CN118395877B - Method, device, equipment and medium for optimally designing microstrip patch antenna - Google Patents
Method, device, equipment and medium for optimally designing microstrip patch antenna Download PDFInfo
- Publication number
- CN118395877B CN118395877B CN202410814332.7A CN202410814332A CN118395877B CN 118395877 B CN118395877 B CN 118395877B CN 202410814332 A CN202410814332 A CN 202410814332A CN 118395877 B CN118395877 B CN 118395877B
- Authority
- CN
- China
- Prior art keywords
- parameter
- antenna
- model
- determining
- microstrip patch
- 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
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000001228 spectrum Methods 0.000 claims abstract description 82
- 238000013461 design Methods 0.000 claims abstract description 59
- 238000005457 optimization Methods 0.000 claims abstract description 43
- 230000002441 reversible effect Effects 0.000 claims abstract description 20
- 238000012545 processing Methods 0.000 claims description 35
- 230000015654 memory Effects 0.000 claims description 34
- 230000003595 spectral effect Effects 0.000 claims description 27
- 230000007246 mechanism Effects 0.000 claims description 25
- 238000012549 training Methods 0.000 claims description 21
- 230000007774 longterm Effects 0.000 claims description 16
- 238000004422 calculation algorithm Methods 0.000 claims description 15
- 238000003860 storage Methods 0.000 claims description 11
- 238000013507 mapping Methods 0.000 claims description 8
- 230000009467 reduction Effects 0.000 claims description 8
- 230000006403 short-term memory Effects 0.000 claims description 6
- 230000007787 long-term memory Effects 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 abstract description 9
- 230000000694 effects Effects 0.000 abstract description 6
- 238000004904 shortening Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 9
- 238000004891 communication Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 238000004836 empirical method Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 5
- 238000003062 neural network model Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000005855 radiation Effects 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000005094 computer simulation Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 230000005672 electromagnetic field Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
- 230000005428 wave function Effects 0.000 description 1
Landscapes
- Waveguide Aerials (AREA)
Abstract
The application discloses an optimization design method, a device, equipment and a medium of a microstrip patch antenna, which relate to the technical field of antenna design, and the method comprises the following steps: when antenna parameters are received, determining the parameter types corresponding to the antenna parameters; determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model; the antenna parameters are processed based on the forward prediction model or the reverse inversion model, and the target S parameter spectrum curve or the target structure parameter of the antenna is determined, so that the technical problems that the error of predicting the performance parameter curve of the antenna based on the structure parameter through experimental data in the related technology is large and the accuracy is low are effectively solved, the technical effects of shortening the antenna design period and improving the antenna design efficiency and accuracy are achieved.
Description
Technical Field
The present application relates to the field of antenna design technologies, and in particular, to a method, an apparatus, a device, and a medium for optimally designing a microstrip patch antenna.
Background
In the field of wireless communication, an antenna is a device for transmitting and receiving electromagnetic waves, the performance of which directly affects the quality of the entire communication system, and the design of the antenna has attracted more and more attention. In the face of rapid development of wireless communication, not only are various novel antennas which meet design indexes required to be provided, but also a scheme capable of rapidly and accurately predicting the antenna performance or designing the antenna structure is required to be provided, which clearly presents great challenges for antenna designers.
In the related art, the antenna design method mainly adopted mainly comprises an empirical method and a numerical method, and the empirical method is mainly used for carrying out antenna design based on experience of a designer and previous successful cases and is suitable for some simple antenna structures or common application scenes; the numerical method utilizes computer simulation and numerical calculation technology to carry out numerical solution on the electromagnetic field of the antenna, thereby designing the antenna meeting the requirements. Common numerical methods include finite element methods, time-domain integral equation methods, time-domain finite difference methods, and the like. These methods are mainly used for designing relatively simple antenna structures. However, these methods have limitations in terms of long design cycle, low accuracy, and the like. Based on this, as deep learning technology is developed, it is beginning to be applied to the field of antenna design. For example, machine learning is used for the design of multiband rectangular helical microstrip antennas, rectangular microstrip patch antennas, and the like. These methods are mostly limited to deriving the performance parameter curves of the antennas from the structural parameters.
However, these methods are limited to predicting the performance parameter curve of the antenna based on the structural parameters by experimental data, and the experimental error is large and the accuracy is low.
Disclosure of Invention
The embodiment of the application solves the technical problems of larger error and lower accuracy of predicting the performance parameter curve of the antenna based on the structural parameters by experimental data in the related technology by providing the optimized design method, the device, the equipment and the medium of the microstrip patch antenna, and realizes the technical effects of shortening the antenna design period and improving the antenna design efficiency and accuracy.
The embodiment of the application provides an optimal design method of a microstrip patch antenna, which comprises the following steps:
when antenna parameters are received, determining the parameter types corresponding to the antenna parameters;
Determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model;
And processing the antenna parameters based on the forward prediction model or the inverse inversion model, and determining a target S parameter spectrum curve or a target structure parameter of the antenna.
Optionally, the step of determining the parameter type corresponding to the antenna parameter when the antenna parameter is received includes:
When receiving an input value corresponding to an antenna parameter, judging the input value;
And determining the parameter type of the antenna parameter as a structural parameter or an S-parameter spectrum curve according to the judging result.
Optionally, the step of determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model includes:
when the parameter type is a structural parameter, determining that the forward prediction model is a long-period memory model and short-period memory model mixed attention mechanism model;
And when the parameter type is an S-parameter spectrum curve, determining the inverse inversion model as a converter model.
Optionally, the step of determining the target S-parameter spectral curve or the target structural parameter of the antenna by processing the antenna parameter based on the forward prediction model or the inverse inversion model includes:
Determining the target S parameter spectrum curve of the antenna based on the forward prediction model processing structure parameters; or alternatively, the first and second heat exchangers may be,
And processing an S-parameter spectrum curve based on the inverse inversion model, and determining target structure parameters of the antenna.
Optionally, the step of determining the target S-parameter spectral curve of the antenna based on the forward prediction model processing structure parameters includes:
Capturing a long-term dependency relationship in the structural parameters based on a gating unit in the long-term and short-term memory model;
determining the time dependence and parameter information corresponding to the structural parameters based on the long-term dependence;
Weighting and summing the hidden states of the long-term dependency relationship based on an attention mechanism, and extracting attention features;
Predicting the target S-parameter spectral curve of the antenna based on the time dependence, the parameter information and the feature of interest.
Optionally, the step of determining the target structural parameter of the antenna based on the inverse inversion model processing S-parameter spectral curves includes:
Taking the S parameter spectrum curve as an input sequence of an encoder in a converter model;
Performing relevance modeling on the input sequence based on a self-attention mechanism, and determining parameter characteristics;
the target structural parameters are determined based on mapping the parametric features to a structural parameter space of the antenna by an output layer of the converter model.
Optionally, after the step of determining the target S-parameter spectral curve or the target structural parameter of the antenna, processing the antenna parameter based on the forward prediction model or the inverse inversion model includes:
Performing data dimension reduction on the antenna parameters and extracting key features;
Learning a nonlinear relationship between the structural parameters and antenna performance based on the key features and a back propagation algorithm;
Training the inverse inversion model based on the nonlinear relationship.
In addition, the application also provides an optimal design device of the microstrip patch antenna, which comprises:
The selection module is used for determining the parameter type corresponding to the antenna parameter when the antenna parameter is received; determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model;
The forward spectrum curve prediction module is used for processing structural parameters based on a forward prediction model and determining the target S parameter spectrum curve of the antenna;
and the inverse structure parameter inversion module is used for processing the S-parameter spectrum curve based on the inverse inversion model and determining the target structure parameter of the antenna.
In addition, the application also provides an optimal design device of the microstrip patch antenna, which comprises a memory, a processor and an optimal design program of the microstrip patch antenna, wherein the optimal design program of the microstrip patch antenna is stored in the memory and can run on the processor, and the processor realizes the steps of the optimal design method of the microstrip patch antenna when executing the optimal design program of the microstrip patch antenna.
In addition, the application also provides a computer readable storage medium, wherein the computer readable storage medium is stored with an optimization design program of the microstrip patch antenna, and the optimization design program of the microstrip patch antenna realizes the steps of the optimization design method of the microstrip patch antenna when being executed by a processor.
One or more technical solutions provided in the embodiments of the present application at least have the following technical effects or advantages:
Due to the adoption of the method, when the antenna parameters are received, the parameter types corresponding to the antenna parameters are determined; determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model; the antenna parameters are processed based on the forward prediction model or the reverse inversion model, and the target S parameter spectrum curve or the target structure parameter of the antenna is determined, so that the technical problems that the error of predicting the performance parameter curve of the antenna based on the structure parameter through experimental data in the related technology is large and the accuracy is low are effectively solved, the technical effects of shortening the antenna design period and improving the antenna design efficiency and accuracy are achieved.
Drawings
Fig. 1 is a schematic flow chart of a first embodiment of a method for optimizing a microstrip patch antenna according to the present application;
Fig. 2 is a schematic flow chart of step S210 in a second embodiment of the method for optimally designing a microstrip patch antenna of the present application;
Fig. 3 is a schematic flow chart of a third embodiment of the method for optimizing a microstrip patch antenna of the present application;
fig. 4 is a schematic diagram of an overall process involved in the optimal design of the microstrip patch antenna in the third embodiment of the method for optimizing a microstrip patch antenna of the present application;
Fig. 5 is a schematic diagram of a hardware structure involved in an embodiment of the microstrip patch antenna optimization design device of the present application.
Detailed Description
In the related art, the antenna design method mainly adopted mainly comprises an empirical method and a numerical method, and the empirical method is mainly used for carrying out antenna design based on experience of a designer and previous successful cases and is suitable for some simple antenna structures or common application scenes; the numerical method utilizes computer simulation and numerical calculation technology to carry out numerical solution on the electromagnetic field of the antenna, thereby designing the antenna meeting the requirements. Common numerical methods include finite element methods, time-domain integral equation methods, time-domain finite difference methods, and the like. These methods are mainly used for designing relatively simple antenna structures. However, these methods have limitations in terms of long design cycle, low accuracy, and the like. Based on this, as deep learning technology is developed, it is beginning to be applied to the field of antenna design. For example, machine learning is used for the design of multiband rectangular helical microstrip antennas, rectangular microstrip patch antennas, and the like. These methods are mostly limited to deriving the performance parameter curves of the antennas from the structural parameters. However, these methods are limited to predicting the performance parameter curve of the antenna based on the structural parameters by experimental data, and the experimental error is large and the accuracy is low. The embodiment of the application adopts the main technical scheme that: when antenna parameters are received, determining the parameter types corresponding to the antenna parameters;
Determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model;
And processing the antenna parameters based on the forward prediction model or the inverse inversion model, and determining a target S parameter spectrum curve or a target structure parameter of the antenna. Therefore, the technical effects of shortening the antenna design period and improving the antenna design efficiency and accuracy are achieved.
In order to better understand the above technical solution, exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the application to those skilled in the art.
The embodiment of the application discloses an optimal design method of a microstrip patch antenna, and referring to fig. 1, the optimal design method of the microstrip patch antenna comprises the following steps:
Step S110, when antenna parameters are received, determining the parameter types corresponding to the antenna parameters.
In this embodiment, the antenna parameters are related parameters of the microstrip patch antenna, and are classified into structural parameters and spectral curves according to the types of the parameters. Microstrip patch antennas are one type of commonly used planar antenna structure. The device consists of a metal patch and a grounding plate, wherein the metal patch and the grounding plate are separated by an insulating medium. The structural parameters include the size, shape and location of the patch and the thickness of the medium. The spectral profile of a typical microstrip patch antenna may vary according to different design parameters, and generally includes center frequency, bandwidth, and radiation characteristics. The structural parameters include: patch size: the length, width and thickness of the patch; patch shape: common shapes are rectangular, circular, oval, etc.; patch position: the position of the patch relative to the ground plate. The spectral profile is characterized by: center frequency: the working frequency center of the microstrip patch antenna; bandwidth: a frequency range that the antenna is capable of covering; standing wave ratio: the method is used for describing the matching condition of the input port, and a lower standing wave ratio indicates better matching performance; radiation characteristics: the gain, radiation pattern, radiation power and other relevant parameters of the antenna in different directions.
As an alternative embodiment, step S110 includes: when receiving an input value corresponding to an antenna parameter, judging the input value; and determining the parameter type of the antenna parameter as a structural parameter or a frequency spectrum curve according to the judging result.
Inputting structural parameters or S parameters of the microstrip patch antenna into the model, judging the input values, and if the input values are the structural parameters of the microstrip patch antenna, predicting a forward S parameter spectrum curve; otherwise, if the input is the S-parameter spectrum curve, the inverse structure parameter inversion is performed. Wherein the S parameter refers to a scattering parameter (SCATTERING PARAMETERS), also referred to as a network parameter or a matrix description parameter. The S parameter is used to describe the relationship between voltage and current between input and output of a circuit or device at a particular frequency, and is an important parameter in the transmission of electromagnetic waves. For microstrip patch antennas, the common S parameters are S11 and S21, which represent the relationship between the incident wave and the reflected wave and the relationship between the incident wave and the transmitted wave, respectively. S11 represents the reflection coefficient at the antenna input port, i.e. the proportion of the signal emitted from the antenna input port that is partially reflected back. S21 represents the transmission coefficient at the antenna input port, i.e., the proportion of the signal transmitted from the antenna input port that is transmitted through the antenna to the output port.
Step S120, determining a forward prediction model or a reverse inversion model based on the parameter type, where the forward prediction model is a hybrid model.
In this embodiment, the forward prediction model is configured to determine, according to the input structural parameter, an S-parameter spectrum curve of the antenna as a target S-parameter spectrum curve. The inverse inversion model is used for inversely inverting the structural parameters of the antenna according to the input S-parameter spectrum curve.
As an optional implementation manner, when the parameter type is a structural parameter, determining that the forward prediction model is a long-period and short-period memory model mixed attention mechanism model; and when the parameter type is an S-parameter spectrum curve, determining the inverse inversion model as a converter model.
The parameter types, such as the structural parameters and the S-parameter spectral curves, are determined. If the parameter type is a structural parameter, the forward prediction model is a Long Short Term Memory (LSTM) model mixed attention mechanism model. The model combines an LSTM model and an attention mechanism, can predict input structural parameters, and outputs the relevant performance of the antenna. If the parameter type is an S-parameter spectrum curve, the inverse inversion model is a converter model. The model can convert the input S-parameter spectrum curve into the structural parameters of the antenna, thereby realizing inverse inversion. The forward prediction model is an LSTM model combined with the Attention model, and the inverse inversion model is a transducer model.
And step S130, processing the antenna parameters based on the forward prediction model or the inverse inversion model, and determining a target S parameter spectrum curve or a target structure parameter of the antenna.
In the present embodiment, for the forward prediction model (long-short term memory model mixed attention mechanism model): input: known antenna structure parameters, prediction: processing the input structural parameters by using a hybrid model to obtain a target S parameter spectrum curve of the antenna, and outputting: target S-parameter spectral curve of the antenna.
For the inverse inversion model (converter model): input: s-parameter spectral curves of known antennas, inversion: processing the input S-parameter spectrum curve by using a converter model to obtain target structural parameters of the antenna, and outputting: target structural parameters of the antenna.
Due to the adoption of the method, when the antenna parameters are received, the parameter types corresponding to the antenna parameters are determined; determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model; the antenna parameters are processed based on the forward prediction model or the reverse inversion model, and the target S parameter spectrum curve or the target structure parameter of the antenna is determined, so that the technical problems that the error of predicting the performance parameter curve of the antenna based on the structure parameter through experimental data in the related technology is large and the accuracy is low are effectively solved, the technical effects of shortening the antenna design period and improving the antenna design efficiency and accuracy are achieved.
Based on the first embodiment, the second embodiment of the present application provides an optimization design method for a microstrip patch antenna, and step S130 includes:
step S210, determining the target S parameter spectrum curve of the antenna based on the forward prediction model processing structure parameters; or alternatively, the first and second heat exchangers may be,
And step S220, processing an S-parameter spectrum curve based on the inverse inversion model, and determining target structure parameters of the antenna.
In this embodiment, a known antenna structure parameter data set is collected, or a structure parameter is determined according to an input value, including information such as the size and the material of the antenna, and a corresponding target S-parameter spectral curve data set. The data set is preprocessed, including data cleaning, normalization and the like, so that training and prediction of the model are facilitated. And processing the structural parameters of the antenna by using a forward prediction model (such as a long-short-term memory model mixed attention mechanism model in deep learning), and predicting a target S-parameter spectrum curve of the antenna by learning the relation between the structural parameters and the target S-parameter spectrum curve. And performing model evaluation and tuning, performing model training by using a training set and a verification set, performing performance evaluation on a test set, and obtaining a more accurate prediction result by adjusting parameters and structures of the model. The S-parameter spectral curve data set of the known antenna is collected, or the S-parameter spectral curve is determined from the input values, while the corresponding structural parameter data set of the antenna is also required. And performing data preprocessing, including data cleaning, normalization and the like, so as to facilitate training and inversion of the model. And processing the S-parameter spectrum curve by using a reverse inversion model (such as a converter model), and inverting the target structure parameter of the antenna by learning the relation between the S-parameter spectrum curve and the structure parameter. And performing model evaluation and tuning, performing model training by using a training set and a verification set, performing performance evaluation on a test set, and obtaining a more accurate inversion result by adjusting the super-parameters and the structure of the model.
Optionally, referring to fig. 2, step S210 includes:
Step S211, capturing a long-term dependency relationship in the structural parameters based on a gating unit in the long-term and short-term memory model;
step S212, determining the time dependence and parameter information corresponding to the structural parameters based on the long-term dependence;
step S213, carrying out weighted summation on the hidden states of the long-term dependency relationship based on an attention mechanism, and extracting attention features;
Step S214 predicts the target S-parameter spectral curve of the antenna based on the time dependence, the parameter information and the feature of interest.
As an alternative embodiment, the sequence of structural parameters is input into a Long Short Term Memory (LSTM) model as input. The gating unit inside the LSTM model can capture long-term dependency in the input sequence, and can selectively memorize and forget part of history information in a self-adaptive mode. Training is carried out by using an LSTM model, and long-term dependence in the structural parameter sequence is learned. And acquiring hidden state information of each moment according to the trained LSTM model. Attention weights are calculated for these hidden states to capture important time dependencies. And multiplying the hidden state of each moment by the corresponding attention weight in the time dimension and summing to obtain the attention feature. The hidden states are weighted summed using an attention mechanism to extract features of interest. The attention mechanism may assign different weights according to the importance of the hidden state to capture key information in long-term dependencies. The feature of interest is combined with the time-dependent, parameter information. Using the combined features as input, a prediction of the target S-parameter spectral curve is performed via a prediction model, such as a multi-layer perceptron (MLP) model. By training and optimizing the prediction model, the target S parameter spectrum curve of the antenna can be accurately predicted.
As another alternative implementation way, when the LSTM+attention model is used for forward S parameter spectrum curve prediction, the LSTM network can effectively capture long-term dependency relationship in sequence data through a gating unit inside the LSTM network, and keep memorizing previous information in the sequence. The LSTM network takes as input the sequence of structural parameters received for the microstrip patch antenna and extracts the characteristics of these parameters by sequence learning. These features contain time-dependence and related information in the structural parameter sequence so that the model can predict future S-parameter spectral curves. By introducing an Attention mechanism, the model can carry out weighted summation on the LSTM hidden states, so that important features are further extracted, and related information is more focused in the prediction process, thereby improving the prediction accuracy.
Optionally, before step S210, training the model, and when training the lstm+attribute model, inputting an S-parameter spectrum curve and a structural parameter of the microstrip patch antenna into the model, where the input is a structural parameter of the microstrip patch antenna, and the S-parameter spectrum curve is a tag; the LSTM network can effectively capture complex correlation characteristics between S parameter sequence data and structural parameters, and efficiently predicts S parameter spectrum curves corresponding to given microstrip patch antenna structures, and has lower mean square error compared with other traditional neural network algorithms.
Optionally, step S220 includes:
step S221, using the S parameter spectrum curve as an input sequence of an encoder in a converter model;
Step S222, carrying out relevance modeling on the input sequence based on a self-attention mechanism, and determining parameter characteristics;
step S223, mapping the parameter feature to the structural parameter space of the antenna based on the output layer of the converter model, and determining the target structural parameter.
As an alternative embodiment, the S-parameter spectral curves are converted into a sequence form as input to the converter model based on the step of the S-parameter spectral curves as input sequence to the encoder in the converter model. The S-parameter spectral curve may be a vector sequence comprising parameter values at different frequencies, or a time sequence, depending on the application. The step of determining the parameter characteristics is based on a self-attention mechanism modeling the input sequence to capture the correlation between different elements in the sequence. The self-attention mechanism may adaptively assign different weights according to the importance of each element in the input sequence to obtain the parameter characteristics. And a step of determining a target structural parameter by mapping the parameter feature to a structural parameter space of the antenna based on an output layer of the converter model, wherein the parameter feature is mapped to the structural parameter space of the antenna using the output layer of the converter model. The output layer may be a fully connected layer mapping the parameter characteristics to the target structural parameters. In training the converter model, training is required to optimize the parameters of the output layer using sample data with the target structural parameters.
As another alternative, the primary function of the encoder is feature extraction when using a transducer model for inversion of the inverse structure parameters. The encoder in the transducer model consists of multiple layers of identical structure, each of which contains a self-attention mechanism and a feed-forward neural network. In the inverse structure parametric inversion, the input is the S-parameter spectral curve of the microstrip patch antenna, and the encoder in the transducer model takes these S-parameters as the input sequence and models the relevance of each element in the sequence by a self-attention mechanism, thereby extracting the features. These features contain the relevant information in the S-parameter sequence and the dependency relationship with each other. Finally, the extracted features can be mapped into the structural parameter space of the antenna through the output layer of the transducer model, thereby realizing inverse inversion.
Optionally, before step S220, training the inverse inversion model is further included, where the training process is: when training a transducer model, inputting an S-parameter spectrum curve and structural parameters of the microstrip patch antenna into the model, wherein the input data are S-parameter spectrum curves, and the structural parameters are labels; the transducer model processes the structure parameter prediction task of the microstrip patch antenna in a sequence-to-sequence (seq 2 seq) mode, and can directly learn the complex relation between the structure parameters from an S parameter curve; meanwhile, a transducer model introduces a global attention mechanism, so that the model can pay attention to all positions in an S-parameter spectrum curve sequence at the same time, thereby capturing global dependency relations among structural parameters better, and being beneficial to exploring the influence of the whole structure on performance in microstrip patch antenna design; the self-attention mechanism in the transducer model can dynamically adjust the weight according to the interrelationship between each structural parameter in the input sequence, so that the dependency relationship between the structural parameters can be better captured, and the method is very effective for the microstrip patch antenna structural parameter prediction.
In this embodiment, on the one hand, the lstm+attention model is used to rapidly and efficiently predict the S-parameter spectrum curve from the input structural parameters, so that the time for performing simulation calculation by simulation software such as HFS can be saved, and possible operation errors in the simulation modeling process can be avoided. On the other hand, the antenna structure parameters are rapidly and efficiently predicted by using a transducer model through the input S-parameter frequency spectrum curve, so that the time for trial-and-error inversion according to an empirical method can be saved, and the learning requirement of a designer is reduced. Finally, the VAE+BPNN+SWO model is used, and the microstrip patch antenna is optimized according to specific needs of users, so that the designed microstrip antenna has better performance, and a large amount of iterative optimization time can be saved.
Based on the first embodiment or the second embodiment, the third embodiment of the present application provides an optimization design method for a microstrip patch antenna, referring to fig. 3, after step S130, the method includes:
and step S310, performing data dimension reduction on the antenna parameters and extracting key features.
In this embodiment, for the original antenna parameter data, the data may be subjected to dimension reduction by a common dimension reduction technique (such as principal component analysis, feature selection, etc.). The reduced-dimension data may contain fewer key features but with information richness to reduce the complexity and computational effort of the model.
Step S320, learning a nonlinear relationship between the structural parameter and the antenna performance based on the key feature and a back propagation algorithm.
In this embodiment, a neural network model is trained using a back-propagation algorithm using the extracted key features as inputs. The output of the neural network model is the target S parameter spectral curve or target structural parameter of the antenna. The neural network may be a multi-layer perceptron or a reverse inversion model, i.e., a transducer model, that learns the nonlinear relationship through multiple hidden layers and activation functions.
Step S330, training the inverse inversion model based on the nonlinear relationship.
In this embodiment, the nonlinear relationship between the weight and structural parameters of the trained forward predictive model and the antenna performance is used as the input and label for the inverse inversion model. Training is performed by a back propagation algorithm using a back inversion model to learn the mapping from antenna performance to target structural parameters.
As an alternative implementation, the performance optimization design of the microstrip patch antenna is completed with vae+bpnn+swo, where VAE (Variational Autoencoder, variational self-encoder) is used to learn the potential representation of the structural parameters of the microstrip patch antenna to reduce the data dimension and extract key features; the BPNN (Backpropagation Neural Network ) trains the neural network model through a back propagation algorithm using the low-dimensional potential representation learned by the VAE to achieve accurate prediction of structural parameters of the microstrip patch antenna. By means of a back propagation algorithm, the model can learn a complex nonlinear relationship between the structural parameters and the microstrip patch antenna performance. WSO (Sine Wave Optimization, sine wave optimization algorithm) comprehensively considers a plurality of performance indexes of the microstrip patch antenna in a weighted summation mode, so that multi-objective optimization is realized in the optimization process. By adjusting the weight, the balance among different performance indexes can be realized, so that a better microstrip patch antenna design is obtained.
In this embodiment VAE (Variational Autoencoder): the variable self-encoder is a neural network model without supervision learning, and consists of an encoder and a decoder. It can be used to learn and extract potential feature representations of the input data and can be used to generate new samples that are similar to the original data. In microstrip patch antenna designs, VAEs may be used for dimension reduction and feature extraction, learning key features from raw antenna parameters. BPNN (Backpropagation Neural Network): the back propagation neural network is a commonly used machine learning algorithm that can be used to train a neural network model with multiple hidden layers. In microstrip patch antenna designs, BPNN may be used to learn a nonlinear mapping relationship between input features and output targets, such as learning a relationship between structural parameters and antenna performance. SWO (Sine Wave Optimization): the sine wave optimization algorithm is a global optimization algorithm based on sine wave functions and is used for searching an optimal solution of an optimization problem. In microstrip patch antenna designs, SWO algorithms may be used to search for and optimize the values of antenna structural parameters to maximize antenna performance, such as maximizing efficiency or bandwidth.
Optionally, after step S130, the method further includes: and judging whether the S-parameter spectrum curve meets the requirement or not by comparing the Mean Square Error (MSE), and optimizing the S-parameter spectrum curve of the microstrip patch. The S parameter of the general microstrip patch antenna can be regarded as good performance of the antenna under-10 dB of the resonance point, the performance is further optimized, the echo loss is possibly small, and the condition that the microstrip patch antenna meets the optimization completion is judged when the echo loss of the microstrip patch antenna at the resonance point is under-15 dB.
Referring to fig. 4, the overall flow of the method for optimally designing the microstrip patch antenna is as follows: when the structural parameters or S-parameter spectrum curves of the microstrip patch antenna are input by a user, forward S-parameter spectrum curve prediction is carried out, namely the input structural parameters are given, and the S-parameter spectrum curves are output by using an LSTM+attention model. When the input is not the structural parameter but the S-parameter spectrum curve, the structural parameter is output by using a transducer model, HFS simulation is carried out based on the output structural parameter, the S-parameter spectrum curve is output, at the moment, whether the input meets the requirement or the maximum iteration number is reached is judged, and if the input is not the structural parameter, the step of judging whether the input is the structural parameter is executed again. If yes, comparing the S-parameter spectrum curves output by the forward prediction model and the reverse inversion model, judging whether the prediction is good or not by using a mean square error, and if the mean square error is within an acceptable range, indicating that the prediction is successful, namely, the prediction result is output without optimization. If the mean square error is not in the receiving range, the prediction needs to be optimized, at the moment, the performance of the predicted value is optimized through the VAE+BPNN+SWO, whether the optimization requirement is met or the maximum iteration number is achieved is judged, if not, the optimization is executed again, and if yes, the optimization result is output.
Due to the fact that data dimension reduction is carried out on the antenna parameters, and key features are extracted; learning a nonlinear relationship between the structural parameters and antenna performance based on the key features and a back propagation algorithm; and training the inverse inversion model based on the nonlinear relation, so that the balance among different performance indexes is determined based on the selection of a user, and a better microstrip patch antenna design is obtained.
The application further provides an optimal design device of the microstrip patch antenna, and referring to fig. 5, fig. 5 is a schematic structural diagram of the optimal design device of the microstrip patch antenna in a hardware operation environment according to the embodiment of the application.
As shown in fig. 5, the apparatus for optimally designing a microstrip patch antenna may include: a processor 1001, such as a central processing unit (Central ProceSing Unit, CPU), a communication bus 1002, a user interface 1003, a network interface 1004, a memory 1005. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., a wireless FIdelity (WI-FI) interface). The Memory 1005 may be a high-speed random access Memory (Random AcceS Memory, RAM) Memory or a stable Non-Volatile Memory (NVM), such as a disk Memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the structure shown in fig. 5 does not constitute a limitation of the optimally designed device of the microstrip patch antenna, and may include more or fewer components than shown, or may combine certain components, or may have a different arrangement of components.
Optionally, the memory 1005 is electrically connected to the processor 1001, and the processor 1001 may be configured to control operation of the memory 1005, and may also read data in the memory 1005 to implement an optimized design of the microstrip patch antenna.
Optionally, as shown in fig. 5, an operating system, a data storage module, a network communication module, a user interface module, and an optimization design program of the microstrip patch antenna may be included in the memory 1005 as a storage medium.
Optionally, in the optimally designed microstrip patch antenna device shown in fig. 5, the network interface 1004 is mainly used for data communication with other devices; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the microstrip patch antenna optimizing design device of the present application may be disposed in the microstrip patch antenna optimizing design device.
As shown in fig. 5, the optimization design device of the microstrip patch antenna invokes, through the processor 1001, an optimization design program of the microstrip patch antenna stored in the memory 1005, and executes related steps of the optimization design method of the microstrip patch antenna provided by the embodiment of the present application:
when antenna parameters are received, determining the parameter types corresponding to the antenna parameters;
Determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model;
And processing the antenna parameters based on the forward prediction model or the inverse inversion model, and determining a target S parameter spectrum curve or a target structure parameter of the antenna.
Optionally, the processor 1001 may call the optimization design program of the microstrip patch antenna stored in the memory 1005, and further perform the following operations:
When receiving an input value corresponding to an antenna parameter, judging the input value;
And determining the parameter type of the antenna parameter as a structural parameter or an S-parameter spectrum curve according to the judging result.
Optionally, the processor 1001 may call the optimization design program of the microstrip patch antenna stored in the memory 1005, and further perform the following operations:
when the parameter type is a structural parameter, determining that the forward prediction model is a long-period memory model and short-period memory model mixed attention mechanism model;
And when the parameter type is an S-parameter spectrum curve, determining the inverse inversion model as a converter model.
Optionally, the processor 1001 may call the optimization design program of the microstrip patch antenna stored in the memory 1005, and further perform the following operations:
Determining the target S parameter spectrum curve of the antenna based on the forward prediction model processing structure parameters; or alternatively, the first and second heat exchangers may be,
And processing an S-parameter spectrum curve based on the inverse inversion model, and determining target structure parameters of the antenna.
Optionally, the processor 1001 may call the optimization design program of the microstrip patch antenna stored in the memory 1005, and further perform the following operations:
Capturing a long-term dependency relationship in the structural parameters based on a gating unit in the long-term and short-term memory model;
determining the time dependence and parameter information corresponding to the structural parameters based on the long-term dependence;
Weighting and summing the hidden states of the long-term dependency relationship based on an attention mechanism, and extracting attention features;
Predicting the target S-parameter spectral curve of the antenna based on the time dependence, the parameter information and the feature of interest.
Optionally, the processor 1001 may call the optimization design program of the microstrip patch antenna stored in the memory 1005, and further perform the following operations:
Taking the S parameter spectrum curve as an input sequence of an encoder in a converter model;
Performing relevance modeling on the input sequence based on a self-attention mechanism, and determining parameter characteristics;
the target structural parameters are determined based on mapping the parametric features to a structural parameter space of the antenna by an output layer of the converter model.
Optionally, the processor 1001 may call the optimization design program of the microstrip patch antenna stored in the memory 1005, and further perform the following operations:
Performing data dimension reduction on the antenna parameters and extracting key features;
Learning a nonlinear relationship between the structural parameters and antenna performance based on the key features and a back propagation algorithm;
Training the inverse inversion model based on the nonlinear relationship.
In addition, the application also provides an optimal design device of the microstrip patch antenna, which comprises:
The selection module is used for determining the parameter type corresponding to the antenna parameter when the antenna parameter is received; determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model;
The forward spectrum curve prediction module is used for processing structural parameters based on a forward prediction model and determining the target S parameter spectrum curve of the antenna;
and the inverse structure parameter inversion module is used for processing the S-parameter spectrum curve based on the inverse inversion model and determining the target structure parameter of the antenna.
In this embodiment, the forward spectrum curve prediction module predicts the S-parameter spectrum curve of the microstrip patch antenna with high accuracy and high efficiency. The inverse structure parameter inversion module, namely, a user can invert corresponding structure parameters according to the input S-parameter spectrum curve, so that a microstrip patch antenna structure is designed, and the predicted S-parameter spectrum curve and the real S-parameter spectrum curve are compared to verify whether the structural parameters of the inverse performance meet the design standard. And the performance optimization module is used for optimizing the performance of the microstrip patch antenna, namely optimizing an S-parameter spectrum curve of the microstrip patch. The S parameter of the general microstrip patch antenna can be regarded as good performance of the antenna under-10 dB of the resonance point, and the performance is further optimized, so that the echo loss is as small as possible. And setting the condition that the return loss of the microstrip patch antenna at the resonance point is less than-15 dB and is judged to be in accordance with the optimization completion.
Further, the system also comprises an accuracy evaluation module for S-parameter spectrum curve prediction and structural parameter inversion, and whether the system meets the requirements is judged by comparing Mean Square Error (MSE).
In this embodiment, an lstm+attention model is used to rapidly predict a structural parameter input by a user, and at the same time, the structural parameter is compared with an S-parameter spectrum curve obtained by HFS simulation of the set of structural parameters, and a mean square error is used to judge whether the prediction is successful or not, where the mean square error is within a certain acceptable range. And the reverse structure parameter inversion module uses a transducer model to rapidly predict an S-parameter spectrum curve input by a user, simultaneously puts the predicted structure parameter into the HFS for simulation, compares the S-parameter spectrum curves of the S-parameter spectrum curve and the S-parameter spectrum curve, uses a mean square error to judge whether the prediction is successful or not, and indicates that the prediction is successful when the mean square error is within a certain acceptable range. And the performance optimization module is used for performing performance optimization on structural parameters or S-parameter spectrum curves needing to be optimized by using a VAE+BPNN+SWO model. For microstrip patch antennas, the return loss is a major factor in evaluating performance, and in this experiment, the return loss at the resonance point is mainly optimized to be as small as possible, and generally, the return loss is below-10 dB, which is considered to be good in performance. Under the condition of selecting to optimize, the return loss is enabled to be below-15 dB, and the optimization target is reached, namely the optimization is completed.
In addition, the application also provides a computer readable storage medium, wherein the computer readable storage medium stores an optimization design program of the microstrip patch antenna, and the optimization design program of the microstrip patch antenna realizes the steps of the optimization design method of the microstrip patch antenna in each embodiment when being executed by a processor.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flowchart and/or block of the flowchart illustrations and/or block diagrams, and combinations of flowcharts and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The application may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (7)
1. The method for optimally designing the microstrip patch antenna is characterized by comprising the following steps of:
when antenna parameters are received, determining the parameter types corresponding to the antenna parameters;
Determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model;
processing the antenna parameters based on the forward prediction model or the inverse inversion model, and determining a target S parameter spectrum curve or a target structure parameter of the antenna;
The step of determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model, comprises the following steps:
when the parameter type is a structural parameter, determining that the forward prediction model is a long-period memory model and short-period memory model mixed attention mechanism model;
When the parameter type is an S-parameter spectrum curve, determining the inverse inversion model as a converter model;
The step of processing the antenna parameters based on the forward prediction model or the inverse inversion model to determine a target S parameter spectrum curve or a target structure parameter of the antenna comprises the following steps:
Determining the target S parameter spectrum curve of the antenna based on the forward prediction model processing structure parameters; or alternatively, the first and second heat exchangers may be,
Processing an S-parameter spectrum curve based on the inverse inversion model, and determining a target structure parameter of the antenna;
The step of determining the target S parameter spectral curve or the target structural parameter of the antenna includes:
Performing data dimension reduction on the antenna parameters and extracting key features;
Learning a nonlinear relationship between the structural parameters and antenna performance based on the key features and a back propagation algorithm;
Training the inverse inversion model based on the nonlinear relationship.
2. The method for optimizing design of microstrip patch antenna according to claim 1, wherein said step of determining a type of parameter corresponding to an antenna parameter when receiving said antenna parameter comprises:
When receiving an input value corresponding to an antenna parameter, judging the input value;
And determining the parameter type of the antenna parameter as a structural parameter or an S-parameter spectrum curve according to the judging result.
3. The method of optimizing design of a microstrip patch antenna according to claim 1, wherein said step of determining said target S-parameter spectral curve of said antenna based on said forward prediction model processing structure parameters comprises:
Capturing a long-term dependency relationship in the structural parameters based on a gating unit in the long-term and short-term memory model;
determining the time dependence and parameter information corresponding to the structural parameters based on the long-term dependence;
Weighting and summing the hidden states of the long-term dependency relationship based on an attention mechanism, and extracting attention features;
Predicting the target S-parameter spectral curve of the antenna based on the time dependence, the parameter information and the feature of interest.
4. The method of optimizing design of microstrip patch antenna according to claim 1, wherein said step of determining a target structural parameter of said antenna based on said inverse inversion model processing S-parameter spectral curves comprises:
Taking the S parameter spectrum curve as an input sequence of an encoder in a converter model;
Performing relevance modeling on the input sequence based on a self-attention mechanism, and determining parameter characteristics;
the target structural parameters are determined based on mapping the parametric features to a structural parameter space of the antenna by an output layer of the converter model.
5. An apparatus for implementing the method for optimally designing a microstrip patch antenna according to claim 1, said apparatus comprising:
The selection module is used for determining the parameter type corresponding to the antenna parameter when the antenna parameter is received; determining a forward prediction model or a reverse inversion model based on the parameter type, wherein the forward prediction model is a hybrid model;
The forward spectrum curve prediction module is used for processing structural parameters based on a forward prediction model and determining the target S parameter spectrum curve of the antenna;
and the inverse structure parameter inversion module is used for processing the S-parameter spectrum curve based on the inverse inversion model and determining the target structure parameter of the antenna.
6. An apparatus for optimizing a microstrip patch antenna, comprising a memory, a processor, and an optimizing program for the microstrip patch antenna stored in the memory and operable on the processor, wherein the processor performs the steps of the optimizing method for the microstrip patch antenna according to any one of claims 1 to 4 when executing the optimizing program for the microstrip patch antenna.
7. A computer-readable storage medium, wherein the computer-readable storage medium has stored thereon an optimization design program for a microstrip patch antenna, which when executed by a processor, implements the steps of the optimization design method for a microstrip patch antenna according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410814332.7A CN118395877B (en) | 2024-06-24 | 2024-06-24 | Method, device, equipment and medium for optimally designing microstrip patch antenna |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410814332.7A CN118395877B (en) | 2024-06-24 | 2024-06-24 | Method, device, equipment and medium for optimally designing microstrip patch antenna |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118395877A CN118395877A (en) | 2024-07-26 |
| CN118395877B true CN118395877B (en) | 2024-08-20 |
Family
ID=92005928
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410814332.7A Active CN118395877B (en) | 2024-06-24 | 2024-06-24 | Method, device, equipment and medium for optimally designing microstrip patch antenna |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118395877B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107831408A (en) * | 2017-09-15 | 2018-03-23 | 河北省电力建设调整试验所 | A kind of universal design of ultra high-frequency partial discharge sensor, optimization and method of testing |
| CN113587990A (en) * | 2021-07-30 | 2021-11-02 | 中北大学 | Parameter detection method, device and equipment based on microstrip antenna sensor |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114154205A (en) * | 2021-12-06 | 2022-03-08 | 安徽大学 | Frequency reconfigurable filtering antenna design method |
| CN117313558B (en) * | 2023-11-29 | 2024-03-15 | 深圳飞骧科技股份有限公司 | Optimizing method, optimizing system and related equipment for ultra-wideband antenna parameters with gradual change structure |
-
2024
- 2024-06-24 CN CN202410814332.7A patent/CN118395877B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107831408A (en) * | 2017-09-15 | 2018-03-23 | 河北省电力建设调整试验所 | A kind of universal design of ultra high-frequency partial discharge sensor, optimization and method of testing |
| CN113587990A (en) * | 2021-07-30 | 2021-11-02 | 中北大学 | Parameter detection method, device and equipment based on microstrip antenna sensor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118395877A (en) | 2024-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111860982A (en) | Wind power plant short-term wind power prediction method based on VMD-FCM-GRU | |
| CN111461445B (en) | Short-term wind speed prediction method and device, computer equipment and storage medium | |
| CN113627066A (en) | Displacement prediction method for reservoir bank landslide | |
| CN112733997A (en) | Hydrological time series prediction optimization method based on WOA-LSTM-MC | |
| CN114298290A (en) | Neural network coding method and coder based on self-supervision learning | |
| Li et al. | A novel hybrid approach of ABC with SCA for the parameter optimization of SVR in blind image quality assessment | |
| Garbuglia et al. | Bayesian optimization for microwave devices using deep GP spectral surrogate models | |
| CN118395877B (en) | Method, device, equipment and medium for optimally designing microstrip patch antenna | |
| CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
| Fu et al. | Enhanced machine learning model via twin support vector regression for streamflow time series forecasting of hydropower reservoir | |
| CN117155806A (en) | Communication base station flow prediction method and device | |
| CN116709409A (en) | Knowledge distillation-based lightweight spectrum prediction method | |
| CN115604131B (en) | Link flow prediction method, system, electronic device and medium | |
| CN116561569A (en) | Industrial power load identification method based on EO feature selection and AdaBoost algorithm | |
| CN116613732A (en) | Multi-element load prediction method and system based on SHAP value selection strategy | |
| CN115345207A (en) | Self-adaptive multi-meteorological-element prediction method | |
| CN115564466A (en) | Double-layer day-ahead electricity price prediction method based on calibration window integration and coupled market characteristics | |
| CN112616160A (en) | Intelligent short-wave frequency cross-frequency-band real-time prediction method and system | |
| CN116780524B (en) | Industrial enterprise short-term load prediction method based on LSTM deep learning | |
| Peng et al. | An Efficient Antenna Optimization Framework Based on Inverse Radial Basis Function Network | |
| Wu et al. | Surrogate model-based precipitation tuning for CAM5 | |
| CN114418122A (en) | Hyper-parameter configuration method and device of machine learning model and readable storage medium | |
| CN118714585A (en) | Optimization method, system, device and medium for predicting uplink throughput based on RSRP | |
| CN117113818A (en) | Micro-architecture design space exploration method, device and computer equipment | |
| Deng | A Survey: Hardware Neural Architecture Search On FPGA/ASIC |
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 |