectric plate with dielectric constant of 4.4, bottom width w =34 mm 1 Length l 5 Metal floor of 10mm, top consisting of microstrip feed line and one rectangular topologically optimised design area. The width of the microstrip feed line is w 3 Length l =2 mm 3 =14 mm. Width w of region to be optimized 2 Length l =14 mm 2 =18 mm, which is divided into m×n sub-blocks, where m=10, n=8. An example of mapping the antenna topology to a matrix X is given in fig. 3, where grey squares represent metal, 1 in the matrix, white squares represent air, 0 in the matrix. The design objective of the antenna is set to be S in the frequency band of 2.35-2.45,3.45-3.55,5.15-5.25,5.75-5.85GHz 11 The value of is the smallest.
In this example, the convolutional neural network structure adopted in the step (3) is shown in fig. 2, and the convolutional layer, the activation function layer, the pooling layer, the convolutional layer, the activation function layer, the pooling layer and the full connection layer are respectively arranged from front to back. The parameters of the first layer of convolution layer comprise square convolution kernels with the size of 3, the number of the convolution kernels is 16, the zero filling range is set to be 2 in the upper, lower, left and right directions, the parameters of the second layer of convolution layer comprise square convolution kernels with the size of 3, the number of the convolution kernels is 64, and the zero filling range is kept unchanged; the activation function layers all adopt ReLU activation functions; the pooling layers adopt a maximum pooling mode, and the pooling core size is 2; batch normalization layers are used between the convolution layer and the activation function layer.
In this example, step (4) uses a classical BPSO optimization algorithm, whose optimization targets are set as:
wherein f c In order to take into account the frequency points,representing the amplitude of the reflection coefficient of the antenna predicted by the convolutional neural network at all frequency points under the matrix X corresponding to the given antenna topology, wherein the optimization task is to find the matrix X corresponding to the proper antenna topology, so that ∈>Even though its worst value is the smallest.
Fig. 4 shows a comparison of the machine learning aided optimization algorithm of the present invention with a classical BPSO algorithm. Both algorithms were performed 5 times each and the decreasing curves of their fitness functions were averaged. Compared with a classical BPSO algorithm, the antenna topology optimization design method for machine learning auxiliary optimization can greatly accelerate the iteration process. At the iteration to 500 full wave simulations, the algorithm presented by the present invention can reach the in-band worst reflection coefficient of-13.29 dB, whereas classical BPSO can only reach-7.91 dB.
Fig. 5 shows a typical reflection coefficient plot of the optimization result, which can be seen to meet the reflection coefficient requirement of-10 dB in all three frequency bands, compared to the initial value (i.e., all-metal patch of matrix all 1).
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 (5)
1. An antenna topological pixel optimization design method adopting machine learning auxiliary optimization is characterized in that an iterative machine learning auxiliary optimization algorithm architecture is adopted; the algorithm maps the topological structure of the antenna into a matrix formed by 1 and 0 by a method of mapping metal into 1 and mapping air into 0; in the initial step of the algorithm, randomly sampling to generate a series of antenna topological structures represented by a matrix formed by 1 and 0, calculating to obtain antenna performances corresponding to the topological structures by using full-wave simulation software, thereby establishing an initial training set, and entering an iterative flow of the algorithm; introducing a convolutional neural network in each iteration of an algorithm, training based on a training set, and establishing a proxy model mapping between antenna topology and antenna performance; further introducing a binary evolution algorithm to optimize the proxy model to obtain an antenna topological structure capable of optimizing the antenna performance; and then full-wave simulation verification is carried out on the optimization result, whether the algorithm is terminated is judged, if not, the optimization and verification result is added into the data set, the neural network is retrained, and the next algorithm iteration is carried out.
2. The method for optimizing design of antenna topology pixels using machine learning aided optimization of claim 1, wherein the step of constructing an initial training set comprises: firstly, determining a topological area range for antenna design, dividing the topological area range into m multiplied by n subblocks, and respectively corresponding to elements in a binary m multiplied by n matrix; mapping elements in the matrix with 0 as air and 1 as metal; representing the topology structure of the planar antenna by constructing a binary m×n matrix; obtaining k binary m multiplied by n matrixes by using a random sampling, uniform sampling or mixed sampling method, and defining the matrix as an input parameter X of a data set; converting X into an antenna topological structure model, calculating by using full-wave simulation software to obtain a corresponding antenna performance response set, and defining the corresponding antenna performance response set as an output parameter Y of a data set; an initial data set M is thus constructed, which is composed of both the input parameters X and the output parameters Y.
3. The method for optimizing design of topological pixels of an antenna by adopting machine learning auxiliary optimization according to claim 1, wherein the convolutional neural network comprises two convolutional layers, two activation function layers, two pooling layers and one full-connection layer.
4. The method for optimizing design of topological pixels of an antenna by adopting machine learning auxiliary optimization according to claim 1, wherein the binary evolution algorithm is a binary particle swarm algorithm, and the fitness function is set as the combination of reflection, radiation performance and the like of the antenna.
5. The method for optimizing design of topological pixels of an antenna using machine learning aided optimization of claim 2, wherein the algorithm termination condition is to reach a maximum iteration number limit or the true performance Y of the antenna 2 The optimization objective has been met; if the termination condition is met, the algorithm is ended, and if the termination condition is not met, the antenna topology X obtained by the iteration is obtained 1 And the true performance Y of the corresponding antenna 2 And adding the data set M, namely updating the data set, and retraining the convolutional neural network.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310044507.6A CN116341360A (en) | 2023-01-30 | 2023-01-30 | Antenna topology pixel optimization design method adopting machine learning auxiliary optimization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310044507.6A CN116341360A (en) | 2023-01-30 | 2023-01-30 | Antenna topology pixel optimization design method adopting machine learning auxiliary optimization |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116341360A true CN116341360A (en) | 2023-06-27 |
Family
ID=86893684
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310044507.6A Pending CN116341360A (en) | 2023-01-30 | 2023-01-30 | Antenna topology pixel optimization design method adopting machine learning auxiliary optimization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116341360A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117892252A (en) * | 2024-03-18 | 2024-04-16 | 浙江威星电子系统软件股份有限公司 | Intelligent park operation management platform based on big data |
-
2023
- 2023-01-30 CN CN202310044507.6A patent/CN116341360A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117892252A (en) * | 2024-03-18 | 2024-04-16 | 浙江威星电子系统软件股份有限公司 | Intelligent park operation management platform based on big data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Dong et al. | Fast multi-objective optimization of multi-parameter antenna structures based on improved BPNN surrogate model | |
Zhang et al. | Surrogate-assisted quasi-Newton enhanced global optimization of antennas based on a heuristic hypersphere sampling | |
Koziel et al. | Efficient multi-objective simulation-driven antenna design using co-kriging | |
CN109086531B (en) | Antenna design method based on neural network | |
Yuan et al. | Multibranch artificial neural network modeling for inverse estimation of antenna array directivity | |
CN108984985B (en) | Antenna structure design method based on neural network | |
CN111199126B (en) | Quick optimization design method for chip antenna | |
CN110232212B (en) | Antenna design optimization method and system based on bat algorithm | |
CN111985150B (en) | Robust optimization design method for multilayer electronic device by utilizing machine learning auxiliary optimization | |
Zhang et al. | Antenna design by an adaptive variable differential artificial bee colony algorithm | |
CN116341360A (en) | Antenna topology pixel optimization design method adopting machine learning auxiliary optimization | |
CN116562143A (en) | Antenna topology and parameter mixing optimization method based on normalized Gaussian network | |
Koziel et al. | On nature-inspired design optimization of antenna structures using variable-resolution EM models | |
Singh et al. | Microstrip line fed dielectric resonator antenna optimization using machine learning algorithms | |
Ding et al. | Antenna Optimization Based on Auto‐Context Broad Learning System | |
Jin et al. | Efficient radiation pattern prediction of array antennas based on complex-valued graph neural networks | |
Nandal et al. | Fuzzy enhanced image fusion using pixel intensity control | |
Mohammadjafari et al. | Designing mm-wave electromagnetic engineered surfaces using generative adversarial networks | |
CN116595867A (en) | Agent model aided antenna design optimization method based on variable space data constraint | |
CN109117545B (en) | Neural network-based antenna rapid design method | |
Zhang et al. | Nonuniform antenna array design by parallelizing three-parent crossover genetic algorithm | |
Koziel et al. | Cost-efficient electromagnetic-simulation-driven antenna design using co-Kriging | |
CN114021484A (en) | Antenna simulation design optimization method based on CNN stack width learning system | |
CN115719027A (en) | Method for realizing antenna design and related equipment | |
Pietrenko-Dabrowska et al. | Dimensionality-reduced antenna modeling with stochastically established constrained domain |
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 |