Optical parameter calculation, prediction model training and wavelength selection emitter design method

Abstract

The invention relates to the technical field of thin film structure design, and particularly provides an optical parameter calculation method, a prediction model training method and a wavelength selection emitter design method. According to the optical parameter calculation method provided by the invention, the optical parameters of the target material at any preset temperature can be calculated based on crystal structure data of the target material, a density functional theory, a Lorentz phonon resonance model, a first principle molecular dynamics simulation method and a Boltzmann transport equation; an optical database of a material can be perfected, so that the design coverage of the wavelength selection emitter is more comprehensive. And on the basis of the Lorentz phonon resonance model, in the process of calculating the optical parameters of the target material, phonon spectrum infrared activity optical branch transverse and longitudinal wave splitting is considered, so that the accuracy of the calculated optical parameters of the infrared spectrum band can be improved, and the designed wavelength selection emitter is further optimized.

Description

Technical Field

[0001] The present application relates to the technical field of thin film structure design, and more specifically, to optical parameter calculation, prediction model training, and wavelength selective emitter design methods. Background Art

[0002] Wavelength-selective emitters are widely used in the field of high-temperature thermal radiation control. Examples include thermophotovoltaic cells operating at ultra-high temperatures (>1500K), infrared radiation control systems operating at medium- to high-temperatures (600-1200K), high-efficiency solar cells and radiation cooling devices operating at low temperatures (300-600K), and gas sensors and photosynthetic membranes operating at room temperature (approximately 300K).

[0003] However, thermal radiation is closely related to device temperature, but the existing material optical parameters obtained mainly through testing are limited, especially the high-temperature optical parameter testing is difficult, resulting in an incomplete material optical database, which in turn leads to incomplete design coverage of wavelength selective emitters. Summary of the Invention

[0004] In view of this, the purpose of the embodiments of the present application is to provide a method for optical parameter calculation, prediction model training and wavelength selective transmitter design to solve the above technical problems.

[0005] In a first aspect, an embodiment of the present application provides a method for calculating optical parameters, the method comprising:

[0006] Calculating the optical frequency dielectric constant, the Born effective charge, and the initial phonon resonance frequency of the target material at the initial temperature based on the crystal structure data and density functional theory of the target material;

[0007] performing thermodynamic calculations on the target material based on the Born effective charge, the initial phonon resonance frequency, and a first-principles molecular dynamics simulation method to obtain a relaxed crystal structure of the target material at a preset temperature;

[0008] The optical parameters of the target material at the preset temperature are calculated based on the optical frequency dielectric constant, the relaxed crystal structure, the density functional theory, the Boltzmann transport equation, and the Lorentz phonon resonance model.

[0009] In the above implementation process, the optical parameter calculation method calculates the optical frequency dielectric constant, Born effective charge and initial phonon resonance frequency of the target material at the initial temperature based on the crystal structure data of the target material and density functional theory; according to the Born effective charge, initial phonon resonance frequency and first-principles molecular dynamics simulation method, the target material is thermodynamically calculated to obtain the relaxed crystal structure of the target material at a preset temperature; based on the optical frequency dielectric constant, relaxed crystal structure, density functional theory, Boltzmann transport equation and Lorentz phonon resonance model, the optical parameters of the target material at a preset temperature are calculated.

[0010] This optical parameter calculation method can calculate the optical parameters of a target material at any preset temperature based on its crystal structure data, density functional theory, the Lorentz phonon resonance model, first-principles molecular dynamics simulations, and the Boltzmann transport equation. This method can improve the optical database of the material and thus provide a more comprehensive design coverage for wavelength-selective emitters. Furthermore, based on the Lorentz phonon resonance model, the calculation of the target material's optical parameters takes into account the splitting of transverse and longitudinal waves in the infrared-active optical branches of the phonon spectrum, thereby improving the accuracy of the calculated optical parameters in the infrared spectral range. With more accurate optical parameters in the infrared spectral range, the designed wavelength-selective emitter can be further optimized.

[0011] Optionally, in an embodiment of the present application, the optical parameters of the target material at the preset temperature are calculated based on the optical frequency dielectric constant, the relaxed crystal structure, the density functional theory, the Boltzmann transport equation and the Lorentz phonon resonance model, including: calculating the preset phonon resonance frequency and the preset phonon scattering rate of the target material at the preset temperature based on the relaxed crystal structure, the density functional theory and the Boltzmann transport equation; calculating the optical parameters of the target material at the preset temperature according to the optical frequency dielectric constant, the preset phonon resonance frequency, the preset phonon scattering rate and the Lorentz phonon resonance model; wherein the preset phonon resonance frequency includes the branch phonon resonance frequency under different optical branches corresponding to different infrared active phonon modes; the preset phonon scattering rate includes the branch phonon scattering rate under different optical branches corresponding to different infrared active phonon modes.

[0012] In the above-mentioned implementation process, the preset phonon resonance frequency includes the branch phonon resonance frequency under different optical branches corresponding to different infrared active phonon modes, and the preset phonon scattering rate includes the branch phonon scattering rate under different optical branches corresponding to different infrared active phonon modes. Therefore, based on the optical frequency dielectric constant, the preset phonon resonance frequency (that is, the branch phonon resonance frequency under different optical branches corresponding to different infrared active phonon modes), the preset phonon scattering rate (that is, the branch phonon scattering rate under different optical branches corresponding to different infrared active phonon modes) and the Lorentz phonon resonance model, the optical parameters of the target material at a preset temperature can be calculated while taking into account the transverse and longitudinal wave splitting of the infrared active optical branches of the phonon spectrum, thereby improving the calculation accuracy of the optical parameters of the infrared spectrum.

[0013] Optionally, in an embodiment of the present application, the optical parameters of the target material at the preset temperature are calculated according to the optical frequency dielectric constant, the preset phonon resonance frequency, the preset phonon scattering rate, and the Lorentz phonon resonance model, including: calculating the total branch phonon scattering rate under the corresponding optical branch corresponding to the corresponding infrared active phonon mode according to the branch phonon resonance frequency; calculating the total branch phonon scattering rate under the corresponding optical branch corresponding to the corresponding infrared active phonon mode according to the optical frequency dielectric constant, the branch phonon resonance frequency, the total branch phonon scattering rate, and the Lorentz phonon resonance model. Calculate the complex dielectric constant of the target material at the preset temperature; wherein E(ω) represents the complex dielectric constant corresponding to the frequency ω of the infrared wavelength λ, ε ∞ represents the optical frequency dielectric constant, ω j,LO represents the branch phonon resonance frequency under the longitudinal optical branch corresponding to the jth infrared active phonon mode, ω j,TO represents the branch phonon resonance frequency under the transverse optical branch corresponding to the jth infrared active phonon mode, γ j,LO represents the total scattering rate of branch phonons under the longitudinal optical branch corresponding to the jth infrared active phonon mode, γ j,To represents the total scattering rate of branch phonons under the transverse optical branch corresponding to the jth infrared active phonon mode, i is an imaginary unit; according to the complex dielectric constant, N 2 =E(ω) and N=n+ik, determine the optical parameters of the target material at the preset temperature; wherein N represents the complex refractive index of the target material at the preset temperature, n represents the refractive index of the target material at the preset temperature, and k represents the extinction coefficient of the target material at the preset temperature.

[0014] In the above implementation process, the complex dielectric constant of the target material at a preset temperature can be calculated through the optical frequency dielectric constant, branch phonon resonance frequency, branch phonon scattering rate and Lorentz phonon resonance model, and then the optical parameters of the target material at a preset temperature can be determined based on the complex dielectric constant.

[0015] In a second aspect, an embodiment of the present application provides a reflectivity prediction model training method, the method comprising:

[0016] Determine input data based on the optical parameter calculation method described in any one of the first aspects above, the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, and the optical parameters of each of the preselected materials at the preselected temperature;

[0017] Inputting the input data into the model to be trained to obtain output data output by the model to be trained; wherein the output data includes the predicted reflectivity of the multilayer film structure wavelength selective emitter for light of different wavelength bands within the preselected wavelength range at the preselected temperature;

[0018] Determining a loss function value of the model to be trained based on the predicted reflectivity and the actual reflectivity;

[0019] According to the loss function value, the model parameters of the model to be trained are adjusted until the adjusted model meets the model training conditions, thereby obtaining a trained reflectivity prediction model.

[0020] In the above-mentioned implementation process, by using the optical parameter calculation method as described in any one of the first aspects, the optical parameters of the preselected material at any preselected temperature can be calculated, so that the trained reflectivity prediction model can accurately predict the "reflectivity of the multi-layer film structure wavelength selective emitter to different bands of light within the preselected wavelength range at the preselected temperature", thereby improving the design efficiency of the multi-layer film structure wavelength selective emitter.

[0021] Optionally, in an embodiment of the present application, the input data is determined based on the preselected wavelength range, the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, and the optical parameters of each of the preselected materials at the preselected temperature, including: randomly generating W initial arrays of dimensions 2S×1 based on the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter; wherein S represents the number of layers of the film structure, and the initial array includes the material and thickness parameters of each film layer; based on the optical parameters of each of the preselected materials at the preselected temperature, the initial array is dimensionalized to obtain a three-dimensional array of S×(2c+1)×b; wherein b represents the number of the preselected temperatures, and c represents the number of different wavelengths of light within the preselected wavelength range.

[0022] In the aforementioned implementation, an initial array of dimensions 2S × 1 is generated based on the material and thickness of each layer in the multilayer wavelength selective emitter. This initial array is then transformed based on the optical parameters of each preselected material at a preselected temperature to obtain a three-dimensional array of dimensions S × (2c + 1) × b. Because the transformed three-dimensional array includes the material and thickness of each layer, as well as the refractive index and extinction coefficient of each preselected material at a preselected temperature for light of different wavelengths, this transformed three-dimensional array can be used as the aforementioned input data.

[0023] Optionally, in an embodiment of the present application, inputting the input data into the model to be trained includes: dividing the three-dimensional array into b S×(2c+1)×1 arrays; flattening each S×(2c+1)×1 array into a one-dimensional array of (2S×c+S)×1, and inputting the flattened one-dimensional array into the model to be trained.

[0024] In the above implementation, the three-dimensional array is divided into b S×(2c+1)×1 arrays, each S×(2c+1)×1 array is flattened into a (2S×c+S)×1 one-dimensional array, and the flattened one-dimensional array is input into the model to be trained as input data. By converting the input data into a one-dimensional array, the compatibility and robustness of the trained reflectivity prediction model can be improved.

[0025] Optionally, in an embodiment of the present application, the predicted reflectivity is calculated based on input data using a rigorous coupled wave analytical algorithm.

[0026] In the aforementioned implementation, a rigorous coupled-wave analytical algorithm is used to accurately calculate the actual reflectivity of a multilayer film-structured wavelength-selective emitter for light of different wavelengths at different preselected temperatures. The loss function value of the trained model is determined based on the actual and predicted reflectivities. The model parameters of the trained model are then adjusted based on the loss function value to obtain a trained reflectivity prediction model. This trained reflectivity prediction model is then used to improve the efficiency of reflectivity calculations for the multilayer film-structured wavelength-selective emitter.

[0027] Optionally, in an embodiment of the present application, adjusting the model parameters of the model to be trained according to the loss function value includes: using the gradient descent method to calculate the partial derivatives of the loss function value with respect to the model parameters of the model to be trained, and updating the model parameters of the model to be trained through gradient back propagation.

[0028] In a third aspect, an embodiment of the present application provides a method for designing a wavelength selective emitter with a multilayer film structure, the method comprising:

[0029] Generate an initialization population; wherein the initialization population includes a plurality of individuals, each of the individuals corresponds to an initialized multilayer film structure wavelength selective emitter;

[0030] Based on the optical parameter calculation method as described in any one of the first aspects, determining the optical parameters of each layer of preselected material at a preselected temperature in the initialized multi-layer film structure wavelength selective emitter;

[0031] Based on the trained reflectivity prediction model, calculating the initial reflectivity of the initialized multilayer film structure wavelength selective emitter to light of different wavelength bands within the preselected wavelength range at the preselected temperature; wherein the trained reflectivity prediction model is obtained by training based on the reflectivity prediction model training method according to any one of the second aspects;

[0032] Calculating the fitness of each individual according to the initial reflectivity;

[0033] A new population is generated through crossover and mutation operations until preset conditions are met, and the individual with the greatest fitness is determined as the designed multi-layer film structure wavelength selective emitter.

[0034] In the above implementation process, based on the optical parameter calculation method described in any one of the first aspects, the optical parameters of the preselected material at any preselected temperature can be calculated, so that the coverage of the designed multilayer film structure wavelength selective emitter is more comprehensive. Since the design process of the multilayer film structure wavelength selective emitter involves the complex interaction of multiple parameters such as optical parameters, preselected temperature, preselected material, and preselected wavelength range, traditional design methods may fall into a local optimal solution, resulting in poor wavelength selection performance of the emitter. The multilayer film structure wavelength selective emitter design method provided by the present application, based on the reflectivity prediction model combined with the genetic algorithm and the neural network, can search in a wider parameter space to find a better design solution, so that the reflectivity performance indicators of the designed multilayer film structure wavelength selective emitter in different bands are better balanced, thereby improving the wavelength selection accuracy and efficiency of the designed multilayer film structure wavelength selective emitter. In addition, based on the trained reflectivity prediction model, the initial reflectivity of the initialized multilayer film structure wavelength selective emitter for different wavelength bands within the preselected wavelength range is obtained at the preselected temperature, which can improve the efficiency of reflectivity acquisition and thereby improve the design efficiency of the multilayer film structure wavelength selective emitter. Therefore, based on the above-mentioned design method provided in this application, a rapid and universal design of a multi-layer film structure wavelength selective emitter that is compatible with a wide temperature range and full infrared can be achieved.

[0035] Optionally, in an embodiment of the present application, the initial reflectivity of the initialized multi-layer film structure wavelength selective emitter to different wavelengths of light within the preselected wavelength range at the preselected temperature is calculated based on the trained reflectivity prediction model, including: randomly generating W initial arrays of dimension 2S×1 based on the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter; wherein S represents the number of layers of the film structure, and the initial array includes the material and thickness parameters of each film layer; based on the optical parameters of each of the preselected materials at the preselected temperature, performing a dimensionality increase transformation on the initial array to obtain a three-dimensional array of S×(2c+1)×b; wherein b represents the number of the preselected temperatures, and c represents the number of different wavelengths of light within the preselected wavelength range; dividing the three-dimensional array into b S×(2c+1)×1 arrays; flattening each S×(2c+1)×1 array into a one-dimensional array of (2S×c+S)×1, and inputting the flattened one-dimensional array into the trained reflectivity prediction model to obtain the initial reflectivity output by the model.

[0036] In the above implementation process, compared with the strict coupled wave analytical algorithm, obtaining the initial reflectivity based on the trained reflectivity prediction model can improve the efficiency of obtaining the initial reflectivity, thereby improving the design efficiency of the multi-layer film structure wavelength selective emitter.

[0037] The beneficial effects of the present application include at least: the optical parameter calculation method can calculate the optical parameters of the target material at any preset temperature within a limited temperature range based on the crystal structure data of the target material, density functional theory, Lorentz phonon resonance model, first-principles molecular dynamics simulation method and Boltzmann transport equation, and can improve the optical database of the material, thereby making the design coverage of the wavelength selective emitter more comprehensive. Specifically, based on the optical parameter calculation method provided in the present application, the optical parameters of a wide temperature range (300K-1500K) and a full infrared spectrum range (1μm-25μm) can be calculated, and then the design of a wavelength selective emitter compatible with a wide temperature range and full infrared can be realized based on the calculated parameters. And based on the Lorentz phonon resonance model, in the process of calculating the optical parameters of the target material, the transverse and longitudinal wave splitting of the infrared active optical branch of the phonon spectrum is taken into account, which can improve the accuracy of the calculated optical parameters of the infrared spectrum. Based on more accurate optical parameters of the infrared spectrum, the designed wavelength selective emitter can be further optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 A schematic diagram of a flow chart of an optical parameter calculation method provided in an embodiment of the present application;

[0040] Figure 2 A schematic diagram of optical parameters of a target material at a preset temperature provided in an embodiment of the present application;

[0041] Figure 3 A flow chart of a reflectivity prediction model training method provided in an embodiment of the present application;

[0042] Figure 4 A schematic flow chart of a method for designing a wavelength selective emitter with a multilayer film structure provided in an embodiment of the present application;

[0043] Figure 5 A schematic diagram of the fitness function and optimized design time provided in an embodiment of the present application;

[0044] Figure 6 A schematic diagram of the optimized design time corresponding to the design methods for wavelength selective emitters with different multilayer film structures provided in the embodiments of the present application;

[0045] Figure 7 A schematic diagram of a designed spectrum provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] The following embodiments of the technical solution of the present application will be described in detail with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present application and are therefore only examples and are not intended to limit the scope of protection of the present application.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0048] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is two or more, unless otherwise specifically defined.

[0049] See Figure 1 The flowchart of an optical parameter calculation method provided by an embodiment of the present application is shown. The optical parameter calculation method may include the following steps:

[0050] S101. Calculating the optical frequency dielectric constant, the Born effective charge, and the initial phonon resonance frequency at the initial temperature of the target material based on the crystal structure data of the target material and density functional theory;

[0051] S102, performing thermodynamic calculations on the target material based on the Born effective charge, the initial phonon resonance frequency, and a first-principles molecular dynamics simulation method to obtain a relaxed crystal structure of the target material at a preset temperature;

[0052] S103. Calculate the optical parameters of the target material at the preset temperature based on the optical frequency dielectric constant, the relaxed crystal structure, the density functional theory, the Boltzmann transport equation, and the Lorentz phonon resonance model.

[0053] In step S101, the target material can be any inorganic crystalline material, such as aluminum oxide (Al2O3), beryllium oxide (BeO), magnesium fluoride (MgF2), strontium titanate (SrTiO3), or titanium oxide (TiO2). Crystal structure data can provide information such as the atomic positions and lattice parameters of the target material. The crystal structure data of the target material can be obtained based on experiments (for example, X-ray diffraction, neutron diffraction, etc.). The optical frequency dielectric constant is a complex quantity that describes the electric polarization response of the target material within the optical wave frequency range. Its characteristics are mainly dominated by electronic transitions, manifested as significant dispersion and absorption effects. The Born effective charge reflects the strength of the ion's response to the electric field during the polarization process and is an important parameter that describes the degree of polarization of atoms or molecules under the action of an electric field. The initial temperature can be 0K, 100K, or other reasonable temperature lower than the preset temperature. The phonon resonant frequency refers to the phenomenon in which phonons (i.e., quasiparticles in lattice vibrations) resonate at a specific frequency in a phononic crystal. The initial phonon resonance frequency refers to the phonon resonance frequency at the initial temperature. The crystal structure of the target material can be constructed based on the crystal structure data and the corresponding software (for example, VASP, Quantum ESPRESSO or CASTEP, etc.). After selecting the appropriate exchange-correlation functional (for example, generalized gradient approximation or hybrid functional) and pseudopotential (for example, norm conservation pseudopotential), the calculation parameters (for example, k-point grid) are set to optimize the constructed crystal structure to obtain the stable crystal structure of the target material. The dielectric function of the target material is then calculated based on the stable crystal structure of the target material, and the optical frequency dielectric constant of the target material can be extracted from the dielectric function. The crystal structure data of the high-temperature stable crystalline phase of the target material can also be directly derived from the structure database of inorganic crystal materials, and the stable crystal structure of the target material can be constructed based on the crystal structure data of the high-temperature stable crystalline phase. The use of crystal structure data in a high-temperature stable state can more accurately calculate the infrared optical parameters of the target material at high temperature. Static self-consistent calculations can be performed based on the stable crystal structure of the target material to obtain the electronic ground state properties of the target material. Based on the static self-consistent calculation, a non-self-consistent calculation is performed to extract the Bern effective charge of the target material from the non-self-consistent calculation results. The quasi-harmonic approximation can be used to consider the effect of the initial temperature on the lattice vibration. The phonon frequency and lattice constant are calculated based on the quasi-harmonic approximation combined with density functional theory. The initial phonon resonance frequency of the target material at the initial temperature is then obtained through iterative calculation.

[0054] In step S102, first-principles molecular dynamics simulation is a molecular simulation method based on the principles of quantum mechanics. It directly solves the Schrödinger equation to describe the interaction between atomic nuclei and electrons, combined with molecular dynamics to simulate the dynamic behavior of the system. The target material can be subjected to thermodynamic calculations using the NPT and NVT canonical systems with appropriate step sizes and numbers to obtain the relaxed crystal structure of the target material at a preset temperature. The NPT canonical system monitors the system's temperature and pressure during the molecular dynamics simulation and maintains these values by adjusting atomic velocities or applying additional restraining forces. The NVT canonical system monitors the system's temperature and simulated volume during the molecular dynamics simulation and maintains these values by adjusting atomic velocities or applying additional restraining forces. The preset temperature can be 300K, 600K, 900K, 1500K, or other reasonable values within a finite temperature range. Finite temperature refers to temperatures above absolute zero, within which particles in the material are no longer in their lowest energy state but instead experience some thermal motion.

[0055] Wherein, in step S103, the optical parameters may include: the refractive index and / or extinction coefficient of the target material within a preset wavelength range at a preset temperature. The Lorentz phonon resonance model can be determined based on theoretical knowledge related to the transverse and longitudinal wave splitting phenomenon of infrared active optical branches in the phonon spectrum. The transverse and longitudinal wave splitting phenomenon of infrared active optical branches in the phonon spectrum is a frequency splitting phenomenon caused by long-range Coulomb interactions of optical phonon modes in ionic crystals or polar materials, which is specifically manifested as the frequency difference between longitudinal wave optical phonons (LO) and transverse wave optical phonons (TO). In polar materials, optical branch vibrations will produce changes in dipole moment, thereby coupling with electromagnetic waves (such as infrared light), which are called infrared active phonons. TO phonons appear as infrared absorption peaks in the spectrum due to coupling with infrared light; LO phonons usually do not appear directly in the infrared absorption spectrum because they cannot directly couple with transverse electromagnetic waves, but can be observed through Raman spectroscopy or inelastic neutron scattering. The phonon spectrum comprises multiple phonon modes. The splitting of transverse and longitudinal waves in the infrared-active optical branches of the phonon spectrum includes the splitting of transverse and longitudinal waves within different infrared-active phonon modes. By considering this splitting in the calculation of the target material's optical parameters, the accuracy of the calculated optical parameters in the infrared spectrum can be improved. This more accurate set of infrared optical parameters allows for further optimization of the designed wavelength-selective emitter.

[0056] It can be seen that the optical parameter calculation method provided in the embodiment of the present application can calculate the optical parameters of the target material at any preset temperature within a limited temperature range based on the crystal structure data of the target material, density functional theory, Lorentz phonon resonance model, first-principles molecular dynamics simulation method and Boltzmann transport equation, and can improve the optical database of the material, thereby making the design coverage of the wavelength selective emitter more comprehensive. Specifically, based on the optical parameter calculation method provided in the present application, the optical parameters of a wide temperature range (300K-1500K) and a full infrared spectrum range (1μm-25μm) can be calculated, and then a wavelength selective emitter design compatible with a wide temperature range and full infrared can be realized based on the calculated parameters. And based on the Lorentz phonon resonance model, in the process of calculating the optical parameters of the target material, the transverse and longitudinal wave splitting of the infrared active optical branch of the phonon spectrum is taken into account, which can improve the accuracy of the calculated optical parameters of the infrared spectrum. Based on more accurate optical parameters of the infrared spectrum, the designed wavelength selective emitter can be further optimized.

[0057] In some optional embodiments, S103, based on the optical frequency dielectric constant, the relaxed crystal structure, the density functional theory, the Boltzmann transport equation and the Lorentz phonon resonance model, the optical parameters of the target material at the preset temperature are calculated, including: based on the relaxed crystal structure, the density functional theory and the Boltzmann transport equation, the preset phonon resonance frequency and the preset phonon scattering rate of the target material at the preset temperature are calculated; according to the optical frequency dielectric constant, the preset phonon resonance frequency, the preset phonon scattering rate and the Lorentz phonon resonance model, the optical parameters of the target material at the preset temperature are calculated; wherein the preset phonon resonance frequency includes the branch phonon resonance frequency under different optical branches corresponding to different infrared active phonon modes; the preset phonon scattering rate includes the branch phonon scattering rate under different optical branches corresponding to different infrared active phonon modes.

[0058] Among them, the interatomic force constant matrix of the target material at a preset temperature can be first calculated based on the relaxed crystal structure and density functional theory; then, based on the interatomic force constant matrix, density functional theory, and the Boltzmann transport equation, the preset phonon resonance frequency and the preset phonon scattering rate of the target material at a preset temperature can be calculated. Then, based on the optical frequency dielectric constant and the phonon resonance frequency, phonon scattering rate, and Lorentz phonon resonance model of the target material at a preset temperature, the optical parameters of the target material at a preset temperature can be calculated. Based on the optical frequency dielectric constant, the preset phonon resonance frequency (that is, the branch phonon resonance frequency under different optical branches corresponding to different infrared active phonon modes), the preset phonon scattering rate (that is, the branch phonon scattering rate under different optical branches corresponding to different infrared active phonon modes), and the Lorentz phonon resonance model, the optical parameters of the target material at a preset temperature can be calculated while considering the transverse and longitudinal wave splitting of the infrared active optical branches of the phonon spectrum, thereby improving the calculation accuracy of the optical parameters of the infrared spectrum.

[0059] Among them, different optical branches may include transverse optical branches and longitudinal optical branches. The ground state properties of electrons in a periodic potential field can be calculated by relaxing the crystal structure and density functional theory, and the eigenvalues and wave functions of the electrons can be obtained, and then the interatomic force constant matrix can be calculated. The interatomic force constant matrix may include: a second-order interatomic force constant matrix, a third-order interatomic force constant matrix, and a fourth-order interatomic force constant matrix. A dynamic matrix can be constructed based on the interatomic force constant matrix, and the branch phonon resonance frequency of the target material at a preset temperature can be obtained by solving the eigenvalues and eigenvectors of the dynamic matrix. The phonon distribution function can be obtained based on the interatomic force constant matrix and the Boltzmann transport equation, and the branch phonon scattering rate of the target material at a preset temperature can be solved based on the phonon distribution function. Based on the relaxed crystal structure and density functional theory, the interatomic force constant matrix of the target material at a preset temperature is calculated. Furthermore, based on the interatomic force constant matrix, the branched phonon resonant frequencies of the target material at the preset temperature, corresponding to different optical branches of the target material's infrared-active phonon modes, are calculated. Based on the interatomic force constant matrix and the Boltzmann transport equation, the branched phonon scattering rates of the target material at the preset temperature, corresponding to different optical branches of the target material's infrared-active phonon modes, are calculated. Based on the branched phonon resonant frequencies, branched phonon scattering rates, and the Lorentz phonon resonance model of the target material at the preset temperature, the optical parameters of the target material can be more precisely and accurately calculated, further optimizing the designed wavelength-selective emitter.

[0060] In some optional embodiments, the above calculation of the optical parameters of the target material at a preset temperature based on the optical frequency dielectric constant, the preset phonon resonant frequency, the preset phonon scattering rate and the Lorentz phonon resonance model includes: calculating the total branch phonon scattering rate under the corresponding optical branch corresponding to the corresponding infrared active phonon mode based on the branch phonon scattering rate; calculating the total branch phonon scattering rate under the corresponding optical branch corresponding to the corresponding infrared active phonon mode based on the optical frequency dielectric constant, the branch phonon resonant frequency, the branch phonon total scattering rate and the Lorentz phonon resonance model. Calculate the complex dielectric constant of the target material at a preset temperature; where E(ω) represents the complex dielectric constant corresponding to the frequency ω of the infrared wavelength λ, ε ∞ represents the optical frequency dielectric constant, ω j,LO represents the branch phonon resonance frequency under the longitudinal optical branch corresponding to the jth infrared active phonon mode, ω j,TO represents the branch phonon resonance frequency under the transverse optical branch corresponding to the jth infrared active phonon mode, γ j,LO represents the total scattering rate of branch phonons under the longitudinal optical branch corresponding to the jth infrared active phonon mode, γ j,TO represents the total scattering rate of branch phonons under the transverse optical branch corresponding to the jth infrared active phonon mode, i is an imaginary unit; according to the complex dielectric constant, N 2 =E(ω) and N=n+ik, determine the optical parameters of the target material at a preset temperature; wherein N represents the complex refractive index of the target material at the preset temperature, n represents the refractive index of the target material at the preset temperature, and k represents the extinction coefficient of the target material at the preset temperature.

[0061] Among them, based on the complex dielectric constant corresponding to the infrared wavelength λ, the optical parameters of the target material at the infrared wavelength λ at a preset temperature can be calculated. Then, the optical parameters of the target material within the required wavelength range at the preset temperature are calculated. The wavelength interval between different infrared wavelengths can be 1μm, or 0.5μm, 2μm or other reasonable values. This application does not make specific restrictions on this. The interatomic force constant matrix can include a second-order interatomic force constant matrix, a third-order interatomic force constant matrix and a fourth-order interatomic force constant matrix. Based on the second-order interatomic force constant matrix, the third-order interatomic force constant matrix and the fourth-order interatomic force constant matrix, the branch phonon scattering rates corresponding to different optical branches under different infrared active phonon modes of the target material can be calculated at a preset temperature. The branch phonon scattering rate includes the third-order phonon scattering rate and the fourth-order phonon scattering rate. The total branch phonon scattering rate corresponding to different optical branches under the corresponding infrared active phonon mode can be determined based on the sum of the third-order phonon scattering rate and the fourth-order phonon scattering rate of the corresponding mode.

[0062] Please refer to Figure 2 , Figure 2 Schematic diagram of the optical parameters of the target material at a preset temperature provided in an embodiment of the present application. Figure 2 Specifically shown are schematic diagrams of optical parameters when the target material is aluminum oxide Al2O3 and the preset temperatures are 300K, 600K, 900K, 1200K and 1500K respectively. Figure 2 The upper figure in the figure specifically shows the optical parameter of refractive index, and the lower figure specifically shows the optical parameter of extinction coefficient. Figure 2 It can be seen that the refractive index n and extinction coefficient k of Al2O3 material change significantly with temperature; this indicates that the calculation of optical parameters in the high-temperature infrared spectrum range is crucial for the optimal design of wavelength-selective emitters.

[0063] Please refer to Figure 3 , Figure 3 A flow chart of a reflectivity prediction model training method provided in an embodiment of the present application. The reflectivity prediction model training method may include the following steps:

[0064] S201, determining the optical parameters of each preselected material in the multilayer film structure wavelength selective emitter at a preselected temperature;

[0065] S202, determining input data based on a preselected wavelength range, the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, and the optical parameters of each of the preselected materials at the preselected temperature;

[0066] S203, inputting the input data into the model to be trained, and obtaining output data output by the model to be trained; wherein the output data includes the predicted reflectivity of the multi-layer film structure wavelength selective emitter for light of different wavelength bands within the preselected wavelength range at the preselected temperature;

[0067] S204, determining a loss function value of the to-be-trained model based on the predicted reflectivity and the actual reflectivity;

[0068] S205. Adjust the model parameters of the model to be trained according to the loss function value until the adjusted model meets the model training conditions, thereby obtaining a trained reflectivity prediction model.

[0069] In step S201, the optical parameters of each preselected material in the multilayer film structure wavelength selective emitter at a preselected temperature are determined based on the optical parameter calculation method as described in any one of the first aspects. The number of layers of the multilayer film structure wavelength selective emitter can be 10, 12, 15, or other reasonable values. The preselected material can be any inorganic crystalline material, such as aluminum oxide Al2O3, beryllium oxide BeO, magnesium fluoride MgF2, strontium titanate SrTiO3, or titanium oxide TiO2. The preselected temperature can be 300K, 600K, 900K, 1200K, 1500K, or other reasonable values.

[0070] In steps S202 to S205, the preselected wavelength range may be an infrared spectrum range, specifically 1 μm to 25 μm, and the wavelength interval may be 1 μm (or other reasonable values). The thickness of each layer of the membrane structure may be in the thickness range of 100 nm to 1300 nm, and the thickness interval may be 10 nm (or other reasonable values). The model to be trained may be a fully connected neural network model, and may specifically be composed of an input layer containing hundreds of neurons (which may be 588 or other reasonable values), multiple hidden layers each containing hundreds of neurons (which may be 200 or other reasonable values), and an output layer containing dozens of neurons (which may be 24 or other reasonable values). Among them, the input layer and the output layer may use the ReLU activation function, while the hidden layer may use the sigmoid activation function. The predicted reflectivity may be used and the actual reflectivity R i The deviation between Determine the loss function.

[0071] It can be seen that by using the optical parameter calculation method based on any of the first aspects, the optical parameters of the preselected material at any preselected temperature can be calculated, so that the trained reflectivity prediction model can accurately predict the "reflectivity of the multi-layer film structure wavelength selective emitter for light of different wavelength bands within the preselected wavelength range at the preselected temperature", thereby improving the design efficiency of the multi-layer film structure wavelength selective emitter.

[0072] In some optional embodiments, S202, based on the preselected wavelength range, the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, and the optical parameters of each of the preselected materials at the preselected temperature, determine the input data, including: based on the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, randomly generate W initial arrays of dimension 2S×1; wherein S represents the number of layers of the film structure, and the initial array includes the material and thickness parameters of each film layer; based on the optical parameters of each preselected material at the preselected temperature, perform a dimensionality increase transformation on the initial array to obtain a three-dimensional array of S×(2c+1)×b; wherein b represents the number of preselected temperatures, and c represents the number of different wavelengths of light within the preselected wavelength range.

[0073] Among them, a one-dimensional array with a dimension of 2S×1 can be generated based on the material and thickness of each membrane structure: [L1,L2,…,L S ,T1,T2…,T S ]. Where S represents the number of layers of the membrane structure, L S Indicates the material of the S-th layer of membrane structure, T sBased on the optical parameter calculation method described in any one of the first aspects, the thickness of the preselected material at b preselected temperatures (including t1, t2, ..., t b ) in the infrared spectrum range (including c infrared wavelengths; when the infrared spectrum range includes 1 μm to 25 μm and the wavelength interval is 1 μm, c is 25) c ) and extinction coefficient k (including k1, k2, ..., k c ). Based on the calculated refractive index and extinction coefficient, the one-dimensional array can be transformed into a three-dimensional array of S×(2c+1)×b: {[n1(L1,t1),k1(L1,t1),…,n c (L1,t1),k c (L1,t1),T1(t1);n1(L2,t1),k1(L2,t1),…,n c (L2,t1),k c (L2,t1),T2(t1);…;n1(L S ,t1),k1(L S ,t1),…,n c (L S ,t1),k c (L S ,t1),T s (t1)];n1(L1,t2),k1(L1,t2),…,n c (L1,t2),k c (L1,t2),T1(t2);…;[n1(L1,t b ),k1(L1,t b ),…,n c (L1,t b ),k c (L1,t b ),T1(t b );…;n1(L S ,t b ),k1(L S ,t b ),…,n c (L S ,t b ),k c (L S ,t b ),T S (t b)]}. By selecting the material and thickness of each film structure in the multilayer film structure wavelength-selective emitter, an initial array of dimensions 2S×1 is generated. Then, based on the optical parameters of each preselected material at a preselected temperature, the initial array is dimensionalized to obtain a three-dimensional array of dimensions S×(2c+1)×b. The dimensionalized three-dimensional array can include the material and thickness of each film structure, as well as the refractive index and extinction coefficient of each preselected material for light of different wavelengths at a preselected temperature, and the dimensionalized three-dimensional array can be used as the above-mentioned input data.

[0074] In some optional embodiments, inputting input data into the model to be trained includes: dividing the three-dimensional array into b S×(2c+1)×1 arrays; flattening each S×(2c+1)×1 array into a one-dimensional array of (2S×c+S)×1, and inputting the flattened one-dimensional array into the model to be trained.

[0075] The input data can be determined by the material and thickness of each film structure in the multilayer film structure wavelength selective emitter within the preselected wavelength range, as well as the optical parameters of each preselected material at a preselected temperature. Alternatively, the three-dimensional array obtained by the above-mentioned dimensionality increase transformation can be divided into b S×(2c+1)×1 three-dimensional arrays, and each S×(2c+1)×1 three-dimensional array is flattened into a (2S×c+S)×1 one-dimensional array. The flattened (2S×c+S)×1 one-dimensional array is used as the input data of the model. Accordingly, based on the model to be trained, 1×c output data corresponding to each of the b input data can be obtained. The output data includes the predicted reflectivity of the multilayer film structure wavelength selective emitter for c different wavelengths of light within the preselected wavelength range at a preselected temperature. When the model to be trained is implemented by a neural network model, since the neural network has insufficient ability to process high-dimensional nonlinear data, the present application establishes a dimensionally enhanced data set (including the flattened one-dimensional array) by implementing the above steps, which significantly improves the universality and robustness of the neural network in the design of wavelength-selective emitters in a wide temperature range and full infrared band.

[0076] In some optional embodiments, the actual reflectivity is calculated based on input data using a rigorous coupled wave analytical algorithm.

[0077] Among them, the rigorous coupled wave analytical algorithm can simulate the interaction between electromagnetic waves and periodic optical structures; the periodic optical structure can be decomposed into a set of coupled plane waves, and its behavior can be described by Maxwell's equations and Floquet's theorem. Based on the above-obtained three-dimensional array S×(2c+1)×b and the rigorous coupled wave analytical algorithm, the actual reflectivity of the multilayer film structure wavelength selective emitter for c different wavelengths of light within the preselected wavelength range at a preselected temperature can be obtained. Based on the rigorous coupled wave analytical algorithm, a 1×c×b reflectivity array can be obtained first, and then the reflectivity array can be divided into b 1×c arrays, and then the actual reflectivity corresponding to c different wavelengths of light at b different preselected temperatures can be obtained. Through the rigorous coupled wave analytical algorithm, the actual reflectivity of the multilayer film structure wavelength selective emitter corresponding to different wavelengths of light at different preselected temperatures can be accurately calculated. The loss function value of the model to be trained can be determined based on the actual reflectivity and the predicted reflectivity; then, based on the loss function value, the model parameters of the model to be trained are adjusted to obtain a trained reflectivity prediction model; and the reflectivity calculation efficiency of the multi-layer film structure wavelength selective emitter is improved based on the trained reflectivity prediction model.

[0078] In some optional embodiments, the above-mentioned adjusting the model parameters of the model to be trained according to the loss function value includes: using the gradient descent method to calculate the partial derivatives of the loss function value with respect to the model parameters of the model to be trained, and updating the model parameters of the model to be trained through gradient back propagation.

[0079] Based on the gradient descent method, the partial derivatives of the loss function with respect to the current model parameters of the model to be trained can be obtained, and the model parameters of the model to be trained can be optimized and updated through gradient backpropagation to adjust the model parameters of the model to be trained. Model training conditions may include a parameter update number threshold or a loss function threshold, etc., which are not specifically limited in this application.

[0080] Please refer to Figure 4 , Figure 4 A schematic flow chart of a method for designing a wavelength selective emitter with a multilayer film structure provided in an embodiment of the present application. The method for designing a wavelength selective emitter with a multilayer film structure may include the following steps:

[0081] S301, generating an initialization population; wherein the initialization population includes a plurality of individuals, each of which corresponds to an initialized multilayer film structure wavelength selective emitter;

[0082] S302, determining the optical parameters of each layer of preselected material at a preselected temperature in the initialized multi-layer film structure wavelength selective emitter;

[0083] S303. Calculating, based on a trained reflectivity prediction model, the initial reflectivity of the initialized multilayer film structure wavelength selective emitter to light of different wavelength bands within a preselected wavelength range at the preselected temperature; wherein the trained reflectivity prediction model is obtained by training based on the reflectivity prediction model training method described in any one of the second aspects;

[0084] S304, calculating the fitness of each individual according to the initial reflectivity;

[0085] S305 , generating a new population through crossover and mutation operations until the preset conditions are met, and determining the individual with the largest fitness as the designed multi-layer film structure wavelength selective emitter.

[0086] Wherein, in step S301, the number of individuals included in the initialization population can be 150, 200, 300 or other reasonable values. Each individual includes an initialization parameter for initializing the multilayer film structure wavelength selective emitter (specifically, it can include the material and thickness parameters of each layer). In step S302, based on the optical parameter calculation method as described in any one of the first aspects, the optical parameters of each layer of preselected material at the preselected temperature in the multilayer film structure wavelength selective emitter are determined. The preselected temperature can be 300K, 600K, 900K, 1200K, 1500K or other reasonable values. In step S303, the preselected wavelength range can be an infrared spectrum range, which can specifically include 1μm to 25μm, and the wavelength interval is 1μm. In step S304, the fitness of each individual can be calculated based on a preset fitness function (which can be adjusted according to the actual application scenario), for example, the fitness function F1 for the preset spectrum Tar1 of thermophotovoltaic, the fitness function F2 for the preset spectrum Tar2 of mid- and far-infrared radiation regulation, the fitness function F3 for the preset spectrum Tar3 of radiative cooling, and the fitness function F4 for the preset spectrum Tar4 of near-infrared radiation regulation. In step S305, based on the fitness, a preset number of individuals with the highest fitness can be selected from the initialized population as candidate individuals, or individuals with fitness greater than the preset fitness can be selected as candidate individuals. The crossover operation allows two excellent individuals to exchange some genes to generate new offspring; the mutation operation randomly changes some genes in the individual to increase the diversity of the population. The initialized population can be updated by crossover and / or mutation operations to generate a new population. The preset conditions may include the maximum number of generations (corresponding to the number of crossovers or mutations) or the minimum fitness threshold, etc., which are not specifically limited in this application.

[0087] Among them, the multi-layer film structure wavelength selective emitter has become an ideal choice for wavelength selective emitters used in a wide temperature range and full infrared spectrum due to its many advantages such as simple structure, stable performance, mass production, and large-area processing. However, the current optimization design of wavelength selective emitters compatible with a wide temperature range and full infrared still has the following problems. First, the infrared optical parameter data obtained through testing is limited, especially the high-temperature infrared spectrum testing is difficult, resulting in an incomplete infrared parameter database and incomplete device design coverage. Secondly, the existing fast and intelligent design methods are targeted at a single spectral band or a single temperature, making it difficult to achieve a wide temperature range and full infrared compatible design. The multi-layer film structure wavelength selective emitter design method provided in the embodiment of the present application can calculate the optical parameters of the pre-selected material at any pre-selected temperature based on the optical parameter calculation method as described in any of the first aspects, so that the coverage of the designed multi-layer film structure wavelength selective emitter is more comprehensive. The design of a multi-layer film structure wavelength selective emitter compatible with a wide temperature range and full infrared can be achieved. And because the design process of the multi-layer film structure wavelength selective emitter involves the complex interaction of multiple parameters such as optical parameters, pre-selected temperature, pre-selected materials and pre-selected wavelength range, the traditional design method may fall into a local optimal solution, resulting in poor wavelength selection performance of the emitter. The multi-layer film structure wavelength selective emitter design method provided in this application, based on the reflectivity prediction model combining genetic algorithm and neural network, can search in a wider parameter space to find a better design scheme, so that the reflectivity performance index of the designed multi-layer film structure wavelength selective emitter in different bands can achieve a better balance, thereby improving the wavelength selection accuracy and efficiency of the designed multi-layer film structure wavelength selective emitter. In addition, by evaluating and evolving multiple individuals at the same time, more design schemes can be explored in the same time, improving design efficiency and significantly shortening the design cycle of the multi-layer film structure wavelength selective emitter.

[0088] In some optional embodiments, S303, based on the trained reflectivity prediction model, calculate the initial reflectivity of the initialized multilayer film structure wavelength selective emitter to different wavelengths of light within the preselected wavelength range at the preselected temperature, including: based on the material and thickness of each layer of the film structure in the multilayer film structure wavelength selective emitter, randomly generate W initial arrays of dimension 2S×1; wherein S represents the number of layers of the film structure, and the initial array includes the material and thickness parameters of each layer of the film; based on the optical parameters of each preselected material at the preselected temperature, perform a dimensionality upgrade on the initial array to obtain a three-dimensional array of S×(2c+1)×b; wherein b represents the number of preselected temperatures, and c represents the number of different wavelengths of light within the preselected wavelength range; divide the three-dimensional array into b S×(2c+1)×1 arrays; flatten each S×(2c+1)×1 array into a one-dimensional array of (2S×c+S)×1, and input the flattened one-dimensional array into the trained reflectivity prediction model to obtain the initial reflectivity output by the model.

[0089] The trained reflectivity prediction model can be obtained using the reflectivity prediction model training method described in the second aspect. To avoid repetition, the input data corresponding to initializing the multilayer film structure wavelength selective emitter can be determined by referring to the relevant description in step S202. Compared to the rigorous coupled-wave analytical algorithm, obtaining the initial reflectivity based on the trained reflectivity prediction model can improve the efficiency of obtaining the initial reflectivity, thereby improving the design efficiency of the multilayer film structure wavelength selective emitter.

[0090] Please refer to Figure 5 , Figure 5 A schematic diagram of the fitness function and optimization design time provided in an embodiment of the present application. Figure 5 Specifically shown is a schematic diagram of the design of a multi-layer film structure wavelength selective emitter based on the above-mentioned multi-layer film structure wavelength selective emitter design method, and during the design process, a fitness function and optimization design time in the reflectivity prediction case are realized based on the trained reflectivity prediction model, showing how they change with the number of generations. Figure 5 In the figure above, the fitness function F1, F2, F3, and F4 corresponding to the above fitness function changes with the number of generations; the figure below shows the optimization design time corresponding to the above fitness function F1, F2, F3, and F4 corresponding to the above fitness function changes with the number of generations. Figure 5 As shown in the figure, the fitness corresponding to the four fitness functions F1, F2, F3, and F4 gradually increases with the increase of the number of generations, and finally reaches convergence; and the corresponding optimization design time t F1 (corresponding to the fitness function F1), t F2 (corresponding to the fitness function F2), t F3(corresponding to fitness function F3), t F4 (corresponding to the fitness function F4) increases linearly with the number of generations, and when the number of generations is 200, the optimization design time corresponding to the four fitness functions is about 360 seconds.

[0091] Please refer to Figure 6 , Figure 6 Schematic diagram of the optimized design time corresponding to the design methods of wavelength selective emitters with different multilayer film structures provided in the embodiments of the present application. Figure 6 Specifically shown is the change of the optimization design time with the number of generations in the case of realizing the reflectivity calculation based on the rigorous coupled wave analytical algorithm in the design process (corresponding to t GA ); and, based on the above-mentioned multilayer film structure wavelength selective emitter design method, realize the design of a multilayer film structure wavelength selective emitter, and in the design process, realize the change of the optimized design time under the reflectivity prediction condition with the number of generations (corresponding to t F1 ).like Figure 6 As shown, when the number of generations is 100, t F1 is 165 seconds, t GA It is 19381 seconds, indicating that in the design process, compared with the reflectivity calculation based on the rigorous coupled wave analytical algorithm, the above-mentioned method of realizing reflectivity prediction based on the trained reflectivity prediction model provided by the present application can greatly improve the design efficiency of the multi-layer film structure wavelength selective emitter.

[0092] Please refer to Figure 7 , Figure 7 A schematic diagram of a designed spectrum provided in an embodiment of the present application. Figure 7 Specifically, the following are shown: during the design process, the reflectivity prediction is realized based on the trained reflectivity prediction model, and the optimized design spectra (P1, P2, P3, P4) are obtained; and during the design process, the reflectivity calculation is realized based on the rigorous coupled wave analytical algorithm, and the analytical design spectrum (P1 ′ 、P2 ′ 、P3 ′ 、P4 ′ ).like Figure 7 As shown, the optimized design spectrum has a high degree of overlap with the analytical design spectrum, indicating that the above-mentioned method of realizing reflectivity prediction based on the trained reflectivity prediction model provided by this application can greatly improve the design efficiency of the multi-layer film structure wavelength selective emitter while ensuring the design accuracy, and has good universality and robustness.

[0093] It can be seen that based on the trained reflectivity prediction model, obtaining the initial reflectivity of the initialized multilayer film structure wavelength selective emitter for light of different bands within the preselected wavelength range at a preselected temperature can improve the efficiency of obtaining the reflectivity, thereby improving the design efficiency of the multilayer film structure wavelength selective emitter.

[0094] In the several embodiments provided in the embodiments of the present application, it should be understood that the disclosed devices / systems and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to the multiple embodiments of the embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0095] In addition, the functional modules in each embodiment of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0096] The above description is only an optional implementation method of the embodiment of the present application, but the protection scope of the embodiment of the present application is not limited to this. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in the embodiment of the present application, and they should all be covered by the protection scope of the embodiment of the present application.

Claims

1. A method for calculating optical parameters, characterized in that: The method comprises: Calculating the optical frequency dielectric constant, the Born effective charge, and the initial phonon resonance frequency of the target material at the initial temperature based on the crystal structure data and density functional theory of the target material; performing thermodynamic calculations on the target material based on the Born effective charge, the initial phonon resonance frequency, and a first-principles molecular dynamics simulation method to obtain a relaxed crystal structure of the target material at a preset temperature; The optical parameters of the target material at the preset temperature are calculated based on the optical frequency dielectric constant, the relaxed crystal structure, the density functional theory, the Boltzmann transport equation, and the Lorentz phonon resonance model.

2. The method according to claim 1, characterized in that The calculating the optical parameters of the target material at the preset temperature based on the optical frequency dielectric constant, the relaxed crystal structure, the density functional theory, the Boltzmann transport equation, and the Lorentz phonon resonance model includes: Calculating a preset phonon resonance frequency and a preset phonon scattering rate of the target material at the preset temperature based on the relaxed crystal structure, the density functional theory, and the Boltzmann transport equation; Calculating the optical parameters of the target material at the preset temperature based on the optical frequency dielectric constant, the preset phonon resonance frequency, the preset phonon scattering rate, and the Lorentz phonon resonance model; Among them, the preset phonon resonance frequency includes the branch phonon resonance frequency under different optical branches corresponding to different infrared active phonon modes; the preset phonon scattering rate includes the branch phonon scattering rate under different optical branches corresponding to different infrared active phonon modes.

3. The method according to claim 2, characterized in that Calculating the optical parameters of the target material at the preset temperature according to the optical frequency dielectric constant, the preset phonon resonance frequency, the preset phonon scattering rate, and the Lorentz phonon resonance model includes: Calculating the total branch phonon scattering rate under the corresponding optical branch corresponding to the corresponding infrared active phonon mode according to the branch phonon scattering rate; According to the optical frequency dielectric constant, the branch phonon resonance frequency, the branch phonon total scattering rate and the Lorentz phonon resonance model Calculate the complex dielectric constant of the target material at the preset temperature; wherein E(ω) represents the complex dielectric constant corresponding to the frequency ω of the infrared wavelength λ, ε ∞ represents the optical frequency dielectric constant, ω j,LO represents the branch phonon resonance frequency under the longitudinal optical branch corresponding to the jth infrared active phonon mode, ω j,TO represents the branch phonon resonance frequency under the transverse optical branch corresponding to the jth infrared active phonon mode, γ j,lo represents the total scattering rate of branch phonons under the longitudinal optical branch corresponding to the jth infrared active phonon mode, γ j,TO represents the total branch phonon scattering rate under the transverse optical branch corresponding to the jth infrared active phonon mode, where i is an imaginary unit; According to the complex dielectric constant, N 2 =E(ω) and N=n+ik, determine the optical parameters of the target material at the preset temperature; wherein N represents the complex refractive index of the target material at the preset temperature, n represents the refractive index of the target material at the preset temperature, and k represents the extinction coefficient of the target material at the preset temperature.

4. A reflectivity prediction model training method, characterized in that: The method comprises: Determine the optical parameters of each preselected material at a preselected temperature in a multilayer film structure wavelength selective emitter based on the optical parameter calculation method according to any one of claims 1 to 3; Determining input data based on a preselected wavelength range, the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, and the optical parameters of each of the preselected materials at the preselected temperature; Inputting the input data into the model to be trained to obtain output data output by the model to be trained; wherein the output data includes the predicted reflectivity of the multi-layer film structure wavelength selective emitter for light of different wavelength bands within the preselected wavelength range at the preselected temperature; Determining a loss function value of the model to be trained based on the predicted reflectivity and the actual reflectivity; According to the loss function value, the model parameters of the model to be trained are adjusted until the adjusted model meets the model training conditions, thereby obtaining a trained reflectivity prediction model.

5. The method according to claim 4, characterized in that The step of determining input data based on a preselected wavelength range, the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, and the optical parameters of each of the preselected materials at the preselected temperature includes: Based on the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, randomly generating W initial arrays of dimensions 2S×1; where S represents the number of layers of the film structure, and the initial array includes the material and thickness parameters of each film layer; Based on the optical parameters of each of the preselected materials at the preselected temperature, the initial array is subjected to a dimensionality increase transformation to obtain a three-dimensional array of S×(2c+1)×b; wherein b represents the number of the preselected temperatures, and c represents the number of different wavelengths of light within the preselected wavelength range.

6. The method according to claim 5, characterized in that Inputting the input data into the model to be trained includes: Divide the three-dimensional array into b arrays of S×(2c+1)×1; Each S×(2c+1)×1 array is flattened into a one-dimensional array of (2S×c+S)×1, and the flattened one-dimensional array is input into the model to be trained.

7. The method according to claim 4, characterized in that The actual reflectivity is calculated based on input data using a rigorous coupled wave analytical algorithm.

8. The method according to claim 4, characterized in that The adjusting the model parameters of the to-be-trained model according to the loss function value includes: The gradient descent method is used to calculate the partial derivatives of the loss function value with respect to the model parameters of the model to be trained, and the model parameters of the model to be trained are updated by gradient back propagation.

9. A method for designing a wavelength selective emitter with a multilayer film structure, characterized in that: The design method includes: Generate an initialization population; wherein the initialization population includes a plurality of individuals, each of the individuals corresponds to an initialized multilayer film structure wavelength selective emitter; Determine the optical parameters of each layer of the preselected material at the preselected temperature in the initialized multilayer film structure wavelength selective emitter based on the optical parameter calculation method according to any one of claims 1 to 3; Based on the trained reflectivity prediction model, calculating the initial reflectivity of the initialized multilayer film structure wavelength selective emitter to light of different wavelength bands within the preselected wavelength range at the preselected temperature; wherein the trained reflectivity prediction model is obtained by training based on the reflectivity prediction model training method according to any one of claims 4 to 8; Calculating the fitness of each individual according to the initial reflectivity; A new population is generated through crossover and mutation operations until preset conditions are met, and the individual with the greatest fitness is determined as the designed multilayer film structure wavelength selective emitter.

10. The method according to claim 9, characterized in that The calculating, based on the trained reflectivity prediction model, the initial reflectivity of the initialized multilayer film structure wavelength selective emitter to light of different wavelengths within the preselected wavelength range at the preselected temperature comprises: Based on the material and thickness of each film structure in the multi-layer film structure wavelength selective emitter, randomly generating W initial arrays of dimensions 2S×1; where S represents the number of layers of the film structure, and the initial array includes the material and thickness parameters of each film layer; Based on the optical parameters of each layer of the preselected material at the preselected temperature, the initial array is subjected to a dimensionality-enhancing transformation to obtain a three-dimensional array of S×(2c+1)×b; wherein b represents the number of the preselected temperatures, and c represents the number of different wavelengths of light within the preselected wavelength range; Divide the three-dimensional array into b arrays of S×(2c+1)×1; Each S×(2c+1)×1 array is flattened into a one-dimensional array of (2S×c+S)×1, and the flattened one-dimensional array is input into the trained reflectivity prediction model to obtain the initial reflectivity output by the model.

Images