Scattering parameter prediction method for element atoms and element equipment construction method

Abstract

The invention discloses a scattering parameter prediction method for element atoms and a construction method of element equipment, and relates to the field of element atoms. According to the scattering parameter prediction method, by constructing an initial network model comprising a splitting module, a first channel, a second channel, a stacking module, a comprehensive feature channel and a calculation module, element atom design can be effectively decomposed, and spatial features and physical features of the element atom design can be effectively extracted; meanwhile, the initial network model is trained through a plurality of meta-atom design samples with different design structures and design parameters, the generalization ability of the model is greatly improved, the real part and imaginary part of the transmission coefficient corresponding to the meta-atoms can be accurately predicted in a short time, the scattering parameters are further calculated, and the calculation accuracy is improved. The efficiency and accuracy of element atomic scattering parameter prediction are greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of meta-atoms, and in particular to a scattering parameter prediction method for meta-atoms and a meta-device construction method. Background Art

[0002] Modeling and characterization tools play a vital role in the design process of metasurfaces and meta-atoms. In order to meet the stringent requirements of free shapes and versatility for next-generation components, researchers are committed to developing more reliable and efficient modeling tools. An early approach was to simplify the simulation process of metamaterials by establishing analytical effective medium models, such as the Lewin model and the GEM model. However, these models are only applicable to long-wavelength approximations under specific conditions and can only handle metamaterials with micro-spherical shapes. Another commonly used method relies on iterative full-wave numerical simulation techniques, including the finite element method (FEM), the finite-difference time-domain method (FDTD), and the finite integral method (FIT). Although this method can provide highly accurate results, its high computational cost limits its practicality.

[0003] Recently, data-driven modeling tools based on deep neural networks (DNNs) have brought new breakthroughs to this field. Compared with traditional methods, this new tool has not only demonstrated its accuracy and timeliness, but also shown significant advantages in predicting the optical response of nanophotonic structures. For example, DNN models using fully connected layers (FCLs) have been successfully applied to predict the amplitude response of cylindrical, elliptical cylindrical, spherical, and rod-shaped meta-atoms. More importantly, some studies have also shown that DNNs can be used to predict phase responses, which is critical for most optical applications that require precise control of incident light. After sufficient training, these models are able to generate electromagnetic responses within milliseconds, greatly accelerating the design process of meta-atoms / meta-surfaces.

[0004] However, existing FCL-based deep neural network models also have certain limitations. First, they are mainly applicable to meta-atom structures that can be simply described by 3 to 5 parameters, which limits the possibility of achieving high efficiency and wide phase coverage, especially in applications of composite meta-surfaces. Second, these models usually work within a very limited design space, where the key design parameters involved (such as lattice size, meta-atom thickness, and material properties) are fixed. Once these parameters change, data needs to be re-collected and the model retrained, which is a time-consuming process. Due to these weaknesses, current deep neural network-based methods are limited in their generalization capabilities, affecting their potential for widespread application. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a method for predicting scattering parameters of meta-atoms, comprising: Construct an initial network model including a split module, a first channel, a second channel, a stacking module, a comprehensive feature channel, and a calculation module; wherein: The splitting module is used to decompose the meta-atom design into a two-dimensional image tensor and a one-dimensional attribute tensor; the meta-atom design includes a design structure and a design parameter; the design structure corresponds to the two-dimensional image tensor, and the design parameter corresponds to the one-dimensional attribute tensor; The first channel is used to extract the spatial features of the two-dimensional image tensor through convolution and pooling operations to obtain an image feature vector; The second channel is used to extract the physical features of the one-dimensional attribute tensor through nonlinear transformation and spatial tiling operations to obtain an attribute feature vector; The stacking module is used to stack the image feature vector and the attribute feature vector to obtain a splicing feature; The comprehensive feature channel is used to process the splicing features through convolution and pooling operations to obtain electromagnetic response features; The calculation module is used to obtain the real part and the imaginary part of the transmission coefficient corresponding to the meta-atom design according to the electromagnetic response characteristics, and calculate the scattering parameters based on the real part and the imaginary part of the transmission coefficient; wherein the scattering parameters include phase and amplitude; Generate multiple meta-atom designs with different design structures and design parameters, calculate the scattering parameters of the meta-atom designs, and label the corresponding meta-atom designs with the scattering parameters to obtain meta-atom design samples, and form a data set through the meta-atom design samples; the design parameters include: refractive index, thickness and lattice size; The initial network model is trained by the data set to obtain a scattering parameter prediction model; the scattering parameters of the meta-atom design to be tested are predicted by the scattering parameter prediction model.

[0006] Furthermore, the first channel includes: The first convolution layer is used to perform a first convolution calculation on the two-dimensional image tensor to obtain a first spatial feature; A first maximum pooling layer is used to perform a maximum pooling operation on the first spatial feature to obtain a pooling feature; The second convolution layer is used to perform a second convolution calculation on the pooled features to obtain the second spatial features; The second maximum pooling layer is used to perform a maximum pooling operation on the second spatial feature to obtain an image feature vector.

[0007] Furthermore, the second channel comprises: The nonlinear transformation layer is used to perform nonlinear transformation on the one-dimensional attribute tensor to obtain the transformation features; The spatial tiling layer is used to perform a spatial tiling operation on the transformed features to obtain an attribute feature vector.

[0008] Furthermore, the comprehensive feature channel includes sequentially connected: The third convolutional layer is used to perform convolution operations on the concatenated features to obtain the initial convolution features; The third maximum pooling layer is used to perform a maximum pooling operation on the initial convolutional features to obtain the initial pooling features; The fourth maximum pooling layer is used to perform the maximum pooling operation on the initial pooling features again to obtain the target pooling features; The fourth convolutional layer is used to perform convolution calculation on the target pooling feature and flatten the result obtained by the convolution calculation to obtain the target convolution feature; The first fully connected layer is used to perform a fully connected operation on the target convolutional features to obtain the initial fully connected features; The second fully connected layer is used to transform the initial fully connected features to obtain electromagnetic response features.

[0009] Furthermore, in the comprehensive feature channel, the third convolutional layer, the third maximum pooling layer, the fourth maximum pooling layer, the fourth convolutional layer and the first fully connected layer all include a batch normalization unit and a ReLU activation function; The batch normalization unit is used to perform batch normalization processing on the features obtained after the respective layers perform corresponding operations; The ReLU activation function is used to perform a nonlinear transformation on the result after batch normalization processing, and finally obtain the output feature corresponding to the layer; the output feature is the initial pooling feature, the target pooling feature, the target convolution feature or the initial fully connected feature.

[0010] Furthermore, the generation of a plurality of meta-atom designs with different design structures and design parameters is specifically as follows: Select a square canvas of preset size as the design space for each meta-atom; For each meta-atom, a preset rectangular strip is designed based on randomly selected parameters in the upper left quadrant of the canvas; The randomly selected parameters include: the number, size and position of the rectangular bars; each rectangular bar represents a set type of high refractive index material, and the rectangular bars correspond to the types of high refractive index materials one by one; Mirror the design in the upper left quadrant along the x-axis and y-axis to form the complete design of each meta-atom; By adjusting the randomly selected parameters and repeating the above steps, a plurality of meta-atom designs with different design structures and design parameters are generated.

[0011] Furthermore, the scattering parameters of the calculated meta-atom design are specifically: Setting boundary conditions: Use unit cell boundary conditions to simulate the electromagnetic response of the meta-atom design in a square lattice structure, and apply open boundary conditions in the direction perpendicular to the plane of the meta-atom design, i.e., in the z direction, to ensure that the wavefront can propagate freely; Define the illumination method: Use x-polarized plane waves to illuminate the meta-atom design from the substrate side to simulate the incident light conditions in actual application scenarios; Determine the frequency range: Calculate the electromagnetic response of the meta-atom design in the 30 to 60 THz frequency range, including transmittance and phase shift information; Based on the set boundary conditions, illumination mode and frequency range, the transmittance and phase shift of the meta-atom design are calculated using full-wave electromagnetic simulation tools; Based on the obtained transmittance and phase shift data, the phase and amplitude of the meta-atom design are calculated.

[0012] The present invention also proposes a method for constructing a meta-device, comprising: S1: Design a metasurface phase mask and decompose the metasurface phase mask into the performance requirements of each unit cell; a unit cell of the metasurface phase mask represents a meta-atom; S2: Set a serial number for the cell, the initial value of the serial number is equal to 1; S3: Input the cell performance requirements corresponding to the current sequence number into the pre-trained GAN model; S4: Generate a meta-atom design set through the GAN model; the meta-atom design set includes multiple meta-atom designs; S5: inputting the meta-atom designs in the meta-atom design set into the scattering parameter prediction model described above in sequence for prediction; obtaining the optimal meta-atom design of the corresponding cell according to the model prediction result corresponding to each meta-atom design in the meta-atom design set; S6: Determine whether the current sequence number is greater than or equal to the maximum sequence number of the cell. If not, add 1 to the sequence number and return to step S3; if yes, proceed to the next step; S7: Assemble into meta-devices through optimal meta-atom design of each unit cell.

[0013] Furthermore, the step S5 specifically includes: S50: Set the target number of iterations and initialize the number of iterations to 1; S51: inputting the meta-atom designs in the meta-atom design set into the scattering parameter prediction model as described above in sequence for prediction; S52: Based on the model prediction results of each atomic design, calculate the corresponding key performance indicators; S53: Select the best meta-atom design from the meta-atom design set as the current optimal design based on the key performance indicators; determine whether the current number of iterations is greater than or equal to the target number of iterations, if not, increase the number of iterations by 1 and proceed to the next step, if yes, jump to step S55; S54: Feedback the current optimal design to the GAN model, wherein the GAN model uses the current optimal design as a reference design, regenerates a new meta-atom design set, and returns to step S51; S55: the current optimal design is set as the optimal meta-atom design, and the iteration is terminated.

[0014] Furthermore, the key performance indicators include transmission efficiency and phase shift accuracy.

[0015] The beneficial effects of the embodiments of the present invention include:

[0016] (1) The present invention constructs an initial network model including a splitting module, a first channel, a second channel, a stacking module, a comprehensive feature channel and a calculation module, thereby effectively decomposing the meta-atom design and extracting its spatial features and physical features. At the same time, the initial network model is trained through a plurality of meta-atom design samples with different design structures and design parameters, thereby greatly improving the generalization ability of the model. The method of the present invention can accurately predict the real and imaginary parts of the transmission coefficient corresponding to the meta-atom design in a relatively short time, and further calculate the scattering parameters, thereby greatly improving the efficiency and accuracy of the meta-atom scattering parameter prediction.

[0017] (2) The present invention selects a square canvas of preset size as the design space for each meta-atom, and allows the meta-atom to be designed based on randomly selected parameters (such as the number, size and position of rectangular bars) in the upper left quadrant, so that the design is no longer limited to a few fixed parameters, which greatly increases the diversity and complexity of the design and helps to achieve high efficiency and wide phase coverage. At the same time, the design in the upper left quadrant is mirrored along the x-axis and y-axis to form a complete meta-atom design. This method not only simplifies the design process, but also ensures the consistency and symmetry of the design, which helps to improve the stability of the electromagnetic response.

[0018] (3) In the present invention, since each design can generate a unique meta-atom structure based on different random parameters, this diversified design approach can effectively expand the working range of the model and make it no longer limited to a specific set of design parameters. Even in the face of changing design parameters, it can quickly adapt and generate corresponding designs, reducing the need to re-collect data and retrain the model.

[0019] (4) The present invention generates multiple meta-atom designs using a pre-trained GAN model and evaluates the performance of each design using the above-mentioned scattering parameter prediction model to select the optimal meta-atom design. This process ensures that each unit cell can meet specific performance requirements, thereby improving the design and manufacturing accuracy of the entire meta-device.

[0020] (5) The present invention introduces an iterative optimization mechanism in step S5, which continuously feeds back the current optimal design to the GAN model to generate a new set of meta-atom designs, and repeats the evaluation and selection process until a predetermined number of iterations is reached. This method effectively ensures the superiority of the final selected meta-atom design in key performance indicators, such as transmission efficiency and phase shift accuracy, and maximizes the performance of the meta-device.

[0021] The details of one or more embodiments of the invention are set forth in the following drawings and description so that other features, objects, and advantages of the invention are more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] 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 prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other embodiments can be obtained based on these drawings without creative work.

[0023] Figure 1 is a structural diagram of an initial network model of an embodiment of the present invention;

[0024] Figure 2 is a flow chart of a method for constructing a meta-device according to an embodiment of the present invention;

[0025] Figure 3 is a design structure diagram of a meta-atom design of an embodiment of the present invention;

[0026] Figure 4 1 is the electric field distribution of the transmission metalens at different iteration times according to the embodiment of the present invention. DETAILED DESCRIPTION

[0027] Embodiments of the present embodiment will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present embodiment are shown in the accompanying drawings, it should be understood that the present embodiment can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein, which are instead provided for a more thorough and complete understanding of the present embodiment. It should be understood that the drawings and embodiments of the present embodiment are only for exemplary purposes and are not intended to limit the scope of protection of the present embodiment.

[0028] Embodiment 1

[0029] In order to improve the generalization ability of the model and improve the efficiency and accuracy of meta-atom scattering parameter prediction, an embodiment of the present invention proposes a scattering parameter prediction method for meta-atom, including:

[0030] Construct a system including splitting module, first channel, second channel, stacking module, comprehensive feature channel, and calculation module. Figure 1 The initial network model shown; where:

[0031] The splitting module is used to decompose the meta-atom design into a two-dimensional image tensor (size: 32×32) and a one-dimensional attribute tensor (size: 1×3); the meta-atom design includes a design structure and design parameters; the design structure corresponds to the two-dimensional image tensor, and the design parameters correspond to the one-dimensional attribute tensor;

[0032] The first channel is used to extract the spatial features of the two-dimensional image tensor through convolution and pooling operations to obtain an image feature vector; The first channel comprises: The first convolution layer (the convolution kernel size is 3×3 and the number of output channels is 32) is used to perform the first convolution calculation on the two-dimensional image tensor to obtain the first spatial feature; The first maximum pooling layer is used to perform a maximum pooling operation on the first spatial feature to obtain a pooled feature (size: 16×16×32); The second convolution layer (the convolution kernel size is 3×3 and the number of output channels is 64) is used to perform a second convolution calculation on the pooled features to obtain the second spatial features; The second maximum pooling layer is used to perform a maximum pooling operation on the second spatial feature to obtain an image feature vector (size: 8×8×64).

[0033] The second channel is used to extract the physical features of the one-dimensional attribute tensor through nonlinear transformation and spatial tiling operations to obtain an attribute feature vector; The second channel comprises: The nonlinear transformation layer (the convolution kernel size is 64×3×3 and the number of output channels is 32) is used to perform nonlinear transformation on the one-dimensional attribute tensor to obtain the transformation features; The spatial tiling layer is used to perform a spatial tiling operation on the transformed features to obtain an attribute feature vector (size: 8×8×64).

[0034] The stacking module is used to stack the image feature vector and the attribute feature vector to obtain a splicing feature (size: 8×8×128);

[0035] The comprehensive feature channel is used to process the splicing features through convolution and pooling operations to obtain electromagnetic response features; The comprehensive feature channel includes the following connected in sequence: The third convolution layer (the convolution kernel size is 3×3 and the number of output channels is 128) is used to perform convolution operations on the concatenated features to obtain the initial convolution features; The third maximum pooling layer is used to perform the maximum pooling operation on the initial convolutional features to obtain the initial pooling features (size: 4×4×128); The fourth maximum pooling layer is used to perform the maximum pooling operation on the initial pooling features again to obtain the target pooling features (size: 2×2×256); The fourth convolutional layer (the convolution kernel size is 3×3 and the number of output channels is 256) is used to perform convolution calculation on the target pooling feature and flatten the result of the convolution calculation (the size of the convolution result is kept as 2×2×256) to obtain the target convolution feature (size: 1×1024); The first fully connected layer is used to perform a fully connected operation on the target convolutional features to obtain the initial fully connected features (size: 1×256); The second fully connected layer is used to transform the initial fully connected features to obtain electromagnetic response features (size: 1×51).

[0036] In the comprehensive feature channel, the third convolutional layer, the third maximum pooling layer, the fourth maximum pooling layer, the fourth convolutional layer and the first fully connected layer all include a batch normalization unit and a ReLU activation function; The batch normalization unit is used to perform batch normalization processing on the features obtained after the respective layers perform corresponding operations; The ReLU activation function is used to perform a nonlinear transformation on the result after batch normalization processing, and finally obtain the output feature corresponding to the layer; the output feature is the initial pooling feature, the target pooling feature, the target convolution feature or the initial fully connected feature.

[0037] The calculation module is used to obtain the real part and the imaginary part of the transmission coefficient corresponding to the meta-atom design according to the electromagnetic response characteristics, and calculate the scattering parameters based on the real part and the imaginary part of the transmission coefficient; wherein the scattering parameters include phase and amplitude;

[0038] The "needle drop method" is used to generate multiple meta-atom designs with different design structures and design parameters, the scattering parameters of the meta-atom designs are calculated, and the corresponding meta-atom designs are labeled with the scattering parameters to obtain meta-atom design samples, and a data set is formed through the meta-atom design samples; the design parameters include: refractive index, thickness and lattice size; The generating of a plurality of meta-atom designs with different design structures and design parameters is specifically as follows: Select a square canvas of preset size (32x32 pixels) as the design space for each meta-atom; For each meta-atom, a preset number (3 to 7) of rectangular strips are designed based on randomly selected parameters in the upper left quadrant of the canvas, so that each meta-atom has a unique initial design; the minimum resolution of each rectangular strip is 1 pixel; The randomly selected parameters include: the number, size and position of the rectangular bars; each rectangular bar represents a set type of high refractive index material, and the rectangular bars correspond to the types of high refractive index materials one by one; Mirror the design in the upper left quadrant along the x-axis and y-axis to form the complete design of each meta-atom; By adjusting the randomly selected parameters and repeating the above steps, a plurality of meta-atom designs with different design structures and design parameters are generated.

[0039] In this example, a total of 53,000 meta-atom designs with different shapes, refractive indices, thicknesses, and lattice sizes were generated. Figure 3 In the figure, four different meta-atom designs are shown. The red lines indicate that the canvas is divided into four quadrants, and different colors (purple, green, blue) represent different rectangular bars.

[0040] The present invention selects a square canvas of preset size as the design space for each meta-atom, and allows the meta-atom to be designed based on randomly selected parameters (such as the number, size and position of rectangular bars) in the upper left quadrant, so that the design is no longer limited to a few fixed parameters, which greatly increases the diversity and complexity of the design and helps to achieve high efficiency and wide phase coverage. At the same time, the design in the upper left quadrant is mirrored along the x-axis and y-axis to form a complete meta-atom design. This method not only simplifies the design process, but also ensures the consistency and symmetry of the design, which helps to improve the stability of the electromagnetic response.

[0041] At the same time, in the present invention, since each design can generate a unique meta-atom structure according to different random parameters, this diversified design method can effectively expand the working range of the model, so that it is no longer limited to a specific set of design parameters. Even in the face of changing design parameters, it can quickly adapt and generate corresponding designs, reducing the need to re-collect data and retrain models.

[0042] The scattering parameters of the calculated meta-atom design are specifically: Setting boundary conditions: Use unit cell boundary conditions to simulate the electromagnetic response of the meta-atom design in a square lattice structure, and apply open boundary conditions in the direction perpendicular to the plane of the meta-atom design, i.e., in the z direction, to ensure that the wavefront can propagate freely; Define the illumination method: Use x-polarized plane waves to illuminate the meta-atom design from the substrate side to simulate the incident light conditions in actual application scenarios; Determine the frequency range: Calculate the electromagnetic response of the meta-atom design in the 30 to 60 THz frequency range, including transmittance and phase shift information; Based on the set boundary conditions, illumination mode and frequency range, the transmittance and phase shift of the meta-atom design are calculated using full-wave electromagnetic simulation tools; Based on the obtained transmittance and phase shift data, the phase and amplitude of the meta-atom design are calculated.

[0043] The initial network model is trained by the data set to obtain a scattering parameter prediction model; the scattering parameters of the meta-atom design to be tested are predicted by the scattering parameter prediction model. Specifically, the initial network model is trained and tested by dividing the data set into a training set and a test set.

[0044] The present invention constructs an initial network model including a splitting module, a first channel, a second channel, a stacking module, a comprehensive feature channel and a calculation module, thereby effectively decomposing the meta-atom design and extracting its spatial features and physical features. At the same time, the initial network model is trained by using a plurality of meta-atom design samples with different design structures and design parameters, thereby greatly improving the generalization ability of the model. The method of the present invention can accurately predict the real part and the imaginary part of the transmission coefficient corresponding to the meta-atom design in a relatively short time, and further calculate the scattering parameters, thereby greatly improving the efficiency and accuracy of the meta-atom scattering parameter prediction.

[0045] Specifically, this embodiment uses the hyperparameters shown in Table 1 below to train the initial network model: Table 1:

[0046] Embodiment 2

[0047] The scattering parameter prediction model proposed in the present invention is not limited to a single application scenario, it can also be used as an optimization tool in meta-surface or meta-atom design methods based on deep neural networks (DNNs). A significant advantage of DNN-based design methods is that they can generate a large number of design options at a very low cost. However, all of these designs require a detailed characterization and evaluation process to determine the best solution, which is often more time-consuming than the generation of the design itself, especially when relying on complex simulation tools for performance verification.

[0048] The scattering parameter prediction model proposed in this invention greatly alleviates the above problems by virtue of its fast response prediction capability. Figure 2 As shown, the embodiment of the present invention also proposes a method for constructing a meta-device, and the method comprises: S1: Design a metasurface phase mask and decompose the metasurface phase mask into the performance requirements of each unit cell; a unit cell of the metasurface phase mask represents a meta-atom; S2: Set a serial number for the cell, the initial value of the serial number is equal to 1; S3: Input the cell performance requirements corresponding to the current sequence number into the pre-trained GAN model; S4: Generate a meta-atom design set through the GAN model; the meta-atom design set includes multiple meta-atom designs; S5: inputting the meta-atom designs in the meta-atom design set into the scattering parameter prediction model described above in sequence for prediction; obtaining the optimal meta-atom design of the corresponding cell according to the model prediction result corresponding to each meta-atom design in the meta-atom design set;

[0049] The step S5 specifically includes:

[0050] S50: Set the target number of iterations and initialize the number of iterations to 1;

[0051] S51: inputting the meta-atom designs in the meta-atom design set into the scattering parameter prediction model as described above in sequence for prediction;

[0052] S52: Based on the model prediction results of each meta-atom design, the corresponding key performance indicators are calculated; the key performance indicators include transmission efficiency and phase shift accuracy.

[0053] S53: Select the best meta-atom design from the meta-atom design set as the current optimal design based on the key performance indicators; determine whether the current number of iterations is greater than or equal to the target number of iterations, if not, increase the number of iterations by 1 and proceed to the next step, if yes, jump to step S55;

[0054] It should be noted that in order to select the best meta-atom design, this embodiment adopts an evaluation method based on key performance indicators. The key performance indicators mainly include transmission efficiency and phase shift accuracy. The transmission efficiency measures the ability of the meta-atom to transmit light waves, while the phase shift accuracy reflects the accuracy of the phase change caused by the meta-atom. In order to comprehensively evaluate the overall performance of each meta-atom design, this embodiment sets corresponding weights for these two indicators. By multiplying the transmission efficiency and phase shift accuracy by their respective weights and taking the weighted sum, the comprehensive score of each meta-atom design can be calculated, and the design with the highest comprehensive score is selected as the current optimal design. This method can not only flexibly balance the importance of different performance indicators, but also ensure that the final selected design achieves optimal or near-optimal performance in multiple dimensions.

[0055] S54: Feedback the current optimal design to the GAN model, wherein the GAN model uses the current optimal design as a reference design, regenerates a new meta-atom design set, and returns to step S51;

[0056] S55: the current optimal design is set as the optimal meta-atom design, and the iteration is terminated.

[0057] The present invention introduces an iterative optimization mechanism in step S5, which continuously feeds back the current optimal design to the GAN model to generate a new set of meta-atom designs, and repeats the evaluation and selection process until a predetermined number of iterations is reached. This method effectively ensures the superiority of the final selected meta-atom design in key performance indicators, such as transmission efficiency and phase shift accuracy, and maximizes the performance of the meta-device.

[0058] S6: Determine whether the current sequence number is greater than or equal to the maximum sequence number of the cell. If not, add 1 to the sequence number and return to step S3; if yes, proceed to the next step;

[0059] S7: Assemble into meta-devices through optimal meta-atom design of each unit cell.

[0060] The meta-device in this embodiment is specifically a transmissive meta-lens. Specifically:

[0061] This embodiment uses a well-trained generative adversarial network (GAN) model to design a transmissive metalens operating at a frequency of 57 THz. The metalens is composed of 50×50 meta-atoms, with a lattice size set to 2.6 μm, forming a device with a total area of ​​130 μm×130 μm. The focal length of this metalens is set to 80 μm, with a numerical aperture (NA) of 0.63, and is composed of meta-atoms made of a dielectric material with a refractive index of 4.7 and a thickness of 0.75 μm, placed on a substrate with a refractive index of 1.4. After calculating the required metasurface phase mask, a GAN model specifically for generating meta-atom designs is further trained to generate specific meta-atom designs for each unit cell within the transmissive metalens. The generated meta-atom designs are then evaluated using the scattering parameter prediction model proposed in the present invention.

[0062] This cascaded design-evaluation process undergoes multiple iterations, aiming to select the optimal meta-atom design for each unit cell and finally assemble it into a complete meta-device. In order to verify the prediction accuracy of the scattering parameter prediction model and the effectiveness of the entire design-evaluation process, e.g. Figure 4 As shown, this embodiment uses this combined network (i.e., the GAN model plus the scattering parameter prediction model) for 1, 2, 5, and 10 iterations to generate four different transmissive metalenses, and their performance is tested using a full-wave simulation tool. The results are shown in Figure 4 , where the transmissive metalens is placed in the xy plane with the optical axis along the z direction. Figure 4The two-dimensional electric field distributions of four different transmission metalenses simulated on the xz plane and the one-dimensional electric field distribution along the x-axis on the focal plane are presented. As the number of optimization iterations increases, the peak electric field intensity at the focal center gradually increases, which indicates that during the optimization process, the meta-atom design with higher transmission efficiency and more accurate phase shift is identified through the scattering parameter prediction model, thus proving the effectiveness and superiority of the method of the present invention. It is also necessary to explain in detail that Figure 4 middle:

[0063] 1. E(V / m) stands for Electric Field, and its unit is volts per meter (V / m). In the curve chart on the left side of the figure, the vertical axis is marked with the electric field strength value, reflecting the electric field distribution at different locations.

[0064] 2. Z (μm) represents the position coordinate along the optical axis, in micrometers (μm). In the curve chart on the left side of the figure, the horizontal axis marks the position of the Z axis, showing how the electric field intensity changes with the Z axis.

[0065] 3. Iterations = 1, 2, 5, 10: correspond to the four sub-graphs in the figure. Each sub-graph shows the two-dimensional electric field distribution of the transmission metalens at different iterations. As the iterations increase, the peak electric field amplitude at the focal center increases, indicating the effectiveness of the optimization process.

[0066] 4. X, Y, Z:

[0067] X and Y: represent the horizontal and vertical directions in the spatial coordinate system, which are used to describe the position of the transmissive metalens in the plane.

[0068] Z: represents the direction of the optical axis, that is, the direction in which light propagates.

[0069] The X, Y, and Z axes are marked in the sub-figure in the lower right corner to help understand the spatial orientation of the electric field distribution.

[0070] 5. Gradient color bar: Indicates the magnitude of the electric field strength. The color changes from light to dark, corresponding to the electric field strength from low to high. The distribution of the electric field strength at different locations is intuitively displayed through the change of color.

[0071] 6. In the color bar in the upper left corner of the image: (1) Blue (marked as 1): represents the electric field intensity distribution curve when the iteration number is 1, that is, the change of the electric field intensity in the focal area; (2) Orange (marked as 2): represents the electric field intensity distribution curve when the number of iterations is 2; (3) Yellow (marked as 5): represents the electric field intensity distribution curve when the number of iterations is 5; (4) Purple (marked as 10): represents the electric field intensity distribution curve when the number of iterations is 10.

[0072] The present invention generates multiple meta-atom designs using a pre-trained GAN model and evaluates the performance of each design using the above scattering parameter prediction model to select the optimal meta-atom design. This process ensures that each unit cell can meet specific performance requirements, thereby improving the design and manufacturing accuracy of the entire meta-device.

[0073] It should be noted that the term "including" and its variations used in the embodiments of the present invention are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". The modifications of "one" and "multiple" mentioned in the embodiments of the present invention are illustrative rather than restrictive. Those skilled in the art should understand that unless the context clearly indicates otherwise, it should be understood as "one or more".

[0074] The term "embodiment" in this specification refers to specific features, structures or characteristics described in conjunction with the embodiment that can be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. The various embodiments in this specification are described in a related manner, and the same or similar parts between the various embodiments refer to each other. In particular, for the device, equipment, and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts refer to the partial description of the method embodiment.

[0075] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the attached claims.

Claims

1. A method for predicting scattering parameters of meta-atoms, characterized in that: include: Construct an initial network model including a split module, a first channel, a second channel, a stacking module, a comprehensive feature channel, and a calculation module; wherein: The splitting module is used to decompose the meta-atom design into a two-dimensional image tensor and a one-dimensional attribute tensor; the meta-atom design includes a design structure and a design parameter; the design structure corresponds to the two-dimensional image tensor, and the design parameter corresponds to the one-dimensional attribute tensor; The first channel is used to extract the spatial features of the two-dimensional image tensor through convolution and pooling operations to obtain an image feature vector; The second channel is used to extract the physical features of the one-dimensional attribute tensor through nonlinear transformation and spatial tiling operations to obtain an attribute feature vector; The stacking module is used to stack the image feature vector and the attribute feature vector to obtain a splicing feature; The comprehensive feature channel is used to process the splicing features through convolution and pooling operations to obtain electromagnetic response features; The calculation module is used to obtain the real part and the imaginary part of the transmission coefficient corresponding to the meta-atom design according to the electromagnetic response characteristics, and calculate the scattering parameters based on the real part and the imaginary part of the transmission coefficient; wherein the scattering parameters include phase and amplitude; Generate multiple meta-atom designs with different design structures and design parameters, calculate the scattering parameters of the meta-atom designs, and label the corresponding meta-atom designs with the scattering parameters to obtain meta-atom design samples, and form a data set through the meta-atom design samples; the design parameters include: refractive index, thickness and lattice size; The initial network model is trained by the data set to obtain a scattering parameter prediction model; the scattering parameters of the meta-atom design to be tested are predicted by the scattering parameter prediction model.

2. A method for predicting scattering parameters of meta-atoms according to claim 1, characterized in that: The first channel comprises: The first convolution layer is used to perform a first convolution calculation on the two-dimensional image tensor to obtain a first spatial feature; A first maximum pooling layer is used to perform a maximum pooling operation on the first spatial feature to obtain a pooling feature; The second convolution layer is used to perform a second convolution calculation on the pooled features to obtain the second spatial features; The second maximum pooling layer is used to perform a maximum pooling operation on the second spatial feature to obtain an image feature vector.

3. A method for predicting scattering parameters of meta-atoms according to claim 2, characterized in that: The second channel comprises: The nonlinear transformation layer is used to perform nonlinear transformation on the one-dimensional attribute tensor to obtain the transformation features; The spatial tiling layer is used to perform a spatial tiling operation on the transformed features to obtain an attribute feature vector.

4. A method for predicting scattering parameters of meta-atoms according to claim 3, characterized in that: The comprehensive feature channel includes the following connected in sequence: The third convolutional layer is used to perform convolution operations on the concatenated features to obtain the initial convolution features; The third maximum pooling layer is used to perform a maximum pooling operation on the initial convolutional features to obtain the initial pooling features; The fourth maximum pooling layer is used to perform the maximum pooling operation on the initial pooling features again to obtain the target pooling features; The fourth convolutional layer is used to perform convolution calculation on the target pooling feature and flatten the result obtained by the convolution calculation to obtain the target convolution feature; The first fully connected layer is used to perform a fully connected operation on the target convolutional features to obtain the initial fully connected features; The second fully connected layer is used to transform the initial fully connected features to obtain electromagnetic response features.

5. A method for predicting scattering parameters of meta-atoms according to claim 4, characterized in that: In the comprehensive feature channel, the third convolutional layer, the third maximum pooling layer, the fourth maximum pooling layer, the fourth convolutional layer and the first fully connected layer all include a batch normalization unit and a ReLU activation function; The batch normalization unit is used to perform batch normalization processing on the features obtained after the respective layers perform corresponding operations; The ReLU activation function is used to perform a nonlinear transformation on the result after batch normalization processing, and finally obtain the output feature corresponding to the layer; the output feature is the initial pooling feature, the target pooling feature, the target convolution feature or the initial fully connected feature.

6. A method for predicting scattering parameters of meta-atoms according to claim 1, characterized in that: The generating of a plurality of meta-atom designs with different design structures and design parameters is specifically as follows: Select a square canvas of preset size as the design space for each meta-atom; For each meta-atom, a preset rectangular strip is designed based on randomly selected parameters in the upper left quadrant of the canvas; The randomly selected parameters include: the number, size and position of the rectangular bars; each rectangular bar represents a set type of high refractive index material, and the rectangular bars correspond to the types of high refractive index materials one by one; Mirror the design in the upper left quadrant along the x-axis and y-axis to form the complete design of each meta-atom; By adjusting the randomly selected parameters and repeating the above steps, a plurality of meta-atom designs with different design structures and design parameters are generated.

7. A method for predicting scattering parameters of meta-atoms according to claim 6, characterized in that: The scattering parameters of the calculated meta-atom design are specifically: Setting boundary conditions: Use unit cell boundary conditions to simulate the electromagnetic response of the meta-atom design in a square lattice structure, and apply open boundary conditions in the direction perpendicular to the plane of the meta-atom design, i.e., in the z direction, to ensure that the wavefront can propagate freely; Define the illumination method: Use x-polarized plane waves to illuminate the meta-atom design from the substrate side to simulate the incident light conditions in actual application scenarios; Determine the frequency range: Calculate the electromagnetic response of the meta-atom design in the 30 to 60 THz frequency range, including transmittance and phase shift information; Based on the set boundary conditions, illumination mode and frequency range, the transmittance and phase shift of the meta-atom design are calculated using full-wave electromagnetic simulation tools; Based on the obtained transmittance and phase shift data, the phase and amplitude of the meta-atom design are calculated.

8. A method for constructing a meta-device, characterized in that: include: S1: Design a metasurface phase mask and decompose the metasurface phase mask into the performance requirements of each unit cell; a unit cell of the metasurface phase mask represents a meta-atom; S2: Set a serial number for the cell, the initial value of the serial number is equal to 1; S3: Input the cell performance requirements corresponding to the current sequence number into the pre-trained GAN model; S4: Generate a meta-atom design set through the GAN model; the meta-atom design set includes multiple meta-atom designs; S5: inputting the meta-atom designs in the meta-atom design set into the scattering parameter prediction model in the method according to any one of claims 1 to 7 in sequence for prediction; According to the model prediction result corresponding to each meta-atom design in the meta-atom design set, the optimal meta-atom design of the corresponding cell is obtained; S6: Determine whether the current sequence number is greater than or equal to the maximum sequence number of the cell. If not, add 1 to the sequence number and return to step S3; If yes, proceed to the next step; S7: Assemble into meta-devices through optimal meta-atom design of each unit cell.

9. A method for constructing a metadevice according to claim 8, characterized in that: The step S5 specifically includes: S50: Set the target number of iterations and initialize the number of iterations to 1; S51: inputting the meta-atom designs in the meta-atom design set into the scattering parameter prediction model in the method according to any one of claims 1 to 7 in sequence for prediction; S52: Based on the model prediction results of each atomic design, calculate the corresponding key performance indicators; S53: Select the best meta-atom design from the meta-atom design set as the current optimal design based on the key performance indicators; determine whether the current number of iterations is greater than or equal to the target number of iterations, if not, increase the number of iterations by 1 and proceed to the next step, if yes, jump to step S55; S54: Feedback the current optimal design to the GAN model, wherein the GAN model uses the current optimal design as a reference design, regenerates a new meta-atom design set, and returns to step S51; S55: the current optimal design is set as the optimal meta-atom design, and the iteration is terminated.

10. A method for constructing a metadevice according to claim 9, characterized in that: The key performance indicators include transmission efficiency and phase shift accuracy.

Images