Molecular hole recombination energy prediction method and device, medium and product

Abstract

The invention discloses a molecular hole recombination energy prediction method and device, a medium and a product. The method comprises the following steps: acquiring a first molecule; the feature value of the first molecule is input to a target model, hole recombination energy of the first molecule is obtained, the target model is obtained through training according to a plurality of training samples, each training sample comprises a feature value of a second molecule and hole recombination energy obtained through calculation according to the second molecule, and the hole recombination energy of the second molecule is obtained according to the feature value of the second molecule. The second molecules are hole transport layer molecules. According to the method, after training of the target model is completed, the target model is adopted to predict the hole recombination energy of the first molecule, compared with direct calculation of the hole recombination energy of the first molecule, consumed calculation power is small, the prediction speed is high, and the obtaining efficiency of the hole recombination energy of the first molecule can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of molecules, and in particular to a molecular hole reorganization energy prediction method, device, medium and product. BACKGROUND

[0002] In order to evaluate the transport ability of the hole transport layer molecule, the hole reorganization energy of the hole transport layer molecule needs to be calculated. At present, the quantum chemical density functional theory (DFT) is used to calculate the hole reorganization energy of the molecule. However, the calculation of the hole reorganization energy by this method takes a long time and has low calculation efficiency. SUMMARY

[0003] The present application provides a molecular hole reorganization energy prediction method, device, medium and product to improve the acquisition efficiency of the hole reorganization energy of the hole transport layer molecule.

[0004] In a first aspect, the present application provides a molecular hole reorganization energy prediction method, comprising:

[0005] obtaining a first molecule;

[0006] inputting the characteristic value of the first molecule into a target model to obtain the hole reorganization energy of the first molecule, wherein the target model is trained according to a plurality of training samples, each training sample comprising a characteristic value of a second molecule and a hole reorganization energy calculated according to the second molecule, and the second molecule is a hole transport layer molecule.

[0007] In this embodiment, after the target model is trained, the target model is used to predict the hole reorganization energy of the first molecule. Compared with directly calculating the hole reorganization energy of the first molecule, the algorithm power consumed is smaller, the prediction speed is faster, and the acquisition efficiency of the hole reorganization energy of the first molecule can be improved.

[0008] In an embodiment of the present application, before the characteristic value of the first molecule is inputted into the target model to obtain the hole reorganization energy of the first molecule, the method further comprises:

[0009] obtaining a plurality of end groups and a plurality of intermediate molecular fragments;

[0010] generating a plurality of second molecules based on the plurality of end groups and the plurality of intermediate molecular fragments;

[0011] extracting features of each second molecule to obtain a characteristic value corresponding to each second molecule;

[0012] calculating the hole reorganization energy of each second molecule;

[0013] According to the characteristic value corresponding to each second molecule and the hole reorganization energy, a plurality of training samples are constructed.

[0014] In the embodiment, the plurality of training samples are constructed in the above manner, the target model is obtained by using the plurality of training samples for model training, and the hole reorganization energy of the first molecule is predicted by using the target model, so that the hole reorganization energy of the first molecule is quickly obtained.

[0015] In an embodiment of the present application, the hole reorganization energy of each second molecule is calculated, including:

[0016] For any second molecule, the sum of the first value and the second value is taken as the hole reorganization energy of the second molecule, the first value is the energy difference of different neutral state structures in the second molecule, and the second value is the energy difference of different cation state structures in the second molecule.

[0017] In the embodiment, for the second molecule, the hole reorganization energy of the second molecule is obtained by calculation, and the training sample is constructed to train the target model. Although the calculation process uses a large amount of calculation power, the amount of calculation power can be estimated and expended in advance because the number of training samples is limited. Once the target model is trained, the hole reorganization energy of the first molecule can be predicted by using the target model, and the efficiency of obtaining the hole reorganization energy is improved.

[0018] In an embodiment of the present application, the number of characteristic values corresponding to each second molecule is a plurality;

[0019] The feature extraction is performed on each second molecule to obtain a characteristic value corresponding to each second molecule, including:

[0020] The feature extraction is performed on each second molecule to obtain a plurality of initial characteristic values of each second molecule, and each initial characteristic value is used to describe the physical and chemical information of the second molecule.

[0021] For any first characteristic value in the plurality of initial characteristic values, the standard deviation and the mean value of the first characteristic value are calculated according to the first characteristic value of each second molecule.

[0022] For any second molecule, a characteristic value of the second molecule is calculated according to the first characteristic value of the second molecule, the standard deviation and the mean value of the first characteristic value.

[0023] In the embodiment, by adopting the above standardization processing on each initial feature value, each feature value of the second molecule obtained by processing is located in a standard normal distribution range with a mean value of 0 and a variance of 1, which can eliminate the dimensional difference between different features, make different features comparable, thereby accelerating the convergence speed of the model and improving the target model precision.

[0024] In an embodiment of the present application, a plurality of training samples are constructed according to the feature value corresponding to each second molecule and the hole recombination energy, including:

[0025] A plurality of molecular samples are constructed according to the feature value corresponding to each second molecule and the hole recombination energy, and each molecular sample includes the feature value corresponding to a second molecule and the hole recombination energy;

[0026] The plurality of molecular samples are randomly divided to obtain the plurality of training samples and a plurality of test samples;

[0027] The plurality of initial models are trained by using the plurality of training samples to obtain a prediction model corresponding to each initial model;

[0028] The plurality of prediction models are screened by using the plurality of test samples to determine the target model.

[0029] It should be noted that the feature value corresponding to each second molecule can be multiple.

[0030] In the embodiment, the plurality of training samples and a plurality of test samples are obtained by randomly dividing the plurality of molecular samples, the plurality of initial models are trained by using the plurality of training samples to obtain a prediction model corresponding to each initial model, and the plurality of prediction models are screened by using the plurality of test samples to determine the target model, so that a target model with better performance can be obtained, and the accuracy of the target model in predicting the hole recombination energy of the first molecule is improved.

[0031] In an embodiment of the present application, the plurality of prediction models are screened by using the plurality of test samples to determine the target model, including:

[0032] For any prediction model, the feature values of the plurality of test samples are input into the prediction model to obtain a plurality of predicted hole recombination energies;

[0033] According to the hole recombination energies of the plurality of test samples and the plurality of predicted hole recombination energies, a determination coefficient of the prediction model is calculated;

[0034] The prediction model corresponding to the determination coefficient with the largest value is taken as the target model.

[0035] In this embodiment, the target model is determined by comparing the sizes of the determination coefficients of the plurality of prediction models, so that the target model screened is the model with the best fitting degree among all the prediction models, thereby improving the accuracy of the target model in predicting the hole recombination energy of the first molecule.

[0036] In an embodiment of the present application, after the characteristic value of the first molecule is input into the target model to obtain the hole recombination energy of the first molecule, the method further comprises:

[0037] According to the hole recombination energy of the first molecule, the hole mobility of the first molecule is calculated.

[0038] In the case where the hole mobility is greater than a preset threshold, the first molecule is added to a material library, and the molecules in the material library are used to make a target material.

[0039] In this embodiment, the first molecule can be screened according to the hole mobility of the first molecule in the above manner, and the first molecule is added to the material library in the case where the hole mobility is greater than a preset threshold, so that the molecules meeting the use requirements are added to the material library, facilitating subsequent use.

[0040] In a second aspect, an embodiment of the present application provides a device for predicting hole recombination energy of a molecule, comprising:

[0041] A first obtaining module is configured to obtain a first molecule.

[0042] A prediction module is configured to input a characteristic value of the first molecule into a target model to obtain a hole recombination energy of the first molecule, wherein the target model is trained according to a plurality of training samples, each of the training samples comprises a characteristic value of a second molecule and a hole recombination energy calculated according to the second molecule, and the second molecule is a hole transport layer molecule.

[0043] In this embodiment, after the target model is trained, the hole recombination energy of the first molecule is predicted by using the target model, which consumes less computing power and has a faster prediction speed than directly calculating the hole recombination energy of the first molecule, thereby improving the efficiency of obtaining the hole recombination energy of the first molecule.

[0044] In an embodiment of the present application, the device further comprises:

[0045] A second obtaining module is configured to obtain a plurality of end groups and a plurality of intermediate molecular fragments.

[0046] A generating module is configured to generate a plurality of second molecules based on the plurality of end groups and the plurality of intermediate molecular fragments.

[0047] a feature extraction module, configured to perform feature extraction on each of the second molecules to obtain a feature value corresponding to each of the second molecules;

[0048] a first calculation module, configured to calculate a hole reorganization energy of each of the second molecules;

[0049] a construction module, configured to construct a plurality of training samples according to the feature value corresponding to each of the second molecules and the hole reorganization energy.

[0050] In this embodiment, the plurality of training samples are constructed in the above manner, the target model can be obtained by using the plurality of training samples for model training, and the hole reorganization energy of the first molecule can be predicted by using the target model, so that the hole reorganization energy of the first molecule can be quickly obtained.

[0051] In an embodiment of the present application, the first calculation module specifically includes:

[0052] For any of the second molecules, the sum of a first value and a second value is taken as the hole reorganization energy of the second molecule, the first value is the energy difference of different neutral state structures in the second molecule, and the second value is the energy difference of different cation state structures in the second molecule.

[0053] In this embodiment, for the second molecule, the hole reorganization energy of the second molecule is obtained by calculation, and the training sample is constructed to train the target model. Although the calculation process requires a large amount of calculation, the amount of calculation can be estimated and paid in advance because the number of training samples is limited. Once the target model is trained, the hole reorganization energy of the first molecule can be predicted by using the target model, and the efficiency of obtaining the hole reorganization energy is improved.

[0054] In an embodiment of the present application, the number of feature values corresponding to each of the second molecules is multiple;

[0055] The feature extraction module includes:

[0056] The first acquisition submodule is configured to perform feature extraction on each of the second molecules to obtain a plurality of initial feature values of each of the second molecules, and each of the initial feature values is used to describe the physical and chemical information of the second molecule.

[0057] The second acquisition submodule is configured to, for any first feature value in the plurality of initial feature values, calculate the standard deviation and the mean value of the first feature value according to the first feature value of each of the second molecules.

[0058] The third acquisition submodule is configured to, for any of the second molecules, calculate a feature value of the second molecule according to the first feature value of the second molecule, the standard deviation and the mean value of the first feature value.

[0059] In the embodiment, by adopting the above normalization processing on each initial feature value, each feature value of the second molecule obtained by processing is located in a standard normal distribution range with a mean value of 0 and a variance of 1, which can eliminate the dimensional difference between different features, make different features comparable, thereby accelerating the convergence speed of the model and improving the target model precision.

[0060] In an embodiment of the present application, the construction module comprises:

[0061] The construction submodule is configured to construct a plurality of molecular samples according to the feature value and the hole recombination energy corresponding to each second molecule, wherein each molecular sample comprises the feature value and the hole recombination energy corresponding to a second molecule.

[0062] The fourth acquisition submodule is configured to randomly divide the plurality of molecular samples to obtain the plurality of training samples and the plurality of test samples.

[0063] The training submodule is configured to train each initial model using the plurality of training samples to obtain a prediction model corresponding to each initial model.

[0064] The screening submodule is configured to screen the plurality of prediction models using the plurality of test samples to determine the target model.

[0065] In the embodiment, the plurality of molecular samples are randomly divided to obtain the plurality of training samples and the plurality of test samples, the plurality of training samples are used to train each initial model to obtain a prediction model corresponding to each initial model, and the plurality of test samples are used to screen the plurality of prediction models to determine the target model, so that a target model with better performance can be obtained, and the accuracy of the target model in predicting the hole recombination energy of the first molecule is improved.

[0066] In an embodiment of the present application, the screening submodule comprises:

[0067] The first acquisition unit is configured to input the feature values of the plurality of test samples into the prediction model to obtain a plurality of predicted hole recombination energies for any prediction model.

[0068] The second acquisition unit is configured to calculate the determination coefficient of the prediction model according to the hole recombination energies of the plurality of test samples and the plurality of predicted hole recombination energies.

[0069] The screening unit is configured to take the prediction model corresponding to the determination coefficient with the largest value as the target model.

[0070] In this embodiment, the target model is determined by comparing the sizes of the determination coefficients of the plurality of prediction models, so that the target model screened is the model with the best fitting degree among all the prediction models, thereby improving the accuracy of the target model in predicting the hole recombination energy of the first molecule.

[0071] In an embodiment of the present application, the device further comprises:

[0072] The second calculation module is configured to calculate the hole mobility of the first molecule according to the hole recombination energy of the first molecule.

[0073] The adding module is configured to add the first molecule into a material library if the hole mobility is greater than a preset threshold, wherein the molecules in the material library are used to make a target material.

[0074] In this embodiment, the first molecule can be screened according to the hole mobility of the first molecule, and the first molecule is added into the material library if the hole mobility is greater than a preset threshold, so that the molecules meeting the use requirements are added into the material library, which is convenient for subsequent use.

[0075] In a third aspect, the embodiments of the present application provide a readable storage medium, and the readable storage medium stores programs or instructions, and the programs or instructions are executed by a processor to implement the steps of the hole recombination energy prediction method of the first aspect.

[0076] In a fourth aspect, the embodiments of the present application provide a computer program product stored in a storage medium, and the computer program product is executed by at least one processor to implement the steps of the hole recombination energy prediction method of the first aspect.

[0077] The above description is only a summary of the technical solutions of the present application, in order to more clearly understand the technical means of the present application, the specific embodiments of the present application can be implemented according to the content of the description, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0078] The features, advantages and technical effects of the exemplary embodiments of the present application will be described below with reference to the accompanying drawings.

[0079] Figure 1 The flowchart of the hole recombination energy prediction method provided by the embodiments of the present application is shown in the figure.

[0080] Figure 2 The flowchart of the hole recombination energy prediction method provided by the embodiments of the present application is shown in the figure.

[0081] Figure 3A part flowchart of a method for predicting molecular hole reorganization energy is provided in an embodiment of the present application.

[0082] Figure 4 A comparison diagram of a true value and a predicted value of a target model output is provided in an embodiment of the present application.

[0083] Figure 5 A structure diagram of a device for predicting molecular hole reorganization energy is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0084] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0085] Unless otherwise defined, all technical and scientific terms used in the present application have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs; the terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms “include” and “have” and any variations thereof in the description of the present application and the claims and the above description of drawings are intended to cover non-exclusive inclusion. The terms “first”, “second” and the like in the description of the present application and the claims or the above description of drawings are used to distinguish different objects, and are not used to describe a specific order or primary and secondary relationship.

[0086] Figure 1 A flowchart of a method for predicting molecular hole reorganization energy is provided in an embodiment of the present application, as shown in Figure 1 The method comprises steps 101-102, wherein:

[0087] In step 101, a first molecule is obtained, and the first molecule is a hole transport layer molecule to be predicted.

[0088] In step 102, a characteristic value of the first molecule is input into a target model to obtain a hole reorganization energy of the first molecule, wherein the target model is obtained by training a plurality of training samples, each of the training samples comprises a characteristic value of a second molecule and a hole reorganization energy calculated according to the second molecule, the second molecule is a hole transport layer molecule, and the characteristic value of the second molecule is used to describe physical and chemical information of the second molecule.

[0089] In the above, the first molecule and the second molecule are both hole transport layer molecules, the characteristic value of the first molecule is used to describe the physical and chemical information of the first molecule, the characteristic value of the first molecule and the characteristic value of the second branch are the same in number, and the acquisition methods can also be the same, for example, the RDKit molecular toolkit is used to extract features of the first molecule and the second branch, and after feature extraction is performed on each molecule, 209 features are obtained. The first molecule and the second molecule can both be described using a molecular structure file.

[0090] The target model is trained in advance using a plurality of training samples, each training sample including a characteristic value of a second molecule and a hole reorganization energy calculated according to the second molecule. It should be noted that the hole reorganization energy of the second branch used for training is calculated, for example, using density functional theory.

[0091] In the present embodiment, although the hole reorganization energy in the training sample used to train the target model is calculated, due to the limited number of training samples, the computing power can be estimated and expended in advance. After the target model is trained, the hole reorganization energy of the first molecule is predicted using the target model. Compared with directly calculating the hole reorganization energy of the first molecule, the computing power consumed is smaller, the prediction speed is faster, and the efficiency of obtaining the hole reorganization energy of the first molecule can be improved.

[0092] In an embodiment of the present application, before the characteristic value of the first molecule is input into the target model to obtain the hole reorganization energy of the first molecule, the method further includes:

[0093] Obtaining a plurality of end groups and a plurality of intermediate molecular fragments;

[0094] Generating a plurality of second molecules based on the plurality of end groups and the plurality of intermediate molecular fragments;

[0095] Extracting features of each second molecule to obtain a characteristic value corresponding to each second molecule;

[0096] Calculating the hole reorganization energy of each second molecule;

[0097] According to the characteristic value and the hole reorganization energy corresponding to each second molecule, a plurality of training samples are constructed.

[0098] In the above, a plurality of second molecules are automatically generated in batches based on a plurality of end groups and a plurality of intermediate molecular fragments, and the RDKit molecular toolkit is used to extract features of each second molecule. Each molecule can extract 209 characteristic values to describe the physical and chemical information of the corresponding second molecule.

[0099] The hole reorganization energy of each second molecule is calculated by using density functional theory, for example, the hole reorganization energy of each second molecule is calculated by using Gaussian quantum chemistry calculation software based on density functional theory.

[0100] Specifically, the hole reorganization energy of each second molecule is calculated by using density functional theory, for example, the hole reorganization energy of each second molecule is calculated by using Gaussian quantum chemistry calculation software based on density functional theory.

[0101] For any second molecule, the sum of a first value and a second value is taken as the hole reorganization energy of the second molecule, the first value is the energy difference between different neutral state structures of the second molecule, and the second value is the energy difference between different cation state structures of the second molecule, and the hole reorganization energy of the second molecule can be calculated according to formula (1):

[0102]

[0103] wherein λ0 represents the first value, λ + represents the second value, E0 is the energy obtained after optimization of the neutral molecular structure, is the cation energy under the neutral molecular geometry, the value of λ0; is the energy of the neutral molecule under the cation geometry, E + is the energy obtained after optimization of the cation structure, the value of λ + .

[0104] In the above, for the second molecule, the hole reorganization energy of the second molecule is obtained by calculation, and a training sample is constructed to train a target model. Although the calculation process requires a large amount of computing power, since the number of training samples is limited, this part of computing power can be estimated and expended in advance, and once the target model is trained, the hole reorganization energy of the first molecule can be predicted using the target model, thereby improving the efficiency of obtaining the hole reorganization energy.

[0105] Each training sample includes a feature value and a hole reorganization energy of a second molecule. By constructing a plurality of training samples in the above manner, the plurality of training samples can be used for model training to obtain a target model, which facilitates subsequent prediction of the hole reorganization energy of the first molecule using the target model to quickly obtain the hole reorganization energy of the first molecule.

[0106] In another embodiment of the present application, the number of feature values corresponding to each second molecule is multiple;

[0107] The feature extraction of each second molecule to obtain the feature value corresponding to each second molecule includes:

[0108] characteristic extraction is performed on each of the second molecules to obtain a plurality of initial characteristic values of each of the second molecules, each of the initial characteristic values being used to describe physical and chemical information of the second molecule;

[0109] For any first characteristic value in the plurality of initial characteristic values, a standard deviation and a mean value of the first characteristic value are calculated according to the first characteristic value of each of the second molecules;

[0110] For any of the second molecules, a characteristic value of the second molecule is calculated according to the first characteristic value of the second molecule, the standard deviation and the mean value of the first characteristic value.

[0111] For example, each characteristic value x n respectively, and the standardization processing is shown in the following formula (2);

[0112]

[0113] wherein x is the first characteristic value, μ is the mean value of the first characteristic value, σ is the standard deviation of the first characteristic value, and x * is a characteristic value of the second molecule calculated according to the first characteristic value, the first characteristic value x is centered on the mean value μ and scaled according to the standard deviation σ, and the data x * obeys the standard normal distribution with a mean value of 0 and a variance of 1. For each initial characteristic value, the above-mentioned manner is adopted for processing to obtain a characteristic value of the second molecule.

[0114] In the embodiment, by adopting the above-mentioned manner for standardization processing on each initial characteristic value, each characteristic value of the processed second molecule can be located within the range of the standard normal distribution with a mean value of 0 and a variance of 1, which can eliminate the dimensional difference between different characteristics, make different characteristics comparable, thereby accelerating the convergence speed of the model and improving the precision of the target model.

[0115] In an embodiment of the present application, a plurality of training samples are constructed according to the characteristic value and the hole recombination energy corresponding to each of the second molecules, including:

[0116] A plurality of molecular samples are constructed according to the characteristic value and the hole recombination energy corresponding to each of the second molecules, each of the molecular samples including the characteristic value and the hole recombination energy corresponding to a second molecule;

[0117] The plurality of molecular samples are randomly divided to obtain the plurality of training samples and a plurality of test samples;

[0118] The plurality of initial models are trained using the plurality of training samples to obtain a prediction model corresponding to each of the initial models;

[0119] Screening the plurality of prediction models by using the plurality of test samples to determine the target model.

[0120] For example, the plurality of molecular samples are randomly divided, 80% of the molecular samples are taken as training samples, and 20% of the molecular samples are taken as test samples. The plurality of initial models can be selected according to actual conditions, which are not limited herein. In this embodiment, three initial models, i.e., a random forest (RF) model, a gradient boosting decision tree (GBDT) model, and an eXtreme gradient boosting (XGBoost) model, are taken as examples for illustration.

[0121] The random forest model, the gradient boosting decision tree model, and the eXtreme gradient boosting model are trained by using the plurality of training samples to obtain three prediction models. The prediction performance of the three prediction models can be superior or inferior. The plurality of prediction models are screened by using the plurality of test samples to obtain a target model. The target model is the model with the best prediction performance among the plurality of prediction models.

[0122] In this embodiment, the plurality of training samples and the plurality of test samples are obtained by randomly dividing the plurality of molecular samples. The plurality of training samples are used to train the plurality of initial models to obtain a prediction model corresponding to each initial model. The plurality of prediction models are screened by using the plurality of test samples to determine the target model. In this way, the target model with better performance can be obtained, and the accuracy of the target model in predicting the hole recombination energy of the first molecule is improved.

[0123] In an embodiment of the present application, screening the plurality of prediction models by using the plurality of test samples to determine the target model includes:

[0124] For any prediction model, the characteristic values of the plurality of test samples are input into the prediction model to obtain a plurality of predicted hole recombination energies.

[0125] The determination coefficient of the prediction model is calculated according to the hole recombination energies of the plurality of test samples and the plurality of predicted hole recombination energies.

[0126] The prediction model corresponding to the determination coefficient with the largest value is taken as the target model.

[0127] For example, the characteristic values of the plurality of test samples are input into the prediction model corresponding to the random forest model to obtain a plurality of predicted hole recombination energies. The determination coefficient R of the prediction model corresponding to the random forest model is calculated according to the hole recombination energies of the plurality of test samples and the plurality of predicted hole recombination energies. 2.

[0128] Similarly, the characteristic values of the plurality of test samples are input into the prediction model corresponding to the gradient boosting tree model to obtain a plurality of predicted hole reorganization energies, and the determination coefficient R 2 .

[0129] The characteristic values of the plurality of test samples are input into the prediction model corresponding to the extreme gradient boosting tree model to obtain a plurality of predicted hole reorganization energies, and the determination coefficient R 2 .

[0130] The three determination coefficients obtained by the above calculation are used for model screening, for example, the prediction model corresponding to the determination coefficient with the largest value is taken as the target model.

[0131] The determination coefficient represents the explanatory power of the model on the dependent variable, and the value range of the determination coefficient is from 0 to 1, and the closer to 1 indicates that the fitting degree of the model on the data is better. In the present example, by comparing the sizes of the determination coefficients of a plurality of prediction models, the target model is determined, so that the target model screened is the model with the best fitting degree among all the prediction models, thereby improving the accuracy of the target model in predicting the hole reorganization energy of the first molecule.

[0132] Further, the mean absolute error (MAE) and the root mean square error (RMSE) of each prediction model can also be obtained through a plurality of test samples. In the case where the determination coefficients of two prediction models are the same, and the value of the determination coefficient is the largest among all the prediction models, the target model can be determined by comparing the mean absolute error or the root mean square error of each prediction model, that is, the prediction model corresponding to the determination coefficient with the largest value and the mean absolute error (or root mean square error) with the smallest value is selected as the target model.

[0133] In an embodiment of the present application, after the characteristic values of the first molecule are input into the target model to obtain the hole reorganization energy of the first molecule, the method further comprises:

[0134] According to the hole reorganization energy of the first molecule, the hole mobility of the first molecule is calculated;

[0135] In the case where the hole mobility is greater than a preset threshold, the first molecule is added to a material library, and the molecules in the material library are used to make a target material.

[0136] Specifically, the hole mobility of the first molecule can be calculated according to the Marcus theoretical model, for example, according to formula (3):

[0137]

[0138] where h is the Planck constant, v is the transfer integral, l is the hole recombination energy of the first molecule, K B is the Boltzmann constant, and T is the temperature in Kelvin. The smaller the hole recombination energy l h , the higher the hole mobility K

[0139] The preset threshold value can be set according to actual conditions and is not limited herein. If the hole mobility of the first branch is greater than the preset threshold value, it indicates that the first molecule can meet the application requirement, and the first molecule is added to the material library for subsequent use, for example, the first molecule is used to make a target material.

[0140] In the above manner, the first molecule can be screened according to the hole mobility of the first molecule, and the first molecule is added to the material library in the case where the hole mobility is greater than the preset threshold value, so that molecules meeting the use requirement are added to the material library for subsequent use.

[0141] The following illustrates the method for predicting the hole recombination energy of a molecule provided in the present application. As Figure 2 shown is a flowchart of the method for predicting the hole recombination energy of a molecule provided in an embodiment of the present application, as Figure 2 shown, the method comprises the following steps:

[0142] Step 201, molecule generation. Batch automatic generation of hole transport layer molecule sample space. Input end group and intermediate molecule fragment, and automatically generate a large number of molecules.

[0143] Step 202, calculation of hole recombination energy. An automatic script is written for the molecules generated in step 201, and the hole recombination energy of each molecule is calculated according to DFT. This part of the calculation can use Gaussian quantum chemistry calculation software.

[0144] The calculation formula of the hole recombination energy l is shown in formula (1) above.

[0145] Step 203, training of machine learning model: machine learning modeling process is performed according to the hole recombination energy dataset calculated in step 202, and the specific process is as shown in Figure 3 .

[0146] Step 2031, division of dataset. The hole recombination energy dataset calculated in step 202 is divided into data, 80% of which is a training set and 20% of which is a test set;

[0147] Step 2032, molecular feature extraction. The RDKit molecular toolkit is used for feature extraction of the molecule, a total of 209 features are used to describe the physical and chemical information of the molecule, and the feature value is denoted as x n (n = 1, 2, 3, …, 209);

[0148] Step 2033, data preprocessing. The feature values x n extracted from the molecule are standardized, and the formula is shown in equation (2).

[0149] For each feature value extracted in step 2032, the above method is used for processing, and a feature value of the molecule is obtained.

[0150] Step 2034, model training and verification. Three machine learning methods, random forest (RF), gradient boosting tree (GBDT) and extreme gradient boosting tree (XGBoost), are used to train the data set, and the R2, MAE and RMSE model evaluation indexes of each model are obtained, as shown in Table 1. n_estimators represents the number of trees.

[0151] Table 1

[0152]

[0153] Step 204, screening to obtain a target model. The model with the best effect in the example is selected as the model for predicting new molecules. As shown in Table 1, the Xgboost model has the best effect when n_estimators is 600. As shown in Figure 4 , the graph shows that Tr represents the training set, Te represents the test set, the horizontal coordinate is the calculated hole recombination energy in the sample, and the value in the sample is taken as the true value. The vertical coordinate is the hole recombination energy predicted by the Xgboost model, and the value is the predicted value. The closer the true value and the predicted value, the closer the coordinate point to the dashed line in the graph.

[0154] Step 205, using the target model to predict the hole recombination energy of the new molecule.

[0155] Through the above method, the trained machine learning model can be used to quickly predict the hole recombination energy of the new molecule, and then judge the hole transport ability of the molecule.

[0156] Please refer to Figure 5 , which is a structural schematic diagram of a molecular hole recombination energy prediction device provided by the embodiment of the present application, as shown in Figure 5 , the molecular hole recombination energy prediction device 500 comprises:

[0157] A first acquisition module 501 is configured to acquire a first molecule.

[0158] The prediction module 502 is configured to input the characteristic value of the first molecule into a target model to obtain the hole recombination energy of the first molecule, wherein the target model is trained according to a plurality of training samples, each of which includes a characteristic value of a second molecule and a hole recombination energy calculated according to the second molecule, and the second molecule is a hole transport layer molecule.

[0159] In this embodiment, after the target model is trained, the hole recombination energy of the first molecule is predicted by using the target model, which consumes less computing power and is faster than directly calculating the hole recombination energy of the first molecule, and the efficiency of obtaining the hole recombination energy of the first molecule is improved.

[0160] In an embodiment of the present application, the molecular hole recombination energy prediction device 500 further includes:

[0161] The second acquisition module is configured to acquire a plurality of end groups and a plurality of intermediate molecular fragments.

[0162] The generation module is configured to generate a plurality of second molecules based on the plurality of end groups and the plurality of intermediate molecular fragments.

[0163] The feature extraction module is configured to perform feature extraction on each of the second molecules to obtain a characteristic value corresponding to each of the second molecules.

[0164] The first calculation module is configured to calculate the hole recombination energy of each of the second molecules.

[0165] The construction module is configured to construct a plurality of training samples according to the characteristic value and the hole recombination energy corresponding to each of the second molecules.

[0166] In this embodiment, a plurality of training samples are constructed by the above method, and the target model can be obtained by using the plurality of training samples for model training, which facilitates subsequent prediction of the hole recombination energy of the first molecule by using the target model to quickly obtain the hole recombination energy of the first molecule.

[0167] In an embodiment of the present application, the first calculation module specifically includes:

[0168] For any second molecule, the sum of a first value and a second value is taken as the hole recombination energy of the second molecule, the first value is the energy difference of different neutral state structures in the second molecule, and the second value is the energy difference of different cationic state structures in the second molecule.

[0169] In the embodiment, for the second molecules, the hole reorganization energy of the second molecules is obtained by calculation, a training sample is constructed, and the target model is trained. Although the calculation process consumes a large amount of calculation power, because the number of training samples is limited, the calculation power can be estimated and consumed in advance. Once the target model is trained, the hole reorganization energy of the first molecules can be predicted by using the target model, and the efficiency of obtaining the hole reorganization energy is improved.

[0170] In an embodiment of the present application, the number of feature values corresponding to each of the second molecules is multiple;

[0171] The feature extraction module comprises:

[0172] The first obtaining sub-module is configured to extract features of each of the second molecules to obtain multiple initial feature values of each of the second molecules, and each of the initial feature values is used to describe the physical and chemical information of the second molecule.

[0173] The second obtaining sub-module is configured to, for any first feature value in the multiple initial feature values, calculate the standard deviation and the mean of the first feature value according to the first feature value of each of the second molecules.

[0174] The third obtaining sub-module is configured to, for any of the second molecules, calculate a feature value of the second molecule according to the first feature value of the second molecule, the standard deviation and the mean of the first feature value.

[0175] In the embodiment, by using the above method to standardize each initial feature value, each feature value of the second molecule obtained by processing can be located in the range of standard normal distribution with a mean of 0 and a variance of 1, which can eliminate the dimensional difference between different features, make different features comparable, thereby accelerating the convergence speed of the model and improving the accuracy of the target model.

[0176] In an embodiment of the present application, the construction module comprises:

[0177] The construction sub-module is configured to construct multiple molecular samples according to the feature value corresponding to each of the second molecules and the hole reorganization energy, and each molecular sample comprises the feature value corresponding to a second molecule and the hole reorganization energy.

[0178] The fourth obtaining sub-module is configured to randomly divide the multiple molecular samples to obtain the multiple training samples and multiple test samples.

[0179] The training sub-module is configured to train each of the initial models by using the multiple training samples to obtain a prediction model corresponding to each of the initial models.

[0180] The screening submodule is configured to screen the plurality of prediction models by using the plurality of test samples, and determine the target model.

[0181] In this embodiment, the plurality of training samples and the plurality of test samples are obtained by randomly dividing the plurality of molecular samples, the plurality of initial models are trained by using the plurality of training samples, and the prediction model corresponding to each initial model is obtained; the plurality of prediction models are screened by using the plurality of test samples, and the target model is determined, so that the target model with better performance can be obtained, and the accuracy of predicting the hole recombination energy of the first molecule by the target model is improved.

[0182] In an embodiment of the present application, the screening submodule comprises:

[0183] The first obtaining unit is configured to, for any prediction model, input the characteristic values of the plurality of test samples into the prediction model, and obtain a plurality of predicted hole recombination energies.

[0184] The second obtaining unit is configured to calculate the determination coefficient of the prediction model according to the hole recombination energies of the plurality of test samples and the plurality of predicted hole recombination energies.

[0185] The screening unit is configured to take the prediction model corresponding to the determination coefficient with the maximum value as the target model.

[0186] In this embodiment, the target model is determined by comparing the determination coefficients of the plurality of prediction models, so that the target model screened is the model with the best fitting degree among all the prediction models, and the accuracy of predicting the hole recombination energy of the first molecule by the target model is improved.

[0187] In an embodiment of the present application, the molecular hole recombination energy prediction device 500 further comprises:

[0188] The second calculation module is configured to calculate the hole mobility of the first molecule according to the hole recombination energy of the first molecule.

[0189] The adding module is configured to, in a case where the hole mobility is greater than a preset threshold, add the first molecule into a material library, and the molecules in the material library are used to make a target material.

[0190] In this embodiment, the first molecule can be screened according to the hole mobility of the first molecule by the above method, and the first molecule is added into the material library in a case where the hole mobility is greater than a preset threshold, so that the molecules meeting the use requirements are added into the material library, and subsequent use is facilitated.

[0191] In addition, an embodiment of the present application can be implemented by a computer storage medium. The computer storage medium stores computer program instructions. The computer program instructions are executed by a processor to implement any of the above-mentioned molecular hole recombination energy prediction methods.

[0192] An embodiment of the present application provides a computer program product stored in a storage medium. The computer program product is executed by at least one processor to implement the processes of the above-mentioned method embodiments, and achieve the same technical effects. To avoid repetition, details are not described here.

[0193] It should be noted that the present application is not limited to the specific configurations and processes described above and shown in the drawings. For the sake of brevity, detailed descriptions of well-known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method processes of the present application are not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications and additions, or change the order of steps, after understanding the spirit of the present application.

[0194] The functional blocks shown in the above structural block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, etc. When implemented in software, the elements of the present application are program or code segments used to perform the required tasks. The program or code segments can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link. The "machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segments can be downloaded via a computer network such as the Internet, an intranet, etc.

[0195] It should also be noted that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be executed in the order mentioned in the embodiments, or in an order different from the embodiments, or several steps can be executed simultaneously.

[0196] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0197] Finally, it should be noted that the above-described embodiments are merely intended to illustrate the technical solutions of the present application, but not to limit the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified, or some or all of the technical features can be replaced by equivalent replacements; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should be covered in the scope of the claims and the specification of the present application. In particular, each technical feature mentioned in each embodiment can be combined in any manner. The present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for predicting molecular cavity reorganization energy, characterized by, The method comprises: obtaining a first molecule; inputting a characteristic value of the first molecule into a target model to obtain a hole reorganization energy of the first molecule, wherein the target model is trained according to a plurality of training samples, each of the training samples comprising a characteristic value of a second molecule and a hole reorganization energy calculated according to the second molecule, and the second molecule is a hole transport layer molecule.

2. The method of claim 1, wherein the molecular cavity reorganization energy is calculated by the following equation: ###0001### where: E = the molecular cavity reorganization energy; R = the radius of the cavity; and n = the number of atoms in the cavity. Before the inputting of the characteristic value of the first molecule into the target model to obtain the hole reorganization energy of the first molecule, the method further comprises: obtaining a plurality of end groups and a plurality of intermediate molecular fragments; generating a plurality of second molecules based on the plurality of end groups and the plurality of intermediate molecular fragments; extracting features of each of the second molecules to obtain a characteristic value corresponding to each of the second molecules; calculating a hole reorganization energy of each of the second molecules; constructing a plurality of training samples according to the characteristic value and the hole reorganization energy corresponding to each of the second molecules.

3. The method of claim 2, wherein the molecular cavity is a cavity of a protein. The calculating of the hole reorganization energy of each of the second molecules comprises: for any of the second molecules, taking a sum of a first value and a second value as the hole reorganization energy of the second molecule, the first value being an energy difference of different neutral state structures in the second molecule, and the second value being an energy difference of different cationic state structures in the second molecule.

4. The method of claim 2, wherein the molecular cavity is a cavity of a protein. The number of the characteristic values corresponding to each of the second molecules is a plurality; The extracting of features of each of the second molecules to obtain a characteristic value corresponding to each of the second molecules comprises: extracting features of each of the second molecules to obtain a plurality of initial characteristic values of each of the second molecules, each of the initial characteristic values being used to describe physical and chemical information of the second molecule; for any of the initial characteristic values, calculating a standard deviation and a mean value of the first characteristic value according to the first characteristic value of each of the second molecules; for any of the second molecules, calculating a characteristic value of the second molecule according to the first characteristic value of the second molecule, the standard deviation and the mean value of the first characteristic value.

5. The method of claim 2, wherein the molecular cavity is a cavity of a protein. The constructing of a plurality of training samples according to the characteristic value and the hole reorganization energy corresponding to each of the second molecules comprises: constructing a plurality of molecular samples according to the characteristic value and the hole reorganization energy corresponding to each of the second molecules, each of the molecular samples comprising a characteristic value and a hole reorganization energy corresponding to a second molecule; randomly dividing the plurality of molecular samples to obtain the plurality of training samples and a plurality of test samples; training a plurality of initial models using the plurality of training samples to obtain a prediction model corresponding to each of the initial models; screening a plurality of the prediction models using the plurality of test samples to determine the target model.

6. The method of claim 5, wherein the molecular cavity is a cavity of a protein. The screening of a plurality of the prediction models using the plurality of test samples to determine the target model comprises: for any of the prediction models, inputting the characteristic values of the plurality of test samples into the prediction model to obtain a plurality of predicted hole reorganization energies; calculating a determination coefficient of the prediction model according to the hole reorganization energies of the plurality of test samples and the plurality of predicted hole reorganization energies; taking a prediction model corresponding to a determination coefficient with a maximum value as the target model.

7. The method of predicting molecular cavity recombinant energy according to any one of claims 1-6, wherein, After the characteristic value of the first molecule is input into a target model to obtain a hole recombination energy of the first molecule, the method further comprises: calculating a hole mobility of the first molecule according to the hole recombination energy of the first molecule; in a case where the hole mobility is greater than a preset threshold, adding the first molecule into a material library, wherein the molecules in the material library are used to make a target material.

8. A molecular cavity reorganization energy prediction apparatus, comprising: a cavity reorganization energy prediction module configured to predict a cavity reorganization energy of a molecule based on a molecular structure of the molecule. comprise: a first acquisition module configured to acquire a first molecule; a prediction module configured to input a characteristic value of the first molecule into a target model to obtain a hole recombination energy of the first molecule, wherein the target model is trained according to a plurality of training samples, each of the training samples comprises a characteristic value of a second molecule and a hole recombination energy calculated according to the second molecule, and the second molecule is a hole transport layer molecule.

9. A readable storage medium, characterized by, The program or instruction is stored on the readable storage medium, and the program or instruction is executed by the processor to implement the steps of the hole recombination energy prediction method according to any one of claims 1 to 7.

10. A computer program product, characterised in that, The computer program product is stored in a storage medium, and when executed by at least one processor, the computer program product performs the steps of the hole recombination energy prediction method according to any one of claims 1 to 7.

Images