CN108920841B - Antenna design method - Google Patents
Antenna design method Download PDFInfo
- Publication number
- CN108920841B CN108920841B CN201810729086.XA CN201810729086A CN108920841B CN 108920841 B CN108920841 B CN 108920841B CN 201810729086 A CN201810729086 A CN 201810729086A CN 108920841 B CN108920841 B CN 108920841B
- Authority
- CN
- China
- Prior art keywords
- antenna
- neural network
- antenna design
- model
- input
- 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
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Neurology (AREA)
- Variable-Direction Aerials And Aerial Arrays (AREA)
Abstract
The invention discloses an antenna design method, which comprises the steps of constructing an antenna initial model; selecting an input sample to input into an antenna initial model to obtain an antenna model response; adopting a neural network structure to simulate to obtain an antenna agent model; constructing antenna design parameter variables and an objective function; inputting the antenna design parameter variables into an antenna proxy model to obtain a response and calculating a target function value; and judging the objective function value to obtain the final antenna design parameter. The method adopts a hybrid optimization algorithm to jointly optimize the network structure and the initial structure parameters of the neural network, simplifies the neural network structure, reduces the calculation cost of the neural network, improves the prediction precision of the neural network, then utilizes the neural network as a proxy model to fit the electromagnetic simulation data of the antenna design parameter sample, replaces the electromagnetic simulation with great time consumption to realize the instantaneous approximate calculation from the antenna structure parameters to the electromagnetic response, reduces the electromagnetic simulation times, greatly reduces the calculation cost and obviously improves the antenna design efficiency.
Description
Technical Field
The invention particularly relates to an antenna design method.
Background
With the development of economic technology and the improvement of the living standard of people, communication becomes an essential link in production and life of people.
As an energy conversion device between a guided electromagnetic wave and a free space electromagnetic wave, an antenna has wide applications in the fields of mobile communication, radar, satellite communication, and the like. The development of modern wireless communication systems not only requires antennas that are lightweight, low cost, easy to manufacture and easy to integrate, but also places unprecedented demands on miniaturized, broadband, multi-band, conformal and integrated designs for antennas.
The design of a conventional antenna is generally based on a regular structure, and the design experience and the physical measurement and debugging of an antenna engineer are combined by using the existing empirical formula. By adopting the antenna design method, the antenna design period is often longer; more importantly, the conventional antenna design methods are unable to design antennas with irregular structures, novel structures and high performance requirements. And when the antenna structure with multiple parameters and multiple targets is optimally designed, the design process is long, and the optimization capability and efficiency are poor.
Intelligent optimization algorithms can be considered as simple and general objective optimization strategies, typically mimicking various biological or social phenomena (e.g., population intelligence, genetic processes, etc.). These algorithms enable automatic adjustment of antenna structure parameters and simultaneous optimization of multiple design goals. Nevertheless, with the benefit of population-based intelligent optimization algorithms, one significant drawback of such algorithms is that the optimization process requires a significant number of model evaluations. A single evaluation of a real antenna model usually takes several minutes to tens of minutes, and in practical applications, there are often more than one evaluation models, so the calculation cost is very high, which also hinders the direct application of an intelligent optimization algorithm in the design process, and at the same time indirectly leads to the development of various strategies aiming at reducing the calculation cost. On the other hand, the problem of high computational cost may be partially solved with large-scale computing resources in the form of supercomputers with multiple CPU or GPU units and multiple secondary computing design software (in particular EM solvers) licenses. However, such hardware configurations are not widely used, they provide a very low acceleration-to-cost ratio, and are therefore not practical.
Disclosure of Invention
The invention aims to provide an antenna design method which can greatly reduce the calculation cost during antenna design so as to ensure the high efficiency and reliability of the antenna design.
The antenna design method provided by the invention comprises the following steps:
s1, constructing an antenna initial model according to design requirements of an antenna;
s2, selecting a plurality of groups of antenna design parameter values in the antenna design space as input samples, and inputting the samples into the antenna initial model obtained in the step S1 so as to obtain antenna model responses corresponding to the input samples;
s3, simulating the mapping relation between the input sample obtained in the step S2 and the antenna model response by adopting a neural network, so as to obtain a corresponding antenna agent model;
s4, constructing a plurality of groups of antenna design parameter variables, and constructing a plurality of antenna design objective functions according to antenna design requirements;
s5, inputting the antenna design parameter variables obtained in the step S4 into the antenna proxy model obtained in the step S3, so as to obtain responses corresponding to each group of antenna design parameters, and calculating objective function values corresponding to each antenna design parameter according to the obtained responses;
s6, judging the objective function value obtained in the step S5:
if the objective function value meets the preset requirement, the antenna design parameter variable corresponding to the objective function value is determined as the final antenna design parameter;
if all the objective function values do not meet the preset requirements, generating a plurality of new antenna design parameter variables, and repeating the steps S5-S6 until the objective function values meet the preset requirements, thereby obtaining the final antenna design parameters.
And S2, selecting a plurality of groups of antenna design parameter values in the antenna design space, specifically selecting a plurality of groups of antenna design parameter values in the antenna design space by adopting a Latin hypercube sampling method.
And S2, inputting the samples into the antenna initial model to obtain the antenna model response corresponding to each input sample, specifically, performing simulation solution on the antenna initial model to which the samples are input by using an electromagnetic simulation tool to obtain the antenna model response corresponding to each input sample.
And S3, simulating the mapping relation between the obtained input sample and the antenna model response by adopting the neural network, specifically, optimizing the neural network structure and parameters by adopting a hybrid optimization algorithm according to the input sample and the antenna model response corresponding to the input sample, so as to obtain the neural network structure capable of simulating the input sample and the antenna model response corresponding to the input sample, and taking the neural network structure as a final antenna agent model.
The method adopts a hybrid optimization algorithm to optimize the neural network structure and parameters, and specifically adopts the following steps to optimize:
A. respectively determining the number n of input neurons of the neural network according to the antenna design parameter variables and the corresponding response vectors thereof i And number of output neurons n o ；
B. Determining a number n of hidden layer neurons of a neural network h ；
C. Respectively encoding the neural network structure and the initial structure parameters, and initializing a mixed particle swarm;
D. constructing a fitness function f for representing a prediction error between a prediction response and a real response of the neural network structure to the input sample;
E. calculating a fitness function value of each mixed particle;
G. and optimizing the neural networks corresponding to the neuron data of different hidden layers to obtain the prediction error of the neural networks, and selecting the neural network with the minimum prediction error as a final antenna agent model.
And C, respectively encoding the neural network structure and the initial structure parameters and initializing the mixed particle swarm, specifically, encoding and initializing by adopting the following steps:
(1) Binary coding is carried out on the neural network structure, real number coding is carried out on the initial structure parameters of the neural network, and the dimensionality d of binary and real number particles is calculated by adopting the following formula:
d＝n i ×n h +n h +n h ×n o +n o
in the formula n i Number of input neurons being a neural network, n o To export the neuron number, n h Hiding the number of layer neurons for the neural network;
(2) Generating d-dimensional binary particles representing a link switch between an input layer and a hidden layer of the neural network, a link switch between the hidden layer and an output layer and link thresholds of the hidden layer and the output layer, wherein 1 represents that the link exists, and 0 represents that the link does not exist;
(3) Generating real number particles in d dimension (0, 1) representing weights between input layer and hidden layer of neural network, weights between hidden layer and output layer, and thresholds of hidden layer and output layer;
(4) And sequentially arranging the real number particles and the binary particles to form 2 d-dimensional mixed particles representing the neural network structure and the initial structure parameters, and initializing the mixed particle swarm.
And D, constructing a fitness function f, specifically adopting the following formula to construct the fitness function f:
where err is the absolute mean error and
where q is the number of input samples and k represents 1 to n 0 Index of variable between, Y k (t) is the response value of each set of input samples, y k And (t) is the predicted response value of the neural network for each set of input samples.
Step E, calculating the fitness function value of each mixed particle, specifically, calculating by using the following steps:
1) Correspondingly multiplying the d-dimensional real number part of each mixed particle with the binary system part to obtain a structural parameter of the neural network, and then constructing the neural network by using the structural parameter;
2) And inputting each group of input samples serving as input data into the neural network to obtain a prediction response vector corresponding to the input samples, and solving a prediction error of the neural network on the input samples, wherein the prediction error is a fitness function value.
And G, optimizing the neural networks corresponding to the neuron data of different hidden layers, specifically, optimizing by using a Hybrid real-binary particle swarm optimization (HPSO) algorithm.
The neural network is a BP neural network, a perceptron neural network or a linear neural network; the hybrid optimization algorithm is a hybrid particle swarm algorithm or a hybrid Taguchi genetic algorithm; and S6, generating a plurality of new antenna design parameter variables, specifically, generating a plurality of new antenna design parameter variables by adopting a multi-target intelligent algorithm, wherein the multi-target intelligent algorithm is a multi-target evolutionary algorithm based on decomposition, a non-dominated sorting evolutionary algorithm, a multi-target genetic algorithm or a multi-target particle swarm algorithm.
According to the antenna design method provided by the invention, the network structure and the initial structure parameters of the neural network are jointly optimized by adopting a hybrid optimization algorithm, the neural network structure is simplified, the calculation cost of the neural network is reduced, the prediction precision of the neural network is improved, then the neural network is used as a proxy model to fit the electromagnetic simulation data of the antenna design parameter sample, the electromagnetic simulation with great time consumption is replaced to realize the instantaneous approximate calculation from the antenna structure parameters to the electromagnetic response, the electromagnetic simulation times are reduced, and the calculation cost is greatly reduced; the method effectively combines a multi-target intelligent algorithm, an agent model and antenna design, can obviously improve the antenna design efficiency, and particularly solves the problem of complex high-dimensional multi-target antenna design, and has more obvious advantages.
Drawings
FIG. 1 is a process flow diagram of the process of the present invention.
Fig. 2 is a schematic diagram of an initial antenna model constructed in accordance with an embodiment of the invention.
Fig. 3 is a return loss curve diagram of 6 antennas meeting design objectives designed according to an embodiment of the present invention.
Detailed Description
FIG. 1 shows a flow chart of the method of the present invention: the antenna design method provided by the invention comprises the following steps:
s1, constructing an antenna initial model according to design requirements of an antenna;
s2, selecting a plurality of groups of antenna design parameters in an antenna design space as input samples (the selection can be performed by adopting a Latin hypercube sampling method), inputting the samples into the antenna initial model obtained in the step S1, and performing simulation solution on the antenna initial model with the input samples by adopting an electromagnetic simulation tool so as to obtain antenna model responses corresponding to the input samples (the responses are all performance indexes of the antenna, including an antenna return loss value, gain or standing-wave ratio and the like);
s3, simulating the mapping relation between the input sample obtained in the step S2 and the antenna model response by adopting a neural network, so as to obtain a corresponding antenna agent model; specifically, a neural network structure and parameters are optimized by adopting a hybrid optimization algorithm according to an input sample and an antenna model response corresponding to the input sample, so that the neural network structure capable of simulating the input sample and the antenna model response corresponding to the input sample is obtained, and the neural network structure is used as a final antenna agent model; specifically, the method comprises the following steps:
A. respectively determining the number n of input neurons of the neural network according to the antenna design parameter variables and the corresponding response vectors thereof i And number of output neurons n o ；
B. Determining a number n of hidden layer neurons of a neural network h ；
C. Respectively encoding the neural network structure and the initial structure parameters, and initializing a mixed particle swarm; specifically, the following steps are adopted for coding and initializing:
(1) Binary coding is carried out on the neural network structure, real number coding is carried out on the initial structure parameters of the neural network, and the dimensionality d of binary and real number particles is calculated by adopting the following formula:
d＝n i ×n h +n h +n h ×n o +n o
in the formula n i Number of input neurons being a neural network, n o To export the neuron number, n h Number of hidden layer neurons for neural network;
(2) Generating d-dimensional binary particles representing a link switch between an input layer and a hidden layer of the neural network, a link switch between the hidden layer and an output layer and link thresholds of the hidden layer and the output layer, wherein 1 represents that the link exists, and 0 represents that the link does not exist;
(3) Generating real number particles in d dimension (0, 1) representing weights between input layer and hidden layer of the neural network, weights between the hidden layer and output layer, and thresholds of the hidden layer and output layer;
(4) Sequentially arranging real number particles and binary particles to form 2 d-dimensional mixed particles representing a neural network structure and initial structure parameters, and initializing mixed particle groups;
D. constructing a fitness function f for representing a prediction error between a prediction response and a real response of the neural network structure to the input sample; specifically, a fitness function f is constructed by adopting the following formula:
where err is the absolute average error and
where q is the number of input samples and k is 1 to n 0 Index of variable in between, Y k (t) is the response value of each set of input samples, y k (t) is the predicted response value of the neural network for each set of input samples;
E. calculating a fitness function value of each mixed particle; specifically, the following steps are adopted for calculation:
1) Correspondingly multiplying the d-dimensional real number part of each mixed particle with the binary part to obtain the structural parameters of the neural network, and then constructing the neural network by using the structural parameters;
2) Inputting each group of input samples as input data into a neural network to obtain a prediction response vector corresponding to the input samples, and solving a prediction error of the neural network on the input samples, wherein the prediction error is a fitness function value;
G. optimizing neural networks corresponding to different hidden layer neuron data (for example, optimizing by using a Hybrid real-binary particle swarm optimization (HPSO) algorithm) to obtain a prediction error of the neural network, and selecting the neural network with the minimum prediction error as a final antenna agent model;
in specific implementation, the neural network can adopt a BP neural network, a perceptron neural network or a linear neural network and the like; the hybrid optimization algorithm can adopt a hybrid particle swarm algorithm or a hybrid Taguchi genetic algorithm and the like;
s4, constructing a plurality of groups of antenna design parameter variables, and constructing a plurality of antenna design objective functions according to antenna design requirements;
s5, inputting the antenna design parameter variables obtained in the step S4 into the antenna proxy model obtained in the step S3, so as to obtain responses corresponding to each group of antenna design parameters, and calculating objective function values corresponding to each antenna design parameter according to the obtained responses;
s6, judging the objective function value obtained in the step S5:
if the objective function value meets the preset requirement, the antenna design parameter variable corresponding to the objective function value is determined as the final antenna design parameter;
if all the objective function values do not meet the preset requirements, generating a plurality of new antenna design parameter variables, and repeating the steps S5-S6 until the objective function values meet the preset requirements, thereby obtaining the final antenna design parameters; the method can be used for generating a plurality of new antenna design parameter variables by adopting a multi-target intelligent algorithm, wherein the multi-target intelligent algorithm is a multi-target evolutionary algorithm based on decomposition, a non-dominated sorting evolutionary algorithm, a multi-target genetic algorithm or a multi-target particle swarm algorithm.
The process of the invention is further illustrated below with reference to a specific example:
the target is to design a two-target plane multi-band antenna; the neural network adopts a BP neural network model, the Hybrid Particle Swarm Optimization (HPSO) algorithm is adopted in the hybrid optimization algorithm, the multi-objective intelligent algorithm adopts a multi-objective evolutionary algorithm (MOEA/D) based on decomposition, and the HFSS is adopted in the electromagnetic simulation tool.
The specific design process is as follows:
as shown in fig. 2, the design space of the antenna model, i.e. the constraint condition of the antenna model, is the size limit of 10 antenna parameters, as shown in table 1 below:
TABLE 1 antenna modeling constraints (unit: mm)
| Parameter(s) | L | L1 | L2 | L3 | L4 |
| Range | [36.4,40] | [16,19] | [10,12.5] | [8.5,10.5] | [2.8,3.9] |
| Parameter(s) | L5 | W | W1 | W2 | g |
| Range | [9.5,11.5] | [19,24] | [6.5,8.3] | [8.7,11.2] | [1.8,2.1] |
200 groups of antenna design parameter variables are selected as input samples in an antenna design space by using a Latin hypercube sampling method, an electromagnetic simulation tool is called to solve response vectors of all groups of antenna design parameter variables, namely return loss values of 15 frequency sampling points are used as output samples, and 200 groups of antenna design parameter variables and return loss values of all frequency sampling points corresponding to the antenna design parameter variables form a sample set for constructing a proxy model.
Respectively determining the number n of input neurons and output neurons of the BP neural network proxy model according to the antenna design parameter variable and the return loss values of the corresponding frequency sampling points i ＝10，n o ＝15；
Determining a range of the number of hidden layer neurons of the BP neural network proxy model [10,20];
calculating the binary and real particle dimensions d =25 xn h +15, respectively initializing d-dimensional binary and real number particles, combining the d-dimensional binary and real number particles into mixed particles, and initializing particle swarms;
constructing a fitness function of an HPSO algorithm optimized BP neural network structure and initial structure parameters:
Y k (t) solving the return loss value y of each group of antenna design parameter variables by calling an electromagnetic simulation tool to simulate k (t) the return loss values of each set of antenna design parameter variables predicted by using the BP neural network surrogate model;
correspondingly multiplying the real part of each mixed particle with the binary part to obtain structural parameters of a non-fully-connected neural network proxy model, constructing the non-fully-connected BP neural network proxy model by using the structural parameters and a sample set, inputting each group of antenna design parameter variables into the non-fully-connected BP neural network proxy model as input data, predicting to obtain response vectors of the antenna design parameter variables, and solving prediction errors of the proxy model on the antenna design parameter variables;
using HPSO to optimize the non-fully-connected BP neural network proxy models corresponding to different hidden layer neuron numbers 1000 times and outputting prediction errors, and selecting the proxy model with the minimum error as the plane multiband antenna;
randomly initializing 40 sets of antenna design parameter variables x in antenna design space 1 ,x 2 ,...,x 40 The method comprises the following steps of (1) constructing 2 antenna design targets according to antenna design requirements as an initial population of an MOEA/D algorithm;
objective function 1: return loss value S in three frequency bands of 2.40-2.60GHz, 3.30-3.80GHz and 5.00-5.90 GHz 11 ＜-10dB；
Where n is the number of sampling points in the above-mentioned 3 frequency bands, f i Is the frequency of the sampling point in the frequency band, S 11 (f i ) Is a frequency f i A return loss value of (d);
the objective function 2: the size of the antenna;
F 2 ＝w×l
respectively taking 40 groups of antenna design parameter variables as input values, calling a non-fully-connected BP neural network model to predict return loss values of each frequency sampling point of each group of antenna design parameter variables, and solving an objective function value F according to the return loss values 1 Solving the objective function value F according to the design parameters 2 ；
Judging whether the objective function value obtained by the solution in the step 5 meets the antenna design requirement, if so, entering a step 7, otherwise, generating 40 new groups of antenna design parameter variables by utilizing MOEA/D (metal oxide arrester/metal oxide arrester) updating, and returning to the step 5 until the antenna design parameters meeting the design requirement are obtained or the iteration frequency set by MOEA/D is reached;
if the antenna design result meets 2 antenna design targets, the iteration is ended.
The design parameters obtained by applying the method of the invention are shown in table 2, the reflection curve graphs of the 6 antennas meeting the design target are shown in fig. 3, the return loss values of the antennas under different area parameters in three frequency bands of 2.33-2.63 GHz, 3.17-3.92 GHz and 4.97-5.99 GHz are all less than-10 dB, and the design performance requirements of the antennas are met.
Table 2 table of 6 antenna sizes designed to meet design objectives
| Design of | x (1) | x (2) | x (3) | x (4) | x (5) | x (6) |
| F 1 (dB) | -13.26 | -12.61 | -12.40 | -12.05 | -11.73 | -11.21 |
| F 2 (mm) | 885.92 | 839.22 | 825.10 | 794.22 | 777.48 | 726.93 |
| L | 39.2 | 39.4 | 37.0 | 36.6 | 37.2 | 36.9 |
| L1 | 17.2 | 18.3 | 18.9 | 16.8 | 17.4 | 18.2 |
| L2 | 11.1 | 10.6 | 10.3 | 10.7 | 12.5 | 12.2 |
| L3 | 9.2 | 9.9 | 9.2 | 9.5 | 8.9 | 9.4 |
| L4 | 3.6 | 3.4 | 3.5 | 3.1 | 3.0 | 3.1 |
| L5 | 11.3 | 10.3 | 10.9 | 11.4 | 11.5 | 10.5 |
| W | 22.6 | 21.3 | 22.3 | 21.7 | 20.9 | 19.7 |
| W1 | 8.1 | 6.9 | 8.2 | 7.2 | 6.7 | 6.5 |
| W2 | 10.5 | 9.8 | 10.4 | 10.1 | 9.6 | 9.1 |
| g | 1.8 | 1.9 | 2.0 | 2.0 | 2.0 | 1.8 |
And then, designing the antenna by respectively using a traditional electromagnetic simulation (EM) design method, MOEA/D combined BP network models only optimizing structural parameters and MOEA/D combined non-fully-connected BP neural network models, wherein the total calculation cost comparison result of the antenna is shown in a table 3.
Table 3 comparison of computational costs for antenna design methods
Finally, for 6 groups of antenna design parameter variables obtained by design, predicting response values and calculating an objective function F by using a BP neural network model direct prediction (prediction result 1) and a non-full-connection BP neural network model respectively 1 (prediction result 2), and then directly uses the simulation response value to calculate the target function F 1 The error rate comparison is shown in Table 4.
TABLE 4 precision comparison of the prediction methods
| Design of | x (1) | x (2) | x (3) | x (4) | x (5) | x (6) |
| Prediction result 1 | -15.03 | -13.93 | -13.95 | -11.07 | -13.76 | -10.11 |
| Prediction result 2 | -13.26 | -12.61 | -12.40 | -12.05 | -11.73 | -11.21 |
| Simulation result | -13.12 | -12.88 | -12.68 | -12.24 | -12.07 | -11.71 |
| |
14.56％ | 8.15％ | 10.02％ | 9.56％ | 14.00％ | 13.66 |
| Error rate | ||||||
| 2 | 1.83％ | 2.10％ | 2.21％ | 1.56％ | 2.82％ | 4.27％ |
Claims (10)
1. An antenna design method includes the following steps:
s1, constructing an antenna initial model according to design requirements of an antenna;
s2, selecting a plurality of groups of antenna design parameters in an antenna design space as input samples, and inputting the samples into the antenna initial model obtained in the step S1 so as to obtain antenna model responses corresponding to the input samples;
s3, simulating the mapping relation between the input sample obtained in the step S2 and the antenna model response by adopting a neural network, so as to obtain a corresponding antenna agent model;
s4, constructing a plurality of groups of antenna design parameter variables, and constructing a plurality of antenna design objective functions according to antenna design requirements;
s5, inputting the antenna design parameter variables obtained in the step S4 into the antenna proxy model obtained in the step S3, so as to obtain responses corresponding to each group of antenna design parameters, and calculating objective function values corresponding to each antenna design parameter according to the obtained responses;
s6, judging the objective function value obtained in the step S5:
if the objective function value meets the preset requirement, the antenna design parameter variable corresponding to the objective function value is determined as the final antenna design parameter;
if all the objective function values do not meet the preset requirements, generating a plurality of new antenna design parameter variables, and repeating the steps S5-S6 until the objective function values meet the preset requirements, thereby obtaining the final antenna design parameters.
2. The method of claim 1, wherein the selecting step S2 selects a plurality of antenna design parameters in the antenna design space, specifically selecting a plurality of antenna design parameters in the antenna design space by using a latin hypercube sampling method.
3. The antenna design method according to claim 1, wherein the samples are input into the antenna initial model in step S2 to obtain the antenna model response corresponding to each input sample, and specifically, an electromagnetic simulation tool is used to perform simulation solution on the antenna initial model into which the samples are input to obtain the antenna model response corresponding to each input sample.
4. The antenna design method according to one of claims 1 to 3, wherein the input samples and the corresponding antenna model responses thereof are simulated by using a neural network structure in step S3, specifically, the neural network structure and parameters are optimized by using a hybrid optimization algorithm according to the input samples and the corresponding antenna model responses thereof, so as to obtain the neural network structure capable of simulating the input samples and the corresponding antenna model responses thereof, and the neural network structure is used as a final antenna proxy model.
5. The antenna design method according to claim 4, wherein the optimization of the neural network structure and parameters by the hybrid optimization algorithm is specifically performed by the following steps:
A. respectively determining the number n of input neurons of the neural network according to the antenna design parameter variables and the corresponding response vectors thereof i And number of output neurons n o ；
B. Determining the number n of hidden layer neurons of a neural network h ；
C. Respectively encoding the neural network structure and the initial structure parameters, and initializing a mixed particle swarm;
D. constructing a fitness function f for representing a prediction error between a prediction response and a real response of the neural network structure to the input sample;
E. calculating a fitness function value of each mixed particle;
G. and optimizing the neural networks corresponding to the neuron data of different hidden layers to obtain the prediction error of the neural networks, and selecting the neural network with the minimum prediction error as a final antenna agent model.
6. The antenna design method according to claim 5, wherein the step C of encoding the neural network structure and the initial structure parameters and initializing the hybrid particle swarm respectively comprises the following steps:
(1) Binary coding is carried out on the neural network structure, real number coding is carried out on initial structure parameters of the neural network, and the dimensionality d of binary and real number particles is calculated by adopting the following formula:
d＝n i ×n h +n h +n h ×n o +n o
in the formula n i Number of input neurons being a neural network, n o Number of output neurons, n h Hiding the number of layer neurons for the neural network;
(2) Generating d-dimensional binary particles representing a link switch between an input layer and a hidden layer of the neural network, a link switch between the hidden layer and an output layer and link thresholds of the hidden layer and the output layer, wherein 1 represents that the link exists, and 0 represents that the link does not exist;
(3) Generating real number particles in d dimension (0, 1) representing weights between input layer and hidden layer of the neural network, weights between the hidden layer and output layer, and thresholds of the hidden layer and output layer;
(4) And sequentially arranging the real number particles and the binary particles to form 2 d-dimensional mixed particles representing the neural network structure and the initial structure parameters, and initializing the mixed particle swarm.
7. The antenna design method according to claim 5, wherein the constructing of the fitness function f in step D is specifically to construct the fitness function f by using the following formula:
where err is the absolute average error and
where q is the number of input samples and k is 1 to n 0 Index of variable between, Y k (t) is the response value of each set of input samples, y k And (t) is the predicted response value of the neural network for each set of input samples.
8. The antenna design method according to claim 5, wherein the fitness function value of each hybrid particle is calculated in step E by:
1) Correspondingly multiplying the d-dimensional real number part of each mixed particle with the binary part to obtain the structural parameters of the neural network, and then constructing the neural network by using the structural parameters;
2) And inputting each group of input samples serving as input data into the neural network to obtain a prediction response vector corresponding to the input samples, and solving a prediction error of the neural network on the input samples, wherein the prediction error is a fitness function value.
9. The antenna design method according to claim 5, wherein the neural networks corresponding to the different hidden layer neuron data in step G are optimized, specifically, optimized by using HPSO algorithm.
10. The antenna design method according to claim 4, wherein the neural network is a BP neural network, a perceptron neural network or a linear neural network; the hybrid optimization algorithm is a hybrid particle swarm algorithm or a hybrid Taguchi genetic algorithm; and S6, generating a plurality of new antenna design parameter variables, specifically, generating a plurality of new antenna design parameter variables by adopting a multi-target intelligent algorithm, wherein the multi-target intelligent algorithm is a multi-target evolutionary algorithm based on decomposition, a non-dominated sorting evolutionary algorithm, a multi-target genetic algorithm or a multi-target particle swarm algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810729086.XA CN108920841B (en) | 2018-07-05 | 2018-07-05 | Antenna design method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810729086.XA CN108920841B (en) | 2018-07-05 | 2018-07-05 | Antenna design method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108920841A CN108920841A (en) | 2018-11-30 |
| CN108920841B true CN108920841B (en) | 2023-02-14 |
Family
ID=64425570
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810729086.XA Active CN108920841B (en) | 2018-07-05 | 2018-07-05 | Antenna design method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108920841B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111368384B (en) * | 2018-12-07 | 2022-06-10 | 华为技术有限公司 | Method and equipment for predicting antenna engineering parameters |
| CN110765723B (en) * | 2019-11-15 | 2022-06-10 | 苏州浪潮智能科技有限公司 | Routing modeling optimization method and device based on BP neural network |
| CN111460665B (en) * | 2020-04-01 | 2023-12-22 | 江苏科技大学 | Antenna design method based on kriging agent model and multi-target genetic algorithm |
| CN112199770B (en) * | 2020-09-17 | 2024-04-09 | 中国运载火箭技术研究院 | Test agent model construction method and system based on test parameter characteristics |
| CN113065268A (en) * | 2021-03-30 | 2021-07-02 | 中国科学院光电技术研究所 | Rapid multi-objective space reflector optimization design method based on proxy model |
| CN113642168B (en) * | 2021-08-09 | 2024-07-05 | 佛山科学技术学院 | Antenna design method based on structural parameter prediction and crowd optimization |
| CN118153449B (en) * | 2024-04-02 | 2024-08-30 | 电子科技大学 | Antenna pattern robust design method based on sequential optimization |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106021813A (en) * | 2016-06-12 | 2016-10-12 | 中南大学 | Fast multi-target antenna design method based on dynamic updating of agent model |
| CN107357962A (en) * | 2017-06-19 | 2017-11-17 | 西安电子科技大学 | A kind of antenna house rib cross-sectional size optimization method based on Adaptive proxy model |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6973622B1 (en) * | 2000-09-25 | 2005-12-06 | Wireless Valley Communications, Inc. | System and method for design, tracking, measurement, prediction and optimization of data communication networks |
-
2018
- 2018-07-05 CN CN201810729086.XA patent/CN108920841B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106021813A (en) * | 2016-06-12 | 2016-10-12 | 中南大学 | Fast multi-target antenna design method based on dynamic updating of agent model |
| CN107357962A (en) * | 2017-06-19 | 2017-11-17 | 西安电子科技大学 | A kind of antenna house rib cross-sectional size optimization method based on Adaptive proxy model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108920841A (en) | 2018-11-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108920841B (en) | Antenna design method | |
| CN109271695B (en) | Multi-target antenna design method based on neural network | |
| CN109086531B (en) | Antenna design method based on neural network | |
| CN106021813B (en) | A kind of fast multi-target antenna design method based on dynamic renewal agency model | |
| CN108984985B (en) | Antenna structure design method based on neural network | |
| CN111460665B (en) | Antenna design method based on kriging agent model and multi-target genetic algorithm | |
| Wang et al. | Improved prediction method of PV output power based on optimised chaotic phase space reconstruction | |
| CN113688960A (en) | Grey wolf optimization GHFCM-based residential power data clustering method and device | |
| CN113962145A (en) | Parameter uncertainty quantitative modeling method under interval data sample condition | |
| CN109117545B (en) | Neural network-based antenna rapid design method | |
| CN111651936A (en) | FOA-GRNN-based modeling design method for ultra-wideband antenna with double trap characteristics | |
| Kashani et al. | A new metaheuristic approach to task assignment problem in distributed systems | |
| Huang et al. | Microscale searching algorithm for coupling matrix optimization of automated microwave filter tuning | |
| Bao et al. | Restricted Boltzmann Machine‐Assisted Estimation of Distribution Algorithm for Complex Problems | |
| CN104376227A (en) | Method for achieving remote control and maintenance assembly sequence planning of parts in radiation environment | |
| Liu et al. | Automatic coupler design using data-driven-based modular neural network | |
| CN117313542A (en) | Antenna pattern prediction method based on residual convolution neural network model | |
| CN111967184A (en) | Sequence model-based multi-target antenna design method | |
| CN114372539B (en) | Machine learning framework-based classification method and related equipment | |
| BHARADWAJ et al. | Invasive weed optimization for economic dispatch with valve point effects | |
| Wu et al. | AI Deep Learning Optimization for Compact Dual-Polarized High-Isolation Antenna Using Backpropagation Algorithm | |
| CN109002877A (en) | Multiple dimensioned collaboration differential evolution optimization method | |
| Qin et al. | Fast multi-objective antenna design based on BPNN surrogate model | |
| CN114239439A (en) | Automatic filter design method based on tribal algorithm | |
| Gong et al. | Application of CNN-LSTM based hybrid neural network in power load forecasting |
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 |