A method for synthesizing optoelectronic materials based on a physical model and a machine learning model

Abstract

The application discloses a photoelectric material synthesis method based on a physical model and a machine learning model, and comprises the following steps: determining photoelectric data types, and constructing a physical model; screening the photoelectric data types through the physical model, and constructing a photoelectric database; constructing a machine learning model, and guiding photoelectric material synthesis through the machine learning model. The scheme uses the physical model to perform data screening, reduces the difficulty of constructing the photoelectric database, and improves the fitting degree of the machine learning model when the data amount is small.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of photoelectric material synthesis, and particularly relates to a photoelectric material synthesis method based on a physical model and a machine learning model. BACKGROUND

[0002] Machine learning is a big data processing technology that has been widely applied, and in recent years, it has been applied in the fields of materials and biomedical medicine. Its main feature is to use a large amount of data for comprehensive data processing to realize data prediction, rule summary and other functions. However, the data set of machine learning has the characteristics of a large amount of data, which is difficult to realize in the field of materials, but too little data will cause distortion in the analysis process, so a method for synthesizing photoelectric materials by using a neural network without distortion with less data in the field of materials is needed. SUMMARY

[0003] The main purpose of the application is to overcome the shortcomings and deficiencies of the prior art, and to provide a photoelectric material synthesis method based on a physical model and a machine learning model.

[0004] In order to achieve the above purpose, the following technical solutions are adopted in the application:

[0005] A photoelectric material synthesis method based on a physical model and a machine learning model, the photoelectric material synthesis method comprising the following steps:

[0006] S1, determining the data types of photoelectric materials, including determining the data types of photoelectric materials, the synthesis formula parameters of photoelectric materials, the optical and electrical parameters of photoelectric materials, and the optical and electrical parameters of optical devices constructed by photoelectric materials;

[0007] S2, using COMSOL software and photoelectric data to perform multi-physical field simulation to construct a physical model;

[0008] S3, screening and removing redundant data types through the physical model, reducing the data scale, and constructing a photoelectric database; specifically, using the physical model constructed in step S2 to exclude the synthesis formula parameters of photoelectric materials, the optical and electrical parameters of photoelectric materials, and the optical and electrical parameters of optical devices constructed by photoelectric materials with low correlation for the performance parameters of photoelectric devices

[0009] S4, constructing a machine learning model;

[0010] S5, predicting the synthesis of photoelectric materials through the machine learning model.

[0011] Further, the data types of photoelectric materials include the name of the material itself, the concentration of the material, and the proportion of the material;

[0012] The optical and electrical parameters of the photoelectric material refer to the transmittance, absorption, band gap, and conductivity of the material;

[0013] The optical and electrical parameters of the optical device constructed by the photoelectric material include the transmittance, response time, threshold voltage and lifetime of the device.

[0014] Further, the step S2 is as follows:

[0015] S201, COMSOL software constructs a 2D or 3D simulation model of the device according to the device structure parameters, and the photoelectric device structure parameters include boundary length, device layer width and device layer number;

[0016] S202, material parameters are given to the 2D or 3D simulation model, and the material parameters include material transmittance, absorption, band gap and conductivity;

[0017] S203, according to the working principle of the device, the physical fields involved in the device are selected, including diffusion, drift motion of charged particles, particle recombination at the interface, and optical-electric conversion;

[0018] S204, according to the selected physical field, the physical formula involved is selected.

[0019] The physical formula includes the electric flux formula:

[0020] Total flux N i = ∇(−D i ∇c i −z i μ i Fc i ∇ϕ i ) + uc i

[0021] In the formula, c 𝑖 represents the concentration of the substance, D 𝑖 is the diffusion coefficient, φ 𝑖 is the electric potential, F is the Faraday constant, z 𝑖 is the charge number, μ i is the electromigration number, u represents the fluid flow rate, and ∇ represents the gradient operator.

[0022] The physical formula also includes the Butler-Volmer formula:

[0023] j loc =j0{exp (-nαFη / RT)-exp[n(1-α)Fη}

[0024] Where j l𝑜𝑐 and j0 are the local current density and exchange current density, respectively, α is the transfer coefficient, n is the number of transferred electrons, R is the gas constant, T is the reaction temperature, and η is the overpotential.

[0025] Further, the step S3 utilizes the physical model to exclude the photoelectric material synthesis formula parameters, the photoelectric material optical and electrical parameters with lower correlation for the photoelectric device performance parameters.

[0026] Further, the step S3 constructs the database by actually collecting the photoelectric material synthesis formula parameters, the photoelectric material optical and electrical parameters filtered by the above physical model, and the process is as follows:

[0027] S301, normalize all selected data, that is, eliminate the dimension and convert to pure numbers, and the normalization method is

[0028]

[0029] For the normalized data, For the original data, For the minimum value in the original data of this type, For the maximum value in the original data of this type;

[0030] S302, the selected data is composed of a group of data of material synthesis formula parameters, photoelectric material optical and electrical parameters, and photoelectric device parameters, and the specific number of data in the group is specified in advance, which is determined according to the order of magnitude of the database size. All group data constitutes a data set.

[0031] S303, the selected data set is divided into K subsets (K>5), one of which is used as a test set, and the rest is used as a training set.

[0032] Further, the step S4 machine learning model includes linear regression (LR), Lasso regression (Lasso), kernel ridge regression (KRR), Gaussian process regression (GPR), support vector regression (SVR), k-neighborhood regression (KNR), decision tree (DTR) regression, a total of 7 kinds of machine learning models, and the machine learning model is directly imported from the python database; The step S4 process is as follows:

[0033] S401, import the selected test set in step S3 into the machine learning model, select the iteration number N (N>10), and output the performance evaluation index; wherein, the selected performance evaluation index includes root mean square error RMSE and determination coefficient R 2 , as follows:

[0034] R 2 The square of the correlation coefficient between the predicted value and the true value, the closer to 1, the better the fitting effect of the machine learning model, and the calculation expression is:

[0035]

[0036] Wherein, is the true value of the i-th sample, is the predicted value of the i-th sample, is the mean of the sample true values, m is the total number of samples, Var is the variance of the sample true values, the root mean square error RMSE is the square root of the mean square difference between the predicted value and the true value, the smaller the value, the better the fitting effect of the machine learning model, and the calculation expression is:

[0037]

[0038] S402, cross-validation is repeated 7 times, the data set is divided into K subsets, and each time one subset is used as a test set and the rest is used as a training set, and the model performance is evaluated with the test set after each training. Repeat 7 times to obtain the performance evaluation index of 7 machine learning models, and then take the average value as the final evaluation result.

[0039] Further, the step S5 process is as follows:

[0040] S501, select the machine learning model with the best fitting effect as the prediction model to predict the synthesis of optoelectronic materials;

[0041] S502, change the input data type of the prediction model, wherein the change of the input data refers to any one or more of the optoelectronic material synthesis formula parameters, optoelectronic material optical and electrical parameters, and optoelectronic device parameters, and the output data is one or more of the remaining optoelectronic device parameters, that is, any one or more of the optoelectronic device parameters is used as input, and the output of the machine learning model for predicting the synthesis of optoelectronic materials is the optoelectronic device optical and electrical parameters except the input optoelectronic device optical and electrical parameters, and the predicted output of the other optoelectronic device optical and electrical parameters is obtained. The optoelectronic device optical and electrical parameters obtained by inputting the optoelectronic device optical and electrical parameters and the predicted output of the optoelectronic device optical and electrical parameters complete the output of the optoelectronic device synthesis, and by obtaining the fixed optoelectronic device optical and electrical parameters, the synthesis prediction of the optoelectronic material is obtained.

[0042] Compared with the prior art, the present application has the following advantages and beneficial effects:

[0043] (1) The optoelectronic material synthesis method based on a physical model and a machine learning model disclosed in the present application designs a technical scheme of screening data types by a physical model and then constructing a database machine learning, which screens the data types by a physical model, reduces the dimension of the database, and thus reduces the amount of data required for machine learning.

[0044] (2) The photoelectric material synthesis method based on the physical model and the machine learning model discloses a technical scheme of designing a physical model to filter data types and then constructing a database machine learning, using the physical model to establish the correlation between data, and then increasing the fitting degree of the machine learning model under a small database.

[0045] (3) The photoelectric material synthesis method based on the physical model and the machine learning model, i.e. the physical model filtering data types, establishing a small database, and then performing the machine learning steps, can be applied to data prediction in other material fields, and has good generality. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0047] Figure 1 is a flow chart of the photoelectric material synthesis method based on the physical model and the machine learning model disclosed in embodiment 1 of the present application;

[0048] Figure 2 is a physical model schematic diagram of the liquid optoelectronic device disclosed in the embodiment of the present application;

[0049] Figure 3 is a parameter diagram between the formula and performance of ammonium metatungstate-ferric chloride in the embodiment of the present application;

[0050] Figure 4 is a parameter diagram between the formula and performance of ammonium metatungstate-ferric sulfate in the embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to make those skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0052] Reference to an“embodiment” in this application means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive of other embodiments. It is expressly understood that the described embodiments of the application are merely example structures selected for the purposes of explanation and are presented to provide what is believed to be the most useful and readily understood description of the principles and conceptual aspects of the application.

[0053] Embodiment 1

[0054] As shown in the following Figure 1 , the embodiment discloses a method for synthesizing optoelectronic materials based on physical models and machine learning models, Figure 1 The method for synthesizing optoelectronic materials includes the following steps:

[0055] S1, determine the data types of optoelectronic materials, including determining the data types of optoelectronic materials, the synthesis formula parameters of optoelectronic materials, the optical and electrical parameters of optoelectronic materials, and the optical and electrical parameters of optical devices constructed by optoelectronic materials; wherein the data types of optoelectronic materials include the name of the material itself, the concentration of the material, and the ratio of the material; the optical and electrical parameters of optoelectronic materials refer to the transmittance, absorption, band gap, and conductivity of the material; the optical and electrical parameters of optical devices constructed by optoelectronic materials include the transmittance, response time, threshold voltage, and lifetime of the device

[0056] S2, use COMSOL software to perform multi-physics simulation with optoelectronic data to construct a physical model; in this embodiment, the process of step S2 is as follows:

[0057] S201, COMSOL software constructs a 2D or 3D simulation model according to device structure parameters, including boundary length, device layer width, and device layer number;

[0058] S202, assign material parameters to the 2D or 3D simulation model, including material transmittance, absorption, band gap, and conductivity;

[0059] S203, select the physical fields involved in the device according to the working principle of the device, including the diffusion, drift motion of charged particles, particle recombination at the interface, and optical-electrical conversion;

[0060] S204, select the physical formulas involved according to the selected physical fields.

[0061] In this embodiment, COMSOL software is used to construct a physical model of a liquid optoelectronic device as shown in Figure 2 , boundary 1 applies a cathode potential; boundary 3 applies an anode potential; boundaries 4, 5, 6, and 7 are electrically insulated; the lithium fluxes of boundaries 1, 3, 4, 5, 6, and 7 are 0.

[0062] S3, screening out redundant data types through a physical model, reducing the data size, and constructing an optoelectronic database; specifically, using the physical model constructed in step S2 to exclude the lower correlation of the optoelectronic material synthesis formula parameters, the optical and electrical parameters of the optoelectronic material for the performance parameters of the optoelectronic device.

[0063] In this embodiment, the input simulation variables and parameters are shown in Table 1, and the electrical-optical multi-physical field coupling is performed after selecting the electrical physical field formula; the data set type is selected as (ammonium paratungstate concentration, ferrocene concentration, modulation amplitude).

[0064] Table 1. Simulation variables and parameters

[0065]

[0066] S4, constructing a machine learning model; wherein the machine learning model includes 7 kinds of machine learning models, including linear regression (LR), Lasso regression (Lasso), kernel ridge regression (KRR), Gaussian process regression (GPR), support vector regression (SVR), k-neighborhood regression (KNR), and decision tree (DTR) regression, and the machine learning model is directly imported from the python database; the process of step S4 in this embodiment is as follows:

[0067] S401, importing the selected test set in step S3 into the machine learning model, selecting the number of iterations N (N>10), and outputting the performance evaluation index; wherein the selected performance evaluation index includes the root mean square error RMSE and the determination coefficient R 2 , as follows;

[0068] R 2 represents the square of the correlation coefficient between the predicted value and the true value, and the closer the value is to 1, the better the fitting effect of the machine learning model, and the calculation expression is:

[0069]

[0070] wherein, is the true value of the i-th sample, is the predicted value of the i-th sample, is the mean value of the sample true value, m is the total number of samples, Var is the variance of the sample true value, and the root mean square error RMSE is the square root of the mean square difference between the predicted value and the true value, and the smaller the value, the better the fitting effect of the machine learning model, and the calculation expression is:

[0071]

[0072] S402, repeat 7 times for cross-validation, the data set is divided into K subsets, each time one subset is used as the test set and the rest as the training set, and the model performance is evaluated with the test set after each training. Repeat 7 times to get the performance evaluation index of 7 machine learning models, and then take the average as the final evaluation result.

[0073] In this embodiment, the database is divided into 10 groups after the data is normalized, and 7 machine learning models are selected, including linear regression (LR), Lasso regression (Lasso), kernel ridge regression (KRR), Gaussian process regression (GPR), support vector regression (SVR), k-neighborhood regression (KNR), and decision tree (DTR) regression. After importing the data, cross-validation is performed to obtain the determination coefficient R 2 values are shown in Table 2.

[0074] Table 2. Determination coefficient R 2 value

[0075]

[0076] S5, predicting the synthesis of optoelectronic materials by machine learning model.

[0077] In this embodiment, the process of step S5 is as follows:

[0078] S501, selecting the machine learning model with the best fitting effect as the prediction model to predict the synthesis of optoelectronic materials;

[0079] S502, changing the input data type of the prediction model, wherein the change of input data refers to any one or more of the optoelectronic material synthesis formula parameters, optoelectronic material optical and electrical parameters, and optoelectronic device parameters, and the output data is one or more of the remaining optoelectronic device parameters.

[0080] In this embodiment, the SVR model is finally selected for data prediction, with an accuracy of up to 99.35%.

[0081] Comparative Example 1

[0082] Referring to steps S1 to S5 of the optoelectronic material synthesis method based on physical model and machine learning model disclosed in Comparative Example 1, this comparative example further discloses an optoelectronic material synthesis method based on physical model and machine learning model.

[0083] The data set type is (ammonium metatungstate concentration, ferrocene concentration, device thickness, solution transmittance, modulation amplitude). After collecting and normalizing the data, a database is constructed, the database is divided into 10 groups, and 7 machine learning models are selected, including linear regression (LR), Lasso regression (Lasso), kernel ridge regression (KRR), Gaussian process regression (GPR), support vector regression (SVR), k-nearest neighbor regression (KNR), and decision tree (DTR) regression. After importing the data, cross-validation is performed to obtain the determination coefficient R of different models 2 The final selected LR model has a data prediction accuracy of 39.25%.

[0084] Embodiment 2

[0085] Referring to each step of the physical model and machine learning model-based synthesis method of the optoelectronic material in Embodiment 1, this embodiment specifically discloses a method for synthesizing two kinds of liquid optoelectronic devices, ammonium metatungstate-ferrous chloride and ammonium metatungstate-ferrous sulfate, respectively. The parameter changes between the formulations and performances of the two kinds of liquid optoelectronic devices, ammonium metatungstate-ferrous chloride and ammonium metatungstate-ferrous sulfate, are compared.

[0086] In this embodiment, each device is divided into four ammonium metatungstate concentration gradients and five ferrous chloride (or ferrous sulfate) concentration gradients, and a total of 20 devices are produced in each group. Then, the transmittance changes of each device under different current densities are measured to obtain the respective light modulation amplitude, coloring time, and bleaching time. After that, the measured data are used as a training set, and 7 different machine learning models, including linear regression, k-nearest neighbor regression, decision tree, logistic forest regression, kernel ridge regression, Gaussian process regression, and support vector regression, are used to construct the optoelectronic models of the two kinds of liquid optoelectronic devices, ammonium metatungstate-ferrous chloride and ammonium metatungstate-ferrous sulfate, and to predict the solution composition and concentration when the device performance is best, so as to finally form the mutual conversion model between the formulation, solution concentration, light modulation amplitude, coloring time, and bleaching time. The parameter graph between the formulation and performance is shown in FIGS. 1 and 2. Figure 3 and Figure 4

[0087] The technical features of the above embodiments can be combined in any way. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not contradict, they should be considered within the scope of the present disclosure.

[0088] The above embodiments are the preferred embodiments of the present application, but the embodiments of the present application are not limited to the above embodiments, and any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present application should be equivalent replacement methods and should be within the scope of protection of the present application.​

Claims

1.A method for synthesizing optoelectronic materials based on a physical model and a machine learning model, characterized in that, The photoelectric material synthesis method comprises the following steps: S1, determining the data type of the photoelectric material, including determining the photoelectric material data type, the photoelectric material synthesis formula parameter, the photoelectric material optical and electrical parameter, and the optical and electrical parameter of the optical device constructed by the photoelectric material; S2, using COMSOL software and photoelectric data to perform multi-physical field simulation to construct a physical model; the step S2 process is as follows: S201, the COMSOL software constructs a 2D or 3D simulation model according to the device structure parameters, and the device structure parameters include boundary length, device layer width, and device layer number; S202, the 2D or 3D simulation model is given material parameters, and the material parameters include material transmittance, absorption, band gap, and conductivity; S203, according to the working principle of the device, the physical field involved in the device is selected, including the diffusion, drift motion of charged particles, particle recombination at the interface, and optical-electric conversion; S204, according to the selected physical field, the physical formula involved is selected; the physical formula includes the electric flux formula: Total flux N i = ∇(−D i ∇c i −z i μ i Fc i ∇ϕ i ) + uc i where c 𝑖 represents the concentration of a substance, D 𝑖 is the diffusion coefficient, φ 𝑖 is the electric potential, F is the Faraday constant, z 𝑖 is the number of charges, μ i is the electro-mobility, u represents the fluid velocity, and ∇ represents the gradient operator; The physical formula also includes the Butler-Volmer formula: j loc = j0 {exp (-nαFη / RT )-exp[n(1-α)Fη} where, j l𝑜𝑐 and j0are the local current density and exchange current density, respectively, a is the transfer coefficient, n is the number of electrons transferred, R is the gas constant, T is the reaction temperature, and η is the overpotential; S3, filtering out redundant data types through the physical model, reducing the data scale, and constructing a photoelectric database; S4, constructing a machine learning model; the machine learning model includes linear regression, Lasso regression, kernel ridge regression, Gaussian process regression, support vector regression, k-neighborhood regression, and decision tree regression, a total of 7 kinds of machine learning models, which are directly imported from the python database, and the process is as follows: S401, import the test set into the machine learning model, select the number of iterations N, N>10, output the performance evaluation index; wherein the selected performance evaluation index includes the root mean square error RMSE and the coefficient of determination R 2 , as follows; R 2 The square of the correlation coefficient between the predicted value and the true value is represented, and the value closer to 1 indicates that the fitting effect of the machine learning model is better, and the calculation expression is: wherein, is the true value of the i-th sample, is the predicted value of the i-th sample, is the mean of the true values of the samples, m is the total number of samples, Var is the variance of the true values of the samples, the root mean square error RMSE is the mean square root of the square difference between the predicted value and the true value, the smaller the value, the better the fitting effect of the machine learning model, and the calculation expression is: S402, repeating 7 times for cross-validation, the data set is divided into K subsets, each time one subset is used as a test set, and the rest is used as a training set, the machine learning model performance is evaluated with the test set after each training, and the process is repeated 7 times, the performance evaluation index of 7 machine learning models is obtained, and then the average value is taken as the final evaluation result; S5, predicting the photoelectric material synthesis through the machine learning model. 2.The method of claim 1, wherein, The photoelectric material data type includes the material itself name, material concentration, and material ratio; The photoelectric material optical and electrical parameter refers to the transmittance, absorption, band gap, and conductivity of the material; The optical and electrical parameter of the optical device constructed by the photoelectric material includes the transmittance, response time, threshold voltage, and lifetime of the device. 3.The method of claim 1, wherein the method further comprises: determining a plurality of physical parameters of the photoelectric material; and determining a plurality of machine learning parameters of the machine learning model. The step S3 process is as follows: S301, normalizing all selected data; S302, the selected data constitute a group of data in terms of the photoelectric material synthesis formula parameter, the photoelectric material optical and electrical parameter, and the photoelectric device parameter, and all group data constitute a data set; S303, the selected data set is divided into K subsets, K>5, one subset is used as a test set, and the rest is used as a training set. 4.The method of claim 1, wherein the method further comprises: determining a plurality of physical parameters of the photoelectric material; and determining a plurality of machine learning parameters of the machine learning model. The step S5 process is as follows S501, selecting the machine learning model with the optimal fitting effect as the prediction model to predict the photoelectric material synthesis; S502, change the input data category of the prediction model, wherein the change of the input data refers to any one or more of the photoelectric material synthesis formula parameters, the photoelectric material optical and electrical parameters, and the photoelectric device parameters, and the output data is one or more of the remaining photoelectric device parameters.

Images