Absorption rate curve calculation method, design method, device, equipment and medium

Abstract

The present invention discloses an absorptivity curve calculation method, design method, device, equipment and medium, which form an autoencoder by combining an inverse design model and a target absorptivity curve calculation model, and make the inverse design model act as an encoder to perform residual network operation on the acquired enhanced spectrum and output it after low-dimensional encoding, so as to complete the training of the inverse design model and the target absorptivity curve calculation model from the structural parameters input by the target absorptivity curve calculation model as a decoder, and then complete the residual network operation in the target absorptivity curve calculation model by inputting a standard data set into the target absorptivity curve calculation model, and finally output the target absorptivity curve. Without relying on manual experience and simulation processing, absorptivity curve with a specified frequency band and curve shape can be manually designed to obtain a metasurface structure with such absorptivity characteristics, rather than using a known absorptivity curve to reversely obtain the corresponding structure, thereby simplifying the design process and improving design efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic absorber design and manufacturing, and in particular to an absorption rate curve calculation method, design method, device, equipment and medium. Background Art

[0002] In recent years, metasurface technology has attracted much attention. Electromagnetic metasurface is a kind of artificially designed quasi-two-dimensional material, usually composed of a periodic array of sub-wavelength units. Due to its flexible structural design, metasurfaces can achieve electromagnetic response characteristics that many natural materials cannot achieve, such as negative refractive index, anomalous refraction, perfect absorption, etc., and can fully control electromagnetic waves in the frequency domain, time domain, and spatial domain. At the same time, due to its compact structure, metasurface devices can achieve higher functional integration than traditional optical devices. Electromagnetic wave absorbers are an important branch of metasurfaces, and the absorption characteristics of electromagnetic waves can be artificially designed in a specified band. Metasurface electromagnetic wave absorbers have great application potential in the fields of photoelectric detection, photovoltaic devices, etc., and have gradually attracted people's attention.

[0003] Existing approaches to designing metasurface electromagnetic wave absorbers typically rely on manual experience to design the metasurface structure and corresponding parameters, then validate the model using numerical simulation software, and perform iterative optimization to obtain an absorption curve. This approach relies heavily on manual design experience and, due to the precision requirements of numerical simulations, typically consumes significant computing resources and iteration time, making it difficult to achieve rapid reverse design of metasurface electromagnetic wave absorbers with specified absorption characteristics. Summary of the Invention

[0004] The main purpose of the present invention is to provide an absorption rate curve calculation method, design method, device, equipment and medium, aiming to solve the technical problem that related technologies are difficult to achieve rapid reverse design of metasurface electromagnetic wave absorbers with specified absorption characteristics.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a method for calculating an absorptivity curve, comprising the following steps:

[0006] Obtain a training data set; wherein the training data set includes multiple training samples collected from the standard model, each training sample includes an initial absorption curve and structural parameters, the standard model includes a metal substrate and a plurality of meta-structure units spaced and arrayed on the substrate, each meta-structure unit has a cylindrical structure, the meta-structure unit includes a first absorption module and at least one second absorption module stacked in sequence from the substrate toward away from the substrate, the first absorption module includes a first organic layer and a first metal layer, the second absorption module includes a second organic layer and a second metal layer, the second organic layer is adjacent to the first metal layer, and the structural parameters include a first diameter of the first absorption module, a second diameter of the second absorption module, a first thickness of the first organic layer, a second thickness of the second organic layer, and a center-to-center distance between any two adjacent meta-structure units;

[0007] The training data set is used to train an initialized absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model; wherein, the initialized absorption rate curve calculation model includes five residual blocks, and the five residual blocks each include a first convolutional layer, a second convolutional layer, a first batch normalization layer, and a second batch normalization layer. The first convolutional layer is used to perform a convolution operation on the input structural parameters, the first batch normalization layer is used to normalize the operation results output by the first convolutional layer, the second convolutional layer is used to perform a convolution operation on the data after normalization by the first batch normalization layer, and the second batch normalization layer is used to normalize the operation results at the output end of the second convolutional layer.

[0008] Optionally, the step of using the training data set to train an initial absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, includes:

[0009] The first convolutional layers are used to receive the training data set and train the initialized absorption rate curve calculation model in combination with a first loss function to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model; wherein the first loss function is a mean square error loss function.

[0010] Optionally, the step of using the training data set to train an initial absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, further includes:

[0011] The first convolutional layers are used to receive the training data set and train the initialized absorption rate curve calculation model in combination with a second loss function to obtain an inverse design model, so as to obtain the structural parameters when the current data set is input into the inverse design model; wherein the second loss function is a mean square error loss function.

[0012] Optionally, the step of using the training data set to train an initial absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, further includes:

[0013] An autoencoder is constructed using the inverse design model as an encoder and the target absorption rate curve calculation model as a decoder, and the steps of using each of the first convolutional layers to receive the training data set and training the initialized absorption rate curve calculation model in combination with the second loss function are sequentially performed in combination with the third loss function to obtain a reverse design model, so as to obtain the structural parameters when the current data set is input into the inverse design model, and the step of using each of the first convolutional layers to receive the training data set and training the initialized absorption rate curve calculation model in combination with the first loss function to obtain a target absorption rate curve calculation model, so as to obtain the target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, so as to train the autoencoder.

[0014] Optionally, the third loss function is obtained by weighting the first loss function and the second loss function, and the third loss function is described by Formula 1, which is:

[0015] ,

[0016] L is the third loss function, is the second loss function, is the first loss function, is the weight.

[0017] Optionally, the step of obtaining a training data set includes:

[0018] The initial absorption curves of a plurality of arbitrary frequency bands and the structural parameters corresponding to the initial absorption curves are collected on the standard model to obtain the training data set.

[0019] Based on the same technical concept, in a second aspect, the present invention proposes a method for designing an electromagnetic wave absorber, comprising the following steps:

[0020] Constructing the electromagnetic wave absorber model; wherein the electromagnetic wave absorber model includes a metal substrate and a plurality of meta-structure units spaced apart and arranged in an array on the substrate, each meta-structure unit having a cylindrical structure, and the meta-structure unit including a first organic layer and at least one second organic layer stacked sequentially from the substrate toward a direction away from the substrate, and a metal layer is sandwiched between the first organic layer and the second organic layer and on a side of the second organic layer away from the first organic layer;

[0021] acquiring the standard data set according to the initial structural parameters of the electromagnetic wave absorber model;

[0022] Performing residual network calculation on the standard data set using the absorption rate curve calculation method described in the first aspect to obtain the target absorption rate curve;

[0023] The target absorptivity curve is compared with a preset absorptivity curve to obtain a design result of the electromagnetic wave absorber.

[0024] Based on the same technical concept, in a third aspect, the present invention further proposes a device for designing an electromagnetic wave absorber, comprising:

[0025] A modeling module, used for constructing the electromagnetic wave absorber model;

[0026] A data acquisition module, configured to acquire the standard data set according to the initial structural parameters of the electromagnetic wave absorber model;

[0027] a calculation module, configured to perform a residual network operation on the standard data set using the absorption rate curve calculation method described in the first aspect to obtain the target absorption rate curve;

[0028] The result output module is used to compare the target absorptivity curve with a preset absorptivity curve to obtain a design result of the electromagnetic wave absorber.

[0029] Based on the same technical concept, in the fourth aspect, the present invention also proposes an electromagnetic wave absorber design device, which includes a processor and a memory, and the memory stores an electromagnetic wave absorber design program. When the electromagnetic wave absorber design program is executed by the processor, the electromagnetic wave absorber design method described in the second aspect is implemented.

[0030] Based on the same technical concept, in the fifth aspect, the present invention also proposes a computer storage medium, on which a computer program is stored. When the computer program is executed by one or more processors, the electromagnetic wave absorber design method described in the second aspect is implemented.

[0031] The technical solution of the present invention forms an autoencoder by combining an inverse design model and a target absorptivity curve calculation model, and enables the inverse design model to act as an encoder to perform residual network operations on the acquired enhanced spectrum and output low-dimensional encoding to complete the training of the inverse design model and the target absorptivity curve calculation model from the structural parameters input by the target absorptivity curve calculation model as a decoder, thereby forming a target absorptivity curve calculation model. Then, the residual network operations are completed by inputting the standard data set into the target absorptivity curve calculation model, and the target absorptivity curve is finally output. This allows the present invention to obtain a metasurface structure with such absorptivity characteristics by manually designing an absorptivity curve of a specified frequency band and curve shape without relying on manual experience and simulation processing. This further enables the present invention to reversely design a metasurface electromagnetic wave absorber structure based on the obtained absorptivity curve, thereby simplifying the design process and improving the design efficiency, thereby achieving the purpose of rapid reverse design of a metasurface electromagnetic wave absorber structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0033] Figure 1 A flowchart of a method for calculating an absorbance curve according to an embodiment of the present invention;

[0034] Figure 2 A schematic diagram of the structure of a standard model exemplified by the present invention;

[0035] Figure 3 Schematic diagram of the structure of the residual block of the present invention;

[0036] Figure 4 A schematic diagram of the reverse design model according to an example of the present invention;

[0037] FIG5( a ) is a spectrum diagram of any frequency band of a training data set according to an example of the present invention;

[0038] FIG5( b ) is a schematic diagram of enhanced data of multiple frequency bands according to an example of the present invention;

[0039] Figure 6 A schematic diagram of an example of the present invention using an autoencoder structure for model training;

[0040] FIG7( a ) is a spectrum diagram showing the test and verification results of the absorption spectrum of one frequency band using the reverse design model according to an example of the present invention;

[0041] FIG7( b ) is a spectrum diagram showing the test and verification results of the absorption spectrum of another frequency band using the reverse design model according to an example of the present invention;

[0042] FIG7( c ) is a spectrum diagram of an absorption spectrum predicted when the absorptivity of one frequency band is 1 using the target absorptivity curve calculation model according to an example of the present invention;

[0043] FIG7( d ) is a spectrum diagram of an absorption spectrum predicted when the absorption rate of another frequency band is 1 using the target absorption rate curve calculation model according to an example of the present invention;

[0044] Figure 8 A flow chart of a method for designing an electromagnetic wave absorber according to an example of the present invention;

[0045] Figure 9 This is a schematic diagram of the structure of an electromagnetic wave absorber design device according to an example of the present invention. Description of the drawings:

[0047]

[0048] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0050] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between various mechanisms under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0051] In the present invention, unless otherwise specified or limited, the terms "connection" and "fixation" should be understood in a broad sense. For example, "fixation" can mean fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two elements or interaction between two elements, unless otherwise specified. Those skilled in the art will be able to understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0052] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the meaning of "and / or" appearing throughout the text includes three parallel schemes. Taking "A and / or B" as an example, it includes scheme A, or scheme B, or a scheme in which A and B are satisfied at the same time. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in this field to implement. When the combination of technical solutions is mutually contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0053] The inventive concept of the present invention is further described below with reference to some specific embodiments.

[0054] The present invention provides an absorption rate curve calculation method, design method, device, equipment and medium.

[0055] like Figures 1 to 9 As shown, the present invention discloses an embodiment of an absorption rate curve calculation method, design method, device, equipment and medium.

[0056] In this embodiment, please refer to Figures 1-6 The calculation method of this type of absorption rate curve includes the following steps:

[0057] S100, obtaining a training data set; wherein the training data set includes multiple training samples collected from the standard model, each of the training samples includes an initial absorption curve and structural parameters, the standard model includes a metal substrate 100 and a plurality of meta-structure units spaced and arrayed on the substrate, each of the meta-structure units is a cylindrical structure, the meta-structure unit includes a first absorption module and at least one second absorption module stacked in sequence from the substrate toward away from the substrate, the first absorption module includes a first organic layer 200 and a first metal layer 400, the second absorption module includes a second organic layer 300 and a second metal layer 500, the second organic layer 300 is adjacent to the first metal layer 400, and the structural parameters include a first diameter of the first absorption module, a second diameter of the second absorption module, a first thickness of the first organic layer 200, a second thickness of the second organic layer 300, and a center-to-center distance between any two adjacent meta-structure units;

[0058] In this embodiment, it is particularly important to specify that the first thickness is T2 (T2 ranges from 50 mm to 200 mm), the second thickness is T1 (T1 ranges from 50 mm to 200 mm), the first diameter is d1 (d1 ranges from 100 mm to 475 mm), the second diameter is d2 (d2 ranges from 125 mm to 500 mm), the center-to-center spacing between any two adjacent meta-units is P (P ranges from 150 mm to 550 mm), the height of the metal substrate 100 is h (h is 100 mm), the heights of the first metal layer 400 and the second metal layer 500 are both t (t is 15 nm), and the second diameter is greater than the first diameter. Furthermore, the metal substrate 100 is preferably made of gold, and the first and second organic layers 200 and 300 are preferably made of polymethyl methacrylate (PMMA). The standard model used in this embodiment refers to a metasurface electromagnetic wave absorber model. It should be further clarified that when the number of the second absorption modules in the example exceeds two, the multiple second absorption modules are arranged in sequence from the first absorption module toward the direction away from the metal substrate, and the second diameters of the second absorption modules can be the same or can be gradually decreasing.

[0059] In an exemplary embodiment, the specific process of obtaining the training data set can be performed as follows:

[0060] After establishing the metasurface electromagnetic wave absorber model, multiple sets of structural parameters including T1, T2, d1, d2, and P are collected on the metasurface electromagnetic wave absorber model. After completing the collection of structural parameters, a training data set is constructed. The specific process of constructing the training data set is as follows:

[0061] The structural parameters T1, T2, d1, d2, and P of the metasurface electromagnetic wave absorber model are used as labels, and 100 sampling points are set according to the set parameter range. The absorptivity curves of the metasurface in the wavelength range of 400~1600nm are calculated under different structural parameter combinations using the finite-difference time-domain (FDTD) method to characterize its electromagnetic wave absorption characteristics. The data are then sampled and collected to construct the required training dataset.

[0062] To design a metasurface electromagnetic wave absorber with specified absorption characteristics within the 400-1600 nm wavelength range, data augmentation is performed on the existing training dataset. A random band within the 400-1600 nm wavelength range is retained, and the absorptivity of the remaining bands is set to -1 to indicate neglect. See Figure 5(b) for a schematic diagram of data augmentation to obtain the preset dataset. The augmented dataset is used to train the inverse design model, while the unaugmented dataset is used to calculate the labels for the target absorptivity curve model.

[0063] The absorptivity curve calculated using the FDTD method includes all absorptivity characteristics within the 400-1600nm wavelength range. However, in the actual design phase, only the absorptivity characteristics of the specified wavelength band are focused on, while other wavelengths are ignored. To prevent the model from being interfered with by information outside the specified wavelength band and to enhance the model's generalization capabilities, this paper uses a data augmentation dataset for inverse design and training of the target absorptivity curve calculation model.

[0064] After completing the construction of the training data set, the reverse design model of the metasurface electromagnetic wave absorber and the target absorption rate curve calculation model are constructed. The specific process of constructing the reverse design model and the target absorption rate curve calculation model is as follows:

[0065] The inverse design model and the target absorptivity curve calculation model are two opposing tasks. The input of the inverse prediction model is the absorptivity curve of the metasurface electromagnetic wave absorber, and the output is the possible metasurface structural parameters corresponding to the curve.

[0066] The target absorptivity curve calculation model is a fitting of the traditional numerical simulation method. The input is the structural parameters of the metasurface electromagnetic wave absorber, and the structural parameters are mapped to its absorptivity curve.

[0067] Compared with traditional numerical simulation methods, the target absorption rate curve calculation model can quickly verify the designed structural parameters.

[0068] A random subset of the original dataset is created using data that has undergone data augmentation. The data in the random subset is then used to train the inverse design network for the metasurface electromagnetic wave absorber, enabling the network to focus on the absorption rate characteristics of a specified band while ignoring other bands, helping the model achieve better generalization capabilities.

[0069] The supervised residual connection model used is labeled with the hypersurface structure parameters corresponding to the original dataset.

[0070] The network contains a total of 5 residual blocks, each of which contains one or more basic convolution calculation processes. The structure of the residual block is as follows Figure 3 As shown. In the residual block, there are first two convolutional layers 11 and 12 (Conv) with the same number of output channels. The first convolutional layer 11 is connected to the first batch normalization layer ( ), the second convolutional layer 12 is connected to the second batch normalization layer 14 ( ) and, the first batch normalization layer 13 and the second batch normalization layer 14 are both connected to a ReLU activation function 30. Skip connections are used to skip convolution operations, adding the input structural parameters to the final ReLU activation function 30. Each residual block performs a downsampling operation through maximum pooling or one-dimensional convolution to halve the size of the feature map. Specifically, the formula for the residual block is as follows:

[0071] (1),

[0072] Among them, x is the input, y is the output, and F represents the main part of the residual block, which can be expressed as:

[0073] (2),

[0074] in, is the convolution kernel of the first convolutional layer 11, is the convolution kernel of the second convolutional layer 12, σ is the activation function 30, For the first batch normalization layer 13, This is the second batch normalization layer 14. Typically, the activation function 30 can be a nonlinear function such as Sigmoid, ReLU, or Tanh, which performs a nonlinear transformation on each element in the matrix. After extracting the data features of the absorption rate curve through the residual network, the network is connected to a multilayer perceptron consisting of a ReLU activation function 30 and a linear layer to map the data features into the desired metasurface structure parameters.

[0075] It should be particularly and clearly stated that the multi-layer perceptron used in this embodiment is also the fully connected layer 19 mentioned later, and the target absorption rate curve calculation model used in the embodiment is also the target absorption rate curve calculation model.

[0076] The reverse design model and target absorbance curve calculation model used in this embodiment adopt the same model structure. Since the residual network can accept input sequences of any length, the forward prediction and reverse design models differ only in the output dimension of the final output layer. In fact, the absorbance curve of the metasurface is obtained by sampling at a specific wavelength. The data itself is a one-dimensional sequence, but this sequence has no temporal connection, so it is not suitable for recurrent neural network networks. At the same time, although the Transformer-type network structure has better accuracy, it requires a lot of computational cost. For this embodiment, the improved accuracy is not enough to eliminate the increase in computational cost.

[0077] S200. Use the training data set to train an initialized absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model; wherein, the initialized absorption rate curve calculation model includes five residual blocks, and the five residual blocks each include a first convolutional layer 11, a second convolutional layer 12, a first batch normalization layer 13 and a second batch normalization layer 14, the first convolutional layer 11 is used to perform a convolution operation on the input structural parameters, the first batch normalization layer 13 is used to normalize the operation results output by the first convolutional layer 11, the second convolutional layer 12 is used to perform a convolution operation on the data after normalization by the first batch normalization layer 13, and the second batch normalization layer 14 is used to normalize the operation results at the output end of the second convolutional layer 12.

[0078] In this embodiment, the overall process of obtaining the absorption rate curve can be performed as follows:

[0079] After completing the construction of the reverse design model and the target absorption rate curve calculation model, the forward prediction and reverse design models are trained through the autoencoder structure:

[0080] An autoencoder is an unsupervised learning model that aims to learn a compressed representation of data by compressing input data into a low-dimensional code and then reconstructing the encoded data into output data similar to the original input data. It consists of two parts: an encoder 20 that compresses the input data into a low-dimensional code and a decoder 30 that maps the low-dimensional code back to the original data space.

[0081] Specifically, the encoder 20 maps the input data x into a low-dimensional code , usually expressed as , where f is a mapping function. The decoder converts the low-dimensional code Map back to the original data space to obtain reconstructed data , usually expressed as , where g is another mapping function.

[0082] During the training process, the autoencoder learns the parameters of the encoder 20 and decoder 30 by minimizing the reconstruction error so that the error between the reconstructed data and the original data is as small as possible. Specifically, the reconstruction error can be expressed as the mean square error between the input data and the reconstructed data, that is:

[0083] (3);

[0084] By backpropagating the reconstruction error, the parameters of the encoder 20 and decoder 30 can be optimized, thereby learning a compressed representation of the data. In this embodiment, the inverse design model converts the high-dimensional absorption rate into low-dimensional model structure parameters, while the target absorption rate curve calculation model maps the parameters to the corresponding absorption rate curve. Therefore, the low-dimensional model structure parameters can be regarded as a feature of the absorption rate curve and used as the intermediate layer output of the autoencoder. The inverse design model is used as the encoder 20 part of the autoencoder, and the target absorption rate curve calculation model is used as the decoder 30 part for joint training.

[0085] During the training phase, the input of the model is the absorption rate curve x' after data enhancement. After the features are extracted by the residual network, they are mapped to the low-dimensional code of the middle layer through the multi-layer perceptron. , and then input the low-dimensional code into the target absorption rate curve calculation model to reconstruct the complete predicted absorption rate curve The mean square error (MSE) loss function is used to minimize the error between the complete absorption rate curve and the predicted value. By jointly training the autoencoder of the two models, the model acquires the ability to extract data features and enables the inverse model to focus on the absorption rate of the specified band. However, since the low-dimensional encoding is not trained during the training process, the low-dimensional encoding is not trained. Constraints are often not possible during training. It is associated with the designed model structure parameter y, so the MSE loss function is minimized during training. The error between y and y constrains the output of the intermediate layer.

[0086] (4);

[0087] Loss function for joint training of autoencoders and the loss function constrained by the intermediate layer For joint optimization, the total loss function is the weighted sum of the two:

[0088] (5);

[0089] in is the weight of the loss function.

[0090] It should be noted that, in this embodiment, the example is the second loss function, for example is the first loss function, the example is the third loss function.

[0091] Verification of the reverse design model of metasurface electromagnetic wave absorber:

[0092] The trained inverse design model was tested and validated. First, arbitrary absorptivity data from the test set was used as input. By modifying the randomly initialized seed and the ratio of retained data during the data augmentation process, complex absorptivity curves designed under different circumstances were simulated. The structural parameters of the metasurface electromagnetic wave absorber were reversely designed. The obtained structural parameters were then substituted into the target absorptivity curve calculation model (or FDTD simulation model) to calculate the corresponding metasurface absorptivity curve and compare it with the input absorptivity curve. Some of the comparison results are shown in Figures 7(a) to 7(d). Similarly, the absorptivity of a specified band within the range of 400-1600nm was manually set to 1, and the absorptivity of the remaining bands was set to -1. This absorptivity curve was then used as input to test the inverse design model. Some of the output results are compared with the input as shown in Figures 7(a) to 7(d).

[0093] According to the test verification results, the inverse design model can accurately reverse design the structural parameters of the metasurface electromagnetic wave absorber based on the input absorptivity curve, so that it meets the manually set absorptivity characteristics in the specified band within the range of 400~1600nm.

[0094] In this embodiment, by constructing and training a reverse design model and a target absorptivity curve calculation model, the present invention can reversely design an absorptivity curve for representing the absorption characteristics of an electromagnetic wave absorber without relying on manual experience and simulation processing. This also enables the present invention to use an absorptivity curve of a manually designed specified frequency band and curve shape to obtain a metasurface structure with such absorptivity characteristic, thereby simplifying the design process and improving design efficiency, thereby achieving the purpose of rapidly reverse designing a metasurface electromagnetic wave absorber structure.

[0095] In some exemplary embodiments, step S200 includes:

[0096] S210. Utilize each of the first convolutional layers 11 to receive the training data set and train the initialized absorption rate curve calculation model in combination with a first loss function to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model; wherein, the first loss function is a mean square error loss function.

[0097] In this embodiment, when used specifically, the data received by the first convolutional layer 11 should be structural parameters after feature extraction.

[0098] Specifically, step S210 includes:

[0099] S220. Utilize each of the first convolutional layers 11 to receive the training data set and train the initialized absorption rate curve calculation model in combination with a second loss function to obtain an inverse design model, so as to obtain the structural parameters when the current data set is input into the inverse design model; wherein the second loss function is a mean square error loss function.

[0100] Specifically, step 210 further includes:

[0101] An autoencoder is constructed using the inverse design model as the encoder 20 and the target absorption rate curve calculation model as the decoder 30, and the steps of using each of the first convolutional layers 11 to receive the training data set and training the initialized absorption rate curve calculation model in combination with the second loss function are sequentially performed in combination with the third loss function to obtain the inverse design model, so as to obtain the structural parameters when the current data set is input into the inverse design model, and the step of using each of the first convolutional layers 11 to receive the training data set and training the initialized absorption rate curve calculation model in combination with the first loss function to obtain the target absorption rate curve calculation model, so as to obtain the target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, so as to train the autoencoder.

[0102] It should be noted that the third loss function is obtained by weighting the first loss function and the second loss function. The third loss function is described by Formula 1, which is:

[0103] ,

[0104] L is the third loss function, is the second loss function, is the first loss function, is the weight.

[0105] In some specific embodiments, step S100 includes:

[0106] S110 , collecting the initial absorption curves of a plurality of arbitrary frequency bands and the structural parameters corresponding to the initial absorption curves on the standard model to obtain the training data set.

[0107] Based on the same technical concept, in a second aspect, the present invention proposes a method for designing an electromagnetic wave absorber, comprising the following steps:

[0108] A100. Constructing the electromagnetic wave absorber model; wherein the electromagnetic wave absorber model includes a metal substrate 100 and a plurality of meta-structure units spaced apart and arranged in an array on the substrate, each meta-structure unit having a cylindrical structure, and including a first organic layer 200 and at least one second organic layer 300 stacked sequentially from the substrate toward a direction away from the substrate, with a metal layer interposed between the first organic layer 200 and the second organic layer 300 and on a side of the second organic layer 300 away from the first organic layer 200;

[0109] In this embodiment, the exemplified electromagnetic wave absorber model is preferably an electromagnetic wave absorber model.

[0110] A200, acquiring the standard data set according to the initial structural parameters of the electromagnetic wave absorber model;

[0111] A300, performing a residual network operation on the standard data set using the absorption rate curve calculation method exemplified in the previous embodiment to obtain the target absorption rate curve;

[0112] A400: Compare the target absorptivity curve with a preset absorptivity curve to obtain a design result of the electromagnetic wave absorber.

[0113] In this embodiment, the initial structural parameters of the electromagnetic wave absorber model are first collected from the constructed electromagnetic wave absorber to obtain a standard data set. Then, the absorptivity curve calculation method exemplified in the previous embodiment is used to perform residual network operations on the standard data set to obtain a target absorptivity curve. The target absorptivity curve is then compared with the preset absorptivity curve to obtain a design result of the electromagnetic wave absorber. This allows the present invention to reversely design an absorptivity curve for representing the absorption characteristics of the electromagnetic wave absorber without relying on manual experience and simulation processing when in use. This also enables the present invention to use an absorptivity curve of a specified frequency band and curve shape designed manually to obtain a metasurface structure with such absorptivity characteristics, thereby simplifying the design process and improving design efficiency, thereby achieving the purpose of rapid reverse design of a metasurface electromagnetic wave absorber structure.

[0114] Based on the same technical concept, in a third aspect, the present invention further proposes a device for designing an electromagnetic wave absorber, comprising:

[0115] A modeling module, used for constructing the electromagnetic wave absorber model;

[0116] A data acquisition module, configured to acquire the standard data set according to the initial structural parameters of the electromagnetic wave absorber model;

[0117] A calculation module, configured to perform residual network calculation on the standard data set using the absorption rate curve calculation method described in the preceding embodiment to obtain the target absorption rate curve;

[0118] The result output module is used to compare the target absorptivity curve with a preset absorptivity curve to obtain a design result of the electromagnetic wave absorber.

[0119] It should be noted that the functions that can be realized by each module in the electromagnetic wave absorber design device provided in this embodiment and the corresponding technical effects achieved can refer to the description of the specific implementation methods in each embodiment of the electromagnetic wave absorber design method of the present invention. For the sake of brevity of the description, they will not be repeated here.

[0120] Based on the same technical concept, in a fourth aspect, the present invention also proposes an electromagnetic wave absorber design device, which includes a processor and a memory. The memory stores an electromagnetic wave absorber design program. When the electromagnetic wave absorber design program is executed by the processor, the electromagnetic wave absorber design method exemplified in the previous embodiment is implemented.

[0121] Electromagnetic wave absorber design equipment refers to terminal equipment or network equipment that can achieve network connection. Electromagnetic wave absorber design equipment can be terminal equipment such as mobile phones, computers, tablet computers, embedded industrial computers, etc., or it can be network equipment such as servers and cloud platforms.

[0122] like Figure 9 FIG. 1 is a schematic diagram of the hardware structure of an electromagnetic wave absorber design device. The electromagnetic wave absorber design device may include: a processor 1001 , such as a CPU (Central Processing Unit), a communication bus 1002 , a user interface 1003 , a network interface 1004 , and a memory 1005 .

[0123] The memory is used to store various types of data, which may include, for example, instructions for any application or method in the electromagnetic wave absorber design device, as well as data related to the application. The memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), random access memory (RAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. Optionally, the memory can also be a storage device independent of the processor.

[0124] The processor is used to call the electromagnetic wave absorber design program stored in the memory and execute the electromagnetic wave absorber design method as described above. The processor can be an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor, or other electronic component, and is used to execute all or part of the steps of each embodiment of the electromagnetic wave absorber design method as described above.

[0125] Those skilled in the art will understand that Figure 9 The hardware structure shown in the figure does not constitute a limitation on the electromagnetic wave absorber design device of the present invention, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0126] Specifically, the communication bus 1002 is used to implement connection and communication between these components;

[0127] The user interface 1003 is used to connect to the client and perform data communication with the client. The user interface 1003 may include an output unit, such as a display screen, and an input unit, such as a keyboard;

[0128] The network interface 1004 is used to connect to the backend server and perform data communication with the backend server. The network interface 1004 may include an input / output interface, such as a standard wired interface or a wireless interface, such as a Wi-Fi interface.

[0129] The memory 1005 is used to store various types of data. These data may include, for example, instructions for any application or method in the electromagnetic wave absorber design device, as well as data related to the application. The memory 1005 may be a high-speed RAM memory or a stable memory such as a disk memory. Optionally, the memory 1005 may also be a storage device independent of the processor 1001. Figure 9 , the memory 1005 may include an operating system, a network communication module, a user interface module and an electromagnetic wave absorber design program;

[0130] The processor 1001 is configured to call the electromagnetic wave absorber design program stored in the memory 1005 and perform the following operations:

[0131] A100. Construct an electromagnetic wave absorber model.

[0132] A200. Obtain a standard data set based on the initial structural parameters of the electromagnetic wave absorber model.

[0133] A300 , using the absorption rate curve calculation method exemplified in the previous embodiment to perform residual network operation on the standard data set to obtain a target absorption rate curve.

[0134] A400, comparing the target absorptivity curve with the preset absorptivity curve to obtain a design result of the electromagnetic wave absorber.

[0135] Based on the same technical concept, in a fifth aspect, the present invention further proposes a computer storage medium having a computer program stored thereon. When the computer program is executed by one or more processors, the electromagnetic wave absorber design method exemplified in the above embodiments is implemented.

[0136] Based on the same inventive concept, this embodiment provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic memory, a magnetic disk, an optical disk, a server, etc. A computer program is stored on the storage medium, and the computer program can be executed by one or more processors. When the computer program is executed by the processor, all or part of the steps of each embodiment of the non-absorbed body design method of the present invention can be implemented.

[0137] The above descriptions are merely optional embodiments of the present invention and do not limit the scope of the present invention. All equivalent structural transformations made using the contents of the present invention's description and drawings, or direct / indirect applications in other related technical fields within the scope of the present invention are included in the protection scope of the present invention.

Claims

1. A method for calculating an absorbance curve, characterized in that: The steps include: Obtain a training data set; wherein the training data set includes multiple training samples collected from a standard model, each training sample includes an initial absorption curve and structural parameters, the standard model includes a metal substrate and a plurality of meta-structure units spaced and arrayed on the substrate, each meta-structure unit has a cylindrical structure, the meta-structure unit includes a first absorption module and at least one second absorption module stacked sequentially from the substrate toward away from the substrate, the first absorption module includes a first organic layer and a first metal layer, the second absorption module includes a second organic layer and a second metal layer, the second organic layer is adjacent to the first metal layer, and the structural parameters include a first diameter of the first absorption module, a second diameter of the second absorption module, a first thickness of the first organic layer, a second thickness of the second organic layer, and a center-to-center distance between any two adjacent meta-structure units; The training data set is used to train an initialized absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model; wherein the initialized absorption rate curve calculation model includes five residual blocks, and each of the five residual blocks includes a first convolutional layer, a second convolutional layer, a first batch normalization layer, and a second batch normalization layer, the first convolutional layer is used to perform a convolution operation on the input structural parameters, the first batch normalization layer is used to normalize the operation results output by the first convolutional layer, the second convolutional layer is used to perform a convolution operation on the data after normalization by the first batch normalization layer, and the second batch normalization layer is used to normalize the operation results at the output end of the second convolutional layer; The step of using the training data set to train an initial absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, further includes: The first convolutional layers are used to receive the training data set and train the initialized absorption rate curve calculation model in combination with a second loss function to obtain an inverse design model, so as to obtain the structural parameters when the current data set is input into the inverse design model; wherein the second loss function is a mean square error loss function.

2. The method for calculating the absorption rate curve according to claim 1, wherein: The step of using the training data set to train an initial absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, includes: The first convolutional layers are used to receive the training data set and train the initialized absorption rate curve calculation model in combination with a first loss function to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model; wherein the first loss function is a mean square error loss function.

3. The method for calculating the absorption rate curve according to claim 1, wherein: The step of using the training data set to train an initial absorption rate curve calculation model to obtain a target absorption rate curve calculation model, so as to obtain a target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, further includes: An autoencoder is constructed using the inverse design model as an encoder and the target absorption rate curve calculation model as a decoder, and the steps of using each of the first convolutional layers to receive the training data set and training the initialized absorption rate curve calculation model in combination with the second loss function are sequentially performed in combination with the third loss function to obtain a reverse design model, so as to obtain the structural parameters when the current data set is input into the inverse design model, and the step of using each of the first convolutional layers to receive the training data set and training the initialized absorption rate curve calculation model in combination with the first loss function to obtain a target absorption rate curve calculation model, so as to obtain the target absorption rate curve when the current data set is input into the target absorption rate curve calculation model, so as to train the autoencoder.

4. The method for calculating the absorption rate curve according to claim 3, wherein: The third loss function is obtained by weighting the first loss function and the second loss function. The third loss function is described by Formula 1, which is: ， L is the third loss function, is the second loss function, is the first loss function, is the weight.

5. The method for calculating an absorbance curve according to any one of claims 1 to 4, wherein: The step of obtaining a training data set includes: The initial absorption curves of a plurality of arbitrary frequency bands and the structural parameters corresponding to the initial absorption curves are collected on the standard model to obtain the training data set.

6. A method for designing an electromagnetic wave absorber, characterized in that: The steps include: Constructing an electromagnetic wave absorber model; wherein the electromagnetic wave absorber model includes a metal substrate and a plurality of meta-structure units spaced apart and arranged in an array on the substrate, each meta-structure unit having a cylindrical structure, and the meta-structure unit including a first organic layer and at least one second organic layer stacked sequentially from the substrate toward a direction away from the substrate, and a metal layer is sandwiched between the first organic layer and the second organic layer and on a side of the second organic layer away from the first organic layer; Acquiring a standard data set according to the initial structural parameters of the electromagnetic wave absorber model; Performing a residual network operation on the standard data set using the absorption rate curve calculation method according to any one of claims 1 to 5 to obtain the target absorption rate curve; The target absorptivity curve is compared with a preset absorptivity curve to obtain a design result of the electromagnetic wave absorber.

7. An electromagnetic wave absorber design device, characterized in that: include: Modeling module, used to build electromagnetic wave absorber model; A data acquisition module, configured to acquire a standard data set according to the initial structural parameters of the electromagnetic wave absorber model; a calculation module, configured to perform a residual network operation on the standard data set using the absorption rate curve calculation method according to any one of claims 1 to 5, so as to obtain the target absorption rate curve; The result output module is used to compare the target absorptivity curve with a preset absorptivity curve to obtain a design result of the electromagnetic wave absorber.

8. An electromagnetic wave absorber design device, characterized in that: The electromagnetic wave absorber design device includes a processor and a memory, wherein an electromagnetic wave absorber design program is stored in the memory. When the electromagnetic wave absorber design program is executed by the processor, the electromagnetic wave absorber design method according to claim 6 is implemented.

9. A computer storage medium, characterized in that The storage medium stores a computer program, and when the computer program is executed by one or more processors, the method for designing an electromagnetic wave absorber according to claim 6 is implemented.

Images