JP7610095B2 - Polymer raw material data analysis device and polymer raw material data analysis method - Google Patents

Description

The present invention relates to a polymer raw material data analysis device and a polymer raw material data analysis method used to identify the physical properties of polymer raw materials, and in particular to a polymer raw material data analysis device and a polymer raw material data analysis method that identify the physical property information of a polymer raw material from the structural information of the polymer raw material.

Currently, attempts are being made to identify the composition or structure of a particular substance from the data obtained by various analytical instruments, such as mass spectrometers, gas chromatographs (GC), liquid chromatographs (LC), spectrometers, X-ray fluorescence analyzers, and nuclear magnetic resonance (NMR) instruments, by performing multivariate analysis on the data obtained from the analytical instruments.

On the other hand, there are also attempts to predict the microstructure, such as the composition or structure, of a specific substance from its mechanical properties and chemical properties.
For example, Patent Document 1 describes a method for searching for a material structure. Patent Document 1 describes that, for a candidate material to be searched for, (i) data on at least one of reflection, transmission, and scattering of the material obtained by actual measurement or simulation using an image is databased, (ii) data on the structure of the material determined or estimated from actual measurement or simulation using an analytical method is databased, (iii) data on the characteristics obtained by actual measurement or simulation using physical properties is databased by actual measurement or computer simulation, and (iv) data on the correlated interrelated information accumulated including the mutual linking of the above data is databased, and using the databases from (i) to (iv), one or a combination of the image (i), structure (ii), and physical properties (iii) that are close to the material to be searched for is determined or extracted based on the mutual information distance information (iv). Patent Document 1 further describes that the method for searching for a material structure includes calculating the most effective set of material variables that will lead to the material to be searched for from a group of material variables including atoms, molecular composition, crystallinity, orientation, texture, and mixture ratio that define the material.

International Publication No. 2018/025618

As in the material structure search method described in Patent Document 1 above, attempts have been made to calculate the most effective set of material variables leading to the material to be searched for, for a group of material variables including atoms, molecular composition, crystallinity, orientation, texture, and mixture ratio that define the material. However, the accuracy of prediction of the material to be searched for from structural information such as the group of material variables that define the material is insufficient, and further improvement of prediction accuracy is desired.

The object of the present invention is to provide a polymer raw material data analysis device and a polymer raw material data analysis method that predicts the physical property information of a polymer raw material with high prediction accuracy from the structural information of the polymer raw material.

In order to achieve the above-mentioned object, the first aspect of the present invention provides a polymer raw material data analysis device that uses a data set having structural information and characteristic information for each polymer raw material of multiple types of polymer raw materials, and creates a learning model with the structural information of the polymer raw material as an explanatory variable and the physical property information of the polymer raw material as a target variable, a calculation unit that executes the learning model, and a judgment unit that judges the result of the learning model by the calculation unit, and the learning model creation unit repeatedly creates a learning model until the judgment unit judges that a predetermined judgment condition is satisfied, and creates a prediction model that predicts the physical property information of the polymer raw material from the structural information of the polymer raw material from the learning model that satisfies the judgment condition, and the calculation unit uses the prediction model to predict the physical property information of the polymer raw material from the structural information of the polymer raw material.

The second aspect of the present invention provides a polymer raw material data analysis device that uses a data set having structural information and characteristic information for each of a plurality of types of polymer raw materials, and performs clustering processing on the structural information of the plurality of types of polymer raw materials; a learning model creation unit that creates a learning model for each cluster obtained by the clustering processing, with the structural information of the polymer raw materials as explanatory variables and the physical property information of the polymer raw materials as objective variables; a calculation unit that executes the learning model created for each cluster; and a judgment unit that judges the results of the learning model created for each cluster by the calculation unit, where the learning model creation unit repeatedly creates learning models until the judgment unit judges that a predetermined judgment condition is satisfied, and creates a prediction model that predicts the physical property information of the polymer raw materials from the structural information of the polymer raw materials from the learning models that satisfy the judgment conditions, and the calculation unit uses the prediction model to predict the physical property information of the polymer raw materials from the structural information of the polymer raw materials.

It is preferable that the learning model creation unit changes the learning model based on structural information of the polymer raw material.
The system has a data setting unit that creates a data set for polymer raw materials, acquires structural information for multiple types of polymer raw materials, and compiles the structural information for each polymer raw material into a database, and acquires property information for multiple types of polymer raw materials, and compiles the property information for each polymer raw material into a database.Furthermore, in the database of structural information, it is preferable that structural information for the same polymer raw material among the multiple types of polymer raw materials is consolidated into one.
It is preferable that the system has a calculation unit that calculates a correlation coefficient by correlation analysis using structural information of the polymer raw material and characteristic information of the polymer raw material, and that calculates the correlation coefficient by correlation analysis using the structural information and physical property information of the polymer raw material, and that the learning model creation unit creates a learning model in which the structural information used in the calculation of the correlation coefficient is an explanatory variable and the physical property information of the polymer raw material is an objective variable.

It is preferable that the calculation unit determines multiple correlation coefficients by correlation analysis and validates structural information among the multiple correlation coefficients that is equal to or greater than a predetermined value, and the learning model creation unit creates a learning model in which the structural information used in calculating the valid correlation coefficients is used as an explanatory variable and the physical property information of the polymer raw material is used as a target variable.
The polymeric raw material is preferably a diene-based rubber.
The diene rubber is a styrene butadiene rubber or a butadiene rubber, or a diene rubber blend, preferably a styrene butadiene rubber blend or a styrene butadiene rubber/butadiene rubber blend.
It is preferable that the structural information of the polymer raw material is an actual measurement value obtained using an analytical device or a value obtained by simulation or machine learning, and the physical property information of the polymer raw material is an actual measurement value obtained by material testing or a value obtained by simulation.

The third aspect of the present invention provides a method for analyzing polymer raw material data, comprising the steps of: using a data set having structural information and characteristic information for each of a plurality of types of polymer raw materials, creating a learning model in which the structural information of the polymer raw materials is used as an explanatory variable and the physical property information of the polymer raw materials is used as a target variable; executing the learning model and repeatedly creating the learning model until it is determined that the result of the learning model satisfies a predetermined judgment condition; creating a prediction model that predicts the physical property information of the polymer raw materials from the structural information of the polymer raw materials from the learning model that satisfies the judgment condition; and predicting the physical property information of the polymer raw materials from the structural information of the polymer raw materials using the prediction model.

The fourth aspect of the present invention provides a method for analyzing polymer raw material data, comprising the steps of: using a data set having structural information and characteristic information for each of a plurality of types of polymer raw materials, performing a clustering process on the structural information of the plurality of types of polymer raw materials; creating a learning model for each cluster obtained by the clustering process, in which the structural information of the polymer raw materials is used as an explanatory variable and the physical property information of the polymer raw materials is used as a target variable; executing the learning model and repeatedly creating the learning model until it is determined that the result of the learning model satisfies a predetermined judgment condition; creating a prediction model that predicts the physical property information of the polymer raw materials from the structural information of the polymer raw materials from the learning model that satisfies the judgment condition; and predicting the physical property information of the polymer raw materials from the structural information of the polymer raw materials using the prediction model.

The step of creating a learning model preferably changes the learning model based on structural information of the polymer raw material.
It is preferable that the method for producing a data set of polymer raw materials includes a step of acquiring structural information for a plurality of types of polymer raw materials and compiling the structural information for each polymer raw material into a database, a step of acquiring property information for the plurality of types of polymer raw materials and compiling the property information for each polymer raw material into a database, and a step of aggregating structural information for the same polymer raw material among the plurality of types of polymer raw materials into one in the database of structural information.
The method preferably includes a process of determining a correlation coefficient by correlation analysis using structural information of the polymer raw material and characteristic information of the polymer raw material, and a process of determining a correlation coefficient by correlation analysis using the structural information and physical property information of the polymer raw material, and the process of creating a learning model is preferably a process of creating a learning model in which the structural information used in calculating the correlation coefficient is an explanatory variable and the physical property information of the polymer raw material is an objective variable.

It is preferable that the step of calculating the correlation coefficient involves calculating multiple correlation coefficients by correlation analysis, and validating structural information among the multiple correlation coefficients that is equal to or greater than a predetermined value, and that the step of creating a learning model involves creating a learning model in which the structural information used in calculating the valid correlation coefficients is used as an explanatory variable, and the physical property information of the polymer raw material is used as a target variable.
The polymeric raw material is preferably a diene-based rubber.
The diene rubber is a styrene butadiene rubber or a butadiene rubber, or a diene rubber blend, preferably a styrene butadiene rubber blend or a styrene butadiene rubber/butadiene rubber blend.
It is preferable that the structural information of the polymer raw material is an actual measurement value obtained using an analytical device or a value obtained by simulation or machine learning, and the physical property information of the polymer raw material is an actual measurement value obtained by material testing or a value obtained by simulation.

According to the present invention, it is possible to predict the physical property information of a polymer raw material with high prediction accuracy from the structural information of the polymer raw material.

FIG. 1 is a schematic diagram showing an example of a polymer raw material data analysis device according to an embodiment of the present invention. 1 is a graph showing an example of the relationship between structural information and physical property information of a polymer raw material. 1 is a flowchart showing a first example of a polymer raw material data analysis method according to an embodiment of the present invention in the order of steps. 1 is a flowchart showing a modified example of the first example of the polymer raw material data analysis method according to the embodiment of the present invention in the order of steps. 1 is a flowchart showing a second example of a polymer raw material data analysis method according to an embodiment of the present invention in the order of steps. 10 is a flowchart showing a modified example of the second example of the polymer raw material data analysis method according to the embodiment of the present invention in the order of steps.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the polymer raw material data analysis method and tire model creation device of the present invention will be described in detail based on the preferred embodiments shown in the accompanying drawings.
It should be noted that the drawings described below are illustrative for explaining the present invention, and the present invention is not limited to the drawings shown below.

[Polymer raw material data analysis device]
FIG. 1 is a schematic diagram showing an example of a polymer raw material data analysis device according to an embodiment of the present invention.
The polymer raw material data analysis device 10 is configured using hardware such as a computer. For example, the polymer raw material data analysis method uses an analysis device 10 shown in Fig. 1, but as long as the polymer raw material data analysis method can be executed using hardware and software such as a computer, the device is not limited to the polymer raw material data analysis device 10, and may be a program for causing a computer or the like to execute each step of the polymer raw material data analysis method as a procedure. Furthermore, the polymer raw material data analysis device 10 is not limited to the configuration shown in Fig. 1. Hereinafter, the polymer raw material data analysis device 10 is also simply referred to as the analysis device 10.

The analysis device 10 has a processing unit 12, an input unit 14, and a display unit 16. The processing unit 12 has an acquisition unit 20, a data setting unit 21, a processing unit 22, a calculation unit 23, a learning model creation unit 24, an operation unit 25, a determination unit 26, a display control unit 27, a memory 28, and a control unit 29. The analysis device 10 also has a ROM and the like, which are not shown in the figure.
The processing unit 12 is controlled by a control unit 29. In the processing unit 12, the acquisition unit 20, the data setting unit 21, the processing unit 22, the calculation unit 23, the learning model creation unit 24, the operation unit 25, the judgment unit 26, and the display control unit 27 are connected to a memory 28, and data of the acquisition unit 20, the data setting unit 21, the processing unit 22, the calculation unit 23, the learning model creation unit 24, the operation unit 25, the judgment unit 26, and the display control unit 27 can be stored in the memory 28.

The analysis device 10 functionally forms each of the parts, namely, the acquisition unit 20, the data setting unit 21, the processing unit 22, the calculation unit 23, the learning model creation unit 24, the calculation unit 25, and the judgment unit 26, by executing a program (computer software) stored in a ROM or the like in the control unit 29. As described above, the analysis device 10 may be configured by a computer in which each part functions by executing a program, or may be a dedicated device in which each part is configured with a dedicated circuit.

The analysis device 10 aims to predict the physical property information of a polymer raw material with high prediction accuracy from the structural information of the polymer raw material. The analysis device 10 measures and obtains physical property information such as the ratio of cis, trans, vinyl, and styrene for, for example, newly synthesized styrene-butadiene rubber. Using the measured ratios of cis, trans, vinyl, and styrene, the analysis device predicts the physical property information of the newly synthesized styrene-butadiene rubber, such as the elastic modulus and glass transition temperature, with high prediction accuracy.
Each part of the analysis device 10 will be described below.

The input unit 14 is a variety of input devices such as a mouse and keyboard for inputting various information at the instruction of an operator. The display unit 16 displays, for example, the results obtained by the polymer raw material data analysis method, and various known displays are used. The display unit 16 also includes devices such as a printer for displaying various information on an output medium.

In the acquisition unit 20, structural information and physical property information of the polymer raw materials for multiple types of polymer raw materials are input from outside the analysis device 10, for example, by the input unit 14, and are stored. The structural information and physical property information of the multiple types of polymer raw materials stored in the acquisition unit 20 are read out to the data setting unit 21.

The data setting unit 21 creates a data set of polymer raw materials. The data set of polymer raw materials is stored in the memory 28.
The data set of polymer raw materials includes structure information and characteristic information for each of a plurality of types of polymer raw materials. The data set of polymer raw materials is input to a learning model to learn the relationship between the physical property information of the polymer raw materials and the structure information of the polymer raw materials.
The data setting unit 21 acquires structural information for multiple types of polymer raw materials, compiles the structural information for each polymer raw material into a database, and acquires property information for multiple types of polymer raw materials, compiles the property information for each polymer raw material into a database.
The structural information of the polymer raw material is an actual measurement value obtained by an analysis device, or a value obtained by simulation or machine learning. The analysis device is, for example, a nuclear magnetic resonance (NMR) device.
The physical property information of the above-mentioned polymer raw material is actual measured value obtained by material testing or value obtained by simulation.

The structural information can be predicted from spectral information, for example, by measuring one-dimensional spectral information such as an NMR spectrum or an NMR chemical shift for a newly synthesized styrene-butadiene rubber.
Here, the spectral information and the structural information include, for example, a step of acquiring one-dimensional spectral information measured for each of a plurality of types of polymer raw materials, a step of normalizing the one-dimensional spectral information for each of the polymer raw materials, a step of acquiring composition information and at least one structural information for each of the plurality of types of polymer raw materials, and a step of calculating a correlation coefficient by correlation analysis using the normalized one-dimensional spectral information of the plurality of polymer raw materials and the normalized structural information of the plurality of polymer raw materials, and a step of creating a learning model in which the one-dimensional spectral information used in the calculation of the correlation coefficient is used as an explanatory variable and at least one of the composition information and the structural information of the polymer raw materials is used as an objective variable.Furthermore, the learning model is executed, and the learning model is repeatedly created until it is determined that the result of the learning model satisfies a predetermined judgment condition, and at least one of the structure and composition of the polymer raw material is predicted using the one-dimensional spectral information based on the learning model that satisfies the judgment condition.In this way, the structural information can be predicted from the spectral information.

When creating a database as described above, there may be multiple pieces of structural information for the same polymer raw material. In this case, due to differences in analytical equipment, etc., the structural information may not be the same even for the same polymer raw material.
The data setting unit 21 further consolidates the structural information of the same polymer raw material among the multiple types of polymer raw materials in the database of structural information into one. For example, a method of using an average value, a method of determining a value to be adopted depending on the number of data, etc. can be used as a method of consolidating the structural information into one.

The processing unit 22 uses a data set having structure information and property information for each of a plurality of types of polymer raw materials, and performs clustering processing on the structure information of the plurality of types of polymer raw materials. Note that if clustering processing is not performed, the processing unit 22 is not necessary.
An example of the clustering process is unsupervised learning using the above-mentioned data set of polymer raw materials. For example, the k-means method is used for unsupervised learning. The clustering process classifies structural information for each same polymer raw material, i.e., for each cluster. The clustering process makes it possible to extract structural information of the polymer raw materials that is more important for prediction.

The structural information of the polymer raw material, in the case of a diene rubber, includes the ratio of cis, trans, vinyl, and styrene in the microstructure. Other structural information includes dyad and triad chains. In the case of styrene-butadiene rubber, the composition information includes the amount of butadiene and the amount of styrene. The composition information of the polymer raw material also includes the mass average molecular weight, viscosity, degree of branching, and number of branches.
The characteristic information of the polymer raw material includes at least one of the mechanical properties and chemical properties of the polymer raw material. The mechanical properties of the polymer raw material are, for example, the elastic modulus, Poisson's ratio, and viscosity. The chemical properties of the polymer raw material are, for example, the glass transition temperature (Tg), the crystallization temperature, etc. The chemical properties are also called chemical properties. The characteristic information of the polymer raw material, for example, the elastic modulus, Poisson's ratio, viscosity, the glass transition temperature (Tg), the crystallization temperature, etc., are, for example, actual measured values obtained by various material tests, but may also be values obtained by simulation.

The calculation unit 23 calculates a correlation coefficient by correlation analysis using the structural information of the polymer raw material and the property information of the polymer raw material. In this case, it is preferable to calculate a plurality of correlation coefficients. The correlation coefficients calculated by the calculation unit 23 are stored in, for example, the memory 28. Note that the method of correlation analysis and calculation of the correlation coefficient is not particularly limited, and various known methods can be used. For example, the Pearson correlation coefficient is used as the correlation coefficient.
By calculating a plurality of correlation coefficients in the calculation unit 23, it is possible to identify which structural information of the polymer raw material has a high correlation with the characteristic information, i.e., it is possible to identify the degree of influence of the structural information on the characteristic information.

Furthermore, the calculation unit 23 validates structural information having a predetermined value or more among the multiple correlation coefficients. This makes it possible to suppress the calculation of correlation coefficients having a small degree of influence. The correlation coefficients are used in the learning model. By appropriately using the correlation coefficients, it is possible to improve the prediction accuracy of the learning model.
In addition, as described above, by validating structural information having a correlation coefficient equal to or greater than a predetermined value among the multiple correlation coefficients, data having a small correlation coefficient can be eliminated from the data used for learning the learning model, thereby reducing the amount of data used for learning the learning model and shortening the learning time. This makes it possible to predict the physical property information of polymer raw materials with high prediction accuracy in a shorter time.
For example, the calculation unit 23 validates correlation coefficients exceeding 0.2, and invalidates correlation coefficients having a correlation coefficient of 0.2 or less.
In addition, if the correlation coefficient is not obtained by correlation analysis, there is no need to provide the calculation unit 23.

The learning model creation unit 24 uses a data set having structural information and characteristic information for each polymer raw material, and creates a learning model in which the structural information of the polymer raw material is used as an explanatory variable and the physical property information of the polymer raw material is used as a target variable. The number of learning models is not limited to one, and multiple models may be used. For example, a learning model is created for each cluster obtained by the clustering process. In this case, multiple learning models are used for each cluster.
In addition, when the calculation unit 23 calculates the correlation coefficient, a learning model is created in which the structural information used to calculate the correlation coefficient is used as an explanatory variable, and the physical property information of the polymer raw material is used as a target variable. If there are multiple correlation coefficients, all of the correlation coefficients may be used. Also, as described above, among the multiple correlation coefficients, correlation coefficients equal to or greater than a predetermined value may be made valid, and only the valid correlation coefficients may be used as target variables.
In addition, when the calculation unit 23 calculates the correlation coefficient, a learning model is created in which the structural information used to calculate the correlation coefficient is used as an explanatory variable, and the physical property information of the polymer raw material is used as a response variable. If there are multiple correlation coefficients, all of the correlation coefficients may be used. Also, as described above, among the multiple correlation coefficients, correlation coefficients equal to or greater than a predetermined value may be made valid, and only the valid correlation coefficients may be used as response variables.

The learning model creation unit 24 repeatedly creates a learning model based on the judgment result of the judgment unit 26 described below, until the judgment unit 26 judges that a predetermined judgment condition is satisfied. When repeatedly creating a learning model, unnecessary data such as outliers are deleted. In addition, the explanatory variables are changed. When changing the explanatory variables, for example, the explanatory variables themselves may be changed, or when there are multiple explanatory variables, the combination of explanatory variables may be changed, or the correlation coefficient threshold may be changed as an explanatory variable, and a correlation coefficient with a high numerical value may be used.

The learning model is not particularly limited, and for example, a neural network can be used as a machine learning method. In addition, various other methods can be used. For example, a convolutional neural network (CNN), a support vector machine, a random forest, a nonlinear discrimination method, etc. may be used.
Since the appropriate method to be used as a learning model varies depending on the properties of the analytical data of the polymer raw material to be analyzed, multiple machine learning methods may be prepared in advance so that the user can select any one of them. In addition, since it is well known that discriminant analysis or regression analysis by machine learning can combine multiple items of the same type or multiple items of different types, such combinations may be used.

For example, as shown in Fig. 2, some polymer raw materials have non-regular changes in their physical properties depending on the amount of structure A. For the physical properties (structural information) of the polymer raw materials shown in Fig. 2, applying a linear model when structure A is small and applying a non-linear model when structure A is large can improve prediction accuracy, and therefore, it is preferable for the learning model creation unit 24 to change the learning model based on the structural information of the polymer raw material. Changing the learning model based on the structural information of the above-mentioned polymer raw material is, for example, determined by the learning model creation unit 24 based on a known program, and a learning model according to the structural information of the polymer raw material is appropriately selected.
The learning model used by the learning model creation unit 24 may be stored in the memory 28, for example, and the learning model creation unit 24 may read out the learning model from the memory 28.

The calculation unit 25 executes learning of the learning model created by the learning model creation unit 24, and obtains a result based on the learning model. The calculation unit 25 stores the result of the learning model in the memory 28, or outputs it to the determination unit 26.
For example, supervised data consisting of a set of explanatory variables (input data) and objective variables (output data) is used to train the learning model. When using supervised data, available data is not used as all supervised data, but is divided into supervised data and validation data. The learning model is trained using the supervised data. The validation data is used to evaluate the predictive performance of the learning model trained using the supervised data.
Methods for dividing the data into supervised data and validation data include, for example, hold-out validation and k-fold cross validation.
In addition, the learning of the learning model is not limited to using supervised data, but unsupervised data in which explanatory variables (input data) and objective variables (output data) are not set can also be used.

The judgment unit 26 judges the result of the learning model obtained by the calculation unit 25 .
The learning model uses, for example, validation data to input structural information of the polymer raw material used as an explanatory variable, and obtains physical property information of the polymer raw material as a target variable. The results of the learning model are judged based on the obtained physical property information of the polymer raw material.
The method for judging the result of the learning model is not particularly limited, and for example, R ² (coefficient of determination), RMSE (mean square error), MAE (mean absolute error), etc. can be used.

In the judgment unit 26, for example, a threshold value is set in advance as a judgment condition corresponding to each of the judgment methods described above. The judgment condition may be stored in the memory 28. The judgment unit 26 reads out the judgment condition from the memory 28 and evaluates the learning model based on the judgment condition.
If the judgment conditions are satisfied, the model is determined to be the final learning model. The judgment unit 26 creates a prediction model that predicts the physical property information of the polymer raw material from the structural information of the polymer raw material from the learning model, which is the final learning model. The judgment unit 26 stores the prediction model in the calculation unit 25 or memory 28. The calculation unit 25 uses the prediction model to predict the physical property information of the polymer raw material from the structural information of the polymer raw material, and stores the measurement results in, for example, the memory 28 and creates a database.
The prediction model is created from a learning model in which structural information and physical property information of the polymer raw material are associated. As long as the prediction model can predict the physical property information of the polymer raw material from the structural information of the polymer raw material as described above, the creation method of the prediction model is not particularly limited as long as the learning model is used.

For example, when the polymer raw material is styrene-butadiene rubber and the structural information is the ratio of cis, trans, vinyl, and styrene, the calculation unit 25 can predict the glass transition temperature and the like based on the prediction model, for example, using the ratio of cis, trans, vinyl, and styrene of newly synthesized styrene-butadiene rubber. The glass transition temperature predicted using the prediction model is stored in the memory 28.
If the judgment condition is not satisfied, the learning model creation unit 24 creates the learning model again. The recreation of the learning model by the learning model creation unit 32 is the same as the creation of the learning model described above.

The display control unit 27 causes the display unit 16 to display the cis, trans, vinyl, and styrene ratios for each of the multiple types of polymer raw materials acquired by the acquisition unit 20, the correlation coefficients calculated by the calculation unit 23, and information on the explanatory variables and target variables set by the learning model creation unit 24.
Moreover, the physical property information of the polymer raw material predicted by the calculation unit 25 is displayed on the display unit 16 .
When the display control unit 27 causes the display unit 16 to display various information, the display control unit 27 may read out various information from the memory 28 and display the information. The display control unit 27 may also cause the display unit 16 to display various information input via the input unit 14.

The analysis device 10 creates the above-mentioned simulation tire model by executing a program (computer software) stored in a ROM or the like in the control unit 29. The analysis device 10 may be configured as a computer in which each part functions by executing a program as described above, or may be a dedicated device in which each part is configured with a dedicated circuit.

[First Example of Polymer Raw Material Data Analysis Method]
3 is a flow chart showing the steps of a first example of a polymer raw material data analysis method according to an embodiment of the present invention. As described above, the polymer raw material data analysis method uses, for example, the analysis device 10.
In the first example of the polymer raw material data analysis method described below, various processes are performed in each part of the processing section 12 of the analysis device 10, although this is not specifically described. Also, in the following description, the control section 29 controls the series of processes in each part of the processing section 12, although a description is omitted.
In the polymer raw material data analysis method, for example, a plurality of types of styrene-butadiene rubbers, which differ in the amount of butadiene and the amount of styrene, and in the ratios of cis, trans, vinyl, and styrene, are used as polymer raw materials. The structural information of the polymer raw materials is the ratios of cis, trans, vinyl, and styrene. Furthermore, the physical property information of the polymer raw materials will be described using the glass transition temperature as an example. In this case, the data set of the polymer raw materials includes the ratios of cis, trans, vinyl, and styrene as structural information, and the glass transition temperature as physical property information.

First, the above-mentioned data set of polymer raw materials is acquired (step S10).
The method of acquiring the data set of the polymer raw material is not particularly limited, and for example, actually measured structural information is acquired for each of a plurality of types of styrene-butadiene rubber (polymer raw material). For each of the plurality of types of styrene-butadiene rubber, information on the amount of butadiene and the amount of styrene, as well as the ratio of cis, trans, vinyl, and styrene is acquired.
Also, for example, the physical property information of a plurality of types of polymer raw materials is obtained by material testing. Specifically, a material test is performed on styrene-butadiene rubber to obtain the state of the glass transition temperature.

Here, as described above, structural information can be predicted from one-dimensional spectral information. The one-dimensional spectral information of the multiple types of styrene butadiene rubber is, for example, measured using a nuclear magnetic resonance (NMR) apparatus, and is, for example, a ¹³ C-NMR spectrum or a chemical shift of ¹³ C-NMR. However, the one-dimensional spectral information is not limited to this, and the measurement results of a GC (gas chromatograph) and an IR (infrared absorption) spectrum are also included in the actually measured one-dimensional spectral information. The one-dimensional spectral information may be a mixture of information actually measured by different analytical devices. The one-dimensional spectral information may be a value obtained by simulation or machine learning in addition to an actual measurement value using an analytical device.
In this case, the reference may be set according to the analysis conditions at least when the one-dimensional spectrum information is actually measured. When the one-dimensional spectrum information is, for example, an NMR spectrum, if CDCl ₃ solvent is used during measurement, 77.0 ppm is set as the reference. The above-mentioned reference may also be set according to the polymer raw material. Note that the nuclear magnetic resonance apparatus has a function of automatically setting a reference based on the solvent, etc.

The one-dimensional spectral information may be normalized for each of the multiple types of polymer raw materials. For example, in the case of NMR spectra, the spectral intensity is normalized for each polymer raw material. The method for normalizing the one-dimensional spectral information is not particularly limited. For example, in one-dimensional spectral information, one of the multiple spectral values is normalized based on a standard value. In addition to the one-dimensional spectral information, structural information and physical property information may also be normalized.

The maximum and minimum values of the spectra of the multiple pieces of one-dimensional spectral information may be unified. For example, the NMR chemical shifts included in each piece of one-dimensional spectral information may be unified to a specific range. In this case, the chemical shifts are unified to a range of, for example, -20 to 220 ppm, that is, the maximum value is unified to 220 ppm and the minimum value is unified to -20 ppm.
In each one-dimensional spectrum information, the chemical shift value is adjusted so that the chemical shift is in the range of −20 to 220 ppm. The method of adjusting the value is not particularly limited, and any known method can be used as appropriate.
The timing for unifying the maximum and minimum values of the spectrum is not particularly limited, and may be before or after normalization of the one-dimensional spectral information.

The process of creating a dataset of polymer raw materials includes, for example, a process of acquiring structural information for multiple types of polymer raw materials and compiling the structural information for each polymer raw material into a database, a process of acquiring characteristic information for multiple types of polymer raw materials and compiling the characteristic information for each polymer raw material into a database, and a process of consolidating structural information for the same polymer raw material among the multiple types of polymer raw materials into one in the structural information database.
The polymer raw material data set is created by creating a database of structural information and a database of physical property information, and aggregating structural information for the same polymer raw material among multiple types of polymer raw materials into one. As a result, for example, structural information may have different values depending on the analysis device, but this difference between the analysis devices is corrected and one piece of structural information is associated with one polymer raw material. This allows accurate learning. The method of aggregating structural information into one is as described above, so a detailed explanation will be omitted.

Next, using a dataset of polymer raw materials, a learning model is created in which the structural information of the polymer raw materials (the proportions of cis, trans, vinyl, and styrene) are used as explanatory variables and the physical property information of the polymer raw materials (glass transition temperature) is used as the objective variable (step S12).
As described above, the learning model is not particularly limited, and a neural network can be used as described above. Learning is performed using the learning model.

For example, supervised data consisting of a set of explanatory variables (input data) and objective variables (output data) is used for learning the learning model. Specifically, the above-mentioned data set, for example, data in which the relationship between the cis, trans, vinyl, and styrene ratios of styrene-butadiene rubber and the glass transition temperature is known, is used. When using supervised data, instead of using all the data in the data set as described above, a portion of the data is used as supervised data and the rest is used as validation data. The validation data is used to determine the predictive performance of the learning model.
As described above, methods for dividing the data into supervised data and validation data include, for example, hold-out validation and k-fold cross validation.

Next, the learning model is executed and the result of the learning model is judged (step S14). In step S14, as described above, the validation data is input to the learning model, and as a result, the learning model obtains, for example, the glass transition temperature.
Based on this result, the result of the learning model is judged in order to evaluate the prediction performance of the learning model, for example, the prediction accuracy. The method of judging the result of the learning model can use the various methods described above.
Regarding the prediction accuracy, a threshold is set in advance as a judgment condition, and if the prediction accuracy exceeds the threshold, the learning model is judged to satisfy the judgment condition (step S14).
In step S14, if the determination condition is satisfied, the learning model created in step S12 is set as the final learning model.
Next, the final learning model is used to create a prediction model that predicts the physical property information of the polymer raw material from the structural information of the polymer raw material (step S16).

Next, the prediction model is used to predict the physical property information of the polymer raw material from the structural information of the polymer raw material (step S18). Specifically, in step S18, for example, the prediction model is used to predict the glass transition temperature from the proportions of cis, trans, vinyl, and styrene. In this case, the glass transition temperature can be predicted with high prediction accuracy. In step S18, the glass transition temperature predicted using the prediction model is stored, for example, in memory 28.

On the other hand, if the judgment condition is not satisfied in step S14, the learning model created in step S12 is recreated, and the learning model is repeatedly created until the judgment condition is satisfied. In this case, the learning model is reset (step S20). In step S20, for example, as described above, unnecessary data such as outliers are deleted. Also, the explanatory variables are changed. When changing the explanatory variables, for example, the explanatory variables themselves may be changed, or if there are multiple explanatory variables, the combination of explanatory variables may be changed.

The learning model is recreated, and if the judgment condition is satisfied in step S14, the recreated learning model is set as the final learning model. As described above, a prediction model for predicting the physical property information of the polymer raw material is created from the final learning model (step S16).
For example, the glass transition temperature is predicted from the ratio of cis, trans, vinyl, and styrene of a newly synthesized styrene-butadiene rubber using a prediction model (step S18). Even in this case, the glass transition temperature can be predicted with high prediction accuracy.

[Modification of the first example of the polymer raw material data analysis method]
4 is a flow chart showing the process sequence of a modified example of the first example of the polymer raw material data analysis method according to the embodiment of the present invention. In FIG. 4, detailed explanations of the same processes as those in the first example of the polymer raw material data analysis method shown in FIG. 3 are omitted.
The modified example of the first example of the polymer raw material data analysis method is different from the first example of the polymer raw material data analysis method in that a correlation coefficient is used, and other steps are the same as those of the first example of the polymer raw material data analysis method. Between the step of acquiring a data set of polymer raw materials (step S10) and the step of creating a learning model (step S12), a step of calculating a correlation coefficient (step S22) and a step of setting explanatory variables (step S24) are provided.
In a modified example of the first example of the polymer raw material data analysis method, after acquiring a data set of the polymer raw material (step S10), a correlation coefficient is determined by correlation analysis using structural information of the polymer raw material and characteristic information of the polymer raw material, and a correlation coefficient is determined by correlation analysis using structural information and physical property information of the polymer raw material (step S22).

In step S22, specifically, a correlation coefficient is calculated by correlation analysis using the cis, trans, vinyl, and styrene ratios and the glass transition temperature. The calculated correlation coefficient may be one or more, but it is preferable to calculate more than one. As described above, by calculating multiple correlation coefficients, the degree of influence on the structural information of styrene-butadiene rubber (polymer raw material) can be identified. The degree of influence of the cis, trans, vinyl, and styrene ratios on the glass transition temperature of styrene-butadiene rubber (polymer raw material) can be identified.

Next, the setting of explanatory variables (step S24) includes a process of validating structural information (proportions of cis, trans, vinyl, and styrene) having a predetermined value or more among the multiple correlation coefficients obtained in step S22.
Furthermore, in the learning model creation step (step S12), which is the next step after step S24, the structural information (the ratio of cis, trans, vinyl, and styrene) used in calculating the effective correlation coefficient is used as an explanatory variable, and a learning model is created as described above. After step S12, the learning model is judged (step S14) in the same manner as in the first example of the polymer raw material data analysis method described above, and if the judgment conditions are met, a prediction model is created as the final learning model (step S16), and the glass transition temperature is predicted from the ratio of cis, trans, vinyl, and styrene of a newly synthesized styrene-butadiene rubber using the prediction model (step S18).
By setting explanatory variables (step S24), data with small correlation coefficients are excluded from the data used to train the learning model. Therefore, compared to when explanatory variables are not set, the number of data used to train the learning model can be reduced, and the learning time can be shortened. As a result, the physical property information of polymer raw materials can be predicted with high prediction accuracy in a shorter time.

In a modified example of the first example of the polymer raw material data analysis method, explanatory variables are set (step S24) to eliminate data with small correlation coefficients from the data used to train the learning model. Therefore, compared to the first example of the polymer raw material data analysis method, the amount of data used to train the learning model can be reduced, and the learning time can be shortened. This makes it possible to predict the physical property information of polymer raw materials with high prediction accuracy in a shorter time.

[Second Example of Polymer Raw Material Data Analysis Method]
5 is a flow chart showing the order of steps of a second example of the polymer raw material data analysis method according to the embodiment of the present invention. In the second example of the polymer raw material data analysis method, for example, the analysis device 10 is used.
The second example of the polymer raw material data analysis method is similar to the first example of the polymer raw material data analysis method except that, compared to the first example of the polymer raw material data analysis method, the second example of the polymer raw material data analysis method has a step of performing a clustering process (step S30) between the step of acquiring a data set of polymer raw materials (step S10) and the step of creating a learning model (step S12). Therefore, detailed explanation is omitted.

In the above-mentioned clustering process (step S30), a clustering process is performed on the structural information of multiple types of polymer raw materials using the above-mentioned polymer raw material dataset. Unsupervised learning is used for the clustering process, for example, the k-means method is used. The structural information is classified for each identical polymer raw material, that is, for each cluster, by the clustering process. Next, a learning model is created for each cluster obtained by the clustering process, with the structural information of the polymer raw material as the explanatory variable and the physical property information of the polymer raw material as the objective variable (step S12). Therefore, in step S12, if there are multiple clusters, multiple learning models are created.
After step S12, the learning model is judged (step S14) in the same manner as in the first example of the polymer raw material data analysis method described above. If the learning model satisfies the judgment conditions, a prediction model is created as the final learning model (step S16), and the glass transition temperature is predicted from the cis, trans, vinyl, and styrene ratios of, for example, a newly synthesized styrene-butadiene rubber using the prediction model (step S18).
By performing clustering processing, it is possible to extract information on the structure of polymer raw materials that is more important for prediction.

If the learning model does not satisfy the judgment condition, the learning model is reset (step S20) as described above, and the learning model is repeatedly created until the judgment condition is satisfied.
The learning model is recreated, and if the judgment condition is satisfied in step S14, the recreated learning model is set as the final learning model. As described above, a prediction model for predicting the physical property information of the polymer raw material is created from the final learning model (step S16).
A learning model is created for each cluster, and if there are multiple learning models, a learning model judgment (step S14) is performed for each learning model, and the learning model that satisfies the judgment is determined as the final learning model. On the other hand, for a learning model that does not satisfy the judgment condition among the multiple learning models, the learning model is recreated until the judgment condition is satisfied, and if the judgment condition is satisfied, the recreated learning model is determined as the final learning model. The learning models are recreated until all of the multiple learning models satisfy the judgment condition, and the above-mentioned process is performed until the final learning model is obtained.
In the second example of the polymer raw material data analysis method, for example, a prediction model is used to predict the glass transition temperature from the proportions of cis, trans, vinyl, and styrene of a newly synthesized styrene-butadiene rubber (step S18). In this case, too, the glass transition temperature can be predicted with high prediction accuracy.

In step S12, a learning model was created in which the structural information of the polymer raw material for each cluster is used as the explanatory variable and the physical property information of the polymer raw material is used as the objective variable. However, this is not limited to this, and for example, another learning model may be created in which one-dimensional spectral information and structural information are used as explanatory variables and the physical property information of the polymer raw material is used as the objective variable.
For example, the other learning model uses the validation data to input one-dimensional spectral information (e.g., NMR chemical shifts) and structural information of the polymer raw material used as the explanatory variables to obtain the physical property information of the polymer raw material as the objective variable. Based on the obtained physical property information of the polymer raw material, the results of the other learning model can be judged.

[Modification of the second example of the polymer raw material data analysis method]
6 is a flow chart showing the process sequence of a modified example of the second example of the polymer raw material data analysis method according to the embodiment of the present invention. In FIG. 6, detailed explanations of the same steps as those in the second example of the polymer raw material data analysis method shown in FIG. 5 are omitted.
The modified example of the second example of the polymer raw material data analysis method is different from the second example of the polymer raw material data analysis method in that a correlation coefficient is used, and other steps are the same as those of the second example of the polymer raw material data analysis method. Between the clustering process step (step S30) and the learning model creation step (step S12), a correlation coefficient calculation step (step S22) and an explanatory variable setting step (step S24) are provided.

In the modified example of the second example of the polymer raw material data analysis method, after clustering processing using a data set of polymer raw materials (step S30), correlation coefficients are calculated by correlation analysis using structural information of the polymer raw materials and characteristic information of the polymer raw materials, and correlation coefficients are calculated by correlation analysis using structural information and physical property information of the polymer raw materials (step S22). After that, explanatory variables are set (step S24). The calculation of correlation coefficients (step S22) and the setting of explanatory variables (step S24) are the same as those in the modified example of the polymer raw material data analysis method described above, and therefore detailed description thereof will be omitted.
When all correlation coefficients calculated in step S22 are used, it is not necessary to set explanatory variables (step S24) as described above.

After the explanatory variables are set, as in the second example of the polymer raw material data analysis method, from step S12 onwards, a learning model is judged (step S14), and if the judgment conditions are met, a prediction model is created as the final learning model (step S16), and the prediction model is used to predict the glass transition temperature, for example, from the cis, trans, vinyl and styrene ratios of newly synthesized styrene-butadiene rubber (step S18).

If the learning model does not satisfy the judgment condition, the learning model is reset (step S20) as described above, and the learning model is repeatedly created until the judgment condition is satisfied.
The learning model is recreated, and if the judgment condition is satisfied in step S14, the recreated learning model is made the final learning model. As described above, a prediction model for predicting the physical property information of the polymer raw material is created from the final learning model (step S16). For example, the prediction model is used to predict the glass transition temperature from the ratio of cis, trans, vinyl, and styrene of a newly synthesized styrene-butadiene rubber (step S18). Even in this case, the glass transition temperature can be predicted with high prediction accuracy.
In the modified example of the second example, a learning model is also created for each cluster, and if there are multiple learning models, a learning model judgment (step S14) is performed for each learning model, and the learning model that satisfies the judgment is set as the final learning model. On the other hand, for a learning model that does not satisfy the judgment condition among the multiple learning models, the learning model is recreated until the judgment condition is satisfied, and if the judgment condition is satisfied, the recreated learning model is set as the final learning model. The learning models are recreated until all of the multiple learning models satisfy the judgment condition, and the above-mentioned process is performed until the final learning model is obtained.

[Polymer raw materials]
The polymer raw material is, for example, a diene rubber. The diene rubber preferably contains at least one of a styrene-butadiene rubber and a butadiene rubber.
The polymer raw material is not particularly limited, but may have actually measured one-dimensional spectral information, such as an NMR spectrum measured using a nuclear magnetic resonance apparatus, a measurement result of a GC (gas chromatograph), or an IR (infrared absorption) spectrum.

[Diene rubber]
The diene rubber may be a styrene butadiene rubber or a butadiene rubber, or a diene rubber blend, such as a styrene butadiene rubber blend or a styrene butadiene rubber/butadiene rubber blend.
Specific examples of the diene rubber include natural rubber (NR), isoprene rubber (IR), butadiene rubber (BR), styrene butadiene rubber (SBR), acrylonitrile-butadiene copolymer rubber (NBR), butyl rubber (IIR), halogenated butyl rubber (Br-IIR, Cl-IIR), and chloroprene rubber (CR).
The diene rubber preferably contains styrene-butadiene rubber (SBR) or butadiene rubber (BR), and more preferably contains styrene-butadiene rubber (SBR) and butadiene rubber (BR), because the effects of the present invention are more excellent.

The present invention is basically configured as described above. The polymer raw material data analysis device and polymer raw material data analysis method of the present invention have been described in detail above, but the present invention is not limited to the above-mentioned embodiment, and various improvements and modifications may be made without departing from the spirit of the present invention.

Hereinafter, examples of the polymer raw material data analysis method of the present invention will be specifically described.
In this example, learning models of Examples 1 to 3 shown below were created, and the effects of the polymer raw material data analysis method of the present invention were confirmed in terms of prediction accuracy using the created learning models. The results of prediction accuracy for Examples 1 to 3 are shown in Table 1 below.
100 types of styrene-butadiene rubber were used as the polymer raw material. In the data set, the ratio of cis, trans, vinyl, and styrene was used as the structural information. The elastic modulus was used as the physical property information of the polymer raw material.
Of the 100 types of styrene butadiene rubber, 80% was used as supervised data for the learning model, and the remaining 20% was used as validation data. That is, of the 100 types of styrene butadiene rubber, the chemical shift data of 80 types of styrene butadiene rubber was used as supervised data, and the chemical shift data of the remaining 20 types of styrene butadiene rubber was used as validation data.

(Example 1)
In Example 1, a final prediction model was created using a neural network as a learning model based on the first example of the data analysis method described above. In Example 1, the prediction accuracy was calculated using the elastic modulus of the validation data set and the elastic modulus obtained from the validation data.
The prediction accuracy of the prediction model of Example 1 was 82.4. The prediction accuracy was determined using RMS (Root Mean Square). The RMS of Example 1 was taken as 100 for Example 3.
A smaller RMS value means that the error between the observed value and the predicted value is smaller. A smaller RMS value means that the prediction result is better and the learning model is more excellent.

(Example 2)
In Example 2, a final prediction model was created using a neural network as a learning model based on the second example of the data analysis method described above. The prediction accuracy of Example 2 was calculated in the same manner as in Example 1. The prediction accuracy of the prediction model of Example 2 was 80.8. The prediction accuracy was calculated using RMS as in Example 1, and the RMS of Example 2 was calculated by taking the value of Example 3 as 100.
(Example 3)
In Example 3, a learning model was created in which structural information of the polymer raw material was used as an explanatory variable and physical property information of the polymer raw material was used as a target variable. This learning model was trained. A neural network was used for the learning model. The prediction accuracy of the learning model after training, i.e., the learning model of Example 3, was set to 100 as described above. In Example 3, as in Example 1, the prediction accuracy was calculated using the elastic modulus of the data set of the validation data and the elastic modulus obtained from the validation data.

As shown in Table 1, high prediction accuracy was obtained in Examples 1 and 2, which correspond to the working examples. Example 2 was subjected to clustering processing, and the prediction accuracy was higher than that of Example 1.
On the other hand, in Example 3, which corresponds to a comparative example, the prediction accuracy was low.

10 Polymer raw material data analysis device (analysis device)
12 Processing section 14 Input section 16 Display section 20 Acquisition section 21 Data setting section 22 Processing section 23 Calculation section 24 Learning model creation section 25 Calculation section 26 Determination section 27 Display control section 28 Memory 29 Control section 32 Learning model creation section

Claims (14)

a learning model creation unit that uses a data set having structure information and characteristic information for each of a plurality of types of polymer raw materials to create a learning model in which the structure information of the polymer raw materials is used as an explanatory variable and the physical property information of the polymer raw materials is used as a target variable;
A calculation unit that executes the learning model;
A determination unit that determines a result of the learning model by the calculation unit,
The learning model creation unit repeatedly creates the learning model until the determination unit determines that a predetermined determination condition is satisfied,
A prediction model is created from the learning model that satisfies the judgment condition, the prediction model predicting the physical property information of the polymer raw material from the structural information of the polymer raw material,
The calculation unit predicts the physical property information of the polymer raw material from the structural information of the polymer raw material by using the prediction model,
A calculation unit that calculates a correlation coefficient by correlation analysis using the structural information of the polymer raw material and the characteristic information of the polymer raw material, and calculates a correlation coefficient by correlation analysis using the structural information and physical property information of the polymer raw material,
The calculation unit obtains a plurality of correlation coefficients by the correlation analysis, and validates structural information having a correlation coefficient equal to or greater than a predetermined value among the plurality of correlation coefficients;
The learning model creation unit of the polymer raw material data analysis device creates the learning model using the structural information used to calculate the effective correlation coefficient as an explanatory variable and the physical property information of the polymer raw material as an objective variable. a processing unit that uses a data set having structure information and characteristic information for each of a plurality of types of polymer raw materials and performs a clustering process on the structure information of the plurality of types of polymer raw materials;
a learning model creation unit that creates a learning model for each cluster obtained by the clustering process, the structural information of the polymer raw material being an explanatory variable and the physical property information of the polymer raw material being an objective variable;
A calculation unit that executes the learning model created for each cluster;
A determination unit that determines a result of the learning model created for each cluster by the calculation unit,
The learning model creation unit repeatedly creates the learning model until the determination unit determines that a predetermined determination condition is satisfied,
A prediction model is created from the learning model that satisfies the judgment condition, the prediction model predicting the physical property information of the polymer raw material from the structural information of the polymer raw material,
The calculation unit predicts the physical property information of the polymer raw material from the structural information of the polymer raw material using the prediction model,
A calculation unit that calculates a correlation coefficient by correlation analysis using the structural information of the polymer raw material and the characteristic information of the polymer raw material, and calculates a correlation coefficient by correlation analysis using the structural information and physical property information of the polymer raw material,
The calculation unit obtains a plurality of correlation coefficients by the correlation analysis, and validates structural information having a correlation coefficient equal to or greater than a predetermined value among the plurality of correlation coefficients;
The learning model creation unit creates the learning model using the structural information used in the calculation of the effective correlation coefficient as an explanatory variable and the physical property information of the polymer raw material as a target variable.
Polymer raw material data analysis equipment. The polymer raw material data analysis device according to claim 1 or 2, wherein the learning model creation unit changes the learning model based on the structural information of the polymer raw material. A data setting unit that creates the data set of the polymer raw material,
The structural information is acquired for the plurality of types of polymer raw materials, and the structural information is compiled into a database for each of the polymer raw materials; and the property information is acquired for the plurality of types of polymer raw materials, and the property information is compiled into a database for each of the polymer raw materials;
The polymer raw material data analysis device according to any one of claims 1 to 3, further comprising: aggregating the structural information for the same polymer raw material among the plurality of types of polymer raw materials into one in the database of structural information. The polymer raw material data analysis device according to any one of claims 1 to 4, wherein the polymer raw material is a diene rubber. The polymer raw material data analysis device according to claim 5, wherein the diene rubber is a styrene butadiene rubber or a butadiene rubber, or a diene rubber blend, and is a styrene butadiene rubber blend or a styrene butadiene rubber/butadiene rubber blend. The structural information of the polymer raw material is an actual measurement value obtained using an analytical device, or a value obtained by simulation or machine learning,
7. The polymer raw material data analysis device according to claim 1, wherein the physical property information of the polymer raw material is an actual measurement value obtained by a material test or a value obtained by a simulation. A step of creating a learning model for each of a plurality of types of polymer raw materials, using a data set having structure information and characteristic information of the polymer raw materials, with the structure information of the polymer raw materials as explanatory variables and the physical property information of the polymer raw materials as objective variables;
Executing the learning model and repeatedly creating the learning model until it is determined that the result of the learning model satisfies a predetermined determination condition;
A step of creating a prediction model that predicts the physical property information of the polymer raw material from the learning model that satisfies the judgment condition, based on structural information of the polymer raw material;
predicting the physical property information of the polymer raw material from the structural information of the polymer raw material using the prediction model;
The method includes a step of determining a correlation coefficient by correlation analysis using the structural information of the polymer raw material and the characteristic information of the polymer raw material, and a step of determining a correlation coefficient by correlation analysis using the structural information and physical property information of the polymer raw material,
The step of determining the correlation coefficient includes determining a plurality of correlation coefficients by the correlation analysis, and validating structural information having a correlation coefficient equal to or greater than a predetermined value among the plurality of correlation coefficients;
A polymer raw material data analysis method, in which the process of creating the learning model creates the learning model with the structural information used in calculating the effective correlation coefficient as an explanatory variable and the physical property information of the polymer raw material as a target variable. For a plurality of types of polymer raw materials, a data set having structure information and property information for each of the polymer raw materials is used,
A step of performing a clustering process on the structural information of the plurality of types of polymer raw materials;
creating a learning model for each cluster obtained by the clustering process using the structural information of the polymer raw material as an explanatory variable and using the physical property information of the polymer raw material as a target variable;
Executing the learning model and repeatedly creating the learning model until it is determined that the result of the learning model satisfies a predetermined determination condition;
A step of creating a prediction model that predicts the physical property information of the polymer raw material from the structural information of the polymer raw material from the learning model that satisfies the judgment condition;
predicting the physical property information of the polymer raw material from the structural information of the polymer raw material using the prediction model;
The method includes a step of determining a correlation coefficient by correlation analysis using the structural information of the polymer raw material and the characteristic information of the polymer raw material, and a step of determining a correlation coefficient by correlation analysis using the structural information and physical property information of the polymer raw material,
The step of determining the correlation coefficient includes determining a plurality of correlation coefficients by the correlation analysis, and validating structural information having a correlation coefficient equal to or greater than a predetermined value among the plurality of correlation coefficients;
A polymer raw material data analysis method, in which the process of creating the learning model creates the learning model with the structural information used in calculating the effective correlation coefficient as an explanatory variable and the physical property information of the polymer raw material as a target variable. The polymer raw material data analysis method according to claim 8 or 9, wherein the step of creating the learning model changes the learning model based on the structural information of the polymer raw material. creating said data set for said polymeric raw material;
The step of creating the data set of the polymeric raw material comprises:
acquiring the structural information for the plurality of types of polymer raw materials, and compiling the structural information for each polymer raw material into a database;
acquiring the characteristic information for the plurality of types of polymer raw materials, and compiling the characteristic information for each polymer raw material into a database;
The polymer raw material data analysis method according to any one of claims 8 to 10, further comprising a step of aggregating structural information for the same polymer raw material among the plurality of types of polymer raw materials into one in the database of structural information. The polymer raw material data analysis method according to any one of claims 8 to 11, wherein the polymer raw material is a diene rubber. The polymer raw material data analysis method according to claim 12, wherein the diene rubber is a styrene butadiene rubber or a butadiene rubber, or a diene rubber blend, and is a styrene butadiene rubber blend or a styrene butadiene rubber/butadiene rubber blend. The structural information of the polymer raw material is an actual measurement value obtained using an analytical device, or a value obtained by simulation or machine learning,
14. The polymer raw material data analysis method according to claim 8, wherein the physical property information of the polymer raw material is an actual measurement value obtained by a material test or a value obtained by a simulation.