Metasurface design optimization method and system based on task-oriented learning

Abstract

The invention discloses a task-oriented learning-based metasurface design optimization method and system, and relates to the technical field of electromagnetic metamaterial design and optimization. Comprising the following steps: generating metasurface unit structure parameters and corresponding electromagnetic characteristic parameters, and constructing a data set; constructing a neural network taking the unit structure parameters as input and the corresponding electromagnetic characteristic parameters as output, and performing neural network training by using the data set; defining a target feature parameter of the metasurface unit, and constructing an optimization target function taking an output parameter of the neural network and the target feature parameter as independent variables; deriving the target function to obtain the adjustment amount of the unit structure parameters; updating unit structure parameters; and realizing design optimization of the metasurface until the objective function meets the threshold value. A new thought is provided for development of the metasurface technology through the intelligent optimization design method, and the requirement for multifunctional and efficient design of the complex metasurface can be met.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic metamaterial design and optimization, and in particular to a metasurface design optimization method and system based on task-oriented learning. Background Art

[0002] As a new type of artificial electromagnetic material, electromagnetic metasurface technology can achieve precise control of electromagnetic wave propagation behavior, including absorption, reflection, refraction, transmission and polarization conversion, by designing a unit structure with a subwavelength scale. The emergence of metasurfaces provides a revolutionary solution to traditional electromagnetic wave control technology. For example, by controlling the scattering characteristics of electromagnetic waves, metasurfaces can significantly reduce the radar cross section (RCS) of target objects, thereby achieving stealth effects. Compared with traditional stealth materials, metasurface design has higher flexibility and can be customized for the size, shape and material of different targets. It is widely used in military protection, stealth aircraft and drones. In the field of wireless signal enhancement and beam control, metasurface technology can achieve precise control of signal propagation direction, intensity and polarization, thereby improving the signal coverage and reliability of communication systems. In terms of beamforming, it shows unique advantages, and can achieve beam concentration or dispersion in a specified direction through the design of phase distribution. Compared with traditional phased array antennas, metasurface beamforming devices have the characteristics of simple structure, low cost and light weight. In complex electromagnetic environments, metasurface technology can be used as an absorbing material or shielding material to reduce electromagnetic interference and improve electromagnetic compatibility. Its high-efficiency broadband absorption characteristics make it play an important role in RF shielding, electronic equipment protection, and anti-electromagnetic attack scenarios.

[0003] However, the design of metasurfaces usually relies on experience and simulation optimization, and designers need to repeatedly adjust unit structure parameters (such as geometric dimensions and material parameters) to achieve target performance. This process is time-consuming and labor-intensive, especially when dealing with multivariable optimization problems, where efficiency is significantly reduced. Existing design methods are difficult to quickly generate customized metasurface designs that meet different application scenarios when faced with complex scenarios, limiting the promotion of practical applications.

[0004] With the rapid development of machine learning technology, deep neural networks have demonstrated powerful capabilities in nonlinear mapping modeling and optimization problems. By introducing deep learning methods into metasurface design and using neural networks to construct nonlinear mapping relationships between unit structure parameters and electromagnetic properties, the design efficiency can be significantly improved, and global optimization of complex electromagnetic properties can be achieved. However, the metasurface design method based on deep learning is still in the initial exploration stage, and there is still much room for improvement in multi-objective optimization, searching in complex parameter spaces, and adapting to different task requirements.

[0005] Therefore, it is an urgent problem for technical personnel in this field to propose a metasurface design optimization method and system based on task-oriented learning to solve the difficulties existing in the prior art. Summary of the invention

[0006] In view of this, the present invention provides a hypersurface design optimization method and system based on task-oriented learning, which constructs the relationship between target performance and design parameters through a deep neural network, and combines the gradient descent method to quickly optimize unit parameters, providing an efficient and intelligent solution for complex hypersurface design.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] A method for optimizing the design of a metasurface based on task-oriented learning, comprising the following steps:

[0009] S1. Data generation: Generate the metasurface unit structure parameters p and the corresponding electromagnetic characteristic parameters s, and construct a data set And construct an empty correction dataset Used to update network parameters later, where p = {p1, p2, ..., p i}, s={s1, s2, ..., s i}, i represents the index of the sample;

[0010] S2. Network training: Construct a neural network φ with unit structure parameters p as input and corresponding electromagnetic characteristic parameters s as output θ , and use the dataset Perform neural network training, where θ is the network parameter;

[0011] S3. Objective function construction: defining the target characteristic parameters s of the hypersurface unit * , construct the output parameter s of the neural network and the target feature parameter s * is the optimization objective function f(s) of the independent variable * ,s);

[0012] S4. Get the adjustment amount: for f(s * ,s) to obtain the adjustment value Δp of the unit structure parameter p,

[0013] S5. Update parameters: Update unit structure parameters using adjustment amount Δp;

[0014] S6. Design optimization: Repeat S3 to S5 until the objective function meets the threshold, thus achieving the design optimization of the metasurface.

[0015] Optionally, the structural parameters of the metasurface unit in S1 include side length, sheet resistance, material square resistance, unit period length, dielectric constant of the material and dielectric thickness.

[0016] Optionally, the neural network in S2 includes an input layer, a fully connected layer, a reshaping layer, a convolutional layer, a maximum pooling layer, and an output layer that are connected in sequence.

[0017] Optionally, optimize the objective function f(s in S3 * ,s) is calculated as:

[0018]

[0019] Optionally, in S4, the gradient descent method is used to iteratively adjust the unit structure parameters, and f(s * ,s) to obtain the adjustment value Δp of the unit structure parameter p:

[0020]

[0021] Among them, θ represents the derivative, φ θ represents a parameter characteristic prediction neural network with parameter θ;

[0022] The updating unit structure parameters follow the following formula:

[0023] p=p+∈Δp

[0024] Among them, 0<∈≤1 represents the update step size.

[0025] Optionally, in S6, S3 to S5 are repeated until the threshold value satisfied by the objective function is as follows:

[0026] f(s * ,s) <f T

[0027] Among them, f T is the threshold for optimizing the objective function.

[0028] Optional, if f(s * ,s) <f T If it is established, the electromagnetic simulation software is used to simulate and obtain the electromagnetic characteristic parameters corresponding to the structural parameter p at this moment. The objective function at this moment is Calculate the objective function at this moment And the training samples Insert the correction data set

[0029] like Then stop the iteration; otherwise, return to S4 to continue updating and iterating.

[0030] Optionally, the optimizers used in the neural network training process include Adam and RMSprop.

[0031] A supersurface design optimization system based on task-oriented learning, applying any of the above-mentioned supersurface design optimization methods based on task-oriented learning, comprising: a data generation module, a network training module, an objective function construction module, an adjustment amount acquisition module, a parameter update module and a design optimization module;

[0032] A data generation module is connected to the input end of the network training module, and is used to generate the structural parameters of the metasurface unit and the corresponding electromagnetic characteristic parameters, and to construct a data set and a correction data set;

[0033] A network training module is connected to the input end of the objective function building module, and is used to build a neural network with unit structure parameters as input and corresponding electromagnetic characteristic parameters as output, and use the data set to train the neural network;

[0034] An objective function construction module is connected to the input end of the adjustment amount acquisition module, and is used to define the target characteristic parameters of the hypersurface unit and construct an optimization objective function with the output parameters of the neural network and the target characteristic parameters as independent variables;

[0035] An adjustment amount acquisition module is connected to the input end of the parameter update module and is used to derive the objective function to obtain the adjustment amount of the unit structure parameter;

[0036] An updating parameter module, connected to the input end of the design optimization module, for updating the unit structure parameters using the adjustment amount;

[0037] The design optimization module is used to repeatedly construct the objective function, obtain the adjustment amount and update the parameters until the objective function meets the threshold, thereby realizing the design optimization of the metasurface.

[0038] It can be seen from the above technical solution that, compared with the prior art, the present invention provides a supersurface design optimization method and system based on task-oriented learning, which has the following beneficial effects: The present invention provides a new idea for the development of supersurface technology through an intelligent optimization design method, which can meet the needs of multifunctional and efficient design of complex supersurfaces. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0040] Figure 1A flow chart of a supersurface design optimization method based on task-oriented learning provided by the present invention;

[0041] Figure 2 The test result diagram of the indoor experimental flexible stealth device provided in an embodiment of the present invention, wherein 2a is the absorption rate in simulation and experiment, and 2b shows the absorption rate at different incident wave angles. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] Reference Figure 1 As shown, the present invention discloses a metasurface design optimization method based on task-oriented learning, which can be widely used in technical directions such as radar stealth, beam control, wireless signal enhancement and electromagnetic environment management, and is particularly suitable for complex multivariable metasurface unit structure parameter optimization, aiming to achieve fast and efficient metasurface design and performance optimization, including the following steps:

[0044] S1. Data generation: Generate the metasurface unit structure parameters p and the corresponding electromagnetic characteristic parameters s, and construct a data set And construct an empty correction dataset Used to update network parameters later, where p = {p1, p2, ..., p i}, s={s1, s2, ..., s i}, i represents the index of the sample;

[0045] S2. Network training: Construct a neural network φ with unit structure parameters p as input and corresponding electromagnetic characteristic parameters s as output θ , and use the dataset Perform neural network training, where θ is the network parameter;

[0046] S3. Objective function construction: defining the target characteristic parameters s of the hypersurface unit * , construct the output parameter s of the neural network and the target feature parameter s * is the optimization objective function f(s) of the independent variable * ,s);

[0047] S4. Get the adjustment amount: for f(s * ,s) to obtain the adjustment value Δp of the unit structure parameter p,

[0048] S5. Update parameters: Update unit structure parameters using adjustment amount Δp;

[0049] S6. Design optimization: Repeat S3 to S5 until the objective function meets the threshold, thus achieving the design optimization of the metasurface.

[0050] Furthermore, the structural parameters of the metasurface unit in S1 include side length, sheet resistance, material square resistance, unit period length, dielectric constant of the material and dielectric thickness.

[0051] Specifically, by optimizing these structural parameters, it is ensured that the metasurface unit achieves high absorption efficiency in the target frequency range (such as 9–25 GHz) while having angle-insensitivity and polarization-independent characteristics.

[0052] Furthermore, the neural network in S2 includes an input layer, a fully connected layer, a reshaping layer, a convolutional layer, a maximum pooling layer, and an output layer, which are connected in sequence.

[0053] Furthermore, in S3, the objective function f(s * ,s) is calculated as:

[0054]

[0055] Furthermore, in S4, the gradient descent method is used to iteratively adjust the unit structure parameters, and f(s * ,s) to obtain the adjustment value Δp of the unit structure parameter p:

[0056]

[0057] in, represents the derivative, φ θ represents a parameter characteristic prediction neural network with parameter θ;

[0058] The updating unit structure parameters follow the following formula:

[0059] p=p+∈Δp

[0060] Among them, 0<∈≤1 represents the update step size.

[0061] Furthermore, in S6, S3 to S5 are repeated until the threshold value satisfied by the objective function is as follows:

[0062] f(s * ,s) <f T

[0063] Among them, f T is the threshold for optimizing the objective function.

[0064] Furthermore, if f(s * ,s) <f TIf it is established, the electromagnetic simulation software is used to simulate and obtain the electromagnetic characteristic parameters corresponding to the structural parameter p at this moment. The objective function at this moment is Calculate the objective function at this moment And the training samples Insert the correction data set

[0065] like Then stop the iteration; otherwise, return to S4 to continue updating and iterating.

[0066] Furthermore, the optimizers used in the neural network training process include Adam and RMSprop.

[0067] In a specific embodiment, a metasurface design optimization method based on task-oriented learning is applied in a stealth device. The stealth device is designed based on a flexible substrate and a deformable metasurface unit structure, and can adapt to complex and irregular target object surfaces to achieve flexible stealth effects. The stealth device can be molded into a variety of shapes, such as an invisible tent, to shield internal objects from detection by external radars (such as drone SAR radars).

[0068] The application of flexible leather substrate and stretchable conductive ink layer makes the stealth device have excellent ductility. By optimizing key parameters, the target electromagnetic performance (such as broadband high absorption efficiency) can be achieved.

[0069] The key parameters are as follows: conductive pattern: side length l1, l2, sheet resistance m, n; substrate parameters: dielectric thickness h1, h2.

[0070] Construct a neural network with the input as unit structure parameters [h1,h2,l1,l2,m,n] and the output as S parameters. Construct an adaptive objective function f(s * ,s), indicating the s parameter s of the current structure and the target s parameter s * Then, based on the gradient descent method, the objective function is optimized and the structural parameters are adjusted until the error is satisfied. The condition, where f T is the convergence threshold.

[0071] The pseudo code of the Task-Oriented Learning (TOL) algorithm is as follows:

[0072]

[0073]

[0074] In order to fit the function, an embodiment of the present invention proposes a neural network consisting of six layers. The input of the neural network includes six variables. The middle three layers consist of a fully connected layer of 256 units and 2048 units. The 2048 linear outputs are then reshaped into a two-channel 32x32 image. Next, 16 convolutional filters of size 5x5 are applied with a step size of 2 to extract features, followed by a 2x2 pooling layer. The subsequent layer is a convolutional layer with a 3x3 filter and a step size of 1. Finally, two fully connected layers with 512 and 256 units, respectively, are applied to output 256 S parameter values. The network structure parameters are summarized in Table 1.

[0075] Table 1 Network structure parameters

[0076] Layer Type Detailed description Input Layer <![CDATA[Input: [h1, h2, l1, l2, m, n] (6 variables)]]> Fully connected layer 256 units, fully connected Fully connected layer 512 units, fully connected Reshape layer Reshape the 256 cells into a 16x16 image with 2 channels Convolutional Layer 16 filters, 5x5 convolution kernel, stride 2 Max Pooling Layer 3x3 Pooling Convolutional Layer 8 filters, 3x3 convolution kernel, stride 1 Fully connected layer 512 units, fully connected Output Layer 256 units, fully connected

[0077] The network parameters were trained using Python and the CST interface to generate simulated data. A dataset of 20,000 consecutive samples was generated. During the training process, the mean square error (MSE) was used as the loss function, defined as Among them, N B represents the number of samples in each batch, represents the predicted value, s i Indicates the actual value. For efficient network training, the classic ADAM optimizer is used. The training and validation losses converge quickly at the 400th epoch. Figure 2 The test results of the indoor experimental flexible stealth device of this embodiment are shown, where 2a is the absorption rate in simulation and experiment, and 2b shows the absorption rate at different incident wave angles.

[0078] After the stealth device was prepared, the absorption performance was tested in a microwave darkroom using a separate radar transmitting and receiving system. The results showed that the absorption rate exceeded 90% in the 9-25GHz frequency range and exhibited good angle insensitivity in the 0°-60° incident angle range.

[0079] Specifically, the Structural Similarity Index (SSIM) is a method for evaluating the difference in structural information between images. Unlike traditional pixel-by-pixel error metrics, SSIM evaluates images based on brightness, contrast, and structure, providing a more comprehensive image quality assessment.

[0080] Brightness, contrast and structure correlation factors are defined as l(x,y), c(x,y) and s(x,y) respectively and are considered independent of each other. The SSIM index can be defined using the following functions S1, S2:

[0081] SSIM(x,y))=f(l(x,y),c(x,y),s(x,y))

[0082] The brightness correlation factor is derived as follows:

[0083]

[0084] Among them, μ x and μ y are the average values ​​of SAR image data before and after protection, respectively. C1 is a small constant to avoid division by zero and is set small enough to satisfy the Weber-Fechner law. Usually, C1=(K1L) 2 .

[0085] The contrast correlation factor is calculated using the following function:

[0086]

[0087] Among them, σ x 2 and σ y 2 Respectively represent the variance of SAR image data before and after protection, C2 is defined as C2 = (K2L) 2 .

[0088] Structure-related factors:

[0089]

[0090] Among them, σ xy is the covariance of the SAR image data before and after protection, C3 = C2 / 2.

[0091] In summary, the structural similarity index integrates these factors to provide a measure of structural similarity between images as follows:

[0092] SSIM(x,y))=[l(x,y)] α [c(x,y)] β ·[s(x,y)] γ

[0093] Among them, α, γ and β are adjustable parameters that determine the importance of each similarity factor. The range of the SSIM index is [-1, 1], where a value of 1 means that the two images are exactly the same.

[0094] Experimental results: When the stealth device is not covered, the scattering intensity of the target object in the SAR image is obvious and easy to be identified by radar. After the stealth device is covered, the scattering intensity of the target object is significantly reduced and almost merged with the background. The target area image is extracted for SSIM analysis, and the results show that the stealth device effectively suppresses the radar signal of the target object.

[0095] Through the above embodiments, the present invention realizes the design of a wide-band, high-absorption-efficiency SAR stealth device metasurface, and provides a feasible solution for the practical application of the metasurface design optimization method based on task-oriented learning.

[0096] and Figure 1 Corresponding to the method described above, an embodiment of the present invention further provides a super surface design optimization system based on task-oriented learning, which specifically includes: a data generation module, a network training module, an objective function construction module, an adjustment amount acquisition module, a parameter update module and a design optimization module;

[0097] A data generation module is connected to the input end of the network training module, and is used to generate the structural parameters of the metasurface unit and the corresponding electromagnetic characteristic parameters, and to construct a data set and a correction data set;

[0098] A network training module is connected to the input end of the objective function building module, and is used to build a neural network with unit structure parameters as input and corresponding electromagnetic characteristic parameters as output, and use the data set to train the neural network;

[0099] An objective function construction module is connected to the input end of the adjustment amount acquisition module, and is used to define the target characteristic parameters of the hypersurface unit and construct an optimization objective function with the output parameters of the neural network and the target characteristic parameters as independent variables;

[0100] An adjustment amount acquisition module is connected to the input end of the parameter update module and is used to derive the objective function to obtain the adjustment amount of the unit structure parameter;

[0101] An updating parameter module, connected to the input end of the design optimization module, for updating the unit structure parameters using the adjustment amount;

[0102] The design optimization module is used to repeatedly construct the objective function, obtain the adjustment amount and update the parameters until the objective function meets the threshold, thereby realizing the design optimization of the metasurface.

[0103] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0104] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for optimizing the design of a metasurface based on task-oriented learning, characterized in that: The following steps are involved: S1. Data generation: Generate metasurface unit structure parameters p and corresponding electromagnetic characteristic parameters s, and construct a data set And construct an empty correction dataset Used to update network parameters later, where p = {p1, p2, ..., p i }, s={s1, s2, ..., s i }, i represents the index of the sample; S2. Network training: Construct a neural network φ with unit structure parameters p as input and corresponding electromagnetic characteristic parameters s as output θ , and use the dataset Perform neural network training, where θ is the network parameter; S3. Objective function construction: defining the target characteristic parameters s of the hypersurface unit * , construct the output parameter s of the neural network and the target feature parameter s * is the optimization objective function f(s) of the independent variable * ,s); S4. Get the adjustment amount: for f(s * ,s) to obtain the adjustment value Δp of the unit structure parameter p, S5. Update parameters: Update unit structure parameters using adjustment amount Δp; S6. Design optimization: Repeat S3 to S5 until the objective function meets the threshold, thus achieving the design optimization of the metasurface.

2. A method for optimizing the design of a super surface based on task-oriented learning according to claim 1, characterized in that: The structural parameters of the metasurface unit in S1 include side length, sheet resistance, material square resistance, unit period length, dielectric constant of the material and dielectric thickness.

3. The method for optimizing the design of a supersurface based on task-oriented learning according to claim 1, characterized in that: The neural network in S2 includes an input layer, a fully connected layer, a reshaping layer, a convolutional layer, a maximum pooling layer, and an output layer connected in sequence.

4. The method for designing and optimizing a supersurface based on task-oriented learning according to claim 1, characterized in that: In S3, the objective function f(s * ,s) is calculated as:

5. The method for designing and optimizing a supersurface based on task-oriented learning according to claim 1, characterized in that: In S4, the gradient descent method is used to iteratively adjust the unit structure parameters, and f(s * ,s) to obtain the adjustment value Δp of the unit structure parameter p: in, represents the derivative, φ θ represents a parameter characteristic prediction neural network with parameter θ; The updating unit structure parameters follow the following formula: p=p+∈Δp Among them, 0<∈≤1 represents the update step size.

6. The method for designing and optimizing a supersurface based on task-oriented learning according to claim 1, characterized in that: In S6, S3 to S5 are repeated until the threshold value of the objective function is satisfied as follows: f(s * ,s)<f T Among them, f T is the threshold for optimizing the objective function.

7. A method for optimizing the design of a super surface based on task-oriented learning according to claim 1 or 6, characterized in that: If f(s * ,s) <f T If it is established, the electromagnetic simulation software is used to simulate and obtain the electromagnetic characteristic parameters corresponding to the structural parameter p at this moment. The objective function at this moment is Calculate the objective function at this moment And the training samples Insert the correction data set like Then stop the iteration; otherwise, return to S4 to continue updating and iterating.

8. The method for designing and optimizing a supersurface based on task-oriented learning according to claim 1, characterized in that: The optimizers used in the neural network training process include Adam and RMSprop.

9. A metasurface design optimization system based on task-oriented learning, characterized in that: A method for design optimization of a super surface based on task-oriented learning according to any one of claims 1 to 8 is applied, comprising: a data generation module, a network training module, an objective function construction module, an adjustment amount acquisition module, a parameter update module and a design optimization module; A data generation module is connected to the input end of the network training module, and is used to generate the structural parameters of the metasurface unit and the corresponding electromagnetic characteristic parameters, and to construct a data set and a correction data set; A network training module is connected to the input end of the objective function building module, and is used to build a neural network with unit structure parameters as input and corresponding electromagnetic characteristic parameters as output, and use the data set to train the neural network; An objective function construction module is connected to the input end of the adjustment amount acquisition module, and is used to define the target characteristic parameters of the hypersurface unit and construct an optimization objective function with the output parameters of the neural network and the target characteristic parameters as independent variables; An adjustment amount acquisition module is connected to the input end of the parameter update module and is used to derive the objective function to obtain the adjustment amount of the unit structure parameter; An updating parameter module, connected to the input end of the design optimization module, for updating the unit structure parameters using the adjustment amount; The design optimization module is used to repeatedly construct the objective function, obtain the adjustment amount and update the parameters until the objective function meets the threshold, thereby realizing the design optimization of the metasurface.

Images