Method, information processing apparatus, and program for predicting resin physical property value

Abstract

To improve a prediction technique of a resin physical property value.SOLUTION: A method of predicting a resin physical property value to be executed by an information processing apparatus includes: a step of generating a prediction model based on training data in which data of a predetermined wavelength region of spectral data in a resin synthesis step is an explanatory variable and the resin physical property value is an objective variable; and a step of predicting the resin physical property value based on the prediction model, where the predetermined wavelength region is determined using complementary data generated based on the spectral data.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to a method for predicting resin physical property values, an information processing device, and a program. [Background technology]

[0002] In the resin polymerization process, the physical properties of a resin composition are generally measured in a laboratory according to a predetermined procedure. Sampling and physical property testing operations take several hours, and measurement accuracy can vary depending on the operator. If the physical properties of the resin composition do not meet the quality threshold, corrective measures such as adjusting the polymerization conditions are taken. These corrective measures are costly and labor-intensive, and can result in reduced productivity. Furthermore, if the physical properties of a resin composition do not meet the quality threshold, the polymerized product may be discarded. Therefore, a technology has been proposed that uses machine learning to generate a quality prediction model for predicting the physical properties of a resin composition (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-167027 Summary of the Invention [Problem to be solved by the invention]

[0004] To build a machine learning model, a certain amount of actual data is required to be used as training data. However, due to the need to balance with regular work, it is sometimes difficult to secure sufficient actual data, which poses a challenge to practical application. As such, there is room for improvement in the technology for predicting resin physical properties.

[0005] The present disclosure has been made in view of the above circumstances, and aims to improve the technology for predicting resin physical properties. [Means for solving the problem]

[0006] (1) A method according to an embodiment of the present disclosure is a method for predicting a resin physical property value executed by an information processing device, comprising: generating a prediction model based on training data in which data in a predetermined wavelength range of the spectral data in the resin synthesis process is used as an explanatory variable and resin physical property values ​​are used as target variables; predicting resin physical property values ​​based on the prediction model; Including, The predetermined wavelength range is determined using complementary data generated based on the spectral data.

[0007] (2) A method according to one embodiment of the present disclosure is the method described in (1), The spectral data includes data relating to a plurality of resin compositions having different composition ratios of raw materials.

[0008] (3) A method according to one embodiment of the present disclosure is the method described in (2), The spectral data is near-infrared spectral data.

[0009] (4) A method according to one embodiment of the present disclosure is the method according to (3), The spectroscopic sensor for measuring the near-infrared light spectrum data is at least one of a near-infrared spectroscopic sensor and a Raman spectroscopic sensor.

[0010] (5) A method according to one embodiment of the present disclosure is the method according to (3), The method, wherein the explanatory variables further include the mass ratio of raw material to catalyst and temperature.

[0011] (6) A method according to an embodiment of the present disclosure is a method according to any one of (1) to (5), The resin physical property value is the epoxy monomer equivalent.

[0012] (7) A method according to an embodiment of the present disclosure is a method according to any one of (1) to (6), In the step of generating the prediction model, a plurality of prediction models are generated based on the training data using a plurality of machine learning algorithms; In the predicting step, the resin physical property value is predicted based on one prediction model selected from the plurality of prediction models based on a predetermined index.

[0013] (8) An information processing device according to an embodiment of the present disclosure includes: An information processing device including a control unit, The control unit A prediction model is generated based on training data in which data in a predetermined wavelength range of the spectral data in the resin synthesis process is used as an explanatory variable and the resin physical property value is used as a target variable; Predicting resin physical properties based on the prediction model; The data in the predetermined wavelength range is determined using complementary data generated based on the spectral data.

[0014] (9) A program according to an embodiment of the present disclosure includes: On the computer, generating a prediction model based on training data in which data in a predetermined wavelength range of the spectral data in the resin synthesis process is used as an explanatory variable and the resin physical property value is used as a response variable; predicting resin physical property values ​​based on the prediction model; Execute The data in the predetermined wavelength range is determined using complementary data generated based on the spectral data. [Effects of the Invention]

[0015] According to one embodiment of the present disclosure, a technique for predicting resin physical properties is improved. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a block diagram showing a schematic configuration of an information processing device. [Figure 2] 10 is a flowchart illustrating an operation of the information processing device. [Figure 3] 1 shows a schematic diagram of a data set of an example. [Figure 4] 10 is a flowchart illustrating an outline of preprocessing. [Figure 5] 10 is a graph showing prediction results. [Figure 6] 10 is a graph showing prediction accuracy. [Figure 7] 10 is a graph showing prediction results. [Figure 8] 10 is a graph showing prediction accuracy. [Figure 9] 10 is a graph showing prediction results. [Figure 10] 10 is a graph showing prediction accuracy. [Figure 11] 10 is a graph showing prediction results. [Figure 12] 10 is a graph showing prediction accuracy. DETAILED DESCRIPTION OF THE INVENTION

[0017] Hereinafter, embodiments of the present disclosure will be described.

[0018] (Outline of the embodiment)

[0019] A method, an information processing device, and a program for predicting resin physical property values ​​according to embodiments of the present disclosure will be described below with reference to the accompanying drawings. The object of prediction according to embodiments of the present disclosure is resin physical property values ​​in a resin polymerization process. The target resin composition may include a wide range of polymers, such as homopolymers and copolymers. The resin composition may also be a thermoplastic resin or a thermosetting resin. Thermoplastic resins are not particularly limited, but examples include polypropylene (PP), polyethylene (PE), ABS resin, polyvinyl chloride (PVC), acrylic resin, polyester resin, polystyrene resin (PS), urethane resin (PU), and polyphenylene sulfide resin (PPS). Resin physical property values ​​may include NV value, viscosity, residual monomer concentration, molecular weight, etc. In the following embodiments, a case where the resin composition is an epoxy resin will be described as an example.

[0020] In each drawing, the same or corresponding parts are denoted by the same reference numerals. In the description of this embodiment, the description of the same or corresponding parts will be omitted or simplified as appropriate.

[0021] First, an overview of this embodiment will be described, and details will be provided later. The method for predicting resin physical property values ​​in this embodiment is executed by an information processing device 10. The information processing device 10 generates a prediction model based on training data in which a predetermined wavelength range of spectral data in a resin synthesis process is used as an explanatory variable and the resin physical property values ​​are used as a response variable. The information processing device 10 also predicts the resin physical property values ​​based on the prediction model. Here, the predetermined wavelength range is characterized by being determined using complementary data generated based on the spectral data.

[0022] As described above, according to this embodiment, complementary data is generated based on the spectral data. Furthermore, the predetermined wavelength range is determined based on the complementary data. Therefore, even when sufficient performance data cannot be secured, the predetermined wavelength range can be determined by generating complementary data. Furthermore, by using the spectral data in the predetermined wavelength range as explanatory variables, the resin property value prediction technology is improved in that it can predict the resin property values ​​with high accuracy.

[0023] (Configuration of information processing device) As shown in FIG. 1, the information processing device 10 includes a control unit 11, a storage unit 12, an input unit 13, an output unit 14, and a communication unit 15.

[0024] The control unit 11 includes at least one processor, at least one dedicated circuit, or a combination thereof. The processor is a general-purpose processor such as a central processing unit (CPU) or a graphics processing unit (GPU), or a dedicated processor specialized for a specific process. The dedicated circuit is, for example, a field-programmable gate array (FPGA) or an application specific integrated circuit (ASIC). The control unit 11 executes processes related to the operation of the information processing device 10 while controlling each unit of the information processing device 10.

[0025] The storage unit 12 includes at least one semiconductor memory, at least one magnetic memory, at least one optical memory, or a combination of at least two of these. The semiconductor memory is, for example, a random access memory (RAM) or a read only memory (ROM). The RAM is, for example, a static random access memory (SRAM) or a dynamic random access memory (DRAM). The ROM is, for example, an electrically erasable programmable read only memory (EEPROM). The storage unit 12 functions as, for example, a main storage device, an auxiliary storage device, or a cache memory. The storage unit 12 stores data used in the operation of the information processing device 10 and data obtained by the operation of the information processing device 10.

[0026] The input unit 13 includes at least one input interface. The input interface is, for example, a physical key, a capacitance key, a pointing device, or a touch screen integrated with a display. The input interface may also be, for example, a sound sensor that accepts voice input, or a camera that accepts gesture input. The input unit 13 accepts an operation to input data used for the operation of the information processing device 10. The input unit 13 may be connected to the information processing device 10 as an external input device instead of being provided in the information processing device 10. Any connection method may be used, for example, a Universal Serial Bus (USB), a High-Definition Multimedia Interface (HDMI) (registered trademark), or Bluetooth (registered trademark).

[0027] The output unit 14 includes at least one output interface. The output interface is, for example, a display that outputs information as a video, or a speaker that outputs information as a sound. The display is, for example, an LCD (liquid crystal display) or an organic EL (electro luminescence) display. The output unit 14 displays and outputs data obtained by the operation of the information processing device 10. The output unit 14 may be connected to the information processing device 10 as an external output device instead of being provided in the information processing device 10. Any connection method can be used, for example, USB, HDMI (registered trademark), or Bluetooth (registered trademark).

[0028] The communication unit 15 includes at least one external communication interface. The communication interface may be either a wired communication interface or a wireless communication interface. In the case of wired communication, the communication interface is, for example, a LAN (Local Area Network) interface or a USB (Universal Serial Bus). In the case of wireless communication, the communication interface is, for example, an interface compatible with mobile communication standards such as LTE (Long Term Evolution), 4G (4th generation), or 5G (5th generation), or an interface compatible with short-range wireless communication such as Bluetooth (registered trademark). The communication unit 15 receives data used in the operation of the information processing device 10 and transmits data obtained by the operation of the information processing device 10.

[0029] The functions of the information processing device 10 are realized by executing a program according to this embodiment on a processor corresponding to the information processing device 10. That is, the functions of the information processing device 10 are realized by software. The program causes a computer to execute the operations of the information processing device 10, thereby causing the computer to function as the information processing device 10. That is, the computer functions as the information processing device 10 by executing the operations of the information processing device 10 in accordance with the program.

[0030] In this embodiment, the program can be recorded on a computer-readable recording medium. The computer-readable recording medium includes non-transitory computer-readable media, such as a magnetic recording device, an optical disc, a magneto-optical recording medium, or a semiconductor memory. The program can be distributed, for example, by selling, transferring, or lending a portable recording medium, such as a DVD (digital versatile disc) or a CD-ROM (compact disc read only memory), on which the program is recorded. The program can also be distributed by storing the program in the storage of an external server and transmitting the program from the external server to another computer. The program can also be provided as a program product.

[0031] Some or all of the functions of the information processing device 10 may be implemented by a dedicated circuit equivalent to the control unit 11. In other words, some or all of the functions of the information processing device 10 may be implemented by hardware.

[0032] (Operation of information processing device) The operation of the information processing device 10 according to this embodiment will be described with reference to FIG.

[0033] Step S10: The control unit 11 of the information processing device 10 acquires spectral data in the resin synthesis process.

[0034] Any method can be used to acquire the spectral data. For example, the control unit 11 may acquire the spectral data by receiving the spectral data from an external database or the like via the communication unit 21 and a network. The spectral data may be measured either offline or online. The spectral data according to this embodiment may be, for example, near-infrared spectral data. The spectroscopic sensor that measures the near-infrared spectral data may be at least either a near-infrared (NIR) spectroscopic sensor or a Raman spectroscopic sensor.

[0035] Step S20: The control unit 11 performs preprocessing on the spectral data acquired in step S1. The amount of spectral data is enormous, and if all of the spectral data is used as explanatory variables for the prediction model, problems such as over-learning may occur. Therefore, preprocessing is performed to select a predetermined wavelength range from the spectral data that has a large impact on the prediction results. Any method can be used to select the predetermined wavelength range, such as a genetic algorithm, which will be described later.

[0036] Step S30: The control unit 11 generates a prediction model based on the training data in which the data in a predetermined wavelength range of the spectral data in the resin synthesis process is used as an explanatory variable and the resin physical property values ​​are used as response variables. In other words, the control unit 11 generates a prediction model based on the training data in which the preprocessed spectral data is used as an explanatory variable and the resin physical property values ​​are used as response variables.

[0037] The explanatory variables are not limited to the data in the predetermined wavelength region of the above-mentioned spectral data. The explanatory variables may further include the mass ratio of the raw material to the catalyst and the temperature. In this case, in step S10, the control unit 11 also acquires information on the mass ratio of the raw material to the catalyst and the temperature. Any method can be used to acquire this information.

[0038] Step S40: The control unit 11 predicts the resin physical property values ​​from the spectrum data based on the generated prediction model. The control unit 11 may output the prediction results via the output unit .

[0039] In the step of generating a prediction model in step S30, multiple prediction models may be generated based on training data using multiple machine learning algorithms. Examples of algorithms that can be used for the multiple prediction models include Partial Least Squares Regression (PLS), Ridge, Lasso, ElasticNet (EN), Support Vector Regression (SVR) (nonlinear), Random Forest (RF), Gaussian Process Regression_0 (GP_0) (linear), Gaussian Process Regression_1-10 (GP_1-10) (nonlinear), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (Light gbm), and Gradient Boosting Decision Tree (GBDT). In this case, in the prediction step in step S40, one prediction model selected from the multiple prediction models based on a predetermined index may be used to predict the resin physical property values. The predetermined index may be, for example, R2. When the predetermined index is R2, the control unit 11 predicts the resin physical property value using a prediction model in which the value of R2 is closest to 1. Note that the predetermined index is not limited to R2, and the predetermined index may be, for example, an error rate. In this manner, in this embodiment, the control unit 11 may select an optimal algorithm with high prediction accuracy as the algorithm of the prediction model.

[0040] (Example) Examples according to this embodiment will be described below. To obtain data for the examples, four or two lots of three product numbers (product number A, product number B, and product number C) were synthesized using predetermined raw materials and reaction conditions on an experimental scale of 600 g, and spectral data (NIR spectra in this case) and resin physical property values ​​(epoxy monomer equivalent data in this case) were obtained. The NIR spectra were measured every minute. Meanwhile, the epoxy monomer equivalent data was obtained by periodically sampling and analyzing the data.

[0041] 3 shows a schematic diagram of a data set of an example relating to the accuracy verification of the prediction method according to this embodiment. As shown in FIG. 3, the data set includes data on four lots, Lots #1 to #4, for part number A. The data set also includes data on two lots, Lots #1 and #2, for part number B and C, respectively.

[0042] The above-mentioned pre-processing is performed on this data set. An overview of the pre-processing is shown in the flowchart of Figure 4.

[0043] Step S21: The control unit 11 determines whether the number of data points for the spectral data for a certain product number is equal to or greater than a predetermined value. In this embodiment, the number of data points corresponds to the number of lots for each product number. For example, the predetermined value may be 4 (lots). The control unit 11 determines whether a sufficient number of data points has been secured for each product number. If the number of data points for the spectral data is equal to or greater than the predetermined value, the process proceeds to step S22. On the other hand, if the number of data points for the spectral data is less than the predetermined value, the process proceeds to step S22. Specifically, preprocessing is not performed for product number A. On the other hand, preprocessing is performed for product numbers B and C.

[0044] Step S22: The control unit 11 executes noise reduction processing to remove noise from the spectral data. The spectral data may contain information directly used to calculate the resin physical property values ​​and unnecessary information (noise) that may cause errors in the calculation of the resin physical property values. An example of noise reduction processing is smoothing processing. In other words, the control unit 11 reduces noise on the acquired spectral data by smoothing the data. Specifically, the control unit 11 may perform smoothing processing by approximating the data to a polynomial. Preferably, the control unit 11 may perform smoothing using local regression. In this case, varying the local width (kernel width) may result in the loss of necessary information or incomplete removal of noise. Therefore, the control unit 11 may regress the spectral data by selecting an optimal local width. The Savitzky-Golay method (hereinafter also referred to as the SG method), which is based on the least squares method, is known as a representative method of the smoothing processing described above. The control unit 11 may perform noise reduction processing using the SG method.

[0045] Step S23: The control unit 11 selects a predetermined wavelength region based on the spectral data. Any method can be used to select the predetermined wavelength region. For example, a wavelength region from the spectral data that has a large effect on the prediction result may be selected using a genetic algorithm or the like. When using a genetic algorithm, the parameters may be set, for example, as follows: Number of genetic algorithm populations: 100 Number of generations of genetic algorithm: 100 Wavelength selection range: 1~10 Wavelength selection wavelength width: 50 Method: Supervised Learning: Partial Least Squares Regression (PLS) Number of runs: 10

[0046] Step S24: If the number of spectral data points is less than a predetermined value, the control unit 11 generates complementary data based on the acquired spectral data. Any method can be used to generate complementary data. For example, the control unit 11 may generate complementary data based on the acquired spectral data using a method such as linear interpolation or nonlinear interpolation. For example, when there are approximately 10 actual measurement points, the control unit 11 generates approximately 300 complementary data points.

[0047] Step S25: The control unit 11 executes noise reduction processing to remove noise from the spectral data and the complementary data. An example of noise reduction processing is smoothing processing. In other words, the control unit 11 performs noise reduction on the acquired spectral data and complementary data by smoothing the data. Specifically, the control unit 11 may perform smoothing processing by approximating the data to a polynomial. Preferably, the control unit 11 may perform smoothing by local regression. In this case, varying the local width (kernel width) may result in the loss of necessary information or incomplete removal of noise. Therefore, the control unit 11 may regress the spectral data and complementary data by selecting an optimal local width. The control unit 11 may also perform noise reduction processing using the SG algorithm.

[0048] Step S26: The control unit 11 selects a predetermined wavelength region based on the spectral data and the complementary data. Any method can be used to select the predetermined wavelength region. For example, a wavelength region from the spectral data and the complementary data that has a large influence on the prediction result may be selected using a genetic algorithm or the like. When using a genetic algorithm, the parameters may be set, for example, as follows: Number of genetic algorithm populations: 100 Number of generations of genetic algorithm: 100 Wavelength selection range: 1~10 Wavelength selection wavelength width: 50 Method: Supervised Learning: Partial Least Squares Regression (PLS) Number of runs: 10

[0049] Step S27: The control unit 11 deletes the complementary data from the data set. In other words, the control unit 11 uses the acquired spectral data for training a prediction model (described later) and does not use the complementary data for training a prediction model (described later). In other words, the control unit 11 deletes the amplified complementary data and organizes the explanatory variables so that the data consists only of actual measured values.

[0050] Step S28: The control unit 11 merges the data for the product number with other product number data for which the number of data is equal to or greater than a predetermined value. Specifically, in this embodiment, the control unit 11 merges the data for product number B with the data for product number A. The control unit 11 also merges the data for product number C with the data for product number A. In this way, the spectral data in the training data may include data for multiple resin compositions with different raw material composition ratios. This can improve prediction accuracy, as will be described later.

[0051] 5 and 6, the prediction results and prediction accuracy of a prediction model generated using only the data for part number B as training data are shown. Here, Ridge is selected as the algorithm for the prediction model. The horizontal axis of FIG. 5 represents the label number, and the vertical axis represents the epoxy monomer equivalent value. The label number is an identification number associated with each sample of each lot in the synthesis process. The horizontal axis of FIG. 6 represents the actual value, and the vertical axis represents the predicted value corresponding to the actual value. The R2 value of the prediction model calculated from the results of FIG. 6 is 0.641. These results show that even when using a prediction model using only the data for part number B as training data, it is possible to predict the epoxy monomer equivalent data, which is the target variable, with a certain degree of prediction accuracy.

[0052] 7 and 8, the prediction results and prediction accuracy of a prediction model generated using training data combining data for part number B and data for part number A are shown. Here, Ridge is selected as the algorithm for the prediction model. The horizontal axis of FIG. 7 represents the label number, and the vertical axis represents the epoxy monomer equivalent value. The horizontal axis of FIG. 8 represents the actual value, and the vertical axis represents the predicted value corresponding to the actual value. The R2 value of the prediction model calculated from the results of FIG. 8 is 0.829. These results show that by using training data combining data for part number B and data for part number A, it is possible to predict the epoxy monomer equivalent data, which is the target variable, with higher prediction accuracy than when only data for part number B is used as training data.

[0053] 9 and 10, prediction results are shown using a prediction model generated using only the data for part number C as training data. Here, PLS is selected as the algorithm for the prediction model. The horizontal axis in FIG. 9 represents the label number, and the vertical axis represents the epoxy monomer equivalent value. The horizontal axis in FIG. 10 represents the actual value, and the vertical axis represents the predicted value corresponding to the actual value. The R2 value of the prediction model calculated from the results in FIG. 10 is 0.957. These results show that even when using a prediction model that uses only the data for part number C as training data, it is possible to predict the epoxy monomer equivalent data, which is the response variable, with a certain degree of prediction accuracy.

[0054] 11 and 12, prediction results are shown using a prediction model generated using training data combining data for part number C and data for part number A. Here, lasso is selected as the algorithm for the prediction model. The horizontal axis in FIG. 11 represents the label number, and the vertical axis represents the epoxy monomer equivalent value. The horizontal axis in FIG. 12 represents the actual value, and the vertical axis represents the predicted value corresponding to the actual value. The R2 value of the prediction model calculated from the results in FIG. 12 is 0.979. These results show that by using training data combining data for part number C and data for part number A, it is possible to predict the epoxy monomer equivalent data, which is the target variable, with higher prediction accuracy than when only data for part number C is used as training data.

[0055] As described above, the information processing device 10 according to this embodiment generates complementary data based on spectral data. The information processing device 10 also determines a predetermined wavelength range based on the complementary data. This configuration allows the predetermined wavelength range to be determined by generating complementary data even when sufficient performance data is unavailable. Furthermore, by using spectral data in the predetermined wavelength range as explanatory variables, resin property values ​​can be predicted with high accuracy, improving the technology for predicting resin property values.

[0056] Although the present disclosure has been described based on the drawings and examples, it should be noted that those skilled in the art may make various modifications and alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are included in the scope of the present disclosure. For example, the functions included in each component or step can be rearranged so as not to be logically inconsistent, and multiple components or steps can be combined or divided into one.

[0057] Furthermore, for example, the control unit 11 may perform a differentiation process in addition to or instead of the smoothing process in step S22 or step S25. By performing the differentiation process, the control unit 11 can extract spectral information from overlapping peaks and correct the baseline. When the control unit 11 performs the differentiation process, the prediction model described above is generated based on the spectral data on which the differentiation process has been performed. Note that the differentiation process may increase noise in the spectrum. Therefore, depending on the spectral data, it may be better to perform the differentiation process in some cases and not in other cases. Therefore, the control unit 11 may perform a smoothing process on the spectral data and then separately perform a differentiation process to generate a prediction model using optimal spectral data.

[0058] Furthermore, for example, in step S22 or step S25, the control unit 11 may perform logarithmic transformation of the resin physical property values ​​in addition to smoothing and / or differentiating the spectral data. For example, when the relationship between the spectral data and the resin physical property values ​​follows the Arrhenius-type physical law, it is effective to perform logarithmic transformation of the resin physical property values. On the other hand, logarithmic transformation of the resin physical property values ​​may increase the prediction error of the resin physical property values. In other words, performing logarithmic transformation of the resin physical property values ​​may or may not improve accuracy. Therefore, the control unit 11 may output either the resin physical property values ​​that have been logarithmically transformed or the resin physical property values ​​that have not been logarithmically transformed as the resin physical property values. Note that the logarithmic transformation of the resin physical property values ​​is performed independently of the smoothing and differentiation processes. [Explanation of symbols]

[0059] 10. Information processing equipment 11 Control section 12 Storage section 13 Input section 14 Output section 15 Communications Department

Claims

1. A method for predicting a resin physical property value executed by an information processing device, comprising: generating a prediction model based on training data in which data in a predetermined wavelength range of the spectral data in the resin synthesis process is used as an explanatory variable and resin physical property values ​​are used as target variables; predicting resin physical property values ​​based on the prediction model; Including, A method in which the predetermined wavelength range is determined using complementary data generated based on the spectral data.

2. 10. The method of claim 1, The method, wherein the spectral data includes data relating to a plurality of resin compositions having different composition ratios of raw materials.

3. 3. The method of claim 1 or 2, The method, wherein the spectral data is near-infrared spectral data.

4. The method according to claim 3 , wherein the spectroscopic sensor for measuring the near-infrared light spectral data is at least one of a near-infrared spectroscopic sensor and a Raman spectroscopic sensor.

5. 4. The method of claim 3, The method, wherein the explanatory variables further include the mass ratio of raw material to catalyst and temperature.

6. 10. The method of claim 1, The method, wherein the resin physical property value is an epoxy monomer equivalent.

7. 10. The method of claim 1, In the step of generating the prediction model, a plurality of prediction models are generated based on the training data using a plurality of machine learning algorithms; In the predicting step, the resin physical property value is predicted based on one prediction model selected from the plurality of prediction models based on a predetermined index.

8. An information processing device including a control unit, The control unit A prediction model is generated based on training data in which data in a predetermined wavelength range of the spectral data in the resin synthesis process is used as an explanatory variable and the resin physical property value is used as a target variable; Predicting resin physical properties based on the prediction model; The information processing device, wherein the data in the predetermined wavelength region is determined using complementary data generated based on the spectral data.

9. On the computer, generating a prediction model based on training data in which data in a predetermined wavelength range of the spectral data in the resin synthesis process is used as an explanatory variable and the resin physical property value is used as a response variable; predicting resin physical property values ​​based on the prediction model; Execute The data of the predetermined wavelength region is determined using complementary data generated based on the spectral data.

Images