Method and device for predicting glass transition temperature of polymer and electronic equipment

Abstract

The invention provides a method and device for predicting the glass transition temperature of a polymer and electronic equipment, and the method comprises the steps: obtaining a plurality of molecular descriptors of the polymer, obtaining an original database of the characteristic parameters of the polymer, and enabling the molecular descriptors to be used for representing the glass transition temperature distribution of the polymer, the characteristic parameter and the distribution of the polymer glass transition temperature have a preset relationship; according to the feature engineering, removing a plurality of feature parameters in the original database to obtain an optimized database; and constructing a deep learning model at least according to the characteristic parameters in the optimization database, determining a mapping relationship between the characteristic parameters of the polymer and the glass transition temperature according to the deep learning model, the characteristic parameters and an SHAP analysis method, and predicting the glass transition temperature of the polymer according to the mapping relationship. The problem of inaccurate polymer prediction results in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the field of polymer glass transition temperature prediction, and in particular, to a polymer glass transition temperature prediction method, device, computer-readable storage medium and electronic device. Background Art

[0002] In the prior art, the glass transition temperature of polymers has the following methods: generating corresponding molecular descriptors by simplifying molecular linear input specification expressions (SMILES), selecting the most important molecular descriptors, constructing a glass transition temperature prediction model and optimizing the model. However, linear models can only explain linear relationships, and have limited ability to explain complex interaction effects or nonlinear relationships; or by obtaining characteristic data of the polymer monomer to be predicted, the characteristic data includes data files or structure files, and inputting the characteristic data into a pre-trained temperature prediction model to obtain the glass transition temperature of the polymer monomer to be predicted; wherein the temperature prediction model is trained based on a data set sample collected from a database, and the data set sample includes at least the polymer name, the polymer monomer SMILES string and the glass transition temperature. However, the working mechanism of the convolutional neural network is relatively complex and belongs to a "black box" model, which makes it difficult to explain how the model makes predictions and the reliability of the results. Summary of the invention

[0003] The main purpose of the present application is to provide a method, device, computer-readable storage medium and electronic device for predicting the glass transition temperature of a polymer, so as to at least solve the problem of inaccurate prediction results of polymers in the prior art.

[0004] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for predicting the glass transition temperature of a polymer is provided, the prediction method comprising: obtaining multiple molecular descriptors of the polymer, and obtaining an original database of characteristic parameters of the polymer based on the multiple molecular descriptors, the molecular descriptors are used to characterize the distribution of the glass transition temperature of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer; according to feature engineering, multiple characteristic parameters in the original database are removed to obtain an optimized database, and the multiple characteristic parameters have at least one of the following characteristics: a single-valued feature and a highly linearly correlated feature; at least based on the characteristic parameters in the optimized database, a deep learning model is constructed, and based on the deep learning model, the characteristic parameters and a method for explaining the output of a machine learning model based on game theory (SHAP analysis method, SHapleyAdditive exPlanations), a mapping relationship between the characteristic parameters of the polymer and the glass transition temperature is determined, and the glass transition temperature of the polymer is predicted based on the mapping relationship.

[0005] Optionally, the prediction method further comprises: obtaining a plurality of normalized characteristic parameters according to the plurality of characteristic parameters and a first formula, wherein the first formula is: Wherein: X is the original value of the characteristic parameter, x is the value after the original value of the characteristic parameter is normalized, Max is the maximum value of the original value of the characteristic parameter, and Min is the minimum value of the original value of the characteristic parameter.

[0006] Optionally, constructing the deep learning model includes: obtaining multiple learning models and optimizing parameters of the multiple learning models according to a grid search algorithm; and constructing the deep learning model based on the optimized multiple learning models.

[0007] Optionally, the prediction method also includes: obtaining the glass transition temperatures of multiple polymers and constructing a temperature database of the polymer glass transition; judging whether the deep learning model meets a preset accuracy rate based on the deep learning model, data in the temperature database and data in the optimization database; if the judgment result indicates that it is not satisfied, re-optimizing the parameters of the learning model; if the judgment result indicates that it is satisfied, saving the deep learning model.

[0008] Optionally, judging whether the deep learning model meets a preset accuracy rate according to the deep learning model, the database and the optimized database includes: obtaining the prediction accuracy rate of the deep learning model according to a prediction accuracy calculation formula, a predicted polymer glass transition temperature output by the deep learning model according to the characteristic parameters and data of the temperature database, wherein the data of the temperature database includes glass transition temperature data of data points calculated by digital simulation software at the target time, and the accuracy calculation formula is: Wherein, Acc is the prediction accuracy of the deep learning model at the target time, S opi is the glass transition temperature of data point i predicted by the deep learning model at the target time, S ori is the glass transition temperature of data point i calculated by the digital model software at the target time, and n is the number of data points; the classification accuracy of the deep learning model is obtained according to the classification accuracy calculation formula, the number of unit cells of the polymer correctly classified by the deep learning model and the total number of unit cells of the polymer, and the classification accuracy calculation formula is: Wherein, Acc1 is the classification accuracy of the deep learning model, n true is the number of unit cells of the polymer correctly classified by the deep learning model, and n is the total number of unit cells of the polymer; judging whether the prediction accuracy meets the preset prediction accuracy, and whether the classification accuracy meets the preset classification accuracy, and if the prediction accuracy meets the preset prediction accuracy, and the classification accuracy meets the preset classification accuracy, determining that the deep learning model meets the preset accuracy.

[0009] Optionally, determining whether the deep learning model meets a preset accuracy rate further includes: determining whether the deep learning model meets a preset accuracy rate based on multiple evaluation criteria, the evaluation criteria including: multiple ones of determination coefficient, mean absolute error, mean square error, root mean square error and mean absolute relative deviation.

[0010] Optionally, the multiple learning models include: multiple ones of a deep neural network model, a random forest model and a support vector machine model.

[0011] According to another aspect of the present application, a device for predicting the glass transition temperature of a polymer is provided, comprising: an acquisition module, used to acquire multiple molecular descriptors of the polymer, and obtain an original database of characteristic parameters of the polymer based on the multiple molecular descriptors, wherein the molecular descriptors are used to characterize the distribution of the glass transition temperature of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer; a first determination module, used to remove multiple characteristic parameters in the original database according to feature engineering to obtain an optimized database, wherein the multiple characteristic parameters have at least one of the following characteristics: a feature with only a single value and a feature with a high linear correlation; a second determination module, used to construct a deep learning model based on at least the characteristic parameters in the optimized database, and determine the mapping relationship between the characteristic parameters of the polymer and the glass transition temperature based on the deep learning model, the characteristic parameters and the SHAP analysis method, and predict the glass transition temperature of the polymer according to the mapping relationship.

[0012] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for predicting the glass transition temperature of a polymer.

[0013] According to another aspect of the present application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the prediction method of the polymer glass transition temperature.

[0014] Applying the technical solution of the present application, firstly, a plurality of molecular descriptors of the above-mentioned polymer are obtained, and based on the plurality of the molecular descriptors, an original database of characteristic parameters of the above-mentioned polymer is obtained, wherein the molecular descriptors are used to characterize the distribution of the glass transition temperature of the above-mentioned polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the above-mentioned polymer; according to feature engineering, the plurality of characteristic parameters of the above-mentioned original database with at least a single-valued feature and a highly linearly correlated feature are removed to obtain an optimized database; at least based on the characteristic parameters in the above-mentioned optimized database, a deep learning model is constructed, and based on the deep learning model, the characteristic parameters and the SHAP analysis method, the mapping relationship between the characteristic parameters of the above-mentioned polymer and the glass transition temperature is determined, and the glass transition temperature of the above-mentioned polymer is predicted based on the mapping relationship. By using molecular descriptors as a representation method for polymers, the prediction results of the deep learning model are compared with the existing data. The use of a deep learning model and the superior performance based on molecular descriptors as inputs of polymer features highlight its potential for handling the complexity of polymer features. The present application determines the key features associated with high glass transition temperature and low glass transition temperature values ​​through SHAP analysis, improves the reliability of polymer prediction results, and solves the problem of inaccurate polymer prediction results in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0016] Figure 1 A hardware structure block diagram of a mobile terminal for executing a method for predicting the glass transition temperature of a polymer provided in an embodiment of the present application is shown;

[0017] Figure 2 A schematic flow chart of a method for predicting the glass transition temperature of a polymer provided in an embodiment of the present application is shown;

[0018] Figure 3 A schematic flow chart of a deep neural network model structure used in a method for predicting the glass transition temperature of a polymer provided in an embodiment of the present application is shown;

[0019] Figure 4 shows a distribution diagram of a glass transition temperature data set range of a polymer provided according to an embodiment of the present application;

[0020] Figure 5 A schematic structural diagram of a device for predicting the glass transition temperature of a polymer provided according to an embodiment of the present application is shown.

[0021] The above drawings include the following reference numerals:

[0022] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0023] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0024] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] As introduced in the background technology, the linear model in the existing polymer glass transition temperature method can only explain linear relationships, and has limited ability to explain complex interaction effects or nonlinear relationships; or the working mechanism of the convolutional neural network is relatively complex and belongs to a "black box" model, which makes it difficult to explain how the model makes predictions and the reliability of the results.

[0027] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0028] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 1 is a hardware structure block diagram of a mobile terminal of a method for predicting the glass transition temperature of a polymer according to an embodiment of the present invention. Figure 1As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0029] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for predicting the glass transition temperature of a polymer in an embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The transmission device 106 is used to receive or send data via a network. The above-mentioned specific examples of the network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0030] In this embodiment, a method for predicting the glass transition temperature of a polymer running on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0031] Figure 2: is a flow chart of a method for predicting the glass transition temperature of a polymer according to an embodiment of the present application. Figure 2 As shown, the above prediction method includes:

[0032] Step S201, obtaining a plurality of molecular descriptors of the polymer, and obtaining an original database of characteristic parameters of the polymer based on the plurality of molecular descriptors, wherein the molecular descriptors are used to characterize the glass transition temperature distribution of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer;

[0033] Step S202, according to feature engineering, multiple feature parameters in the original database are removed to obtain an optimized database, wherein the multiple feature parameters have at least one of the following characteristics: a single value feature and a highly linearly correlated feature;

[0034] Step S203, constructing a deep learning model at least based on the characteristic parameters in the above-mentioned optimization database, and determining the mapping relationship between the characteristic parameters of the above-mentioned polymer and the above-mentioned glass transition temperature based on the above-mentioned deep learning model, the above-mentioned characteristic parameters and the SHAP analysis method, and predicting the glass transition temperature of the above-mentioned polymer based on the above-mentioned mapping relationship.

[0035] By obtaining multiple molecular descriptors of the above polymers, and based on the multiple molecular descriptors, an original database of characteristic parameters of the above polymers is obtained, the molecular descriptors are used to characterize the distribution of the glass transition temperature of the above polymers, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the above polymers; according to feature engineering, multiple characteristic parameters of at least a single-valued feature and a highly linearly correlated feature in the above original database are removed to obtain an optimized database; at least based on the characteristic parameters in the above optimized database, a deep learning model is constructed, and based on the deep learning model, the characteristic parameters and the SHAP analysis method, the mapping relationship between the characteristic parameters of the above polymer and the glass transition temperature is determined, and the glass transition temperature of the above polymer is predicted based on the mapping relationship. By using molecular descriptors as a representation method for polymers, the prediction results of the deep learning model are compared with existing data. The use of deep learning models and the superior performance based on molecular descriptors as inputs of polymer features highlights its potential for handling the complexity of polymer features. Therefore, the present application determines the key features associated with high glass transition temperature and low glass transition temperature values ​​through SHAP analysis, improves the reliability of polymer prediction results, and solves the problem of inaccurate polymer prediction results in the prior art.

[0036] In the injection molding or extrusion process of plastics and rubber, it is crucial to control the glass transition temperature of the polymer. It is necessary to ensure that the material is in an appropriate state (such as a viscous flow state) at the processing temperature so that it can be smoothly molded. After the glass transition temperature data of the above polymer is predicted according to the above mapping relationship, the injection molding equipment is controlled to adjust the processing conditions, such as temperature, pressure, etc., to optimize the shape and performance of the product.

[0037] In the above optional embodiment, the molecular property description module (MoleculeDescriptors) in the RDKit framework is used to calculate the descriptors, and 208 descriptors are generated for each polymer monomer. The numerical range of the descriptors is wide, from -1.68 to 25661097.01, which may lead to excessive attention to larger values ​​and ignore smaller values, thereby affecting the generalization performance of the model. To solve this problem, descriptor scaling is applied to standardize the values ​​to a mean of 0 and a standard deviation of 1. Through feature engineering, the box plot method is first used to find out the outliers. The outliers include missing values, values ​​of features with only a single value, and values ​​of highly linearly correlated features. Then, the outliers are replaced by the mean of all the values ​​in the molecular descriptor.

[0038] In some optional implementations, the prediction method further includes: obtaining a plurality of normalized characteristic parameters according to the plurality of characteristic parameters and the first formula, wherein the first formula is: Wherein: X is the original value of the above characteristic parameter, x is the value after the original value of the above characteristic parameter is normalized, Max is the maximum value of the original value of the above characteristic parameter, and Min is the minimum value of the original value of the above characteristic parameter.

[0039] In the above optional implementation, a deviation standardization method is used in the normalization process to perform a linear transformation on the original data to reduce the numerical value and improve the accuracy. The specific conversion function is the above first formula.

[0040] In some optional embodiments, the above-mentioned construction of the deep learning model includes: obtaining multiple learning models, and optimizing the parameters of the multiple learning models according to a grid search algorithm; and constructing the above-mentioned deep learning model according to the optimized multiple learning models.

[0041] In the above optional implementation, three machine learning models, deep neural network (DNN), random forest (RF) and support vector machine (SVM), are used to establish a deep learning model for predicting the glass transition temperature of a polymer. The polymer glass transition temperature data is regressed using deep neural network, random forest and support vector machine models, and the molecular descriptor is used to represent the polymer monomer structure as the model input. The three models are regarded as independent frameworks, and grid search is used for parameter optimization. The machine learning model is trained to ensure that it captures the data pattern, and the best parameters are obtained to form a deep learning model.

[0042] In some optional embodiments, the multiple learning models mentioned above include: multiple ones of deep neural network models, random forest models and support vector machine models.

[0043] In the above optional embodiment, for the deep neural network-molecular descriptor model, the optimization involves the number of nodes in the two hidden layers, the learning rate and the activation function, and the range of neurons is set between 50 and 120, with a step size of 10. The learning rate and activation function have specific ranges. All combinations of different hidden layer structures, learning rates and activation functions are tested within these ranges; the best performance of the deep neural network-molecular descriptor occurs when hidden_layer_1 (the number of neurons in hidden layer 1) is 100, hidden_layer_2 (the number of neurons in hidden layer 2) is 90, learning_rate (learning rate) is 0.01 and the activation function is logistic, and the weight calculation equation is:

[0044]

[0045] Among them, w i，j is the weight vector, x i is the input vector (i, j = 1, 2, ..., n), b is the bias, f is the transfer function, and y is the output.

[0046] In the Random Forest-Molecular Descriptor model, the key parameters max_depth (maximum number of splits each tree has), n_estimators (number of trees in the random forest), and max_features (number of features per split tree) were optimized. max_depth and n_estimators were optimized in the range of 1 to 10 with a step size of 1. max_features was optimized between 50 and 100 with a step size of 10.

[0047] The best configuration for Random Forest-Molecular Descriptor is when max_depth is 4, n_estimators is 8, and max_features is 90;

[0048] The support vector machine-molecular descriptor model optimizes three key parameters: penalty coefficient, kernel function coefficient, and tolerance in the insensitive loss function. The above key parameters are all parameters involved in the grid search algorithm;

[0049] The best configuration of the support vector machine-molecular descriptor is when the penalty coefficient is 450, the kernel function coefficient is 0.004, and the tolerance in the insensitive loss function is 4. The comprehensive optimization of these key parameters enables the support vector machine-molecular descriptor model to achieve enhanced performance and generalization ability on the glass transition temperature data;

[0050] The above polymer data after feature engineering and normalization are split into training set, validation set, and test set, with the ratios of 0.8:0.1:0.1 respectively.

[0051] In some optional embodiments, the above-mentioned prediction method also includes: obtaining the glass transition temperatures of the above-mentioned multiple polymers, and constructing a temperature database of the glass transition of the above-mentioned polymers; judging whether the above-mentioned deep learning model meets the preset accuracy rate based on the above-mentioned deep learning model, the data of the above-mentioned temperature database and the data of the above-mentioned optimization database; if the judgment result indicates that it is not satisfied, re-optimizing the parameters of the above-mentioned learning model, and if the above-mentioned judgment result indicates that it is satisfied, saving the above-mentioned deep learning model.

[0052] In the above optional implementation, multiple types of polymers are collected to construct a polymer glass transition temperature database, and 192 types of polymers can be collected, and the glass transition temperatures of polymers range from 178 to 600K. The accuracy of multiple learning models used to construct the deep learning model is calculated to determine whether the output accuracy of the current deep learning model meets the preset requirements. If not, the values ​​of the molecular descriptors need to be re-optimized, and the training sets in the optimized values ​​are re-input into the multiple learning models for re-optimization until it is determined that the output accuracy of the current deep learning model meets the preset requirements. The values ​​of the molecular descriptors of the above-optimized polymers are input into the optimized deep learning model, and the output values ​​are compared with the glass transition temperatures of the collected polymers to obtain the output error of the deep learning model for the glass transition temperature of the polymer. When the above error is greater than the preset threshold, the deep learning model also needs to be re-optimized.

[0053] In some optional embodiments, judging whether the deep learning model meets the preset accuracy according to the deep learning model, the database and the optimized database includes: obtaining the prediction accuracy of the deep learning model according to the prediction accuracy calculation formula, the predicted polymer glass transition temperature output by the deep learning model according to the characteristic parameters and the data of the temperature database, the data of the temperature database includes the glass transition temperature data of the data points calculated by the digital simulation software at the target time, and the accuracy calculation formula is: Where Acc is the prediction accuracy of the deep learning model at the target time, S opi is the glass transition temperature of data point i predicted by the deep learning model at the target time, S ori is the glass transition temperature of data point i calculated by the digital simulation software at the target time, and n is the number of data points; the classification accuracy of the deep learning model is obtained according to the classification accuracy calculation formula, the number of unit cells of the polymer correctly classified by the deep learning model, and the total number of unit cells of the polymer, and the classification accuracy calculation formula is: Among them, Acc1 is the classification accuracy of the above deep learning model, n true is the number of units of the polymer correctly classified by the deep learning model, and n is the total number of units of the polymer; judging whether the prediction accuracy satisfies the preset prediction accuracy, and whether the classification accuracy satisfies the preset classification accuracy. When the prediction accuracy satisfies the preset prediction accuracy, and the classification accuracy satisfies the preset classification accuracy, it is determined that the deep learning model satisfies the preset accuracy.

[0054] In the above optional implementation, the parameters to be adjusted in the deep neural network model mainly include: the number of neural network layers and the number of neurons in each layer, the hidden layer dropout that controls the neuron disconnection ratio of the input linear transformation, the activation function Activation, the loss function loss, the optimizer Optimizer, and the number of iterations epochs and the number of training samples batch_size during model training. The calculation formula for prediction accuracy is as follows:

[0055]

[0056] Where Acc is the prediction accuracy of the deep learning model at the target time, S opi is the glass transition temperature of data point i predicted by the deep learning model at the target time, S oriis the glass transition temperature of data point i calculated by the numerical simulation software at the above target time, and n is the number of data points. The main parameters that need to be optimized in the support vector machine model are: penalty coefficient, kernel function coefficient, and tolerance in the insensitive loss function. The calculation formula for classification accuracy is as follows:

[0057]

[0058] Among them, Acc1 is the classification accuracy of the above deep learning model, n true is the number of unit cells of the above polymer correctly classified by the above deep learning model, and n is the total number of unit cells of the above polymer.

[0059] There are two main types of commonly used activation functions included in the grid search algorithm. An activation function is a logistic function of any variable x, defined as: Another activation function is a tangent function of any variable a, defined as:

[0060] In some optional embodiments, the above-mentioned determination of whether the above-mentioned deep learning model meets the preset accuracy rate also includes: judging whether the above-mentioned deep learning model meets the preset accuracy rate according to multiple evaluation criteria, and the above-mentioned evaluation criteria include: determination coefficient, mean absolute error, mean square error, root mean square error and mean absolute relative deviation.

[0061] In the above optional implementation, a comprehensive set of evaluation indicators is used to evaluate the performance of the above multiple machine learning models, including determination coefficient, mean absolute error, mean square error, root mean square error and mean absolute relative deviation. The calculation equation is as follows:

[0062]

[0063] Among them, y exp represents the experimental value; y pre Indicates that the experimental results are the predicted values ​​of the model; is the average value; n is the total number of data.

[0064] A deep neural network is a machine learning model that is usually composed of multiple neural network layers and is used to solve various complex tasks and problems. Its main feature is that it has multiple layers of neurons that can simulate and learn complex features and representations in input data. Through multi-layer nonlinear mapping, deep neural networks can fit highly complex functions and finally output the selected target value through nonlinear transmission between layers. The random forest model improves the accuracy and robustness of the model by constructing multiple decision trees and combining their prediction results. It is particularly good at dealing with high-dimensional data and missing value problems, and has good interpretability for feature importance assessment. It is widely used in classification and regression tasks. Support vector machines are mainly used for classification and regression analysis. Data classification is achieved by finding a hyperplane that maximizes the classification boundary interval. Deep neural network, random forest and support vector machine models are used to perform regression analysis on polymer glass transition temperature data. Molecular descriptors and Morgan fingerprints are used to represent the polymer monomer structure as model input. These three models are regarded as independent frameworks, and grid search is used for parameter optimization to train machine learning models to ensure that they capture data patterns. The SHAP value of each sample is calculated using the SHAP library to indicate the contribution of each feature to the model output.

[0065] For example, from the known database, we know that the glass transition temperature of polyvinyl bromide is 373K. Figure 2 The prediction method of the glass transition temperature of the polymer was used to obtain the prediction result of the final model, which was 372.968K, with an error of 0.032. From the known database, it is known that the glass transition temperature of polynonamethylene terephthalamide is 394K. Figure 2 The prediction method of the glass transition temperature of the polymer was used to obtain the final model prediction result of 393.971K with an error of 0.029. The structure of the deep neural network model used is as follows Figure 3 As shown in the figure, three results are selected from the overall data set, and the best solution is decided by voting among the three results, and finally the final prediction value is obtained. Figure 4 It is the distribution graph of the glass transition temperature data set range (polymer glass transition temperature K-number of data points).

[0066] Exemplarily, a plurality of molecular descriptors of the polymer are obtained and preprocessed, descriptor scaling is applied, and the values ​​are standardized to a mean of 0 and a standard deviation of 1. The outliers are first found by using a box plot method through feature engineering. The outliers include missing values, values ​​of features with only a single value, and values ​​of features with high linear correlation. The outliers are then replaced by the mean of all the values ​​in the molecular descriptors. After the outliers are replaced, the molecular descriptors are normalized. In the normalization process, the deviation standardization method is used to perform a linear transformation on the original data to reduce the values ​​and improve the accuracy. The data of the normalized molecular descriptors are divided into a training set, a validation set, and a test set, with a ratio of 0.8:0.1:0.1, respectively. A deep neural network model, a random forest model, and A support vector machine model learning model is used, and the parameters of the above learning model are optimized according to the grid search algorithm and the training set data, the above deep learning model is constructed, the training set is sent to the deep neural network for learning, the validation set is used to verify the model accuracy and the configuration of the hyperparameters, and the evaluation score of the test set is used to evaluate the prediction results of the polymer glass transition temperature. A comprehensive evaluation index is used to evaluate the performance of the deep learning model, including the determination coefficient, mean absolute error, mean square error, root mean square error and mean absolute relative deviation; SHAP analysis is used to calculate the SHAP value of each sample using the SHAP library to indicate the contribution of each feature to the model output, determine the mapping relationship between the characteristic parameters of the above polymer and the above glass transition temperature, and predict the glass transition temperature of the above polymer according to the above mapping relationship.

[0067] The present application embodiment also provides a kind of prediction device of polymer glass transition temperature. It should be noted that the prediction device of polymer glass transition temperature of the present application embodiment can be used to perform the prediction method for polymer glass transition temperature provided by the present application embodiment. The device is used to implement the above-mentioned embodiment and preferred embodiment, and the description has been made and will not be repeated. As used below, the term "module" can realize the combination of software and / or hardware of predetermined functions. Although the device described in the following embodiments is preferably implemented with software, the implementation of hardware, or the combination of software and hardware is also possible and conceived.

[0068] The following is an introduction to the device for predicting the glass transition temperature of a polymer provided in the embodiments of the present application.

[0069] Figure 5 Schematic diagram of a device for predicting the glass transition temperature of a polymer according to an embodiment of the present application. Figure 5As shown, the device for predicting the glass transition temperature of a polymer includes an acquisition module 100, which is used to acquire multiple molecular descriptors of the above-mentioned polymer, and obtain an original database of characteristic parameters of the above-mentioned polymer based on the multiple molecular descriptors, wherein the above-mentioned molecular descriptors are used to characterize the distribution of the glass transition temperature of the above-mentioned polymer, and the above-mentioned characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the above-mentioned polymer; a first determination module 200, which is used to remove multiple characteristic parameters in the above-mentioned original database according to feature engineering to obtain an optimized database, wherein the above-mentioned multiple characteristic parameters have at least one of the following characteristics: a feature with only a single value and a feature with high linear correlation; a second determination module 300, which is used to construct a deep learning model based on at least the characteristic parameters in the above-mentioned optimized database, and determine the mapping relationship between the characteristic parameters of the above-mentioned polymer and the above-mentioned glass transition temperature based on the above-mentioned deep learning model, the above-mentioned characteristic parameters and the SHAP analysis method, and predict the glass transition temperature of the above-mentioned polymer according to the above-mentioned mapping relationship.

[0070] The prediction device of the glass transition temperature of the polymer of the present application, firstly, the acquisition module obtains multiple molecular descriptors of the polymer, and obtains the original database of the characteristic parameters of the polymer according to the multiple molecular descriptors, the molecular descriptors are used to characterize the distribution of the glass transition temperature of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer; the first determination module removes the multiple characteristic parameters of the characteristics with at least a single value and a highly linearly correlated characteristic in the original database according to feature engineering, and obtains an optimized database; the second determination module constructs a deep learning model according to at least the characteristic parameters in the optimized database, and determines the mapping relationship between the characteristic parameters of the polymer and the glass transition temperature according to the deep learning model, the characteristic parameters and the SHAP analysis method, and predicts the glass transition temperature of the polymer according to the mapping relationship. By using molecular descriptors as a representation method for polymers, the prediction results of the deep learning model are compared with the existing data. The use of deep learning models and the superior performance based on molecular descriptors as inputs of polymer features highlights its potential for handling the complexity of polymer features. The key features associated with high and low glass transition temperature values ​​were identified through SHAP analysis, which improved the reliability of polymer prediction results and solved the problem of inaccurate polymer prediction results in the prior art.

[0071] In some optional implementations, the prediction device further includes a third determination module, wherein the third determination module is used to obtain a plurality of normalized characteristic parameters according to the plurality of the characteristic parameters and the first formula, wherein the first formula is: Wherein: X is the original value of the above characteristic parameter, x is the value after the original value of the above characteristic parameter is normalized, Max is the maximum value of the original value of the above characteristic parameter, and Min is the minimum value of the original value of the above characteristic parameter.

[0072] In some optional embodiments, the second determination module includes a first sub-construction module, which is used for constructing the deep learning model, including: obtaining multiple learning models and optimizing the parameters of the multiple learning models according to a grid search algorithm; and constructing the deep learning model according to the optimized multiple learning models.

[0073] In some optional embodiments, the prediction device also includes a first sub-acquisition module and a first sub-judgment module, wherein the first sub-acquisition module is used to obtain the glass transition temperatures of the above-mentioned polymers and construct a temperature database of the glass transition of the above-mentioned polymers; the first sub-judgment module is used to judge whether the above-mentioned deep learning model meets the preset accuracy rate based on the above-mentioned deep learning model, the data of the above-mentioned temperature database and the data of the above-mentioned optimization database; if the judgment result indicates that it is not satisfied, the parameters of the above-mentioned learning model are re-optimized, and if the above-mentioned judgment result indicates that it is satisfied, the above-mentioned deep learning model is saved.

[0074] In some optional embodiments, the first sub-judgment module includes a first sub-determination unit, a second sub-determination unit and a first sub-judgment unit, wherein the first sub-determination unit is used to obtain the prediction accuracy of the deep learning model according to the prediction accuracy calculation formula, the predicted polymer glass transition temperature output by the deep learning model according to the characteristic parameters and the data of the temperature database, the data of the temperature database includes the glass transition temperature data of the data point calculated by the digital simulation software at the target time, and the accuracy calculation formula is: Where Acc is the prediction accuracy of the deep learning model at the target time, S opi is the glass transition temperature of data point i predicted by the deep learning model at the target time, S ori is the glass transition temperature of data point i calculated by the digital simulation software at the target time, and n is the number of data points; the second sub-determination unit is used to obtain the classification accuracy of the deep learning model according to the classification accuracy calculation formula, the number of unit cells of the polymer correctly classified by the deep learning model and the total number of unit cells of the polymer, and the classification accuracy calculation formula is: Among them, Acc1 is the classification accuracy of the above deep learning model, n trueis the number of units of the polymer correctly classified by the deep learning model, and n is the total number of units of the polymer; the first sub-judgment unit is used to judge whether the prediction accuracy meets the preset prediction accuracy, and whether the classification accuracy meets the preset classification accuracy. When the prediction accuracy meets the preset prediction accuracy and the classification accuracy meets the preset classification accuracy, it is determined that the deep learning model meets the preset accuracy.

[0075] In some optional embodiments, the first sub-judgment unit is also used to judge whether the above-mentioned deep learning model meets the preset accuracy, and also includes: judging whether the above-mentioned deep learning model meets the preset accuracy according to multiple evaluation criteria, and the above-mentioned evaluation criteria include: determination coefficient, mean absolute error, mean square error, root mean square error and mean absolute relative deviation.

[0076] In some optional implementations, the first sub-building module includes multiple learning models including: multiple ones of a deep neural network model, a random forest model, and a support vector machine model.

[0077] The above-mentioned polymer glass transition temperature prediction device includes a processor and a memory, and the above-mentioned acquisition unit and the like are stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize the corresponding functions. The above-mentioned modules are all located in the same processor; or, the above-mentioned modules are located in different processors in any combination.

[0078] The processor includes a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set, and the kernel parameters are adjusted to solve the problem of security risks in multi-source power data caused by the influence of malicious nodes on the security of power Internet of Things data transmission in the prior art.

[0079] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0080] An embodiment of the present invention provides a computer program product, including computer instructions, which implement the above-mentioned method for predicting the glass transition temperature of a polymer when executed by a processor.

[0081] Specifically, the prediction method of polymer glass transition temperature includes:

[0082] Step S201, obtaining a plurality of molecular descriptors of the polymer, and obtaining an original database of characteristic parameters of the polymer based on the plurality of molecular descriptors, wherein the molecular descriptors are used to characterize the glass transition temperature distribution of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer;

[0083] Step S202, according to feature engineering, multiple feature parameters in the original database are removed to obtain an optimized database, wherein the multiple feature parameters have at least one of the following characteristics: a single value feature and a highly linearly correlated feature;

[0084] Step S203, constructing a deep learning model at least based on the characteristic parameters in the above-mentioned optimization database, and determining the mapping relationship between the characteristic parameters of the above-mentioned polymer and the above-mentioned glass transition temperature based on the above-mentioned deep learning model, the above-mentioned characteristic parameters and the SHAP analysis method, and predicting the glass transition temperature of the above-mentioned polymer based on the above-mentioned mapping relationship.

[0085] pass.

[0086] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for predicting the glass transition temperature of a polymer.

[0087] Specifically, the prediction method of polymer glass transition temperature includes:

[0088] Step S201, obtaining a plurality of molecular descriptors of the polymer, and obtaining an original database of characteristic parameters of the polymer based on the plurality of molecular descriptors, wherein the molecular descriptors are used to characterize the glass transition temperature distribution of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer;

[0089] Step S202, according to feature engineering, multiple feature parameters in the original database are removed to obtain an optimized database, wherein the multiple feature parameters have at least one of the following characteristics: a single value feature and a highly linearly correlated feature;

[0090] Step S203, constructing a deep learning model at least based on the characteristic parameters in the above-mentioned optimization database, and determining the mapping relationship between the characteristic parameters of the above-mentioned polymer and the above-mentioned glass transition temperature based on the above-mentioned deep learning model, the above-mentioned characteristic parameters and the SHAP analysis method, and predicting the glass transition temperature of the above-mentioned polymer based on the above-mentioned mapping relationship.

[0091] An embodiment of the present invention provides a processor, and the processor is used to run a program, wherein the program executes the method for predicting the glass transition temperature of a polymer when it is run.

[0092] Specifically, the prediction method of polymer glass transition temperature includes:

[0093] Step S201, obtaining a plurality of molecular descriptors of the polymer, and obtaining an original database of characteristic parameters of the polymer based on the plurality of molecular descriptors, wherein the molecular descriptors are used to characterize the glass transition temperature distribution of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer;

[0094] Step S202, according to feature engineering, multiple feature parameters in the original database are removed to obtain an optimized database, wherein the multiple feature parameters have at least one of the following characteristics: a single value feature and a highly linearly correlated feature;

[0095] Step S203, constructing a deep learning model at least based on the characteristic parameters in the above-mentioned optimization database, and determining the mapping relationship between the characteristic parameters of the above-mentioned polymer and the above-mentioned glass transition temperature based on the above-mentioned deep learning model, the above-mentioned characteristic parameters and the SHAP analysis method, and predicting the glass transition temperature of the above-mentioned polymer based on the above-mentioned mapping relationship.

[0096] An embodiment of the present invention provides an electronic device, the device comprising a processor, a memory, and a program stored in the memory and executable on the processor, and when the processor executes the program, at least the following steps are implemented:

[0097] Step S201, obtaining a plurality of molecular descriptors of the polymer, and obtaining an original database of characteristic parameters of the polymer based on the plurality of molecular descriptors, wherein the molecular descriptors are used to characterize the glass transition temperature distribution of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer;

[0098] Step S202, according to feature engineering, multiple feature parameters in the original database are removed to obtain an optimized database, wherein the multiple feature parameters have at least one of the following characteristics: a single value feature and a highly linearly correlated feature;

[0099] Step S203, constructing a deep learning model at least based on the characteristic parameters in the above-mentioned optimization database, and determining the mapping relationship between the characteristic parameters of the above-mentioned polymer and the above-mentioned glass transition temperature based on the above-mentioned deep learning model, the above-mentioned characteristic parameters and the SHAP analysis method, and predicting the glass transition temperature of the above-mentioned polymer based on the above-mentioned mapping relationship.

[0100] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0101] Obviously, it should be understood by those skilled in the art that the above-mentioned units and platforms or steps of the present invention can be implemented by a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0102] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0103] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0104] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0106] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0107] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0108] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0109] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0110] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0111] This application uses molecular descriptors as a representation method for polymers to compare the prediction results of the deep learning model with existing data. The deep learning model shows superior performance based on molecular descriptors as input of polymer features, highlighting its potential for handling the complexity of polymer features. The key features associated with high glass transition temperature and low glass transition temperature values ​​are determined through SHAP analysis, which improves the reliability of polymer prediction results and solves the problem of inaccurate polymer prediction results in the prior art.

[0112] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting the glass transition temperature of a polymer, characterized in that: The prediction method comprises: Acquire multiple molecular descriptors of the polymer, and obtain an original database of characteristic parameters of the polymer based on the multiple molecular descriptors, wherein the molecular descriptors are used to characterize the glass transition temperature distribution of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer; According to feature engineering, multiple feature parameters in the original database are removed to obtain an optimized database, wherein the multiple feature parameters have at least one of the following characteristics: a single-valued feature and a highly linearly correlated feature; A deep learning model is constructed based on at least the characteristic parameters in the optimization database, and a mapping relationship between the characteristic parameters of the polymer and the glass transition temperature is determined based on the deep learning model, the characteristic parameters and the SHAP analysis method, and the glass transition temperature of the polymer is predicted based on the mapping relationship.

2. The prediction method according to claim 1, characterized in that: The prediction method further comprises: According to the plurality of characteristic parameters and the first formula, a plurality of normalized characteristic parameters are obtained, wherein the first formula is: Wherein: X is the original value of the characteristic parameter, x is the value after the original value of the characteristic parameter is normalized, Max is the maximum value of the original value of the characteristic parameter, and Min is the minimum value of the original value of the characteristic parameter.

3. The prediction method according to claim 1, characterized in that: The deep learning model is constructed, comprising: Acquire multiple learning models, and optimize parameters of the multiple learning models according to a grid search algorithm; The deep learning model is constructed based on the optimized multiple learning models.

4. The prediction method according to claim 3, characterized in that: The prediction method further comprises: Obtaining the glass transition temperatures of a plurality of the polymers and constructing a temperature database of the glass transition of the polymers; According to the deep learning model, the data of the temperature database and the data of the optimization database, determining whether the deep learning model meets a preset accuracy rate; When the judgment result indicates that the conditions are not met, the parameters of the learning model are optimized again; when the judgment result indicates that the conditions are met, the deep learning model is saved.

5. The prediction method according to claim 4, characterized in that: The determining, based on the deep learning model, the database, and the optimized database, whether the deep learning model meets a preset accuracy rate includes: The prediction accuracy of the deep learning model is obtained according to the prediction accuracy calculation formula, the predicted polymer glass transition temperature output by the deep learning model according to the characteristic parameters, and the data of the temperature database, wherein the data of the temperature database includes the glass transition temperature data of the data point calculated by the digital simulation software at the target time, and the accuracy calculation formula is: Wherein, Acc is the prediction accuracy of the deep learning model at the target time, S opi is the glass transition temperature of data point i predicted by the deep learning model at the target time, S ori is the glass transition temperature of data point i calculated by the numerical simulation software at the target time, and n is the number of data points; According to the classification accuracy calculation formula, the number of unit bodies of the polymer correctly classified by the deep learning model and the total number of unit bodies of the polymer, the classification accuracy of the deep learning model is obtained. The classification accuracy calculation formula is: Wherein, Acc1 is the classification accuracy of the deep learning model, n true is the number of unit cells of the polymer correctly classified by the deep learning model, and n is the total number of unit cells of the polymer; Determine whether the prediction accuracy meets the preset prediction accuracy and whether the classification accuracy meets the preset classification accuracy. If the prediction accuracy meets the preset prediction accuracy and the classification accuracy meets the preset classification accuracy, determine that the deep learning model meets the preset accuracy.

6. The prediction method according to claim 4, characterized in that: The determining whether the deep learning model meets the preset accuracy rate also includes: Whether the deep learning model meets the preset accuracy is determined according to multiple evaluation criteria, wherein the evaluation criteria include multiple ones of determination coefficient, mean absolute error, mean square error, root mean square error and mean absolute relative deviation.

7. The prediction method according to claim 2, characterized in that: The multiple learning models include: multiple deep neural network models, random forest models and support vector machine models.

8. A device for predicting the glass transition temperature of a polymer, characterized in that: include: An acquisition module, used for acquiring a plurality of molecular descriptors of the polymer, and obtaining an original database of characteristic parameters of the polymer according to the plurality of molecular descriptors, wherein the molecular descriptors are used for characterizing the glass transition temperature distribution of the polymer, and the characteristic parameters have a preset relationship with the distribution of the glass transition temperature of the polymer; A first determination module is used to remove multiple feature parameters in the original database according to feature engineering to obtain an optimized database, wherein the multiple feature parameters have at least one of the following characteristics: a feature with only a single value and a feature with high linear correlation; The second determination module is used to construct a deep learning model based on at least the characteristic parameters in the optimization database, and determine the mapping relationship between the characteristic parameters of the polymer and the glass transition temperature based on the deep learning model, the characteristic parameters and the SHAP analysis method, and predict the glass transition temperature of the polymer based on the mapping relationship.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for predicting the glass transition temperature of a polymer according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the prediction method of the polymer glass transition temperature according to any one of claims 1 to 7.

Images